Lung Biology in Health & Disease Volume 215 Ventilator-Induced Lung Injury

Ventilator-Induced Lung Injury LUNG BIOLOGY IN HEALTH AND DISEASE Executive Editor Claude Lenfant Former Director, Na...

Author: Didier Dreyfuss | Georges Saumon | Rolf Hubmayr

71 downloads 1093 Views 7MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Ventilator-Induced Lung Injury

LUNG BIOLOGY IN HEALTH AND DISEASE Executive Editor Claude Lenfant Former Director, National Heart, Lung, and Blood Institute National Institutes of Health Bethesda, Maryland

1. Immunologic and Infectious Reactions in the Lung, edited by C. H. Kirkpatrick and H. Y. Reynolds 2. The Biochemical Basis of Pulmonary Function, edited by R. G. Crystal 3. Bioengineering Aspects of the Lung, edited by J. B. West 4. Metabolic Functions of the Lung, edited by Y. S. Bakhle and J. R. Vane 5. Respiratory Defense Mechanisms (in two parts), edited by J. D. Brain, D. F. Proctor, and L. M. Reid 6. Development of the Lung, edited by W. A. Hodson 7. Lung Water and Solute Exchange, edited by N. C. Staub 8. Extrapulmonary Manifestations of Respiratory Disease, edited by E. D. Robin 9. Chronic Obstructive Pulmonary Disease, edited by T. L. Petty 10. Pathogenesis and Therapy of Lung Cancer, edited by C. C. Harris 11. Genetic Determinants of Pulmonary Disease, edited by S. D. Litwin 12. The Lung in the Transition Between Health and Disease, edited by P. T. Macklem and S. Permutt 13. Evolution of Respiratory Processes: A Comparative Approach, edited by S. C. Wood and C. Lenfant 14. Pulmonary Vascular Diseases, edited by K. M. Moser 15. Physiology and Pharmacology of the Airways, edited by J. A. Nadel 16. Diagnostic Techniques in Pulmonary Disease (in two parts), edited by M. A. Sackner 17. Regulation of Breathing (in two parts), edited by T. F. Hornbein 18. Occupational Lung Diseases: Research Approaches and Methods, edited by H. Weill and M. Turner-Warwick 19. Immunopharmacology of the Lung, edited by H. H. Newball 20. Sarcoidosis and Other Granulomatous Diseases of the Lung, edited by B. L. Fanburg

21. Sleep and Breathing, edited by N. A. Saunders and C. E. Sullivan 22. Pneumocystis carinii Pneumonia: Pathogenesis, Diagnosis, and Treatment, edited by L. S. Young 23. Pulmonary Nuclear Medicine: Techniques in Diagnosis of Lung Disease, edited by H. L. Atkins 24. Acute Respiratory Failure, edited by W. M. Zapol and K. J. Falke 25. Gas Mixing and Distribution in the Lung, edited by L. A. Engel and M. Paiva 26. High-Frequency Ventilation in Intensive Care and During Surgery, edited by G. Carlon and W. S. Howland 27. Pulmonary Development: Transition from Intrauterine to Extrauterine Life, edited by G. H. Nelson 28. Chronic Obstructive Pulmonary Disease: Second Edition, edited by T. L. Petty 29. The Thorax (in two parts), edited by C. Roussos and P. T. Macklem 30. The Pleura in Health and Disease, edited by J. Chrétien, J. Bignon, and A. Hirsch 31. Drug Therapy for Asthma: Research and Clinical Practice, edited by J. W. Jenne and S. Murphy 32. Pulmonary Endothelium in Health and Disease, edited by U. S. Ryan 33. The Airways: Neural Control in Health and Disease, edited by M. A. Kaliner and P. J. Barnes 34. Pathophysiology and Treatment of Inhalation Injuries, edited by J. Loke 35. Respiratory Function of the Upper Airway, edited by O. P. Mathew and G. Sant’Ambrogio 36. Chronic Obstructive Pulmonary Disease: A Behavioral Perspective, edited by A. J. McSweeny and I. Grant 37. Biology of Lung Cancer: Diagnosis and Treatment, edited by S. T. Rosen, J. L. Mulshine, F. Cuttitta, and P. G. Abrams 38. Pulmonary Vascular Physiology and Pathophysiology, edited by E. K. Weir and J. T. Reeves 39. Comparative Pulmonary Physiology: Current Concepts, edited by S. C. Wood 40. Respiratory Physiology: An Analytical Approach, edited by H. K. Chang and M. Paiva 41. Lung Cell Biology, edited by D. Massaro 42. Heart–Lung Interactions in Health and Disease, edited by S. M. Scharf and S. S. Cassidy 43. Clinical Epidemiology of Chronic Obstructive Pulmonary Disease, edited by M. J. Hensley and N. A. Saunders 44. Surgical Pathology of Lung Neoplasms, edited by A. M. Marchevsky

45. The Lung in Rheumatic Diseases, edited by G. W. Cannon and G. A. Zimmerman 46. Diagnostic Imaging of the Lung, edited by C. E. Putman 47. Models of Lung Disease: Microscopy and Structural Methods, edited by J. Gil 48. Electron Microscopy of the Lung, edited by D. E. Schraufnagel 49. Asthma: Its Pathology and Treatment, edited by M. A. Kaliner, P. J. Barnes, and C. G. A. Persson 50. Acute Respiratory Failure: Second Edition, edited by W. M. Zapol and F. Lemaire 51. Lung Disease in the Tropics, edited by O. P. Sharma 52. Exercise: Pulmonary Physiology and Pathophysiology, edited by B. J. Whipp and K. Wasserman 53. Developmental Neurobiology of Breathing, edited by G. G. Haddad and J. P. Farber 54. Mediators of Pulmonary Inflammation, edited by M. A. Bray and W. H. Anderson 55. The Airway Epithelium, edited by S. G. Farmer and D. Hay 56. Physiological Adaptations in Vertebrates: Respiration, Circulation, and Metabolism, edited by S. C. Wood, R. E. Weber, A. R. Hargens, and R. W. Millard 57. The Bronchial Circulation, edited by J. Butler 58. Lung Cancer Differentiation: Implications for Diagnosis and Treatment, edited by S. D. Bernal and P. J. Hesketh 59. Pulmonary Complications of Systemic Disease, edited by J. F. Murray 60. Lung Vascular Injury: Molecular and Cellular Response, edited by A. Johnson and T. J. Ferro 61. Cytokines of the Lung, edited by J. Kelley 62. The Mast Cell in Health and Disease, edited by M. A. Kaliner and D. D. Metcalfe 63. Pulmonary Disease in the Elderly Patient, edited by D. A. Mahler 64. Cystic Fibrosis, edited by P. B. Davis 65. Signal Transduction in Lung Cells, edited by J. S. Brody, D. M. Center, and V. A. Tkachuk 66. Tuberculosis: A Comprehensive International Approach, edited by L. B. Reichman and E. S. Hershfield 67. Pharmacology of the Respiratory Tract: Experimental and Clinical Research, edited by K. F. Chung and P. J. Barnes 68. Prevention of Respiratory Diseases, edited by A. Hirsch, M. Goldberg, J.-P. Martin, and R. Masse 69. Pneumocystis carinii Pneumonia: Second Edition, edited by P. D. Walzer

70. Fluid and Solute Transport in the Airspaces of the Lungs, edited by R. M. Effros and H. K. Chang 71. Sleep and Breathing: Second Edition, edited by N. A. Saunders and C. E. Sullivan 72. Airway Secretion: Physiological Bases for the Control of Mucous Hypersecretion, edited by T. Takishima and S. Shimura 73. Sarcoidosis and Other Granulomatous Disorders, edited by D. G. James 74. Epidemiology of Lung Cancer, edited by J. M. Samet 75. Pulmonary Embolism, edited by M. Morpurgo 76. Sports and Exercise Medicine, edited by S. C. Wood and R. C. Roach 77. Endotoxin and the Lungs, edited by K. L. Brigham 78. The Mesothelial Cell and Mesothelioma, edited by M.-C. Jaurand and J. Bignon 79. Regulation of Breathing: Second Edition, edited by J. A. Dempsey and A. I. Pack 80. Pulmonary Fibrosis, edited by S. Hin. Phan and R. S. Thrall 81. Long-Term Oxygen Therapy: Scientific Basis and Clinical Application, edited by W. J. O’Donohue, Jr. 82. Ventral Brainstem Mechanisms and Control of Respiration and Blood Pressure, edited by C. O. Trouth, R. M. Millis, H. F. Kiwull-Schöne, and M. E. Schläfke 83. A History of Breathing Physiology, edited by D. F. Proctor 84. Surfactant Therapy for Lung Disease, edited by B. Robertson and H. W. Taeusch 85. The Thorax: Second Edition, Revised and Expanded (in three parts), edited by C. Roussos 86. Severe Asthma: Pathogenesis and Clinical Management, edited by S. J. Szefler and D. Y. M. Leung 87. Mycobacterium avium–Complex Infection: Progress in Research and Treatment, edited by J. A. Korvick and C. A. Benson 88. Alpha 1–Antitrypsin Deficiency: Biology • Pathogenesis • Clinical Manifestations • Therapy, edited by R. G. Crystal 89. Adhesion Molecules and the Lung, edited by P. A. Ward and J. C. Fantone 90. Respiratory Sensation, edited by L. Adams and A. Guz 91. Pulmonary Rehabilitation, edited by A. P. Fishman 92. Acute Respiratory Failure in Chronic Obstructive Pulmonary Disease, edited by J.-P. Derenne, W. A. Whitelaw, and T. Similowski 93. Environmental Impact on the Airways: From Injury to Repair, edited by J. Chrétien and D. Dusser 94. Inhalation Aerosols: Physical and Biological Basis for Therapy, edited by A. J. Hickey

95. Tissue Oxygen Deprivation: From Molecular to Integrated Function, edited by G. G. Haddad and G. Lister 96. The Genetics of Asthma, edited by S. B. Liggett and D. A. Meyers 97. Inhaled Glucocorticoids in Asthma: Mechanisms and Clinical Actions, edited by R. P. Schleimer, W. W. Busse, and P. M. O’Byrne 98. Nitric Oxide and the Lung, edited by W. M. Zapol and K. D. Bloch 99. Primary Pulmonary Hypertension, edited by L. J. Rubin and S. Rich 100. Lung Growth and Development, edited by J. A. McDonald 101. Parasitic Lung Diseases, edited by A. A. F. Mahmoud 102. Lung Macrophages and Dendritic Cells in Health and Disease, edited by M. F. Lipscomb and S. W. Russell 103. Pulmonary and Cardiac Imaging, edited by C. Chiles and C. E. Putman 104. Gene Therapy for Diseases of the Lung, edited by K. L. Brigham 105. Oxygen, Gene Expression, and Cellular Function, edited by L. Biadasz Clerch and D. J. Massaro 106. Beta2-Agonists in Asthma Treatment, edited by R. Pauwels and P. M. O’Byrne 107. Inhalation Delivery of Therapeutic Peptides and Proteins, edited by A. L. Adjei and P. K. Gupta 108. Asthma in the Elderly, edited by R. A. Barbee and J. W. Bloom 109. Treatment of the Hospitalized Cystic Fibrosis Patient, edited by D. M. Orenstein and R. C. Stern 110. Asthma and Immunological Diseases in Pregnancy and Early Infancy, edited by M. Schatz, R. S. Zeiger, and H. N. Claman 111. Dyspnea, edited by D. A. Mahler 112. Proinflammatory and Antiinflammatory Peptides, edited by S. I. Said 113. Self-Management of Asthma, edited by H. Kotses and A. Harver 114. Eicosanoids, Aspirin, and Asthma, edited by A. Szczeklik, R. J. Gryglewski, and J. R. Vane 115. Fatal Asthma, edited by A. L. Sheffer 116. Pulmonary Edema, edited by M. A. Matthay and D. H. Ingbar 117. Inflammatory Mechanisms in Asthma, edited by S. T. Holgate and W. W. Busse 118. Physiological Basis of Ventilatory Support, edited by J. J. Marini and A. S. Slutsky 119. Human Immunodeficiency Virus and the Lung, edited by M. J. Rosen and J. M. Beck

120. Five-Lipoxygenase Products in Asthma, edited by J. M. Drazen, S.-E. Dahlén, and T. H. Lee 121. Complexity in Structure and Function of the Lung, edited by M. P. Hlastala and H. T. Robertson 122. Biology of Lung Cancer, edited by M. A. Kane and P. A. Bunn, Jr. 123. Rhinitis: Mechanisms and Management, edited by R. M. Naclerio, S. R. Durham, and N. Mygind 124. Lung Tumors: Fundamental Biology and Clinical Management, edited by C. Brambilla and E. Brambilla 125. Interleukin-5: From Molecule to Drug Target for Asthma, edited by C. J. Sanderson 126. Pediatric Asthma, edited by S. Murphy and H. W. Kelly 127. Viral Infections of the Respiratory Tract, edited by R. Dolin and P. F. Wright 128. Air Pollutants and the Respiratory Tract, edited by D. L. Swift and W. M. Foster 129. Gastroesophageal Reflux Disease and Airway Disease, edited by M. R. Stein 130. Exercise-Induced Asthma, edited by E. R. McFadden, Jr. 131. LAM and Other Diseases Characterized by Smooth Muscle Proliferation, edited by J. Moss 132. The Lung at Depth, edited by C. E. G. Lundgren and J. N. Miller 133. Regulation of Sleep and Circadian Rhythms, edited by F. W. Turek and P. C. Zee 134. Anticholinergic Agents in the Upper and Lower Airways, edited by S. L. Spector 135. Control of Breathing in Health and Disease, edited by M. D. Altose and Y. Kawakami 136. Immunotherapy in Asthma, edited by J. Bousquet and H. Yssel 137. Chronic Lung Disease in Early Infancy, edited by R. D. Bland and J. J. Coalson 138. Asthma’s Impact on Society: The Social and Economic Burden, edited by K. B. Weiss, A. S. Buist, and S. D. Sullivan 139. New and Exploratory Therapeutic Agents for Asthma, edited by M. Yeadon and Z. Diamant 140. Multimodality Treatment of Lung Cancer, edited by A. T. Skarin 141. Cytokines in Pulmonary Disease: Infection and Inflammation, edited by S. Nelson and T. R. Martin 142. Diagnostic Pulmonary Pathology, edited by P. T. Cagle 143. Particle–Lung Interactions, edited by P. Gehr and J. Heyder 144. Tuberculosis: A Comprehensive International Approach, Second Edition, Revised and Expanded, edited by L. B. Reichman and E. S. Hershfield

145. Combination Therapy for Asthma and Chronic Obstructive Pulmonary Disease, edited by R. J. Martin and M. Kraft 146. Sleep Apnea: Implications in Cardiovascular and Cerebrovascular Disease, edited by T. D. Bradley and J. S. Floras 147. Sleep and Breathing in Children: A Developmental Approach, edited by G. M. Loughlin, J. L. Carroll, and C. L. Marcus 148. Pulmonary and Peripheral Gas Exchange in Health and Disease, edited by J. Roca, R. Rodriguez-Roisen, and P. D. Wagner 149. Lung Surfactants: Basic Science and Clinical Applications, R. H. Notter 150. Nosocomial Pneumonia, edited by W. R. Jarvis 151. Fetal Origins of Cardiovascular and Lung Disease, edited by David J. P. Barker 152. Long-Term Mechanical Ventilation, edited by N. S. Hill 153. Environmental Asthma, edited by R. K. Bush 154. Asthma and Respiratory Infections, edited by D. P. Skoner 155. Airway Remodeling, edited by P. H. Howarth, J. W. Wilson, J. Bousquet, S. Rak, and R. A. Pauwels 156. Genetic Models in Cardiorespiratory Biology, edited by G. G. Haddad and T. Xu 157. Respiratory-Circulatory Interactions in Health and Disease, edited by S. M. Scharf, M. R. Pinsky, and S. Magder 158. Ventilator Management Strategies for Critical Care, edited by N. S. Hill and M. M. Levy 159. Severe Asthma: Pathogenesis and Clinical Management, Second Edition, Revised and Expanded, edited by S. J. Szefler and D. Y. M. Leung 160. Gravity and the Lung: Lessons from Microgravity, edited by G. K. Prisk, M. Paiva, and J. B. West 161. High Altitude: An Exploration of Human Adaptation, edited by T. F. Hornbein and R. B. Schoene 162. Drug Delivery to the Lung, edited by H. Bisgaard, C. O’Callaghan, and G. C. Smaldone 163. Inhaled Steroids in Asthma: Optimizing Effects in the Airways, edited by R. P. Schleimer, P. M. O’Byrne, S. J. Szefler, and R. Brattsand 164. IgE and Anti-IgE Therapy in Asthma and Allergic Disease, edited by R. B. Fick, Jr., and P. M. Jardieu 165. Clinical Management of Chronic Obstructive Pulmonary Disease, edited by T. Similowski, W. A. Whitelaw, and J.-P. Derenne 166. Sleep Apnea: Pathogenesis, Diagnosis, and Treatment, edited by A. I. Pack 167. Biotherapeutic Approaches to Asthma, edited by J. Agosti and A. L. Sheffer

168. Proteoglycans in Lung Disease, edited by H. G. Garg, P. J. Roughley, and C. A. Hales 169. Gene Therapy in Lung Disease, edited by S. M. Albelda 170. Disease Markers in Exhaled Breath, edited by N. Marczin, S. A. Kharitonov, M. H. Yacoub, and P. J. Barnes 171. Sleep-Related Breathing Disorders: Experimental Models and Therapeutic Potential, edited by D. W. Carley and M. Radulovacki 172. Chemokines in the Lung, edited by R. M. Strieter, S. L. Kunkel, and T. J. Standiford 173. Respiratory Control and Disorders in the Newborn, edited by O. P. Mathew 174. The Immunological Basis of Asthma, edited by B. N. Lambrecht, H. C. Hoogsteden, and Z. Diamant 175. Oxygen Sensing: Responses and Adaptation to Hypoxia, edited by S. Lahiri, G. L. Semenza, and N. R. Prabhakar 176. Non-Neoplastic Advanced Lung Disease, edited by J. R. Maurer 177. Therapeutic Targets in Airway Inflammation, edited by N. T. Eissa and D. P. Huston 178. Respiratory Infections in Allergy and Asthma, edited by S. L. Johnston and N. G. Papadopoulos 179. Acute Respiratory Distress Syndrome, edited by M. A. Matthay 180. Venous Thromboembolism, edited by J. E. Dalen 181. Upper and Lower Respiratory Disease, edited by J. Corren, A. Togias, and J. Bousquet 182. Pharmacotherapy in Chronic Obstructive Pulmonary Disease, edited by B. R. Celli 183. Acute Exacerbations of Chronic Obstructive Pulmonary Disease, edited by N. M. Siafakas, N. R. Anthonisen, and D. Georgopoulos 184. Lung Volume Reduction Surgery for Emphysema, edited by H. E. Fessler, J. J. Reilly, Jr., and D. J. Sugarbaker 185. Idiopathic Pulmonary Fibrosis, edited by J. P. Lynch III 186. Pleural Disease, edited by D. Bouros 187. Oxygen/Nitrogen Radicals: Lung Injury and Disease, edited by V. Vallyathan, V. Castranova, and X. Shi 188. Therapy for Mucus-Clearance Disorders, edited by B. K. Rubin and C. P. van der Schans 189. Interventional Pulmonary Medicine, edited by J. F. Beamis, Jr., P. N. Mathur, and A. C. Mehta 190. Lung Development and Regeneration, edited by D. J. Massaro, G. Massaro, and P. Chambon 191. Long-Term Intervention in Chronic Obstructive Pulmonary Disease, edited by R. Pauwels, D. S. Postma, and S. T. Weiss

192. Sleep Deprivation: Basic Science, Physiology, and Behavior, edited by Clete A. Kushida 193. Sleep Deprivation: Clinical Issues, Pharmacology, and Sleep Loss Effects, edited by Clete A. Kushida 194. Pneumocystis Pneumonia: Third Edition, Revised and Expanded, edited by P. D. Walzer and M. Cushion 195. Asthma Prevention, edited by William W. Busse and Robert F. Lemanske, Jr. 196. Lung Injury: Mechanisms, Pathophysiology, and Therapy, edited by Robert H. Notter, Jacob Finkelstein, and Bruce Holm 197. Ion Channels in the Pulmonary Vasculature, edited by Jason X.-J. Yuan 198. Chronic Obstuctive Pulmonary Disease: Cellular and Molecular Mechanisms, edited by Peter J. Barnes 199. Pediatric Nasal and Sinus Disorders, edited by Tania Sih and Peter A. R. Clement 200. Functional Lung Imaging, edited by David Lipson and Edwin van Beek 201. Lung Surfactant Function and Disorder, edited by Kaushik Nag 202. Pharmacology and Pathophysiology of the Control of Breathing, edited by Denham S. Ward, Albert Dahan and Luc J. Teppema 203. Molecular Imaging of the Lungs, edited by Daniel Schuster and Timothy Blackwell 204. Air Pollutants and the Respiratory Tract: Second Edition, edited by W. Michael Foster and Daniel L. Costa 205. Acute and Chronic Cough, edited by Anthony E. Redington and Alyn H. Morice 206. Severe Pneumonia, edited by Michael S. Niederman 207. Monitoring Asthma, edited by Peter G. Gibson 208. Dyspnea: Mechanisms, Measurement, and Management, Second Edition, edited by Donald A. Mahler and Denis E. O'Donnell 209. Childhood Asthma, edited by Stanley J. Szefler and Søren Pedersen 210. Sarcoidosis, edited by Robert Baughman 211. Tropical Lung Disease, Second Edition, edited by Om Sharma 212. Pharmacotherapy of Asthma, edited by James T. Li 213. Practical Pulmonary and Critical Care Medicine: Respiratory Failure, edited by Zab Mosenifar and Guy W. Soo Hoo 214. Practical Pulmonary and Critical Care Medicine: Disease Management, edited by Zab Mosenifar and Guy W. Soo Hoo 215. Ventilator-Induced Lung Injury, edited by Didier Dreyfuss, Georges Saumon, and Rolf D. Hubmayr The opinions expressed in these volumes do not necessarily represent the views of the National Institutes of Health.

Ventilator-Induced Lung Injury

Edited by

Didier Dreyfuss Paris 7-Denis Diderot Medical School Paris, France Hôpital Louis Mourier (Assistance Publique-Hôpitaux de Paris) Colombes, France

Georges Saumon Paris 7-Denis Diderot Medical School Paris, France

Rolf D. Hubmayr Mayo Clinic Rochester, Minnesota, U.S.A.

New York London

Taylor & Francis is an imprint of the Taylor & Francis Group, an informa business

Published in 2006 by Taylor & Francis Group 270 Madison Avenue New York, NY 10016 © 2006 by Taylor & Francis Group, LLC No claim to original U.S. Government works Printed in the United States of America on acid-free paper 10 9 8 7 6 5 4 3 2 1 International Standard Book Number-10: 0-8493-3716-X (Hardcover) International Standard Book Number-13: 978-0-8493-3716-1 (Hardcover) Library of Congress Card Number 2005046643 This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use. No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers. For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC) 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data Ventilator-induced lung injury / edited by Didier Dreyfuss, Georges Saumon, Rolf Hubmayr. p. ; cm. -- (Lung biology in health and disease ; v. 215) Includes bibliographical references and index. ISBN-13: 978-0-8493-3716-1 (alk. paper) ISBN-10: 0-8493-3716-X (alk. paper) 1. Lungs--Wounds and injuries. 2. Respiratory distress syndrome, Adult. 3. Artificial respiration. 4. Respirators (Medical equipment) I. Dreyfuss, Didier. II. Saumon, Georges. III. Hubmayr, Rolf. IV. Series. [DNLM: 1. Lung--injuries. 2. Respiratory Distress Syndrome, Adult--etiology. 3. Respiratory Distress Syndrome, Adult--prevention & control. 4. Ventilators, Mechanical--adverse effects. WF 600 V465 2006] RC776.R38V46 2006 616.2'4--dc22

2005046643

Visit the Taylor & Francis Web site at http://www.taylorandfrancis.com Taylor & Francis Group is the Academic Division of Informa plc.

Introduction

But that life may in a manner of speaking be restored to the animal, an opening must be attempted in the trunk of the trachea, into which a tube of reed or cane should be put; you will then blow into this, so that the lung may rise again and the animal take in air. —Andreas Vesalius (1514–1564) (1)

This appears to be the ﬁrst report on artiﬁcial, or assisted, ventilation. Yet, a few years before Vesalius, Paracelus (1493–1541), a Swiss-born philosopher, had theorized the principles of resuscitation. It is unclear whether Vesalius had been inspired by the writings of Paracelus, or whether his demonstration of resuscitation was the result of his own creativity. Irrespective, it took generations for the work of these two luminaries to stimulate the application of artiﬁcial ventilation in humans. In fact, it was the discovery of anesthesia in 1846 that provided the necessary impetus, plus about 60 years, in 1904, when Sauerbruch (2) developed his constant negative pressure chamber in order to prevent lung collapse during pulmonary surgery. Today, mechanical ventilation has come to age, and that it assists, or replaces, spontaneous breathing is universally well recognized. Without doubt, it is the mainstay of intensive care medicine, and in many instances it is one of the essential tools of post-surgical care. The moment of triumph for mechanical ventilation came when acute respiratory distress syndrome was ﬁrst described by Ashbaugh, Bigelow, and Petty in 1967 (3) and when it was established that mechanical ventilation was the essential therapy of the ensuing respiratory failure. Today, ventilators are one of the most used devices in medicine. However, as is often the case with interventional therapies, there are some adverse consequences of mechanical ventilation. They are primarily iii

iv

Introduction

pulmonary, but they can also be more general, for example, impacting the kidney and the circulatory system. In order to realize the full beneﬁt of mechanical ventilation, it is critical to have knowledge of these adverse events, and of their mechanisms. This monograph, titled Ventilator-Induced Lung Injury and edited by Drs. Didier Dreyfuss, Georges Saumon, and Rolf D. Hubmayr, gives the reader a panoramic but detailed view of the pulmonary adverse consequences of mechanical ventilation. The authors, with international experience, are well known for their expertise in both fundamental and clinical investigations related to mechanical ventilation. The series of monographs, Lung Biology in Health and Disease, has published many volumes focusing on lung diseases—especially acute respiratory distress syndrome—requiring mechanical respiratory assistance and several others on the approaches to mechanical respiration and management of ventilators. However, none have focused exclusively on the adverse consequences of mechanical ventilation. Thus, this volume is a most valuable addition to the series, and it should be of great interest to respiratory care physicians. As the overall editor of the series, I am grateful to the editors and authors for giving us the opportunity to add this volume to the series. Claude Lenfant, MD Gaithersburg, Maryland, U.S.A. References 1. Vesalius A. Pulmonis motuum. De Humani Corporis Fabrica Libri Septem. Basel, 1545. 2. Ashbaugh DB, Bigelow DB, Petty TL. Acute respiratory distress in adults. Lancet 1967; 2:319–323. 3. Sauerbruch F. Zur Pathologie des offenen Pneumothorax und die Grundlagen meines Verfahrens ze seiner Ausschaltung. Mitteilung Grenzebeig Med Chir 1904; 13:3990482.

preface Preface

Few experimental ﬁndings have so sharply inﬂuenced the care of critically ill patients as has been the case with ventilator-induced lung injury. This breakthrough stemmed from a conceptual and experimental effort, stimulated by the need for improving the dismal prognosis of acute respiratory distress syndrome. One must remember the fatality rate of more than 90% (1) in initial clinical series and compare it with the 31% mortality rate observed with a lung protective strategy in the recent study from the Acute Respiratory Distress Syndrome Network (2) to realize the importance of the prognostic progress fostered by these experimental studies. The pioneering study was published by Webb and Tierney (3), who showed that high peak airway pressure ventilation of intact rats provokes pulmonary edema. The lung lesions produced by this ventilation closely mimic those observed during acute respiratory distress syndrome (4,5). In other words, mechanical ventilators are potentially able to generate the disease they are supposed to support. Mead and coworkers (6), based on theoretical considerations, stressed that applying high pulmonary transmural pressure by ventilators to unevenly expanded lungs might cause

v

vi

Preface

hemorrhages in hyaline membranes, only several years after the initial description of acute respiratory distress syndrome (7). This book aims to describe the different steps of basic research that allowed the comprehension of ventilator-induced lung injury, their clinical consequences, and the new avenues of basic research that again emerged. Studies on mechanical transduction, lung mechanics, and endothelial and epithelial physiology formed the cornerstone of this better comprehension. This knowledge stimulated clinical research for designing safer ventilator studies, with overwhelming success for some strategies and persisting questions for others. Finally, new research efforts on the biology of inﬂammatory mediators during ventilator-induced lung injury and on gene therapy during acute lung injury set hope for further improvement of the prognosis for acute respiratory distress syndrome. It was both a privilege and a pleasure for the three editors of this book to ask for the contributions of recognized experts in this ﬁeld, and we wish to express our gratitude for their outstanding chapters, which will undoubtedly make this book a success. Didier Dreyfuss Georges Saumon Rolf D. Hubmayr References 1. Zapol WM, Snider MT, Hill JD, et al. Extracorporeal membrane oxygenation in severe acute respiratory failure. JAMA 1979; 242:2193–2196. 2. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network [see comments]. N Engl J Med 2000; 342:1301–1308. 3. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110:556–565. 4. Bachofen M, Weibel ER. Structural alterations of lung parenchyma in the adult respiratory distress syndrome. In: Bone RC, ed. Clinics in Chest Medicine. Vol. 3. Philadelphia: WB Saunders, 1982:35–56. 5. Dreyfuss D, Basset G, Soler P, Saumon G. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir. Dis 1985; 132:880–884. 6. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28:596–608. 7. Ashbaugh DG, Bigelow DB, Petty TL, Levine BE. Acute respiratory distress in adults. Lancet 1967; 2:319–323.

Contributors

Yochai Adir Division of Pulmonary Medicine, Carmel Medical Center, Technion, Institute of Technology, Haifa, Israel Mircea Anghelescu Department of Pathophysiology, University of Medicine and Pharmacy, Timisoara, Timis, Romania Steven M. Banks Critical Care Medicine Department, Warren G. Magnuson Clinical Center, National Institutes of Health, Bethesda, Maryland, U.S.A. John A. Belperio Division of Pulmonary, Critical Care, and Hospitalists, David Geffen School of Medicine at UCLA, Los Angeles, California, U.S.A. Anastacia M. Bilek Center for Devices and Radiological Health, Food and Drug Administration, Rockville, Maryland, U.S.A. Karen J. Bosma Dipartimento di Anestesiologia e Rianimazione, Ospedale S. Giovanni Battista-Molinette, Universita´ di Torino, Torino, Italy Laurent Brochard Medical ICU, Henri Mondor Teaching Hospital, APHP, Paris 12 University, Cre´teil, France Roy G. Brower Pulmonary and Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland, U.S.A. Katherine J. Deans Critical Care Medicine Department, Warren G. Magnuson Clinical Center, National Institutes of Health, Bethesda, Maryland, and Department of Surgery, Massachusetts General Hospital, Boston, Massachusetts, U.S.A. vii

viii

Contributors

Kay C Dee Department of Applied Biology and Biomedical Engineering, Rose-Hulman Institute of Technology, Terre Haute, Indiana, U.S.A. Didier Dreyfuss Paris 7-Denis Diderot Medical School, Paris, and Service de Re´animation Me´dicale, Assistance Publique-Hoˆpitaux de Paris, Hoˆpital Louis Mourier, Colombes, France Peter Q. Eichacker Critical Care Medicine Department, Warren G. Magnuson Clinical Center, National Institutes of Health, Bethesda, Maryland, U.S.A. Phillip Factor Pulmonary, Allergy, and Critical Care Medicine, Columbia University College of Physicians and Surgeons, New York, New York, U.S.A. Vito Fanelli Dipartimento di Anestesiologia e Rianimazione, Ospedale S. Giovanni Battista-Molinette, Universita´ di Torino, Torino, Italy Niall D. Ferguson Interdepartmental Division of Critical Care Medicine and Department of Medicine, Division of Respirology, University Health Network and Mount Sinai Hospital, University of Toronto, Toronto, Ontario, Canada Leopoldo Ferrer Division of Pulmonary and Critical Care Medicine, University of Minnesota, Regions Hospital, St. Paul, Minnesota, U.S.A. James A. Frank Cardiovascular Research Institute, University of California at San Francisco, San Francisco, California, U.S.A. Bradley P. Fuhrman Division of Pediatric Critical Care, Department of Pediatrics, State University of New York at Buffalo and Women’s and Children’s Hospital of Buffalo, Buffalo, New York, U.S.A. Joe G. N. Garcia Division of Pulmonary and Critical Care Medicine, Center for Translational Respiratory Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland, U.S.A. Donald P. Gaver III Department of Biomedical Engineering, Tulane University, New Orleans, Louisiana, U.S.A.

Contributors

ix

Andreas Gu¨nther Department of Internal Medicine, University of Giessen Lung Center (UGLC), Medical Clinic II, Justus-Liebig University School of Medicine, Giessen, Germany David N. Hager Pulmonary and Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland, U.S.A. Rolf D. Hubmayr Division of Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester, Minnesota, U.S.A. Yumiko Imai Institute of Molecular Biotechnology of the Austrian Academy of Sciences (IMBA), Vienna, Austria Anne-Marie Jacob Department of Biomedical Engineering, Tulane University, New Orleans, Louisiana, U.S.A. Jeffrey R. Jacobson Division of Pulmonary and Critical Care Medicine, Center for Translational Respiratory Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland, U.S.A. Brian P. Kavanagh Departments of Critical Care Medicine and Anesthesia, Hospital for Sick Children, and University of Toronto, Toronto, Ontario, Canada Michael P. Keane Division of Pulmonary, Critical Care, and Hospitalists, David Geffen School of Medicine at UCLA, Los Angeles, California, U.S.A. John G. Laffey Department of Anesthesia and Intensive Care Medicine, Clinical Sciences Institute and National Centre for Biomedical Engineering Sciences, National University of Ireland and University College Hospital, Galway, Ireland Emilia Lecuona Pulmonary and Critical Care Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois, U.S.A. Ste´phanie Lehoux Inserm U689, Cardiovascular Research Center Inserm Lariboisie`re, Paris, France Qin Lu Re´animation Chirugicale Polyvalente Pierre Viars, Hoˆpital Pitie´Salpeˆtrie`re, Assistance Publique Hoˆpitaux de Paris, Universite´ Pierre et Marie Curie, Paris, France John J. Marini Division of Pulmonary and Critical Care Medicine, University of Minnesota, Regions Hospital, St. Paul, Minnesota, U.S.A.

x

Contributors

Philipp Markart Department of Internal Medicine, University of Giessen Lung Center (UGLC), Medical Clinic II, Justus-Liebig University School of Medicine, Giessen, Germany Thomas R. Martin Pulmonary Research Laboratories, VA Puget Sound Health Care System, and Division of Pulmonary and Critical Care Medicine, Department of Medicine, University of Washington School of Medicine, Seattle, Washington, U.S.A. Michael A. Matthay Cardiovascular Research Institute, University of California at San Francisco, San Francisco, California, U.S.A. Christian Me´lot ICU, Erasme Teaching Hospital, Free University of Brussels, Brussels, Belgium Alain Mercat

Medical ICU, Angers Teaching Hospital, Angers, France

Peter C. Minneci Critical Care Medicine Department, Warren G. Magnuson Clinical Center, National Institutes of Health, Bethesda, Maryland, and Department of Surgery, Massachusetts General Hospital, Boston, Massachusetts, U.S.A. Takashige Miyahara First Department of Internal Medicine, Shinshu University, Matsumoto, Nagano, Japan Alan H. Morris Pulmonary and Critical Care Divisions, Department of Medicine, LDS Hospital and University of Utah School of Medicine, Salt Lake City, Utah, U.S.A. Go¨khan M. Mutlu Pulmonary and Critical Care Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois, U.S.A. Charles Natanson Critical Care Medicine Department, Warren G. Magnuson Clinical Center, National Institutes of Health, Bethesda, Maryland, U.S.A. Margaret J. Neff Division of Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington, Seattle, Washington, U.S.A. Stephanie A. Nonas Division of Pulmonary and Critical Care Medicine, Center for Translational Respiratory Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland, U.S.A.

Contributors

xi

Se´verine Oudin Division of Medical Intensive Care, University Hospital of Geneva, Geneva, Switzerland James C. Parker Department of Physiology, University of South Alabama College of Medicine, Mobile, Alabama, U.S.A. Je´roˆme Pugin Division of Medical Intensive Care, University Hospital of Geneva, Geneva, Switzerland V. Marco Ranieri Dipartimento di Anestesiologia e Rianimazione, Ospedale S. Giovanni Battista-Molinette, Universita´ di Torino, Torino, Italy Jean-Damien Ricard Service de Re´animation Me´dicale, Assistance Publique-Hoˆpitaux de Paris, Hoˆpital Louis Mourier, Colombes, France, Inserm U 722, Paris 7-Denis Diderot Medical School, Paris, France Jean-Jacques Rouby Re´animation Chirugicale Polyvalente Pierre Viars, Hoˆpital Pitie´-Salpeˆtrie`re, Assistance Publique Hoˆpitaux de Paris, Universite´ Pierre et Marie Curie, Paris, France Gordon D. Rubenfeld Division of Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington, Seattle, Washington, U.S.A. Clemens Ruppert Department of Internal Medicine, University of Giessen Lung Center (UGLC), Medical Clinic II, Justus-Liebig University School of Medicine, Giessen, Germany Georges Saumon EA 3512, IFR 02 Claude Bernard, Paris 7-Denis Diderot Medical School, Paris, France Daniel P. Schuster Departments of Internal Medicine and Radiology, Washington University School of Medicine, St. Louis, Missouri, U.S.A. Werner Seeger Department of Internal Medicine, University of Giessen Lung Center (UGLC), Medical Clinic II, Justus-Liebig University School of Medicine, Giessen, Germany Jeffrey M. Singh Interdepartmental Division of Critical Care Medicine and Department of Medicine, Division of Respirology, University Health Network and Mount Sinai Hospital, University of Toronto, Toronto, Ontario, Canada

xii

Contributors

Arthur S. Slutsky Interdepartmental Division of Critical Care Medicine and Division of Respirology, Department of Medicine, University of Toronto, and Department of Critical Care Medicine, St. Michael’s Hospital, Toronto, Ontario, Canada Thomas E. Stewart Interdepartmental Division of Critical Care Medicine and Department of Medicine, Division of Respirology, University Health Network and Mount Sinai Hospital, University of Toronto, Toronto, Ontario, Canada Robert M. Strieter Division of Pulmonary, Critical Care, and Hospitalists, and Pathology and Pediatrics, Department of Medicine, David Geffen School of Medicine at UCLA, Los Angeles, California, U.S.A. Jacob I. Sznajder Pulmonary and Critical Care Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois, U.S.A. Alain Tedgui Inserm U689, Cardiovascular Research Center Inserm Lariboisie`re, Paris, France

Contents

Introduction Claude Lenfant . . . . iii Preface . . . . v Contributors . . . . vii PART I: ACUTE MANIFESTATIONS OF VILI 1. Shear and Pressure-Induced Mechanotransduction . . . . . . . . 1 Ste´phanie Lehoux and Alain Tedgui I. Introduction . . . . 1 II. Mechanical Forces . . . . 2 III. Membrane Signal Transduction . . . . 6 IV. Intracellular Signal Transduction . . . . 10 V. Conclusion . . . . 14 References . . . . 15 2. Pulmonary Micromechanics of Injured Lungs . . . . . . . . . . Rolf D. Hubmayr I. Introduction . . . . 21 II. Determinants of Regional Pressure and Volume in Health and Disease . . . . 22 III. Micromechanics of the Normal Lung . . . . 26 IV. Alveolar Micromechanics in Injury States . . . . 29 V. Mechanisms by Which Ventilators Injure Lungs . . . . 32 VI. Concluding Remarks . . . . 37 References . . . . 38

21

xiii

xiv

Contents

3. Response of Cellular Plasma Membrane to Mechanical Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rolf D. Hubmayr I. Introduction . . . . 45 II. The Histology of VILI . . . . 46 III. Cellular Stress Failure in Ventilator-Injured Lungs . . . . 48 IV. Determinants of PM Tension . . . . 50 V. Cell Deformation–Associated PM Remodeling . . . . 52 VI. PM Repair . . . . 55 VII. Effects of PM Wounding on Gene Expression and Cell Survival . . . . 57 VIII. Conclusion . . . . 59 References . . . . 59 4. Acute Passive and Active Changes in Microvascular Permeability During Lung Distention . . . . . . . . . . . . . . . . James C. Parker, Takashige Miyahara, and Mircea Anghelescu I. Introduction . . . . 69 II. Passive Effects of Lung Distention . . . . 71 III. Active Endothelial Control of Vascular Permeability . . . . 72 IV. Conclusion . . . . 86 References . . . . 86 5. Hemodynamic Interactions During Ventilator-Induced Lung Injury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . John J. Marini and Leopoldo Ferrer I. Introduction . . . . 97 II. Effect of Pulmonary Expansion on the Pulmonary Vascular Tree . . . . 98 III. Response of the Endothelial Cell to Shear Forces . . . . 100 IV. Interactions Between Airway and Pulmonary Vascular Pressures . . . . 101 V. Mechanisms Disrupting the Blood–Gas Barrier . . . . 103 VI. Behavior of Airway and Vascular Pressures in Heterogeneous Areas . . . . 105

45

69

97

Contents

xv

VII. Role of Vascular Pressure and Flow on Genesis of VILI . . . . 106 VIII. Effect of Respiratory Rate and Flow on Expression of VILI . . . . 107 IX. Cyclic Effect on the Microvascular Environment Induced by Mechanical Ventilation . . . . 110 X. Effect of Postalveolar Vascular Pressure on the Development of VILI . . . . 111 XI. Potential Clinical Implications . . . . 112 XII. Conclusions . . . . 113 References . . . . 114 6. Lung Mechanics and Pathological Features During Ventilation-Induced Lung Injury . . . . . . . . . . . . . . . . . . 119 Didier Dreyfuss, Jean-Damien Ricard, and Georges Saumon I. Introduction . . . . 119 II. Acute Pulmonary Edema Consecutive to High-Lung-Volume Ventilation . . . . 120 III. Respiratory Mechanics and Severity of VILI . . . . 128 IV. Respiratory System PV Curve Changes During Lung Injury . . . . 129 V. Improvement of Lung Mechanical Properties and Protection from VILI . . . . 143 VI. Clinical Considerations . . . . 145 References . . . . 147 7. The Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume Ventilator-Induced Lung Injury . . . . . . . . . Donald P. Gaver III, Anne-Marie Jacob, Anastacia M. Bilek, and Kay C Dee I. Introduction . . . . 157 II. Background . . . . 158 III. Introduction to Pulmonary Fluid–Structure Interactions . . . . 162 IV. Microscale Fluid–Structure Interactions Leading to VILI . . . . 170 V. The Protective Effect of Pulmonary Surfactant . . . . 185 VI. Future Directions . . . . 193 References . . . . 197

157

xvi

Contents

8. Cellular and Molecular Basis for Ventilator-Induced Lung Injury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Je´roˆme Pugin and Se´verine Oudin I. Introduction . . . . 205 II. Ventilator-Induced Lung Inﬂammation . . . . 206 III. Cells Submitted to Mechanical Stress . . . . 208 IV. What Happens to Cells When They Are Submitted to Cyclic Stretch? . . . . 209 V. Mechanosensing . . . . 209 VI. Cyclic Stretch of Lung Epithelial Cells . . . . 211 VII. Cyclic Stretch–Induced Cell Activation . . . . 211 VIII. Synergy Between Cyclic Stretch and Inﬂammatory Stimuli . . . . 212 IX. Genes Activated by Cyclic Stretch . . . . 213 X. Conclusions and Perspectives . . . . 215 References . . . . 216

205

PART II: SUBACUTE VILI 9. The Role of Cytokines During the Pathogenesis of Ventilator-Associated and Ventilator-Induced Lung Injury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . John A. Belperio, Michael P. Keane, and Robert M. Strieter I. Introduction . . . . 223 II. Mechanical Ventilation of the ALI/ARDS Lung . . . . 224 III. Mechanotransduction Leads to Lung Injury . . . . 225 IV. Cytokines and the Pathogenesis of VALI/ VILI . . . . 225 V. The Role of TNF-a During the Pathogenesis of VALI/VILI . . . . 227 VI. The Role of IL-1b During the Pathogenesis of VALI/VILI . . . . 235 VII. The Role of IL-6 During the Pathogenesis of VALI/VILI . . . . 238 VIII. The Role of IFN-g During the Pathogenesis of VALI/VILI . . . . 239 IX. The Role of IL-10 During the Pathogenesis of VALI/VILI . . . . 240

223

Contents

xvii

X. The Role of TGF-b During the Pathogenesis of VALI/VILI . . . . 241 XI. The Role of Chemokines and Chemokine Receptors During the Pathogenesis of VALI/VILI . . . . 242 XII. The Role of CC Chemokines During the Pathogenesis of VALI/VILI . . . . 246 XIII. Conclusion . . . . 247 References . . . . 249 10. Systemic Effects of Mechanical Ventilation . . . . . . . . . . 267 Yumiko Imai and Arthur S. Slutsky I. Introduction . . . . 267 II. Physiological Effects of MV . . . . 269 III. Mechanical Strain–Induced Release of Inﬂammatory Mediators In Vitro . . . . 270 IV. Pulmonary and Systemic Release of Inﬂammatory Mediators in Ex Vivo and In Vivo Models of VILI . . . . 271 V. Passage of Mediators from Lung to Bloodstream . . . . 273 VI. Injurious Ventilatory Strategies Can Enhance End-Organ Dysfunction, Apoptosis, and Inﬂammation . . . . 275 VII. Bacterial Translocation in MV . . . . 275 VIII. Does the Release of Mediators by VILI Have Any Pathophysiologic Relevance? . . . . 276 IX. Pulmonary and Systemic Inﬂammatory Mediators in VILI in Clinical Studies . . . . 276 X. Multiple Organ Dysfunction and VILI in Clinical Studies . . . . 278 XI. Conclusions . . . . 278 References . . . . 279 11. Alveolar Fluid Reabsorption During VILI . . . . . . . . . . . . Go¨khan M. Mutlu, Emilia Lecuona, and Jacob I. Sznajder I. Introduction . . . . 285 II. Alveolar Epithelial Sodium Transport . . . . 286 III. Alveolar Fluid Reabsorption During VILI . . . . 287

285

xviii

Contents

IV. Summary . . . . 288 References . . . . 288 12. Interaction of VILI with Previous Lung Alterations . . . . . 293 Jean-Damien Ricard, Didier Dreyfuss, and Georges Saumon I. Introduction . . . . 293 II. Surfactant Depletion and Deactivation . . . . 294 III. Toxic Lung Injuries . . . . 297 IV. Inﬂammation and Infection: The Importance of Lung Priming and the Two-Hit Theory . . . . 302 V. Consequences of Previous Lung Injury on Lung Mechanics . . . . 306 VI. Counteracting Previous Lung Injury . . . . 306 VII. Clinical Considerations . . . . 309 References . . . . 310 13. Biological Markers of Ventilator-Induced Lung Injury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thomas R. Martin and Michael A. Matthay I. Introduction . . . . 315 II. Rationale for Biological Markers of VILI . . . . 316 III. Recent Progress in Identifying Biological Markers of VILI . . . . 318 IV. Future Approaches to Identifying Markers of VILI . . . . 330 V. Summary and Conclusions . . . . 333 References . . . . 334 14. Modulation of Lung Injury by Hypercapnia . . . . . . . . . . John G. Laffey and Brian P. Kavanagh I. Introduction—Historical Context . . . . 341 II. Hypercapnia—Deﬁnitions and Terminology . . . . 342 III. Hypercapnia—Physiologic Effects . . . . 345 IV. Acute Organ Injury: Evidence That CO2 Is Protective . . . . 347 V. Mechanisms of CO2-Induced Protection . . . . 356 VI. Molecular Mechanisms of Hypercapnia-Induced Tissue Injury . . . . 360

315

341

Contents

VII. VIII. IX. X. XI.

xix

Administration and Dose Response . . . . 362 Role of Buffering . . . . 363 Hypercapnia—Clinical Studies . . . . 365 Future Directions . . . . 366 Summary . . . . 367 References . . . . 367

15. Alveolar Epithelial Function in Ventilator-Injured Lungs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . James A. Frank and Michael A. Matthay I. Introduction . . . . 377 II. Effects of Mechanical Ventilation on Alveolar Epithelial Barrier Function . . . . 379 III. Alveolar Epithelial Ion and Fluid Transport . . . . 385 IV. Effects of Mechanical Strain on Epithelial Inﬂammatory Mediators . . . . 387 V. Consequences of the Loss of Epithelial Barrier Function . . . . 391 VI. Effects of VILI on Surfactants . . . . 392 VII. Summary . . . . 393 References . . . . 394 16. Genomic Insights into Ventilator-Induced Lung Injury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stephanie A. Nonas, Jeffrey R. Jacobson, and Joe G. N. Garcia I. Introduction—VALI and Genome Medicine . . . . 403 II. Challenges to Unraveling the Genetics of VALI . . . . 404 III. Current Status of VALI/VILI Genetics and the Candidate Gene Approach . . . . 406 IV. Gene Expression in Animal Models of VILI . . . . 408 V. Ortholog Gene Database in VALI and Mechanical Stress . . . . 412 VI. Regional Heterogeneity in Ventilator-Associated Mechanical Stress . . . . 413 VII. Pre-B-Cell Colony–Enhancing Factor as an ALI Candidate Gene . . . . 418

377

403

xx

Contents

VIII. Preliminary PBEF Genotyping in ALI Patients . . . . 420 IX. Preliminary IL-6 Genotyping in VALI . . . . 422 X. Summary . . . . 423 References . . . . 424 PART III: CLINICAL IMPLICATIONS AND TREATMENT OF VILI 17. Lung Imaging of Ventilator-Associated Injury . . . . . . . . . Jean-Jacques Rouby and Qin Lu I. Introduction . . . . 431 II. Histological Evidence of Mechanical Ventilation–Induced Lung Distortion/Overinﬂation . . . . 432 III. CT Evidence of Mechanical Ventilation–Induced Lung Distortion/Overinﬂation . . . . 436 References . . . . 442 18. Imaging Ventilator-Induced Lung Injury: Present and Future Possibilities . . . . . . . . . . . . . . . . . . . . . . . . . . . Daniel P. Schuster I. Introduction . . . . 447 II. Anatomic Imaging of VILI: Quantifying Edema Accumulation . . . . 448 III. Functional Imaging of VILI . . . . 456 IV. Molecular Imaging of VILI . . . . 461 V. Summary . . . . 468 References . . . . 468 19. Modulation of the Cytokine Network by Lung-Protective Mechanical Ventilation Strategies . . . . . . . . . . . . . . . . . Vito Fanelli, Karen J. Bosma, V. Marco Ranieri, and Arthur S. Slutsky I. Introduction . . . . 475 II. MV and the Cytokine Network . . . . 476 III. Modulation of the Cytokine Network in ALI: Evidence from Studies . . . . 479 IV. Impact of MV on the Cytokine Network in Healthy Lungs . . . . 489

431

447

475

Contents

xxi

V. Conclusion . . . . 492 References . . . . 492 20. Role of Tidal Volume and PEEP in the Reduction of VILI . . . . . . . . . . . . . . . . . . . . . . . . . . . . David N. Hager and Roy G. Brower I. Introduction . . . . 497 II. Traditional Approach to MV in ALI/ARDS . . . . 498 III. Mechanisms of VILI . . . . 499 IV. Lung-Protective Ventilation . . . . 500 V. Clinical Trials of Lung-Protective MV Strategies . . . . 503 VI. Controversies . . . . 509 VII. Summary . . . . 513 References . . . . 514 21. A Critical Review of RCTs of Tidal Volume Reduction in Patients with ARDS and Their Impact on Practice . . . . . Peter C. Minneci, Katherine J. Deans, Steven M. Banks, Charles Natanson, and Peter Q. Eichacker I. Introduction . . . . 519 II. Randomized, Controlled Trials of Tidal Volume Reduction in ARDS . . . . 521 III. Meta-Analyses of the RCTs of Tidal Volume Reduction During ARDS . . . . 527 IV. Impact of the Low Tidal Volume Trials on Practice Patterns . . . . 532 V. Conclusions . . . . 533 References . . . . 534 22. The Importance of Protocol-Directed Patient Management for Research on Lung-Protective Ventilation . . . . . . . . . . Alan H. Morris I. Introduction . . . . 537 II. Experimental Scientiﬁc Principles . . . . 541 III. Computerized Protocol Experience . . . . 575 IV. Summary . . . . 589 References . . . . 591

497

519

537

xxii

Contents

23. Crossing the Quality Chasm in Critical Care: Changing Ventilator Management in Patients with ALI . . . . . . . . . Margaret J. Neff and Gordon D. Rubenfeld I. Introduction . . . . 611 II. Understanding Current Practice . . . . 613 III. Do We Know Why Clinicians Do Not Follow Practice Guidelines? . . . . 615 IV. Barriers to Changing Practice in the ICU . . . . 616 V. Models of Changing Clinical Practice . . . . 618 VI. Effective Strategies to Change Practice in the ICU . . . . 621 VII. Conclusions . . . . 621 References . . . . 622 24. How to Design Clinical Studies for Preventing Ventilator-Induced Lung Injury . . . . . . . . . . . . . . . . . . . Laurent Brochard, Christian Me´lot, and Alain Mercat I. Introduction—Questions to Be Addressed . . . . 627 II. Inclusion and Exclusion Criteria . . . . 631 III. Outcomes . . . . 632 IV. Study Designs . . . . 634 V. The RCT . . . . 635 VI. Ethical Issues in a Clinical Trial . . . . 638 VII. Understanding the Results of a Clinical Trial . . . . 644 VIII. Nonrandomized Cohort Studies . . . . 645 IX. Evidence-Based Medicine and Hierarchy of Study Designs . . . . 648 References . . . . 649 25. Perﬂuorocarbons and Acute Lung Injury . . . . . . . . . . . . Bradley P. Fuhrman I. Introduction . . . . 655 II. Perﬂuorocarbon Liquids as Media for Breathing . . . . 655 III. Effects of Perﬂuorocarbons on Inﬂammation and Oxidative Injury . . . . 656 IV. In Vitro Effects of Neat Perﬂuorocarbon Liquids Involving Surface Tension . . . . 660

611

627

655

Contents

xxiii

V. Effects of Ventilation with Perﬂuorocarbons on Lung Injury . . . . 661 VI. Mechanical Protection from Lung Injury by Perﬂuorocarbon Ventilation . . . . 665 VII. Conclusions . . . . 668 References . . . . 668 26. Prospects for Reduction of Ventilator-Induced Lung Injury with Surfactant . . . . . . . . . . . . . . . . . . . . . . . . . Clemens Ruppert, Philipp Markart, Werner Seeger, and Andreas Gu€nther I. Introduction—The Pulmonary Surfactant System . . . . 677 II. Surfactant Alterations and Replacement Treatment in ALI/ARDS . . . . 681 III. Role of the Pulmonary Surfactant System in VILI . . . . 684 IV. Conclusions . . . . 689 References . . . . 690

677

27. Rationale for High-Frequency Oscillatory Ventilation in Acute Lung Injury . . . . . . . . . . . . . . . . . . . . . . . . . . 697 Jeffrey M. Singh, Niall D. Ferguson, and Thomas E. Stewart I. Introduction . . . . 697 II. Background . . . . 698 III. Rationale for HFOV . . . . 699 IV. Clinical Experience with HFOV . . . . 704 V. Future Directions in the Application of HFOV . . . . 705 VI. Conclusion . . . . 707 References . . . . 707 28. Gene Therapy for Ventilator-Induced Lung Injury . . . . . . Go€khan M. Mutlu, Yochai Adir, and Phillip Factor I. Introduction . . . . 711 II. Gene Therapy for ALI . . . . 711 III. Gene Therapy for VILI . . . . 716 IV. Conclusions . . . . 716 References . . . . 717 Index . . . . 721

711

This page intentionally left blank

Part I: ACUTE MANIFESTATIONS OF VILI

1 Shear and Pressure-Induced Mechanotransduction

STE´PHANIE LEHOUX and ALAIN TEDGUI Inserm U689, Cardiovascular Research Center Inserm Lariboisie`re Paris, France

I. Introduction Blood vessels are permanently subjected to mechanical forces in the form of stretch, encompassing cyclic mechanical strain due to the pulsatile nature of blood ﬂow, and shear stress. Blood pressure is the major determinant of vessel stretch. It creates radial and tangential forces that counteract the effects of intraluminal pressure and affect all cell types in the vessel. In comparison, ﬂuid shear stress results from the friction of blood against the vessel wall, and it acts in parallel to the vessel surface. Accordingly, shear is sensed principally by endothelial cells, strategically located at the interface between the blood and the vessel wall. Alterations in stretch or shear stress invariably produce transformations in the vessel wall that will aim to accommodate the new conditions and to ultimately restore the basal levels of tensile stress and shear stress (1,2). Hence, while acute changes in stretch or shear stress correlate with transient adjustments in vessel diameter, mediated through the release of vasoactive agonists or change in myogenic tone, chronically altered mechanical forces usually instigate important adaptive alterations of vessel wall shape and composition. 1

2

Lehoux and Tedgui

The concept of vascular remodeling has therefore been used to describe these transformations that occur in vessels undergoing mechanical stresses. II. Mechanical Forces On the basis of observations in chick embryos, Thoma in 1893 hypothesized that the diameter of blood vessels is regulated by the magnitude of blood ﬂow, while the thickness of vessel walls depends on the magnitude of the forces of tension generated by blood pressure. This hypothesis has subsequently been experimentally conﬁrmed. It has been demonstrated, for example, that the diameter of the abdominal aorta of a lamb undergoes a signiﬁcant reduction between the 4th and 14th days postpartum (3). This reduction can be accounted for by a fall of approximately 70% in the blood velocity in the abdominal aorta at the time of delivery, due to the disappearance of the placental circulation, and is associated with apoptosis of vascular cells (4). Concurrently, the diameter of the thoracic aorta increases in parallel with the rise in systemic blood ﬂow. Similarly, the thicknesses of the pulmonary artery and aorta, which are almost identical at birth due to the similarity in pressures in utero in both vascular territories, evolve differently after birth. The pulmonary artery atrophies during development, following the fall in pulmonary pressure postpartum, while the thoracic aorta thickens proportionately to the increase in systemic pressure (5). A. Tension and Tensile Stress

Blood pressure produces strain on the vessel wall in a direction perpendicular to the endoluminal surface. This is counterbalanced by the intraparietal tangential forces in the longitudinal and circumferential directions exerted by different elements of the vessel wall, opposing the distending effects of blood pressure. The force per unit length of the vessel (the parietal tension, T ) is related to the blood pressure (P) and the vessel radius (r) by Laplace’s law: T ¼ Pr The relation between circumferential tension and deformation of the vessel as intraluminal pressure increases depends on both the geometry and the elastic characteristics of its wall. The circumferential tension is actually borne by the total thickness of the arterial wall. Each element of the wall bears only a part of this tension. The tension per unit of thickness represents the stress exerted on the wall in the circumferential direction. It is expressed as: T ¼ Pr=h where h is the thickness of the wall. Numerous studies have demonstrated a direct relationship between the circumferential stress to which the vessel wall is exposed and the

Shear and Pressure-Induced Mechanotransduction

3

structure of the wall itself (Fig. 1). When the stress increases due to an increase in arterial pressure, smooth muscle cell (SMC) hypertrophy and increases in collagen and elastin contents follow. Inversely, when the circumferential stress falls, the wall undergoes atrophy (6). Several physiologic and experimental arguments conﬁrm the relationship between the circumferential stress and the thickness and composition of the vessel wall: i.

ii.

iii.

From one animal species to another, as the diameter of a particular blood vessel increases, the number of lamellar units and the total thickness of the wall increase proportionately, so that the circumferential stress remains constant irrespective of the size of the animal, from the rat to the horse. This ‘‘ideal’’ value is of the order of 2.106 dyne/cm2 in the descending thoracic aorta (7). It varies according to the arterial territory and essentially depends on the structure of the blood vessel concerned. In all experimental models of arterial hypertension, a close correlation is observed between the level of arterial pressure and the frequency of polyploidy and hypertrophy of the SMCs of the arterial wall. SMC hypertrophy in the walls of the major arterial trunks develops only when the distending pressure has reached a threshold level, and never precedes the onset of hypertension, even when the neurohumoral abnormalities responsible for hypertension are already present.

Figure 1 Sequence of vascular responses stemming from increased transmural pressure or shear stress and leading, through sequential events, to vascular remodeling.

4

Lehoux and Tedgui

The effects of mechanical tensile stress on the arterial wall have been extensively described and have been applied to the understanding of hypertension. Numerous animal and human studies have shown that sustained hypertension is associated with structural and functional alterations in both large arteries and arterioles. There is good evidence that hypertension is associated with increased arterial wall thickness (8), mostly due to SMC hypertrophy, accompanied by polyploidism, hyperplasia, and proportional changes in contractile and matrix proteins, leading to altered arterial function (9). According to Laplace’s equation (T ¼ Pr/h), the hypertrophy of the arterial wall compensates for the increase in blood pressure and contributes to maintaining a normal level of circumferential stress. In elastic and large conduit arteries, the adaptive response to hypertension serves to reduce and eventually normalize the tensile stress. On the other hand, constant mechanical stimulation appears to be required for maintenance of normal contractile phenotype of SMC in the arterial wall. Vessels placed in conditions of abnormally low intraluminal pressure (10 mmHg) show decreased content, over three to six days, of smooth muscle marker proteins h-caldesmon and ﬁlamin content, compared with native vessels or aortic segments kept at physiological intraluminal pressure (80 mmHg) (10). Likewise, cyclic stretching of cultured airway SMC increases (in fact, prevents the decrease in) the expression of smooth muscle myosin heavy chains and myosin light chain kinase (11). Loss of stretch, together with loss of extracellular matrix contacts, is probably the major cause of differentiation of SMC in culture. Hence, a certain level of stretch is required to maintain vascular SMC (VSMC) in a quiescent state, but overstretching triggers adaptive processes resulting in increased protein synthesis and hypertrophy. B. Shear Stress

As blood ﬂows, it exerts a frictional force on the endothelial surface. This force is expressed as a shear stress (s) on the endothelium, deﬁned as the product of the blood viscosity and the blood-velocity gradient measured at the vessel wall. The shear stress transmitted to the endothelium by the blood ﬂow tends to displace the endothelium and the intimal layer in the direction of ﬂow (one might equally say that it is because the endothelium is ﬁxed that friction occurs). In the case of laminar ﬂow (where the proﬁle of blood velocity is parabolic), shear stress is expressed as: s ¼ 4lQ=pr3 where l is the viscosity, Q the ﬂow rate, and r the vessel radius. Note that the radius appears at the third power in the denominator. Thus, for a constant volume ﬂow, a slight reduction in vascular diameter produces a much greater increase in shear stress.

Shear and Pressure-Induced Mechanotransduction

5

Shear stress arising from the mechanical effects of blood ﬂow on the vascular endothelium is also a determinant of arterial growth (Fig. 1). Under physiologic conditions, the mean shear stress to which the vascular endoluminal surface is exposed is remarkably constant, close to 10 to 15 dyne/cm2, whatever the part of the arterial network considered, conductance or resistance arteries, and whatever the size of the animal (with the exception of the rat and the mouse in which the values are closer to 30 to 35 dyne/cm2). Shear stress–dependent remodeling can be illustrated by experiments where blood ﬂow is either restricted or enhanced. In rabbits, the reduction in caliber of the developing carotid associated with a reduction in its blood ﬂow is accompanied by a reduction in the elastin content of the carotid arterial wall (12). In contrast, the phenomenon of ﬂow-dependent growth is best exempliﬁed using the arteriovenous ﬁstula model. In carotid–jugular arteriovenous ﬁstulas, the ﬂow rate in the developing carotids can be multiplied by a factor of up to 8. The chronic increase in shear tends to enhance the L-arginine/nitric oxide (NO) pathway in endothelial cells, and chronic inhibition of NO production by Nx-nitro-L-arginine methyl ester (L-NAME) treatment inhibits, at least partially, the adaptive wall shear stress regulation in ﬂow-loaded vessels (2). However, simple relaxation of VSMC alone cannot account for the very signiﬁcant increase in vascular caliber observed, which may almost double in response to large increases in ﬂow. Previous microscopic and ultrastructural studies of the arterial wall proximal to an arteriovenous ﬁstula have shown extensive tears and fragmentation, as well as enlarged fenestrae, in the internal elastic lamina (IEL) (2,13,14), suggesting a potential role for matrix metalloproteinases (MMPs) in matrix digestion and reorganization leading to arterial wall remodeling. Indeed, increased blood ﬂow in the rabbit carotid due to an arteriovenous shunt causes the release of MMP-2 and MMP-9, and chronic MMP inhibition prevents IEL fragmentation and adaptive remodeling of the ﬂow-loaded artery (15). Thus, MMP-induced IEL fenestrations are formed following increased blood ﬂow, contributing to arterial distensibility and resulting in an enhanced arterial diameter. As arterial caliber gradually increases, wall shear stress diminishes and the stimulus for MMP production/activation fades. In summary, vessels are normally exposed to two types of mechanical forces: (a) circumferential stress acting tangentially on the vascular wall and directly related to pressure and dimensions (diameter and thickness) of the vessel, and (b) shear stress acting in a longitudinal direction at the blood– endothelium interface and directly related to the ﬂow-velocity proﬁle. Signiﬁcant variations in mechanical forces, of a physiological or physiopathological nature, occur in vivo. These are accompanied by phenotypical modulation of the SMC and the endothelial cells, producing structural modiﬁcations of the arterial wall. In all the cases, vascular remodeling can be attributed to a modiﬁcation of the tensional strain or shear, and underlies a trend to reestablish baseline mechanical conditions.

6

Lehoux and Tedgui III. Membrane Signal Transduction

Vascular cells are equipped with numerous receptors that allow them to detect and respond to the mechanical forces generated by pressure and shear stress. The cytoskeleton and other structural components have an established role in mechanotransduction, being able to transmit and modulate tension within the cell via focal adhesion sites, integrins, cellular junctions, and the extracellular matrix. The cytoskeleton is composed of three major types of protein ﬁlaments: microtubules, microﬁlaments, and intermediate ﬁlaments. Microﬁlaments are polymers of actin that together with a large number of actin-binding and associated proteins form a continuous, dynamic connection between nearly all cellular structures. The cytoskeletal network changes in response to extracellular stimuli and participates in transmembrane signaling, providing a scaffold for organizing or translocating signaling molecules and organelles. Beyond the structural modiﬁcations incurred, mechanical forces can thus initiate complex signal transduction cascades leading to functional changes within the cell, often triggered by activation of integrins, but also by stimulation of other structures such as G-protein receptors, tyrosine kinase receptors, or ion channels (Fig. 2). A. Integrins

The extracellular matrix is an important contributor to the process of mechanotransduction, containing glycoproteins that are displaced by stretch or shear forces and interact with integrins. The latter proteins contribute not only to cell attachment to the substrate, but also to intracellular transmission of mechanical signals. Mechanical stresses stimulate conformational activation of cell integrins and increase cell binding to the

Figure 2 Schematic representation of receptors involved in initiating signaling cascades in vascular cells stimulated by pressure (stretch) or shear stress.

Shear and Pressure-Induced Mechanotransduction

7

extracellular matrix (16). In fact, the dynamic formation of new integrin– ligand connections is required for stretch- or shear-induced mechanotransduction, because blocking unoccupied extracellular matrix ligand sites with isotype speciﬁc antibodies or RGD peptides (RGD being the principal amino acid sequence on extracellular matrix proteins to which integrins bind) inhibits intracellular signaling induced by mechanical forces (16,17). The cytoplasmic domain of integrins is functionally linked to various intracellular proteins that constitute the cytoskeleton and numerous kinases such as focal adhesion kinase (FAK), a key regulator of biochemical cascades initiated by mechanical forces. Integrins therefore form a signaling interface between the extracellular matrix and the cell. Integrins exist as ab pairings that interact with extracellular matrix components including ﬁbronectin (ligand for a5b1 and avb3), vitronectin (ligand for avb3), and laminin (ligand for a6b1). The capacity of cells to sense mechanical forces and the ensuing responses therefore depend on speciﬁc integrin–extracellular matrix interactions. For example, cyclic stretching of SMC grown on ﬁbronectin or vitronectin induces cellular proliferation, which is prevented by anti-b5 or anti-avb3 antibodies, whereas SMC grown on elastin or laminin do not proliferate under the same conditions (17). In comparison, cyclic stretch induces greater expression of the SM-1 isoform of myosin heavy chain in SMC plated on laminin than in SMC grown on on collagen or ﬁbronectin (18). Finally, in SMC plated on type I collagen, serum induces the expression of c-fos and cell proliferation in stretched cells and unstretched controls equally. However, in SMC grown on elastin matrix, both the serum-induced expression of c-fos and the ensuing cell proliferation are abated by stretch (19). Shear stress also induces integrin-speciﬁc signaling cascades. In endothelial cells plated on ﬁbronectin or vitronectin, but not on collagen or laminin, shear triggers avb3-dependent mechanotransduction and association of the integrin with the adapter protein Shc. In contrast, shear stress causes association of a6b1 with Shc in cells plated on laminin, but not on ﬁbronectin, vitronectin, or collagen (16). In cultured endothelial cells, shear stress activates the nuclear factor NFjB, which, acting as the shear stress response element, can promote the expression of mechanosensitive genes. Incubating endothelial cells with an anti-avb3 antibody prevents the activation of NFjB by shear stress (20). Perhaps most importantly, in isolated coronary arteries, where endothelial cells lie on native extracellular matrix, ﬂow-dependent dilation can be abrogated by addition of RGD peptides to the culture medium (21). Similar results are obtained when anti-b3 antibodies are used (21). Integrins are therefore key sensing elements involved in mechanotransduction in vascular cells. The nature of the mechanical stimulus and the substrate components to which the cells are attached determine which integrin–ligand pairs will be recruited and which downstream intracellular

8

Lehoux and Tedgui

cascades will be activated, and hence the ensuing cell response. In this context, whole vessel preparations are particularly adapted to the study of the role of integrins in mechanotransduction, because cells are then in their original three-dimensional and complex extracellular matrix environment. B. Ion Channels

Two different mechanosensitive channels have been described in vascular cells: shear-activated potassium channels and stretch-activated channels (22). Stretch-activated ionic channels are cation-speciﬁc and have an electric activity mainly detectable at the time of their opening. The activation of these channels leads to calcium (Ca2þ) inﬂux followed by membrane depolarization (22). A role for stretch-activated cation channels in mechanotransduction in SMC was conﬁrmed using the speciﬁc blocker gadolinium (23). Flow-induced smooth muscle marker protein expression was reduced by gadolinium, whereas other calcium channel blockers, such as verapamil, did not inhibit the stimulatory effect of shear. Gadolinium also prevents cell proliferation observed in periodically stretched SMC (24). Exposing endothelial cells in culture to shear stress leads to membrane hyperpolarization due to potassium channel opening (25). Because calcium entry in the cell is dependent on membrane potential, the increase in this potential induced by shear raises Ca2þ intake, resulting in an accumulation of calcium in endothelial cells and an enhancement of calcium-dependent signaling cascades. This interpretation is supported by experiments showing on the one hand that endothelial cells do not possess voltage-dependent calcium channels, and on the other hand that high extracellular potassium concentrations reduce calcium entry into these cells (25). Recently, upregulation and activation of endothelial intermediate-conductance Ca2þ-activated Kþ channels [IK(Ca)] was reported in endothelial cells exposed to laminar shear stress (26). Nevertheless, the mechanisms involved in the control of open/closed ion channel conformations by shear remain obscure. One likely contributor is the cytoskeleton, which by deformation could alter channel activation state. In support of this hypothesis, one study implicates cytoskeleton– G-protein coupling in shear-induced potassium channel opening (27). Another recent work highlights a direct role for gadolinium-sensitive channels in endothelial endothelin-1 expression stimulated by rotating integrin-linking RGD peptide–covered ferromagnetic beads (28), establishing a functional link between integrins, the cytoskeleton, and ion channels. As shown by Davies (29), in areas where ﬂow is alternately laminar and turbulent and where mechanical forces vary within short distances, shear and stretch can induce synergistic or antagonistic effects through differential activation of ion channels. Ultimately, the physiological role of various ion channels, sensitive either to shear stress or to stretch, appears to depend on the balance between these hemodynamic forces in the circulation.

Shear and Pressure-Induced Mechanotransduction

9

C. Heterodimeric G Proteins

G proteins consist of three subunits, a, b, and c, which couple membrane receptors with intracellular signaling cascades. If one considers the crucial role of G proteins in the regulation of the cardiovascular system, it is not surprising to ﬁnd that they participate in the transduction of mechanical forces in the endothelium. Indeed, it has been shown that shear-induced regulation of platelet-derived growth factor (PDGF) gene expression is regulated by a protein kinase C (PKC)-dependent mechanism requiring the presence of calcium and G-protein induction (30). The same authors also reported that shear induces the expression of c-fos via a complex mechanotransduction cascade involving PKC, phospholipase C, G proteins, and calcium (31). Moreover, the direct effect of shear on the activation of Gaq/a11 and Gai3/ao in endothelial cells was demonstrated (32), and the activation of both these G proteins was found to be necessary for the activation of downstream signaling cascades (33). The c subunit of heterodimeric G proteins is reported to be present at integrin-rich focal adhesion sites and adjacent to F-actin ﬁlaments stress ﬁbers (34). Colocalization of G proteins and integrins would even allow for a single signal to activate two transmembrane receptor families simultaneously, G protein–coupled receptors and integrins. Thus, G proteins could be indirectly involved in integrin-mediated signaling. Indeed, G protein inhibition prevents activation of potassium channels stimulated by cell adhesion to the extracellular matrix via integrins (35). Acting on integrins, shear deforms the cytoskeleton and so activates a G protein that opens the potassium channels. Interestingly, there are thus far no indications that mechanical forces can activate G proteins in vascular SMC. D. Receptor Tyrosine Kinases

Another class of membrane proteins, receptor tyrosine kinases, also take part in mechanotransduction. For example, activation and phosphorylation of PDGF receptor-a are observed in SMC exposed to cyclic stretch or shear stress (36). That could be explained by a disturbance of the cellular surface or an alteration of the receptor conformation by mechanical forces (36). However, the participation of gadolinium-sensitive Ca2þ channels cannot be excluded. Indeed, the latter are implicated in the phosphorylation of the EGF receptor by mechanical stimulation (37). The role of the phosphorylation of EGF receptors in mechanotransduction was highlighted when protein synthesis induced in stretched SMC was blocked when the cells were incubated with an EGF receptor antagonist (37). In endothelial cells, shear stress induces the transitory phosphorylation of the VEGF receptor Flk-1 and its association with Shc and avb3 and b1 integrins (38). If the role of Flk-1 in mechanotransduction has not yet been perfectly established, it remains that preventing the association

10

Lehoux and Tedgui

of Shc with Flk-1, or with other proteins, attenuates the downstream activation cascades as well as the gene transcription stimulated by shear (38). E. Oxygen-Free Radicals

Recent data suggest that oxygen-free radicals, as well as endogenous antioxidants, probably have critical signaling functions in cells (39). A signiﬁcant source of vascular oxygen-free radicals is the membrane oxidase NADH/ NADPH, whose activity is controlled by hormones, growth factors, and mechanical forces. The basic product of this enzymatic system is the superoxide anion (O 2 ), which is transformed quickly into H2O2 by superoxide dismutase. The H2O2 is transformed in its turn by two enzymes, catalase and glutathione peroxidase. The breakdown products of the H2O2, including lipid hydroperoxides, are also biologically active. On the whole, oxygenfree radicals thus comprise several potential secondary messengers. The production of oxygen-free radicals has been detected in endothelial cells exposed to a cyclic stretch of 10% to 12% (40), and similarly, applying a 10% cyclic stretch to human coronary artery SMC stimulates the production of O 2 , while a stretch of 6% does not have any signiﬁcant effect (41). The activation of PKC, which is induced by stretch and which can activate NADPH oxidase, could in certain cases precede the generation of O 2 (41). However, 10% cyclic stretch stimulates generation of O2 and downstream signaling independently of PKC in whole vessel preparations (42). It has also been proposed that an increase in H2O2 in endothelial cells can induce the reorganization of F-actin, characterized by the formation of stress ﬁbers and the recruitment of vinculin to focal adhesion sites (43). Furthermore, the endothelial oxidative response to stretch is matrix protein–dependent, and is reduced by coincubation with RGD peptides or blocking antibodies to a2- and b-integrin antibodies (44). Interestingly, NADH oxidase activity is upregulated in endothelial cells exposed to oscillatory shear for 24 hours, whereas steady laminar shear inducesa more transientresponse (45).In fact,at24 hours, steady shear induces superoxide dismutase, unlike oscillatory shear (45), consistent with the atheroprotective quality of laminar ﬂow. IV. Intracellular Signal Transduction A. NO and Akt

One of the early events that occurs in endothelial cells placed under ﬂow is the activation of the endothelial NO synthase (eNOS) and the subsequent release of NO. Recent studies show that the activation of eNOS by shear stress does not require Ca2þ inﬂux in the cell, as is the case for its activation by vasoactive agonists, but rather its phosphorylation by Akt (or protein kinase B) (46), which is itself phosphorylated by phosphatidylinositol-3-kinase (47).

Shear and Pressure-Induced Mechanotransduction

11

The intracellular transduction pathways that link shear with eNOS activation are numerous. On the one hand, eNOS activation by shear can be prevented by a potassium channel blocker and necessitates an intact cytoskeleton. On the other hand, the phosphorylation of eNOS and of Akt in endothelial cells under ﬂow is sensitive to tyrosine kinase inhibitors, indicating a possible implication of receptors for VEGF or insulin (48). Akt activation is also observed in cultured SMC subjected to a cyclic stretch (49). In addition to its role of vasodilator, NO intervenes in the regulation of the vascular remodeling induced by chronic shear stress, because inhibition of this pathway attenuates the increase in diameter observed in arteriovenous ﬁstulas and thus prevents ﬂow-dependent adaptation (2). As a result, the vessel loses its capacity for enlargement and shear levels stay at an abnormally high level. Under this condition, NO plays the role of cofactor, facilitating metalloproteinase activation (15). In addition, Akt activation and the production of NO support the survival of the vascular cells by stimulating antiapoptotic pathways and inhibiting proapoptotic cascades (47). B. Focal Adhesion Kinase

During the stimulation of vascular cells by mechanical factors such as stretch or shear, several signaling events are associated with the formation of focal adhesions, which comprise integrin clusters and cytoskeletal proteins, as well as various tyrosine kinases, including FAK. There are in fact several different proteins that are known to bind the cytoplasmic domain of integrins, and which may also be involved in mechanotransduction. Nevertheless, the role of FAK is particularly well established in the context of mechanotransduction. Indeed, a recent study shows that FAK is activated in stretched pulmonary vessels, in particular in the endothelium (50), and activation of this enzyme was also demonstrated in cultured endothelial cells exposed to shear stress (51). The recruitment of integrins to focal adhesion sites is mediated by their cytoplasmic domains, which bind proteins of the cytoskeleton (52). The proteins present at focal adhesions become phosphorylated on tyrosine when the cells are stimulated, and FAK activation is an indicator in focal adhesion formation, rather than the engine of their assembly (53). c-Src, a tyrosine kinase associated with the membrane, also plays a role in the process of FAK activation. Following its activation by stretch, c-Src is transferred to the focal contacts (54), where it interacts with an autophosphorylation site on FAK and creates an acceptor for the Srchomology-2 domain of Grb2 and thus supports association of FAK with the latter (Fig. 3). Although not shown yet in the context of mechanotransduction, activation of FAK could also involve RhoA, because inhibition of this small G protein by Clostridium botulinum C3 exoenzyme transferase disassembles focal adhesions and reduces phosphorylation of FAK in endothelial cells (55) and VSMCs (56).

12

Lehoux and Tedgui

Figure 3 Diverse pathways potentially involved in the activation of MAP kinases (ERK1/2 in this diagram) by mechanical factors. Abbreviations: MAP kinases, mitogen-activated protein kinases; ERK, extracellular signal-regulated kinase.

C. The Mitogen-Activated Protein Kinase Cascade

The mitogen-activated protein (MAP) kinase cascade is an important pathway whereby signals originating from mechanical forces can lead to gene expression and protein synthesis (57). This pathway implicates the sequential phosphorylation and activation of the cytoplasmic protein kinases MAP kinase kinase kinase (MEKK), map kinase kinase (MEK), and ﬁnally MAP kinase. The MAP kinase cascade comprises in reality three different pathways that are triggered in response to various stimuli and initiate distinct cellular responses. The phosphorylation of one of the MAP kinases, which lies downstream of Raf and is present under two isoforms, ERK1 and 2 (extracellular signal-regulated kinase), leads to the activation of regulatory proteins in the cytoplasm and the nucleus. Other MAP kinases, called stress-activated protein kinases because they are activated by stimuli such as ultraviolet light, heat shock, hypoxia, or hyperosmolarity, include C-jun N-terminal kinases (JNK) (which phosphorylate the amino-terminal of the transcription factor c-jun), and p38. There is ample evidence that MAP kinases are activated in vascular cells exposed to mechanical forces, both in vivo and in vitro. Cyclic stretch activates ERK1/2 and JNK in cultured SMC (58), and ERK1/2 and JNK are transiently activated in the arterial wall by acute hypertension (59). Using aortic segments in organ culture, it was shown that high intraluminal pressure (150 mmHg) induces a biphasic stimulation of ERK1/2, characterized by an acute activation peak with return to baseline at two hours, and a second, more prolonged rise within 24 hours and lasting at least three days (60). A similar phenomenon, though slower in its acute phase, was also observed in vessels exposed to 10% cyclic stretch (42). In the latter model,

Shear and Pressure-Induced Mechanotransduction

13

cyclic stretch also activated p38 (42). Finally, MAP kinase activation pathways were also underscored in endothelial cells, in which shear forces of 12 dyne/cm2 induced the phosphorylation of ERK1/2 and p38, but reduced the activity of JNK (61). The activation of MAP kinases most likely involves integrins as upstream mechanical sensors for several reasons. First, the in vitro response of vascular cells to stretch or shear varies considerably according to the nature of the substrate on which the cells are plated. For example, both ERK1/2 and JNK are activated by cyclic stretch in neonatal SMC grown on pronectine, but if the same cells are grown on laminin, only JNK is stimulated by cyclic stretch (58). Second, in endothelial cells, ERK1/2 activation by shear or following adhesion to ﬁbronectin occurs via a common integrin-dependent pathway sensitive to the c-Src kinase family inhibitor herbimycin A and dependent on PKC (62). Third, overexpression of FAK increases ﬁbronectin-dependent c-Src kinase activity and subsequent activation of ERK2, whereas a dominant negative Ras blocks activation of ERK1/2 without affecting phosphorylation of FAK or c-Src activity (54). Finally, substitution of the c-Src acceptor on FAK blocks the transmission of signals between integrins and ERK1/2 (54). Taken together, these observations highlight a pathway starting with integrin activation, focal adhesion assembly, FAK activation by c-Src, association with Grb2 driving c-Src-dependent activation of Ras, and ultimately activation of ERK1/2 via the MAP kinase cascade (Fig. 2). Pathways other than the ones described above also participate in mechanotransduction. For instance, there is evidence that integrin-dependent activation of MAP kinases can in certain cases bypass FAK. Adhesion to matrix can activate ERK in cells expressing a mutant form of the b1 integrin lacking the cytoplasmic segment necessary for FAK interaction (63). Furthermore, the MAP kinase cascade can also be activated by tyrosine phosphorylation of a, b, and c GTP subunits of G proteins (64), as well as by mechanosensitive phosphorylation of tyrosine kinase type receptors (36–38). As described above, cyclic stretch induces the release of oxygenfree radicals in cultured cells. The activation of Ras by oxygen-free radicals, which in theory precedes activation of Raf and the MAP kinase cascade, was reported (65), in agreement with the observed activation of ERK1/2 by O 2 in SMC (66). Finally, the inhibition of small G protein RhoA or its downstream kinase, RhoA kinase (p160ROCK), completely prevents stretch-induced ERK1/2 activation (67) or shear-induced JNK activation (68). Not surprisingly, different pathways can bridge the gap between mechanical stimulation and ERK1/2 activation in vascular cells. As an example, both high intraluminal pressure (150 mmHg) and 10% cyclic stretch activate ERK1/2 in vessels in organ culture. Nonetheless, c-Src kinase inhibition prevents ERK1/2 activation only in vessels at high pressure, and not in pulsatile vessels. On the other hand, activation of ERK1/2 by cyclic stretch is

14

Lehoux and Tedgui

mediated by the release of oxygen-free radicals (42,60). In comparison, shear-induced ERK1/2 activation in cultured endothelial cells is prevented by inhibition or downregulation of PKC, or inhibition of tyrosine kinase activity, and is probably coupled with the activation of G proteins (69). Hence multiple MAP kinase activation pathways can be induced by stretch or shear in vessels, depending on the nature of the mechanical stimulus and the cell types and the extracellular matrix environment involved. The events that occur downstream of the activation of MAP kinases are numerous and varied. Once phosphorylated, ERK1/2 can transfer to the nucleus, where it interacts with and phosphorylates transcription factors, thus controlling gene expression. Both ERK1/2 and JNK can lead to ternary complex formation with the serum response element, present on several gene promoters, and thus increase transcriptional activity (70). Alternatively, phosphorylation of the protein PHAS-I (phosphorylated heat- and acid-stable protein), a translation regulation factor, supports the dissociation of the PHAS-I–eukaryotic initiation factor (eIF) -4E complex, normally closely apposed when PHAS-I is relatively underphosphorylated, releasing eIF-4E, which in turn initiates translation in the nucleus (71). Another downstream target of ERK1/2 in SMC is the 90-kDa ribosomal S6 kinase, which, by activation of the transfer RNA–binding factor, provides an additional pathway for initiation of translation (71). Finally, ERK1/2 activation leads to enhanced expression of c-fos and c-jun and to activation of the AP-1 transcription factor, and as such is likely to play a signiﬁcant role in the regulation of cell cycle progression and in protein synthesis in SMC (71). The availability of downstream ligands could be a factor that determines the biological response to ERK1/2 activation.

V. Conclusion Blood vessels have autocrine and paracrine hormonal mechanisms that enable them to react immediately to local hemodynamic modiﬁcations involving tangential mechanical stretch (which increases with pressure) or shear stress (which increases with blood ﬂow). Vascular tone is modiﬁed almost immediately to compensate for changes in the environment, and in most cases, this efﬁciently restores mechanical forces to normal levels. Exceptionally, the variations in vasomotor tone are not sufﬁcient to compensate for the new mechanical constraints, and the phenotype of the vascular cells is altered, causing local modiﬁcations in trophicity. At length (over a few days to a few weeks), these adaptative changes also tend to return mechanical forces to their physiological values. Vascular remodeling is observed in various situations where the local pressures and ﬂows are modiﬁed, such as arterial hypertension, atherosclerosis, arteriovenous ﬁstula, stenosis, and aneurysm.

Shear and Pressure-Induced Mechanotransduction

15

Many receptors, present on the surface of endothelial cells and SMC, allow vessels to detect subtle changes in their physical environment. From that point, different mechanotransduction cascades can be initiated according to the nature of the mechanical stimulus perceived. Inside the vascular cells, cytoskeletal proteins transmit and modulate the tension between focal adhesion sites, integrins, and the extracellular matrix. In addition to the structural modiﬁcations induced by the mechanical forces, they may lead to changes in the ionic composition of the cells, mediated by ion channels, stimulate various membrane receptors, and induce complex biochemical cascades. Many intracellular pathways, such as the MAP kinase cascade, are activated by ﬂow or stretch and initiate, via sequential phosphorylations, the activation of transcription factors and subsequent gene expression. Thus, by purely local mechanisms, the blood vessels are capable of a true autonomic regulation, which enables them to adapt to their mechanical environment.

References 1. Glagov S. Intimal hyperplasia, vascular modeling, and the restenosis problem. Circulation 1994; 89:2888–2891. 2. Tronc F, Wassef M, Esposito B, Henrion D, Glagov S, Tedgui A. Role of NO in ﬂow-induced remodeling of the rabbit common carotid artery. Arterioscler Thromb Vasc Biol 1996; 16:1256–1262. 3. Langille BL, Brownlee RD, Adamson SL. Perinatal aortic growth in lambs: relation to blood ﬂow changes at birth. Am J Physiol 1990; 259:H1247–H1253. 4. Cho A, Courtman DW, Langille BL. Apoptosis (programmed cell death) in arteries of the neonatal lamb. Circ Res 1995; 76:168–175. 5. Leung DYM, Glagov S, Mathews MB. A new in vitro system for studying cell response to mechanical stimulation. Exp Cell Res 1977; 109:285–298. 6. Bomberger RA, Zarins CK, Taylor KE, Glagov S. Effect of hypotension on atherogenesis and aortic wall composition. J Surg Res 1980; 28:401–409. 7. Wolinsky H, Glagov S. Comparison of abdominal and thoracic aortic medial structure in mammals. Deviation of man for usual pattern. Circ Res 1969; 25:677–686. 8. Levy BI, Michel JB, Salzmann JL, et al. Effects of chronic inhibition of converting enzyme on the mechanical and structural properties of arteries in rat renovascular hypertension. Circ Res 1988; 63:227–239. 9. Intengan HD, Schiffrin EL. Vascular remodeling in hypertension: roles of apoptosis, inﬂammation, and ﬁbrosis. Hypertension 2001; 38:581–587. 10. Birukov KG, Bardy N, Lehoux S, Merval R, Shirinsky VP, Tedgui A. Intraluminal pressure is essential for the maintenance of smooth muscle caldesmon and ﬁlamin content in aortic organ culture. Arterioscler Thromb Vasc Biol 1998; 18:922–927. 11. Smith PG, Tokui T, Ikebe M. Mechanical strain increases contractile enzyme activity in cultured airway smooth muscle cells. Am J Physiol 1995; 268:L999–L1005.

16

Lehoux and Tedgui

12. Langille BL, Bendeck MP, Keeley FW. Adaptations of carotid arteries of young and mature rabbits to reduced carotid blood ﬂow. Am J Physiol 1989; 256:H931–H939. 13. Jones GT, Stehbens WE. The ultrastructure of arteries proximal to chronic experimental carotid-jugular ﬁstulae in rabbits. Pathology 1995; 27:36–42. 14. Wong LCY, Langille BL. Developmental remodeling of the internal elastic lamina of rabbit arteries – effect of blood ﬂow. Circ. Res 1996; 78:799–805. 15. Tronc F, Mallat Z, Lehoux S, Wassef M, Esposito B, Tedgui A. Role of matrix metalloproteinases in blood ﬂow-induced arterial enlargement. Arterioscler Thromb Vasc Biol 2000; 20:e120–e126. 16. Jalali S, del Pozo MA, Chen K, et al. Integrin-mediated mechanotransduction requires its dynamic interaction with speciﬁc extracellular matrix (ECM) ligands. Proc Natl Acad Sci USA 2001; 98:1042–1046. 17. Wilson E, Sudhir K, Ives HE. Mechanical strain of rat vascular smooth muscle cells is sensed by speciﬁc extracellular matrix/integrin interactions. J Clin Invest 1995; 96:2364–2372. 18. Reusch P, Wagdy H, Reusch R, Wilson E, Ives HE. Mechanical strain increases smooth muscle and decreases nonmuscle myosin expression in rat vascular smooth muscle cells. Circ Res 1996; 79:1046–1053. 19. Spofford CM, Chilian WM. The elastin-laminin receptor functions as a mechanotransducer in vascular smooth muscle. Am J Physiol Heart Circ Physiol 2001; 280:H1354–H1360. 20. Bhullar IS, Li YS, Miao H, et al. Fluid shear stress activation of IkappaB kinase is integrin-dependent. J Biol Chem 1998; 273:30544–30549. 21. Muller JM, Chilian WM, Davis MJ. Integrin signaling transduces shear stressdependent vasodilation of coronary arterioles. Circ Res 1997; 80:320–326. 22. Sackin H. Stretch-activated ion channels. Kidney Int 1995; 48:1134–1147. 23. Shirinsky VP, Birukov KG, Stepanova OV, Tkachuk VA, Hahn AWA, Resink TJ. Mechanical stimulation affects phenotype features of vascular smooth muscles. In: Woodford FP, Davignon J, Sniderman A, eds. Atherosclerosis X. Amsterdam: Elsevier Science, 1995:822–826. 24. Standley PR, Obards TJ, Martina CL. Cyclic stretch regulates autocrine IGF-I in vascular smooth muscle cells: implications in vascular hyperplasia. Am J Physiol 1999; 276:E697–E705. 25. Nakache M, Gaub HE. Hydrodynamic hyperpolarization of endothelial cells. Proc Natl Acad Sci USA 1988; 85:1841–1843. 26. Brakemeier S, Kersten A, Eichler I, et al. Shear stress-induced up-regulation of the intermediate-conductance Ca(2þ)-activated K(þ) channel in human endothelium. Cardiovasc Res 2003; 60:488–496. 27. Ohno M, Gibbons GH, Dzau VJ, Cooke JP. Shear stress elevated endothelial cGMP. Role of a potassium channel and G protein coupling. Circulation 1993; 88:193–197. 28. Chen J, Fabry B, Schiffrin EL, Wang N. Twisting integrin receptors increases endothelin-1 gene expression in endothelial cells. Am J Physiol Cell Physiol 2001; 280:C1475–C1484. 29. Davies PF. How do vascular endothelial cells respond to ﬂow?. New Physiol Sci 1989; 4:22–26.

Shear and Pressure-Induced Mechanotransduction

17

30. Hsieh HJ, Li NQ, Frangos JA. Shear-induced platelet-derived growth factor gene expression in human endothelial cells is mediated by protein kinase-C. J Cell Physiol 1992; 150:552–558. 31. Hsieh H-J, Li N-Q, Frangos JA. Pulsatile and steady ﬂow induces c-fos expression in human endothelial cells. J Cell Physiol 1993; 154:143–151. 32. Gudi SR, Clark CB, Frangos JA. Fluid ﬂow rapidly activates G proteins in human endothelial cells. Involvement of G proteins in mechanochemical signal transduction. Circ Res 1996; 79:834–839. 33. Bao X, Lu C, Frangos JA. Mechanism of temporal gradients in shear-induced ERK1/2 activation and proliferation in endothelial cells. Am J Physiol Heart Circ Physiol 2001; 281:H22–H29. 34. Hansen CA, Schroering AG, Carey DJ, Robishaw JD. Localization of a heterotrimeric G protein c subunit to focal adhesions and associated stress ﬁbers. J Cell Biol 1994; 126:811–829. 35. Arcangeli A, Becchetti A, Mannini G, et al. Integrin-mediated neurite outgrowth in neuroblastoma cells depend on activation of potassium channels. J Cell Biol 1993; 122:1131–1143. 36. Hu Y, Bock G, Wick G, Xu Q. Activation of PDGF receptor alpha in vascular smooth muscle cells by mechanical stress. Faseb J 1998; 12:1135–1142. 37. Iwasaki H, Eguchi S, Ueno H, Marumo F, Hirata Y. Mechanical stretch stimulates growth of vascular smooth muscle cells via epidermal growth factor receptor. Am J Physiol Heart Circ Physiol 2000; 278:H521–H529. 38. Chen KD, Li YS, Kim M, et al. Mechanotransduction in response to shear stress. Roles of receptor tyrosine kinases, integrins, and Shc. J Biol Chem 1999; 274:18393–18400. 39. Hensley K, Robinson KA, Gabbita SP, Salsman S, Floyd RA. Reactive oxygen species, cell signaling, and cell injury. Free Radic Biol Med 2000; 28: 1456–1462. 40. Cheng JJ, Wung BS, Chao YJ, Wang DL. Cyclic strain-induced reactive oxygen species involved in ICAM-1 gene induction in endothelial cells. Hypertension 1998; 31:125–130. 41. Hishikawa K, Oemar BS, Yang Z, Luscher TF. Pulsatile stretch stimulates superoxide production and activates nuclear factor-kappa B in human coronary smooth muscle. Circ Res 1997; 81:797–803. 42. Lehoux S, Esposito B, Merval R, Loufrani L, Tedgui A. Pulsatile stretchinduced extracellular signal-regulated kinase 1/2 activation in organ culture of rabbit aorta involves reactive oxygen species. Arterioscler Thromb Vasc Biol 2000; 20:2366–2372. 43. Huot J, Houle F, Marceau F, Landry J. Oxidative stress-induced actin reorganization mediated by the p38 mitogen-activated protein kinase/heat shock protein 27 pathway in vascular endothelial cells. Circ Res 1997; 80:383–392. 44. Wang DS, Profﬁt D, Tsao PS. Mechanotransduction of endothelial oxidative stress induced by cyclic strain. Endothelium 2001; 8:283–291. 45. De Keulenaer GW, Chappell DC, Ishizaka N, Nerem RM, Alexander RW, Griendling KK. Oscillatory and steady laminar shear stress differentially affect human endothelial redox state: role of a superoxide-producing NADH oxidase. Circ Res 1998; 82:1094–1101.

18

Lehoux and Tedgui

46. Dimmeler S, Fleming I, Fisslthaler B, Hermann C, Busse R, Zeiher AM. Activation of nitric oxide synthase in endothelial cells by Akt-dependent phosphorylation. Nature 1999; 399:601–605. 47. Dimmeler S, Assmus B, Hermann C, Haendeler J, Zeiher AM. Fluid shear stress stimulates phosphorylation of Akt in human endothelial cells: involvement in suppression of apoptosis. Circ Res 1998; 83:334–341. 48. Govers R, Rabelink TJ. Cellular regulation of endothelial nitric oxide synthase. Am J Physiol Renal Physiol 2001; 280:F193–F206. 49. Chen AH, Gortler DS, Kilaru S, Araim O, Frangos SG, Sumpio BE. Cyclic strain activates the pro-survival Akt protein kinase in bovine aortic smooth muscle cells. Surgery 2001; 130:378–381. 50. Tanabe Y, Saito M, Ueno A, Nakamura M, Takeishi K, Nakayama K. Mechanical stretch augments PDGF receptor beta expression and protein tyrosine phosphorylation in pulmonary artery tissue and smooth muscle cells. Mol Cell Biochem 2000; 215:103–113. 51. Ishida T, Peterson TE, Kovach NL, Berk BC. MAP kinase activation by ﬂow in endothelial cells – Role of beta 1 integrins and tyrosine kinases. Circ Res 1996; 79:310–316. 52. Solowska J, Guan JL, Arcantonio EE, Trevithick JE, Buck CA, Hynes RO. Expression of normal and mutant avian integrin subunits in rodent cells. J Cell Biol 1989; 109:853–861. 53. Gilmore AP, Romer LH. Inhibition of focal adhesion kinase (FAK) signaling in focal adhesions decreases cell motility and proliferation. Mol Biol Cell 1996; 7:1209–1224. 54. Schlaepfer DD, Hunter T. Focal adhesion kinase overexpression enhances Rasdependent integrin signaling to ERK2/mitogen-activated protein kinase through interactions with and activation of c-Src. J Biol Chem 1997; 272:13189–13195. 55. Carbajal JM, Schaeffer RC Jr. RhoA inactivation enhances endothelial barrier function. Am J Physiol 1999; 277:C955–C964. 56. Bobak D, Moorman J, Guanzon A, Gilmer L, Hahn C. Inactivation of the small GTPase Rho disrupts cellular attachment and induces adhesion-dependent and adhesion-independent apoptosis. Oncogene 1997; 15:2179–2189. 57. Lehoux S, Tedgui A. Signal transduction of mechanical stresses in the vascular wall. Hypertension 1998; 32:338–345. 58. Reusch HP, Chan G, Ives HE, Nemenoff RA. Activation of JNK/SAPK and ERK by mechanical strain in vascular smooth muscle cells depends on extracellular matrix composition. Biochem Biophys Res Commun 1997; 237:239–244. 59. Xu QB, Liu YS, Gorospe M, Udelsman R, Holbrook NJ. Acute hypertension activates mitogen-activated protein kinases arterial wall. J Clin Invest 1996; 97:508–514. 60. Birukov KG, Lehoux S, Birukova AA, Merval R, Tkachuk VA, Tedgui A. Increased pressure induces sustained PKC-independent herbimycin Asensitive activation of extracellular signal-regulated kinase 1/2 in the rabbit aorta in organ culture. Circ Res 1997; 81:895–903. 61. Surapisitchat J, Hoefen RJ, Pi X, Yoshizumi M, Yan C, Berk BC. Fluid shear stress inhibits TNF-alpha activation of JNK but not ERK1/2 or p38 in human

Shear and Pressure-Induced Mechanotransduction

62.

63.

64. 65.

66. 67.

68. 69. 70.

71.

19

umbilical vein endothelial cells: inhibitory crosstalk among MAPK family members. Proc Natl Acad Sci USA 2001; 98:6476–6481. Takahashi M, Berk BC. Mitogen-activated protein kinase (ERK1/2) activation by shear stress and adhesion in endothelial cells – essential role for a herbimycin-sensitive kinase. J Clin Invest 1996; 98:2623–2631. Barberis L, Wary KK, Fiucci G, et al. Distinct roles of the adaptor protein Shc and focal adhesion kinase in integrin signaling to ERK. J Biol Chem 2000; 275:36532–36540. Crespo PN, Xu N, Simons WF, Gutkind JS. Ras-dependent activation of MAP kinase pathway mediated by G protein bc subunits. Nature 1994; 369:418–420. Abe J, Okuda M, Huang Q, Yoshizumi M, Berk BC. Reactive oxygen species activate p90 ribosomal S6 kinase via Fyn and Ras. J Biol Chem 2000; 275: 1739–1748. Baas AS, Berk BC. Differential activation of mitogen-activated protein kinases by H2O2 and O 2 in vascular smooth muscle cells. Circ Res 1995; 77:29–36. Numaguchi K, Eguchi S, Yamakawa T, Motley ED, Inagami T. Mechanotransduction of rat aortic vascular smooth muscle cells requires RhoA and intact actin ﬁlaments. Circ Res 1999; 85:5–11. Li S, Chen BP, Azuma N, et al. Distinct roles for the small GTPases Cdc42 and Rho in endothelial responses to shear stress. J Clin Invest 1999; 103:1141–1150. Davies PF. Flow-mediated endothelial mechanotransduction. Physiol Rev 1995; 75:519–560. Whitmarsh AJ, Shore P, Sharrocks AD, Davis RJ. Integration of MAP kinase signal transduction pathways at the serum response element. Science 1995; 269:403–407. Proud CG. Turned on by insulin. Nature 1994; 371:747–748.

This page intentionally left blank

2 Pulmonary Micromechanics of Injured Lungs

ROLF D. HUBMAYR Division of Pulmonary and Critical Care Medicine, Mayo Clinic Rochester, Minnesota, U.S.A.

I. Introduction Many controversies about mechanical ventilation–associated injury mechanisms can be traced to uncertainties about the small-scale stress and strain distributions in healthy and diseased lungs. It seems, therefore, prudent to begin by discussing the physical determinants of regional lung volume and ventilation in healthy lungs and only then consider the effects of injury on regional mechanics within this framework. I consider it important to detail certain principles in solid mechanics that are applicable to lung biology, not because the principles are new, but because they are fundamental for dealing with the topic at hand. The reader who wishes to go beyond my brief description of these principles is referred to speciﬁc chapters in the Handbook of Physiology (1–3). Finally, I note that some of my arguments about the distribution of edema in injured lungs, which of course has bearing on alveolar mechanics, have been summarized in a previously published opinion piece (4).

21

22

Hubmayr II. Determinants of Regional Pressure and Volume in Health and Disease

The lung is a tissue network that offers relatively little resistance to shape change. Therefore, when it became known in the middle of the 20th century that there are vertical gradients in regional lung expansion and pleural pressure, it made perfect sense to liken the lungs to a liquid. That analogy generated a number of testable hypotheses: (i) the vertical gradients in pressure and volume are determined by the average density of the lung; (ii) changes in body posture have no effect on the magnitude of the gravitational volume and pressure gradients. Physiologists soon realized that measurements of pleural pressure and regional lung volume in experimental animals were not consistent with these predictions and, therefore, they considered alternative mechanisms. Speciﬁcally, observed vertical pressure and volume gradients failed to scale with lung density, and most importantly, the gradients changed substantially with body posture (5–8). Because the critical care community at that time had not yet appreciated the relevance of regional lung function for the ventilator management of critically ill patients, clinicians did not pay much attention to what seemed to be an esoteric debate. Because the lungs did not behave like a liquid, physiologists began to approach questions about the in situ topographical distribution of pressure and volume as a shape-matching problem between two gravitationally deformed elastic solids: the lungs and the chest wall (2). Figure 1 helps to appreciate this concept. The stress and strain distributions of a cone-shaped sponge that is forced to completely ﬁll a rigid cylinder are shown. Gravity is only relevant insofar as it is a determinant of the shape of the cone before it is forced to assume the cylindrical shape. A useful way to think about the lung/thorax shape mismatch is to imagine what shapes the lungs and chest wall would assume before they are forced to conform to each other. The considerable displacement of the diaphragm–abdomen with posture certainly underscores the importance of gravity on the shape of the chest wall. If, by chance gravity deformed the lungs in exactly the same way, then the lungs would be uniformly expanded in the chest and exposed to exactly the same surface pressure (pleural pressure) everywhere. Because that is not the case (at least in large animals and humans in the supine or upright posture), the topographical distribution of pressure and volume must reﬂect the size of the shape mismatch and the resistance of either structure to a shape change. In material science, this resistance is referred to as shear modulus. It is distinct from other measures of elastic properties such as compliance or bulk modulus, but related to it. In the case of liquids, this resistance is zero. In the case of solids, it may be considerable. Case in point: the hydrostatic pressure gradient in a water glass does not care about the shape of the glass, yet it is difﬁcult to ﬁt a square peg into a round hole.

Pulmonary Micromechanics of Injured Lungs

23

Figure 1 The diagram shows a very simple, but nevertheless instructive shapematching problem (the ﬁtting of an elastic cone into a rigid cylinder). As long as the elastic solid (the cone) resists a shape change (behaves like a solid rather than a liquid), its stress distribution will be shape dependent. Note that the vertical orientation of the stress and strain gradients need not imply a gravitational mechanism. For example, the experiment shown here might well have been conducted in a gravityfree environment. Source: From Ref. 4.

Once these principles were understood, it became clear that the weight of the lung could only be one (possibly minor) determinant of a lung/thorax shape mismatch (9,10). Lung weight will determine by how much the lungs will slump (deform) when they are taken out of the chest and by how much the dependent alveoli are compressed when the lungs are supported on a hard surface. However, until one knows by how much this ‘‘compressed’’ lung must deform to ﬁt into the gravitationally deformed thorax, it is impossible to predict the in situ stress (pressure) and strain (volume) distributions. These principles hold true in health as well as in disease. What makes disease more difﬁcult to deal with is the greatly increased small-scale heterogeneity in mechanical properties (such as local shear moduli), which contribute greatly to local stress distributions (see discussion on interdependence). Experiments on normal animals conducted in the 1970s and 1980s established that the lung weight accounts for no more than 20% of the vertical gradient in pleural pressure and alveolar volume (6,7,11). In other words, under normal conditions, the lung weight is only a minor determinant of the topographical distribution of parenchymal stress and strain. In contrast, the weights of the abdomen and heart greatly inﬂuence the gradients in intrathoracic pressure and volume (7,11,12). Both heart and

24

Hubmayr

diaphragm/abdomen are lung boundary structures and as such their gravitational deformations deﬁne the shape the lungs must deform to. Proof of concept was provided by Bar-Yishay et al., who ﬁlled the heart of upright canine cadavers with mercury and showed a dramatic effect on pleural pressure gradients (12). One of the ﬁrst observations, which raised concerns about the weight of the lung hypothesis, was the lack of a vertical pressure or volume gradient in prone animals. One attractive explanation is the difference in heart position and support between the two postures (13). As shown in Figure 2, in the prone posture, the heart rests on the sternum, whereas in the supine posture, the weight of the heart is balanced by pleural pressure in the mid-chest. During a change from the prone to the supine posture, the heart ‘‘sinks’’ from the sternum toward the spine. As it does so, the lungs enter the substernal space vacated by the heart. The resulting lung deformation and the associated local stress generate a ‘‘suction pressure’’ (negative pleural pressure) that prevents the heart from coming to rest entirely on the spine. In other words, the weight of the heart is an important source of the vertical pleural pressure gradient in supine animals (and

Figure 2 Schematic (upper panel) and CT images of volunteers (lower panel) illustrating the effects of posture on the position of the heart relative to lungs and thorax. Abbreviation: CT, computed tomography. Source: From Ref. 13.

Pulmonary Micromechanics of Injured Lungs

25

presumably humans), and it is thought to contribute to the ‘‘proning’’ related recruitment of the dorsal units of injured lungs (13–15). The ‘‘weight of the lung hypothesis’’ reemerged after computer tomography (CT) images of patients with acute lung injury (ALI) and the acute respiratory distress syndrome (ARDS) showed preferential consolidation of the dependent diaphragm near lung regions (16–21). Dependent atelectases were attributed to compression of the dorsal lung by the increased weight of the edematous superimposed tissue. While it is certainly possible that ﬂuid that invariably accumulates in injured lungs accentuates a mechanism that normally is insigniﬁcant, several arguments were brought forth that challenged the superimposed pressure idea (4). Measurements of tissue dimensions (as opposed to regional air content) in oleic acid–injured dogs failed to demonstrate vertical gradients in regional lung expansion (22). In these studies the regional lung volume was deﬁned as the sum of tissue, blood, edema ﬂuid, and air. The most plausible interpretation of the data was that following injury alveolar air was simply replaced by alveolar ﬂuid. As a result the dimensions of alveolar walls and the local stress (transmural pressure) on them need not change appreciably even though the pleural pressure over dependent lung could have increased dramatically. That is because alveolar pressure in dependent ﬂuid or foam-ﬁlled acini no longer equals the pressure at the airway opening. Once one accepts that most forms of lung injury impair the vascular barrier properties, then the images of edematous lungs published by Bachofen and Weibel in the early 1990s serve as a powerful reminder of the small-scale heterogeneity in interfacial tension and hence of mechanical properties (23,24). Because the determinants of the lung parenchymal stress and strain distributions in the intact thorax depend critically on the lungs’ resistance to a shape change, the effects of injury on lung mechanical properties becomes an important variable. It is not my intent to review the considerable literature on the pressure–volume (PV) curve of patients with injured lungs, because Chapter 6 deals with this topic in considerable detail. Moreover, the information obtained from the whole lung PV measurements is insufﬁcient to characterize the apparent shear modulus of the injured lung. What can be concluded is that an injured lung is less deformable than a normal lung. Several candidate mechanisms exist that readily explain the greater shear modulus of injured lungs. These include interfacial tensions associated with airway closure by liquid bridges and foam, solidiﬁcation (gel formation) of alveolar exudate, increased surface tension and the consequently increased prestress of the axial elastin and collagen ﬁber network (see below under the sections ‘‘Micromechanics of the Normal Lung’’ and ‘‘Alveolar Micromechanics in Injury States’’), interstitial edema and matrix remodeling, and ﬁnally scar formation and ﬁbrosis. In light of the injury effect on the deformation resistance of the lung parenchyma, it cannot be a priori assumed that the greater lung weight is responsible for the consolidation of dependent lung.

26

Hubmayr III. Micromechanics of the Normal Lung

For more than 50 years, it has been appreciated that the topographical distributions of lung parenchymal stress and strain are nonuniform and, as just outlined, the biophysical determinants of this nonuniformity are generally understood. However, with increasing precision in the methods for measuring regional lung function, it is now apparent that there is considerable small-scale heterogeneity in lung parenchymal strain, which cannot be explained by any gravitational mechanism (11,25). The lung parenchyma is a tissue network that is distorted by surface tension (Fig. 3) (26,27). Embedded in this network are airways and blood vessels, whose resistance to deformation exceeds that of the parenchyma by varying degrees. In the late 1970s, Bachofen, Weibel, and coworkers

Figure 3 Scanning electron micrograph of an alveolar duct and adjacent alveoli and the schematic of pulmonary micromechanics demonstrate the effects of surface tension on acinar stress/strain distributions. Source: From Ref. 26.

Pulmonary Micromechanics of Injured Lungs

27

reported on the morphology of rabbit lungs at different lung volumes and for different values of surface tension (28–30). These studies identiﬁed three components of the tissue structure of the lung. One is the peripheral tissue system that includes the pleural membrane and membranes that penetrate the lung and connect to sheaths that surround the airways. This is a selfcontained system that is extended as lung volume increases, but is unaffected by surface tension. It provides the only contribution to recoil in the saline-ﬁlled lung, and its contribution to recoil in the air-ﬁlled lung is the same as to the recoil of the saline-ﬁlled lung. The second component of the tissue structure is the axial system, namely a helical network of elastin and collagen ﬁbers that extend from the terminal conducting airways to form alveolar ducts and the line elements at the alveolar openings (31–34). This second tissue component is tensed by surface tension, i.e., surface tension generates prestress in the axial ﬁber system. The third component of the tissue structure is the ﬁne ﬁbrils of connective tissue that thread through the alveolar walls. This component is assumed to be unstressed except at high lung volumes. Guided by Weibel’s description of the architecture of the lung, the qualitative appearance of the micrographs, and the quantitative data, Wilson and Bachofen (35) constructed a model for the mechanical properties of the acinus. The model alveolar duct consisted of intersecting helical elastic line elements that deﬁned the lumen of the duct and formed the free edges of alveolar walls that extended outward from the helical line elements. The alveolar walls were assumed to carry no tissue stress and to serve only as platforms for surface tension at the air–liquid interface. Tension and length of the line elements were determined by a balance between the hoop stress in the line and surface tension on the alveolar walls. As a consequence, the dimensions of alveolar ducts increase with increasing surface tension, at the expense of alveolar surface area (Fig. 3). Because alveolar walls are the planes along which surface tension acts, any increase in surface tension will also promote tissue buckling at alveolar corners. It should be acknowledged that not all investigators subscribe to the Wilson–Bachofen views of acinar micromechanics. Some view alveoli as a scaffold that simply supports surfactant foam (36), while others have entertained the notion that the alveolar liquid lining could be discontinuous so that surfactant interacts directly with plasma membranes and with ‘‘local puddles’’ of an aqueous subphase (37). Finally, some think that alveoli exist in only one of two states, i.e., expanded and recruited or collapsed and derecruited (38). Indirect support for this hypothesis arises from movies of subpleural lung units of mechanically ventilated animals (39,40). These movies failed to reveal appreciable changes in the alveolar projection images during breathing. Considering the technical challenges and assumptions upon which the different schools of thought base their arguments, the mechanisms by which the lungs change volume remain controversial.

28

Hubmayr

The time-honored method of making detailed morphometric measurements of perfusion-ﬁxed lung tissues has been plagued by uncertainty about artifacts from tissue desiccation and preservation (41). Nevertheless, the derived data clearly form the basis of mainstream thinking (42). Intravital microscopy on the other hand has limited three-dimensional (3-D) resolution and is restricted to alveoli that are anchored to the pleural membrane. Because the pleural membrane area change must scale with the tidal volume to the 2/3 power (43), it is hard to imagine that alveoli, which are anchored to that membrane, would be able to resist expansion in the plane of the membrane. Yet, no such expansion was demonstrated with intravital microscopy (39). This raises the question of an experimental setup that requires that the pleura be brought into apposition to a coverslip by gentle suction and as a result constrains local deformation. The data on lung morphology and the model of acinar micromechanics provide a number of insights and predictions that are relevant for understanding mechanical ventilation–related injury mechanisms. Over much of the lungs’ volume range, the parenchyma simply unfolds as opposed to getting stretched (27,44,45). In other words, the parenchyma and, speciﬁcally, the alveoli behave more like wrinkled cellophane bags than deﬂated rubber balloons. Consequently, the stress acting on cells and on the tissue matrix of alveolar walls is small and more or less constant up to lung volumes of 70% total lung capacity (TLC). Tschumperlin and Margulies traced the lengths of alveolar basement membranes in electron microscopic images of rat lungs and estimated their area change with transpulmonary pressure and volume. Accordingly, the basement membrane area increased by approximately 35% during an inspiratory capacity maneuver, which corresponds to a linear strain of approximately 15% (45). Importantly, the stress–strain (transpulmonary pressure–basement membrane area) relationship was highly nonlinear and suggested that elastic tissue deformation occurred only at high volumes. As will be discussed below, these insights have a bearing on the interpretation of PV curves of injured lungs. One of the ‘‘hallmarks’’ of injury is a rightward shift of the lung PV curve. Because surface tension, as one important determinant of lung recoil, has a very nonlinear effect on alveolar wall stress and strain, a rightward shift of the lungs’ PV curve due to surfactant inactivation need not have any bearing on the probability of deformation injury from tissue failure. Consider a sphere of tissue that is coated with an air–liquid interface. A change in surface tension will alter the pressure required to preserve the volume of the sphere, but this will have no effect whatsoever on the tissue stress itself. Indeed, the Wilson–Bachofen model argues that increases in surface tension unload alveolar walls, while shifting the acinar PV curve to the right (26,35). Until one integrates topographical heterogeneity in surfactant function, in airspace edema, and in local impedance into models of alveolar mechanics, it should be appreciated that a rightward shift of the lung PV curve by itself does not inform

Pulmonary Micromechanics of Injured Lungs

29

about tissue stress and about the probability of tissue stress failure or ‘‘biotrauma’’ from mechanical ventilation.

IV. Alveolar Micromechanics in Injury States Injured lungs possess two attributes that explain why they are at an increased risk for deformation injury. The ﬁrst attribute is that the number of airspaces capable of expanding during inspiration is decreased, an attribute referred to by Gattinoni et al. as ‘‘baby lung’’ (18). Unless tidal volume is reduced in proportion, units that do expand during breathing are exposed to a greater deforming stress. This explains the increased risk of injury from regional overexpansion. The second attribute is that the local impedance to lung expansion is heterogeneous because of the heterogenous distributions of the liquid and the surface tension in distal airspaces. This heterogeneity in lung impedances results in shear stress being generated between neighboring, interdependent units that operate at different volumes (46). The stability of a ﬂuid layer on the wall of an airway has been analyzed (47). The results show that if enough ﬂuid to form a liquid bridge across the airway is available, the bridge will form. However, estimates of the magnitude of the pressure difference that could be supported by foam or liquid bridges in the airways are not available. Figure 4 shows subpleural alveoli of two isolated perfused rat lungs that had been imaged with laser confocal microscopy (4). The image on the left is a 3-D representation of a normal lung. The image on the right is a single optical slice 30 mm below the pleural surface of an injured lung that had been perfused with a solution containing ﬂuorescein-labeled dextran. Edema ﬂuid appears white, the alveolar walls gray, and trapped air black. Note that the alveolar walls of the edematous lung are wavy, that the alveoli are completely or partially ﬂooded, and that they contain air pockets of different sizes and shapes. The observations are reminiscent of those by Bachofen et al. based on electron micrographs of edematous rabbit lungs (23,24). The presence of different sized air pockets with different radii of curvature implies a nonuniform alveolar gas pressure and/or nonuniform surface tensions. Regional differences in the physicochemical properties of the surfactant as suggested by Bachofen et al. could well be the source of the nonuniform surface tension. Maintenance of a nonuniform alveolar gas pressure raises the possibility that the air pockets are trapped by liquid and foam in conducting airways. It opens the possibility that crackles, which are readily heard in edematous lungs, are generated by the collapse of unstable bubbles as opposed to the explosive expansion of a previously collapsed alveolus. In either case, the image conveys the impression that the mechanical impedance to lung inﬂation is heterogenous both within and between small neighboring structures.

30

Hubmayr

Figure 4 Laser confocal micrograph of subpleural airspaces of a normal (A) and an edematous (B) rat lung. Edema ﬂuid contains ﬂuorescein-labeled dextran and appears white. For discussion of mechanisms, refer to text. Source: From Ref. 4.

The most convincing examples of heterogenous lung expansion in injury states have been provided by the team of Nieman, who recorded the volume expansions of subpleural alveoli during mechanical ventilation in different injury models (48–50). In contrast to the normal lung in which the apical regions of subpleural alveoli appeared uniformly expanded, injury states were associated with greatly nonuniform alveolar expansions. Because it is difﬁcult to distinguish between the tissue and edema ﬂuid by light microscopy, the images convey the cyclic appearance and disappearance of gas bubbles at the apices of subpleural alveoli. While this observation is insufﬁcient for characterizing the mechanisms of alveolar recruitment and derecruitment as tissue opening and collapse, it nevertheless does underscore the tremendous heterogeneity in local mechanics. This has an obvious bearing on interdependence as a risk factor for deformation injury in edematous lungs. As anticipated, the application of positive end-expiratory pressure (PEEP) increased the number of aerated subpleural alveoli and restored alveolar mechanics toward normal (49). Wilson et al. modeled the micromechanics of the edematous alveolus and tested the validity of the model assumptions against measurements of regional lung expansion in oleic acid-injured dogs (51). The model depicted in Figure 5 was adapted from the classic Wilson–Bachofen model, which had successfully described the dependence of lung recoil and surface area on lung volume and surface tension (35). In that model, the line elements

Pulmonary Micromechanics of Injured Lungs

31

Figure 5 Model of edematous alveolus (A) and corresponding prediction of the pressure/volume behavior of the edematous lung (B). For discussion of mechanisms, refer to text. Source: From Ref. 51.

at the alveolar openings were pictured as helices. Alveolar walls extended radially outward from the helices, but the geometry of the alveolus was not speciﬁed. In the case of alveolar edema, Wilson modiﬁed that model by describing the alveolar geometry in detail. This detail was required in order to add ﬂuid to the model. By modeling the side walls of the alveolus as a cone, the geometry of the ﬂuid pool and the air–liquid interface could be described simply. However, this model was not as self-consistent as the earlier model. In particular, it was not possible to match a model alveolus with cylindrical symmetry around a vertical axis to a duct with cylindrical symmetry around a horizontal axis. The edema model retains the crucial elements of the original model, namely, the dimensions of the outer boundary of the duct depend on lung volume alone. Tension and length of the line element at the inner boundary are determined by a balance between the hoop stress and surface tension. Surface area is a function of both the outer and inner dimensions and depends on lung volume as well as surface tension. In the model for edema, the ﬂuid in the lung was assumed to be conﬁned to the interior of the alveolus. This assumption is consistent with the micrographs of edematous lungs presented by Bachofen et al. (23,24). In these micrographs, the alveolar ducts and alveolar mouths are open, and the alveolar walls are separate and distinct. With smaller amounts of edema ﬂuid, the ﬂuid is conﬁned to the interior corners of the alveoli. With larger amounts of ﬂuid, the ﬂuid pools extend to the free edges of the alveolar walls, and the air–liquid interfaces are smoothly curved. In the edema model, at the lowest lung volume, ﬂuid ﬁlls the alveolus, and the tangent to the air–liquid interface is orthogonal to the entrance

32

Hubmayr

ring. As a result, the ring is slack and does not contribute to recoil. As lung volume increases, the ﬂuid retreats into the alveolus, the angle between the tangent to the air–liquid interface and the alveolar wall (phi in Fig. 5) decreases, the entrance ring expands, and pressure rises rapidly because of the rapid increase in tension in the entrance ring. Above the volume at which the air bubble is enclosed in the alveolus (phi ¼ 0), tension in the ring is independent of volume and the PV curve is nearly vertical as it is for lungs that have been rinsed with a liquid with high surface tension (52–54). Thus, the model generates a PV curve with a low compliance at low volumes, a pronounced knee, and high compliance at higher volumes. The shape of the curve is like the shape of the PV curves of edematous lungs (51,55), but the explanation for this shape is quite different from that based on the hypothesis that alveoli or airways are in the collapsed state at low volumes and pop open at a critical pressure.

V. Mechanisms by Which Ventilators Injure Lungs Once the critical care community appreciated that mechanical ventilation could damage the lungs in more ways than ‘‘barotraumas,’’ as deﬁned by Macklin and Macklin some 60 years ago (56), literally hundreds of experimental studies were conducted with the aim of establishing cause and effect relationships between speciﬁc ventilator settings and some biologic responses (57). In aggregate, these studies have established four speciﬁc ventilator-induced lung injury (VILI) mechanisms: (a) regional over distension (58–60) caused by the application of a local stress or pressure that forces cells and tissues to assume shapes and dimensions that they do not assume during unassisted breathing; (b) so-called ‘‘low volume injury’’ (61,62) associated with the repeated recruitment and derecruitment of unstable lung units, which causes the abrasion of the epithelial airspace lining by interfacial forces; (c) the inactivation of surfactant (63,64) triggered by large alveolar surface area oscillations that stress surfactant adsorption and desorption kinetics and are associated with surfactant aggregate conversion; and (d) interdependence mechanisms (46) that raise cell and tissue stress between neighboring structures with differing mechanical properties. A. Overdistension Injury

When airspaces are exposed to high luminal pressures, the resulting deformation of the connective tissue matrix is transmitted to endothelial and epithelial cells that line the capillary basement membrane. The deformed cells may lose contact with the matrix and/or experience yielding (fracture) of their stress-bearing elements, i.e., cytoskeleton and plasma membrane (60).

Pulmonary Micromechanics of Injured Lungs

33

Consequently cells die and are cleared by macrophages or heal, express stress response genes, and initiate a proinﬂammatory immune response. Associated changes in vascular permeability promote alveolar ﬂooding, which alters the molecular organization of the surfactant and inhibits its surface tension–lowering activity (65). When the stress is very large, not only cells but also basement membranes fracture, allowing the passage of red blood cells into the alveolar space (66,67). Together these events explain the cardinal clinical manifestations of ventilator-associated lung injury, namely, edema, increased impedance to lung inﬂation (‘‘stiff lungs’’), reduced alveolar gas content, impaired gas exchange, and alveolar hemorrhage, microvascular thrombi, and inﬂammation (57). The cellular physiology and biomechanics as it pertains to stretch or overexpansion injury is reviewed in Chapter 3. The probability of overdistension injury is clearly related to the magnitude of the inﬂation pressure and the corresponding maximal lung volume (59,68). The inspiratory capacity of adults with healthy lungs is several liters. Therefore, it is highly unlikely that even very large machine-delivered tidal volumes would injure normal lungs by an overdistension mechanism. In contrast, diseased lungs are vulnerable to overdistension because fewer lung units are capable of expanding during inspiration. Gattinoni et al. coined the term ‘‘baby-lung’’ to highlight this determinant of deformation risk (17,18). Because it is difﬁcult to measure thoracic gas volume in critically ill patients, most experts accept a plateau pressure of 30 cmH2O as a surrogate threshold of lung stress that produces overdistension. This threshold was, in fact, chosen by the investigators of the ARDSnet low tidal volume trial, which established the efﬁcacy of lung-protective mechanical ventilation (69). End-inspiratory hold or plateau pressure (Pplat) is the elastic recoil pressure of the relaxed respiratory system at end-inﬂation. In normal individuals, the recoil pressure of the chest wall near TLC approximates 10 cmH2O, so that a Pplat of 30 cmH2O corresponds to a lung stress (i.e., lung elastic recoil pressure) of approximately 20 cmH2O. The stiffness (elastic modulus) of normal lungs increases at distending pressures above 20 cmH2O. Many clinicians refer to the part of the inﬂation PV curve at which stiffness begins to increase as the upper inﬂection point and consider the corresponding deformation as one at which the lungs approach their structural limit. While the reasons for relating injury to lung volume, distending pressure, and the shape of the respiratory PV curve are compelling, the evidence in support of a single numeric threshold remains at best circumstantial. As already pointed out, in injury states, the determinants of lung recoil are exceedingly complex and only peripherally related to alveolar wall stress. Consistent with the effects of disease on the inspiratory capacity, in patients with ALI, the tidal volume that generates a Pplat above the upper

34

Hubmayr

inﬂection point may be indeed quite small (70). Yet most experimental and clinical studies have failed to convincingly uncouple overdistension from other injury mechanisms, so that the debate whether tidal volume injures lungs independent of Pplat remains unsettled (57,69,71,72). Moreover, patients’ chest wall mechanical properties are quite variable, and it should not be assumed that Pplat -10 cmH2O (i.e., normal chest wall recoil) equals lung stress (73). This is particularly true in recumbent patients with ascites, ileus, or morbid obesity (74,75). In many of them plateau pressures between 35 and 40 cmH2O may not only be safe, but actually desired (refer to Chapters 6, 7, and 20). Some experts advocate the placement of esophageal balloon catheters to directly estimate pleural pressure (Ppl), and thus chest wall compliance. However, esophageal manometry is invasive and subject to artifacts, and although measurements of Ppl have been reported in critically ill patients, the technique cannot be considered validated in this population (73,76,77). In injured lungs, the topographical distribution of alveolar and pleural pressure is nonuniform. Therefore, there is no guarantee that the measured pressure reﬂects the weighted average of all pressures acting on the chest wall. Quite to the contrary, there is every reason to think that in supine patients, the end-inspiratory transpulmonary pressure, deﬁned as the difference between airway and esophageal pressure, is severely biased (underestimated) due to the weight of the mediastinum on the lower esophagus (4,72,76). Recent analyses of CT images of patients with ALI suggested that injured lungs may be overdistended at Pplat 30 cmH2O (78). Overdistension was inferred from the frequency distributions of pixel Houndsﬁeld units (HU), which are measures of the local gas to liquid (essentially water) ratios. At TLC, a normal lung contains 10% water (tissue plus blood) per unit gas volume. Provided that pixels are sufﬁciently large relative to the scale of the microstructure, pixels with a tissue to gas ratio <10% (corresponding to HU 900) would have had to originate from structures that contained more air than a normal lung at TLC. By deﬁnition, this means that these units were overexpanded, or had a local blood content smaller than that of a normal lung at TLC. Such an occurrence is to be anticipated in patients with bullous emphysema or barotrauma, but the observation is somewhat surprising in the context of noncardiogenic pulmonary edema from ALI (79). In fact, the investigators report that some regions met CT criteria for overinﬂation even at functional residual capacity and many regions did so at a PEEP level of 15 cmH2O. This amount of PEEP would not be expected to distend normal lung structures to TLC, unless local surface tension was extremely low. The most plausible explanation for the ﬁndings is a remodeling and stress adaptation of lung structures that had been preferentially deformed during mechanical ventilation prior to imaging. In the sample of patients who were studied, lesions were primarily observed in the substernal diaphragm-apposed regions of the lung.

Pulmonary Micromechanics of Injured Lungs

35

B. Low Volume Injury

There is strong experimental evidence that PEEP protects the lungs from mechanical injury (61,68,80–83). The responsible mechanism is thought to be the PEEP-mediated increase in volume and elastic recoil, which prevents the repeated opening and closing of the otherwise unstable lung units. At least early in the progression of ALI, unstable units tend to be located in dorsal paraspinal regions of the lungs (16,17). The reasons why dorsal regions appear most susceptible to atelectasis (lack of aeration) can be traced to normal physiology. In the supine posture, dependent regions of the normal lung empty close to their residual volume, undergo large tidal expansions during inspiration, and receive most of the pulmonary blood ﬂow (1,11,84). Therefore, insults to the capillary integrity of the lungs are likely to promote alveolar ﬂooding, closure of airspaces by liquid plugs, surfactant inactivation, and gas absorption atelectasis in this part of the lungs. The injury mechanism associated with the repeated opening and closure of dependent airspaces may be attributed to interfacial forces that are generated by the respiratory movement of air–liquid interfaces across their epithelial lining. Such forces may be generated during the rupture of a liquid microﬁlm that separates the apposing epithelial lining cells of a truly collapsed airspace that is pried open. A similar injury mechanism may be envisioned during the to and fro movement of a liquid or foam plug in an airway with normal dimensions. Modeling approaches to bubble and liquid ﬂow in tubes, while constrained by simplifying assumptions (e.g., rigid tube of uniform diameter, smooth surface, etc.) are beginning to shed some light on the more quantitative aspects of this problem (62,85–88). In experimental systems, these forces are large enough to damage cell membranes and may be reduced by the application of surfactants. Because Chapter 7 is devoted to the biomechanical basis of low volume injury, it will not be reviewed in detail here. PEEP protects the injured lungs by decreasing the probability of liquid bridge formation in small airways. It does so by increasing the space in which lung water may be distributed. While increasing airspace dimensions, PEEP also promotes the translocation of ﬂuid from alveoli into the interstitial space (89). Occlusion of an airway by a liquid bridge promotes gas absorption and ﬂooding behind the occlusion. The resulting deformation of the effected tissue, often referred to as atelectasis or collapse, is in large part determined by local alveolar–capillary barrier properties. If the vasculature is leaky and alveolar ﬂuid absorption impaired, then liquid will simply replace alveolar gas and the original tissue dimensions will remain preserved. However, if local barrier properties are normal, then the subtended region will decrease in size and the stress at its boundary will increase as predicted by interdependence mechanisms (see section on ‘‘The Importance of Interdependence as an Injury Mechanism’’). In either case, the

36

Hubmayr

diffusion distance will have increased between the inspired gas and the capillary blood, and gas exchange will be impaired. But the consequences for local microstrains and immune mechanisms (‘‘biotrauma’’), which are in part determined by the local concentrations of bioactive molecules, could be profoundly different in the two scenarios. While there are no data that clearly speak to this issue, it is self-evident that PEEP or a change in posture will be less effective in ‘‘opening’’ a ‘‘closed’’ airspace once the liquid exudates transform into gel in the later stages of lung injury syndromes (90). C. Effects of Tidal Volume on Surfactant Kinetics and Function

Surfactant is a lipid–protein mixture, which lowers alveolar surface tension, and thereby stabilizes peripheral lung units (the reader is referred to Chapter 26). Structure, molecular composition, and biophysical properties of pulmonary surfactants are profoundly altered in injured lungs (64,91,92). Moreover, mechanical ventilation is apt to stress surfactant kinetics, and thereby contribute to a depletion of bioactive surfactant (93). Because alveolar surface area increases during inspiration, surfactant lipids are adsorbed to the air–liquid interface and are organized there by surfactant proteins. The decrease in alveolar surface area during expiration results in buckling of the lipid ﬁlm and the squeeze-out of surfactant material into the liquid subphase. With each adsorption/squeeze-out cycle some molecules lose function and need to be replenished by newly secreted lamellar bodies. In a sense, breathing ‘‘consumes’’ bioactive surfactants and the rate of consumption is directly proportional to the amplitude of surface area oscillations, i.e., local tidal volume. The loss of function is reﬂected in an increase in the small aggregate surfactant subfraction (64,65). The small aggregate subfraction contains very little protein and many small unilamellar vesicles, which are incapable of lowering surface tension. These mechanisms of surfactant inactivation have obvious bearing on the debate if one may relax tidal volume restrictions in patients with injured lungs, as long as peak and plateau airway pressures are considered to be in the ‘‘safe’’ range. Safe means that inﬂation pressures are below levels at which normal lungs are expanded close to their TLC, which most proponents of this approach place near 30 cmH2O. Implicit in this argument is the belief that lung tissue overdistension is the prevailing deformation injury mechanism, while hyperventilation-associated surfactant inactivation is not. Two observations raise caution against this belief. First, it is possible to injure the lungs of experimental animals by pharmacologic induction of spontaneous hyperventilation (94). More importantly, a secondary analysis of ARDS clinical trials network data, which had established the efﬁcacy of lung protective mechanical ventilation, demonstrated a survival beneﬁt from tidal volume reductions even in the subgroup of patients with the lowest plateau airway pressures (Chapter 20).

Pulmonary Micromechanics of Injured Lungs

37

D. The Importance of Interdependence as an Injury Mechanism

The interconnectedness between elements of a network such as the lung parenchyma promotes uniform expansion of individual units. When a unit, be it an alveolus, a lobule, or a lung segment, resists expansion because the airway that supplies it with gas is obstructed, the neighboring units exert a large inﬂationary stress on it. This is because stress, by deﬁnition, increases when the sum of forces of the surrounding tissue attachments act over a smaller than expected surface area. Moreover, the tension and strain of individual connective tissue elements that insert into the collapsed segment increase out of proportion to those of the more remote network structures. Strain (fractional length change) of a tissue element that attaches to a lung unit resisting deformation decays exponentially from the point of insertion. The relevance of interdependence on lung biology was ﬁrst recognized by Mead et al., who used a simple balloon network to illustrate its cardinal features (46). They showed that to empty a single balloon that is a part of a balloon network, a suction pressure several-fold greater than the average inﬂation pressure had to be applied. Even though Mead ignored the strain gradients in surrounding network structures, their estimates of local stress turned out to be quite accurate. Many observations in lung biology can be traced to interdependence. They include the redistribution of alveolar ﬂuid from the airspace to interstitium in response to PEEP (89) as well as the accumulation of ﬂuid (in the case of edema), red blood cells (in the case of capillary stress failure), and air (in the case of barotrauma) along bronchovascular bundles (95). Large airways, blood vessels, and interstitial connective tissue resist deformation to a greater extent than the surrounding parenchyma. Consequently, lung expansion creates a local tensile stress at the surface of airways, blood vessels, and interstitium, which drives ﬂuid, blood, or air as the case may be toward that space. VI. Concluding Remarks As the appreciation for speciﬁc injury mechanisms has grown, so has the realization that the lungs’ responses to injurious stress can be quite nuanced and nonuniform with respect to space and time. In fact in 1998, Tremblay and Slutsky coined the term ‘‘biotrauma’’ to precisely underscore this point (96). Because it would be naive to assume that the many distinct manifestations of biotrauma contain identical prognostic or mechanistic information, it would seem prudent to differentiate between speciﬁc pulmonary stress responses. After all, the term ‘‘injury’’ has been used to describe biologic responses as diverse as altered gene or protein expressions, abnormal respiratory mechanics, inefﬁcient gas exchange, impaired vascular barrier properties, and the remodeling of lung structures.

38

Hubmayr

There is a delicate interplay between physical injury mechanisms, the focus of this chapter, and biotrauma, the biologic response to the injurious stress. This is because many of the lungs’ ‘‘biotrauma’’ responses amplify their susceptibility to deformation injury. Markers of ‘‘biotrauma’’ are often used in translational research as surrogate endpoints of treatment effect. However, VILI is a dynamic process, which is hard to capture at a single point in time. Hence, the relative ‘‘value’’ of speciﬁc surrogate treatment targets, be they oxygenation, lung aeration, or cytokine concentrations, is likely to change as a function of time as well. In light of this complexity it would be foolish to ignore the many gaps in our knowledge of pulmonary micromechanics. After all we are still debating how a normal lung deforms during a breath. References 1. Rodarte JR, Fung YC. Distribution of stresses within the lung. In: Fishman AP, ed. Handbook of Physiology. Section 3: Respiratory System. Volume III: Mechanics of Breathing, Part I. Baltimore, Maryland: Williams and Wilkins Co., 1986:233–246. 2. Wilson TA. Solid mechanics. In: Fishman AP, ed. Handbook of Physiology. Section 3: Respiratory System. Volume III: Mechanics of Breathing, Part I. Baltimore, Maryland: Williams and Wilkins Co., 1986:35–40. 3. Agostoni E. Mechanics of the pleural space. In: Geiger SR, ed. Handbook of Physiology. Section 3: The Respiratory System. Bethesda, Maryland: American Physiological Society, 1986:531–559. 4. Hubmayr RD. Perspective on lung injury and recruitment: a skeptical look at the opening and collapse story. Am J Respir Crit Care Med 2002; 165(12): 1647–1653. 5. Agostoni E, D’Angelo E. Comparative features of the transpulmonary pressure. Respir Physiol 1970; 11(1):76–83. 6. Agostoni E, D’Angelo E, Bonanni MV. The effect of the abdomen on the vertical gradient of pleural surface pressure. Respir Physiol 1970; 5(3):332–346. 7. D’Angelo E, Michelini S, Agostoni E. Partition of factor contributing to the vertical gradient of transpulmonary pressure. Respir Physiol 1971; 12(1):90–101. 8. Hubmayr RD, Walters BJ, Chevalier PA, et al. Topographical distribution of regional volume in anesthetized dogs. J Appl Physiol 1983; 54 (4):1048–1056. 9. Vawter DL, Matthews FL, West JB. Effect of shape and size of lung and chest wall on stresses in the lung. J Appl Physiol 1975; 39(1):9–17. 10. Rodarte JR, Hubmayr RD, Stamenovic D, et al. Regional lung strain in dogs during deﬂation from total lung capacity. J Appl Physiol 1985; 58(1):164–172. 11. Hubmayr RD, Rodarte JR, Walters BJ, et al. Regional ventilation during spontaneous breathing and mechanical ventilation in dogs. J Appl Physiol 1987; 63(6):2467–2475. 12. Bar-Yishay E, Hyatt RE, Rodarte JR. Effect of heart weight on distribution of lung surface pressures in vertical dogs. J Appl Physiol 1986; 61(2):712–718.

Pulmonary Micromechanics of Injured Lungs

39

13. Albert RK, Hubmayr RD. The prone position eliminates compression of the lungs by the heart. Am J Respir Crit Care Med 2000; 161(5):1660–1665. 14. Martynowicz MA, Walters BJ, Hubmayr RD. Mechanisms of recruitment in oleic acid-injured lungs. J Appl Physiol 2001; 90(5):1744–1753. 15. Gattinoni L, Pelosi P, Vitale G, et al. Body position changes redistribute lung computed-tomographic density in patients with acute respiratory failure. Anesthesiology 1991; 74(1):15–23. 16. Maunder RJ, Shuman WP, McHugh JW, et al. Preservation of normal lung regions in the adult respiratory distress syndrome. Analysis by computed tomography. JAMA 1986; 255(18):2463–2465. 17. Gattinoni L, Presenti A, Torresin A, et al. Adult respiratory distress syndrome proﬁles by computed tomography. J Thorac Imaging 1986; 1(3):25–30. 18. Gattinoni L, Pesenti A, Avalli L, et al. Pressure-volume curve of total respiratory system in acute respiratory failure. Computed tomographic scan study. Am Rev Respir Dis 1987; 136(3):730–736. 19. Gattinoni L, D’Andrea L, Pelosi P, et al. Regional effects and mechanism of positive end-expiratory pressure in early adult respiratory distress syndrome. JAMA 1993; 269(16):2122–2127. 20. Gattinoni L, Pelosi P, Crotti S, et al. Effects of positive end-expiratory pressure on regional distribution of tidal volume and recruitment in adult respiratory distress syndrome. Am J Respir Crit Care Med 1995; 151(6):1807–1814. 21. Pelosi P, D’Andrea L, Vitale G, et al. Vertical gradient of regional lung inﬂation in adult respiratory distress syndrome. Am J Respir Crit Care Med 1994; 149(1):8–13. 22. Martynowicz MA, Minor TA, Walters BJ, et al. Regional expansion of oleic acid-injured lungs. Am J Respir Crit Care Med 1999; 160(1):250–258. 23. Bachofen H, Schurch S, Michel RP, et al. Experimental hydrostatic pulmonary edema in rabbit lungs. Am Rev Respir Dis 1993; 147(4):989–996. 24. Bachofen H, Schurch S, Weibel ER. Experimental hydrostatic pulmonary edema in rabbit lungs: barrier lesions. Am Rev Respir Dis 1993; 147(4):997–1004. 25. Chang H, Lai-Fook SJ, Domino KB, et al. Spatial distribution of ventilation and perfusion in anesthetized dogs in lateral postures. J Appl Physiol 2002; 92(2):745–762. 26. Schurch S, Bachofen H, Possmayer R. Surface activity in situ, in vivo, and in the captive bubble surfactometer. Comp Biochem Physiol A: Mol Integr Physiol 2001; 129(1):195–207. 27. Bachofen H, Schurch S. Alveolar surface forces and lung architecture. Comp Biochem Physiol A: Mol Integr Physiol 2001; 129(1):183–193. 28. Gil J, Bachofen H, Gehr P, et al. Alveolar volume to surface area relationship in air- and saline-ﬁlled lungs ﬁxed by vascular perfusion. J Appl Physiol 1979; 47(5):990–1001. 29. Bachofen H, Gehr P, Weibel ER. Alterations of mechanical properties and morphology in excised rabbit lungs rinsed with a detergent. J Appl Physiol 1979; 47(5):1002–1010. 30. Weibel ER. Functional morphology of lung parenchyma. In: Macklem PT, Mead J, eds. Handbook of Physiology, Section 3: The Respiratory System, III. Bethesda, Maryland: American Physiological Society, 1986:89–112.

40

Hubmayr

31. Bachofen H, Schurch S, Urbinelli M, et al. Relations among alveolar surface tension, surface area, volume, and recoil pressure. J Appl Physiol 1987; 62(5): 1878–1887. 32. Mercer RR, Laco JM, Crapo JD. Three-dimensional reconstruction of alveoli in the rat lung for pressure-volume relationships. J Appl Physiol 1987; 62(4): 1480–1487. 33. Mercer RR, Crapo JD. Spatial distribution of collagen and elastin ﬁbers in the lungs. J Appl Physiol 1990; 69(2):756–765. 34. Oldmixon EH, Hoppin FG Jr. Distribution of elastin and collagen in canine lung alveolar parenchyma. J Appl Physiol 1989; 67(5):1941–1949. 35. Wilson TA, Bachofen H. A model for mechanical structure of the alveolar duct. J Appl Physiol 1982; 52(4):1064–1070. 36. Scarpelli EM, Mautone AJ, DeFouw DO, et al. Intraalveolar bubbles and bubble ﬁlms. II. Formation in vivo through adulthood [Review]. Anat Rec 1996; 246(2):245–270. 37. Hills BA. An alternative view of the role(s) of surfactant and the alveolar model. J Appl Physiol 1999; 87(5):1567–1583. 38. Smaldone GC, Mitzner W, Itoh H. Role of alveolar recruitment in lung inﬂation: inﬂuence on pressure-volume hysteresis. J Appl Physiol 1983; 55(4):1321–1332. 39. Carney DE, Bredenberg CE, Schiller HJ, et al. The mechanism of lung volume change during mechanical ventilation. Am J Respir Crit Care Med 1999; 160(5 Pt 1):1697–1702. 40. Steinberg J, Schiller HJ, Halter JM, et al. Tidal volume increases do not affect alveolar mechanics in normal lung but cause alveolar overdistension and exacerbate alveolar instability after surfactant deactivation. Crit Care Med 2002; 30(12):2675–2683. 41. Bachofen H, Gerber U, Schurch S. Effects of ﬁxatives on function of pulmonary surfactant. J Appl Physiol 2002; 93(3):911–916. 42. Greaves IA, Hildebrandt J, Hoppin FG. Micromechanics of the lungs. In: Fishman AP, ed. Handbook of Physiology. Section 3: Respiratory System. Volume III: Mechanics of Breathing, Part I. Baltimore, Maryland: Williams and Wilkins Co., 1986:217–231. 43. Hajji MA, Wilson TA, Lai-Fook SJ. Improved measurements of shear modulus and pleural membrane tension of the lung. J Appl Physiol 1979; 47(1):175–181. 44. Oldmixon EH, Hoppin FG Jr. Alveolar septal folding and lung inﬂation history. J Appl Physiol 1991; 71(6):2369–2379. 45. Tschumperlin DJ, Margulies SS. Alveolar epithelial surface area-volume relationship in isolated rat lungs. J Appl Physiol 1999; 86(6):2026–2033. 46. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28(5):596–608. 47. Cassidy KJ, Hapern D, Ressler BG, et al. Surfactant effects in model airway closure experiments. J Appl Physiol 1999; 87(1):415–427. 48. Schiller HJ, McCann UG II, Carney DE, et al. Altered alveolar mechanics in the acutely injured lung. Crit Care Med 2001; 29(5):1049–1055. 49. Halter JM, Steinberg JM, Schiller HJ, et al. Positive end-expiratory pressure after a recruitment maneuver prevents both alveolar collapse and recruitment/ derecruitment. Am J Respir Crit Care Med 2003; 167(12):1620–1626.

Pulmonary Micromechanics of Injured Lungs

41

50. Steinberg JM, Schiller HJ, Halter JM, et al. Alveolar instability causes early ventilator-induced lung injury independent of neutrophils. Am J Respir Crit Care Med 2004; 169(1):57–63. 51. Wilson TA, Anaﬁ RC, Hubmayr RD. Mechanics of edematous lungs. J Appl Physiol 2001; 90(6):2088–2093. 52. Smith JC, Stamenovic D. Surface forces in lungs. I. Alveolar surface tensionlung volume relationships. J Appl Physiol 1986; 60(4):1341–1350. 53. Stamenovic D, Smith JC. Surface forces in lungs. II. Microstructural mechanics and lung stability. J Appl Physiol 1986; 60(4):1351–1357. 54. Stamenovic D, Smith JC. Surface forces in lungs. III. Alveolar surface tension and elastic properties of lung parenchyma. J Appl Physiol 1986; 60(4):1358–1362. 55. Martin-Lefevre L, Ricard JD, Roupie E, et al. Signiﬁcance of the changes in the respiratory system pressure-volume curve during acute lung injury in rats. Am J Respir Crit Care Med 2001; 164(4):627–632. 56. Macklin MT, Macklin CC. Malignant interstitial emphysema of the lungs and mediastinum as an important occult complication in many respiratory diseases and other conditions: an interpretation of the clinical literature in the light of laboratory experiment. Medicine 1944; 23:281–358. 57. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157(1):294–323. 58. Parker JC, Townsley MI, Rippe B, et al. Increased microvascular permeability in dog lungs due to high peak airway pressures. J Appl Physiol 1984; 57(6): 1809–1816. 59. Dreyfuss D, Basset G, Soler P, et al. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 1985; 132(4):880–884. 60. Vlahakis NE, Hubmayr RD. Cellular stress failure in ventilator injured lungs. Am J Respir Crit Care Med 2005 (published ahead of print on February 1, 2005 as doi: 10.1164/rccm.200408-1036 SO). 61. Muscedere JG, Mullen JB, Gan JC, et al. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med 1994; 149(5): 1327–1334. 62. Gaver DP III, Kute SM. A theoretical model study of the inﬂuence of ﬂuid stresses on a cell adhering to a microchannel wall. Biophys J 1998; 75(2): 721–733. 63. Savov J, Silbajoris R, Young SL. Mechanical ventilation of rat lung: effect on surfactant forms. Am J Physiol 1999; 277(2 Pt 1):L320–L326. 64. Ito Y, Veldhuizen RA, Yao LJ, et al. Ventilation strategies affect surfactant aggregate conversion in acute lung injury. Am J Respir Crit Care Med 1997; 155(2):493–499. 65. Veldhuizen RA, Welk B, Harbottle R, et al. Mechanical ventilation of isolated rat lungs changes the structure and biophysical properties of surfactant. J Appl Physiol 2002; 92(3):1169–1175. 66. Fu Z, Costello ML, Tsukimoto K, et al. High lung volumes increases stress failure in pulmonary capillaries. J Appl Physiol 1992; 73(1):123–133. 67. West JB. Thoughts on the pulmonary blood-gas barrier. Am J Lung Cell Mol Physiol 2003; 285(3):L501–L513.

42

Hubmayr

68. Dreyfuss D, Soler P, Basset G, et al. High inﬂation pressure pulmonary edema. Respective effects of high airway pressure, high tidal volume, and positive endexpiratory pressure. Am Rev Respir Dis 1988; 137(5):1159–1164. 69. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342(18): 1301–1308. 70. Roupie E, Dambrosio M, Servillo G, et al. Titration of tidal volume and induced hypercapnia in acute respiratory distress syndrome. Am J Respir Crit Care Med 1995; 152(1):121–128. 71. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338(6):347–354. 72. De Chazal I, Hubmayr RD. Novel aspects of pulmonary mechanics in intensive care. Br J Anaesth 2003; 91(1):81–91. 73. Mergoni M, Martelli A, Volpi A, et al. Impact of positive end-expiratory pressure on chest wall and lung pressure-volume curve in acute respiratory failure. Am J Respir Crit Care Med 1997; 156(3 Pt 1):846–854. 74. Pelosi P, Croci M, Ravagnan I, et al. The effects of body mass on lung volumes, respiratory mechanics, and gas exchange during general anesthesia. Anesth Analg 1998; 87(3):654–660. 75. Rouby JJ, Lu Q, Goldstein I. Selecting the right level of positive end-expiratory pressure in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2002; 165(8):1182–1186. 76. Gillespie DJ, Lai YL, Hyatt RE. Comparison of esophageal and pleural pressures in the anesthetized dog. J Appl Physiol 1973; 35(5):709–713. 77. Lai-Fook SJ, Rodarte JR. Pleural pressure distribution and its relationship to lung volume and interstitial pressure. J Appl Physiol 1991; 70(3):967–978. 78. Nieszkowska A, Lu Q, Vieira S, et al. Incidence and regional distribution of lung overinﬂation during mechanical ventilation with positive end-expiratory pressure. Crit Care Med 2004; 32(7):1496–1503. 79. Suki B, Lutchen KR, Ingenito EP. On the progressive nature of emphysema: roles of proteases, inﬂammation, and mechanical forces. Am J Respir Crit Care Med 2003; 168(5):516–521. 80. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110(5):556–565. 81. Argiras EP, Blakeley CR, Dunnhill MS, et al. High PEEP decreases hyaline membrane formation in surfactant deﬁcient lungs. Br J Anesth 1987; 59(10): 1278–1285. 82. Lachmann B. Open up the lung and keep the lung open. Int Care Med 1992; 18(6):319–321. 83. Falke KJ, Pontoppidan H, Kumar A, et al. Ventilation with end-expiratory pressure in acute lung disease. J Clin Invest 1972; 51(9):2315–2323. 84. Jones AT, Hansell DM, Evans TW. Pulmonary perfusion in supine and prone positions: an electron-beam computed tomography study. J Appl Physiol 2001; 90(4):1342–1348.

Pulmonary Micromechanics of Injured Lungs

43

85. Cassidy KJ, Gavriely N, Grotberg JB. Liquid plug ﬂow in straight and bifurcating tubes. J Biomech Eng 2001; 123(6):580–589. 86. Bilek AM, Dee KC, Gaver DP III. Mechanisms of surface-tension-induced epithelial cell damage in a model of pulmonary airway reopening. J Appl Physiol 2003; 94(2):770–783. 87. Ghadiali SN, Gaver DP III. An investigation of pulmonary surfactant physicochemical behavior under airway reopening conditions. J Appl Physiol 2000; 88(2):493–506. 88. Kay SS, Bilek AM, Dee KC, et al. Pressure gradient, not exposure duration, determines the extent of epithelial cell damage in a model of pulmonary airway reopening. J Appl Physiol 2004; 97(1):269–276. 89. Malo J, Ali J, Wood LD. How does positive end-expiratory pressure reduce intrapulmonary shunt in canine pulmonary edema? J Appl Physiol 1984; 57(4):1002–1010. 90. Nakos G, Tsangaris I, Kostanti E, et al. Effect of the prone position on patients with hydrostatic pulmonary edema compared with patients with acute respiratory distress syndrome and pulmonary ﬁbrosis. Am J Respir Crit Care Med 2000; 161(2 Pt 1):360–368. 91. Spragg RG, Lewis JF. Pathology of the surfactant system of the mature lung: Second San Diego Conference. Am J Respir Crit Care Med 2001; 163(1): 280–282. 92. Malloy JL, Veldhuizen RA, Lewis JF. Effects of ventilation on the surfactant system in sepsis-induced lung injury. J Appl Physiol 2000; 88(2):401–408. 93. Veldhuizen RA, Yao LJ, Lewis JF. An examination of the different variables affecting surfactant aggregate conversion in vitro. Exp Lung Res 1999; 25(2):127–141. 94. Mascheroni D, Kolobow T, Fumagalli R, et al. Acute respiratory failure following pharmacologically induced hyperventilation: an experimental animal study. Int Care Med 1988; 15(1):8–14. 95. Lai-Fook SJ, Kallok MJ. Bronchial-arterial interdependence in isolated dog lung. J Appl Physiol 1982; 52(4):1000–1007. 96. Tremblay LN, Slutsky AS. Ventilator-induced injury: from barotrauma to biotrauma. Proc Assoc Am Physicians 1998; 110(6):482–488.

This page intentionally left blank

3 Response of Cellular Plasma Membrane to Mechanical Stress

ROLF D. HUBMAYR Division of Pulmonary and Critical Care Medicine, Mayo Clinic Rochester, Minnesota, U.S.A.

I. Introduction Since the recognition of ventilator-induced lung injury (VILI) as a clinically relevant entity, the effects of deforming stress on lung structure and function have been examined in literally hundreds of experimental studies (1–4). In many instances, diverse morphologic and biologic sequelae of deforming stress have been referred to under the single term ‘‘lung injury,’’ and a number of scoring systems have been devised to grade its severity (5–8). However, as the appreciation for speciﬁc injury mechanisms has grown, so has the realization that the lungs’ responses to injurious stress can be quite nuanced and nonuniform with respect to space and time. In fact, in 1998, Tremblay and Slutsky coined the term ‘‘biotrauma’’ to precisely underscore this point (9). Because it would be naive to assume that the many distinct manifestations of biotrauma contain identical prognostic or mechanistic information, it would seem prudent to differentiate between speciﬁc pulmonary stress responses. After all, the term ‘‘injury’’ has been used to describe biologic responses ranging from altered gene expressions, protein synthesis and release, abnormal respiratory mechanics, inefﬁcient 45

46

Hubmayr

gas exchange, and impaired vascular barrier properties to the remodeling of lung structures and scar formation. In this chapter, I review how the stress associated with mechanical ventilation may cause wounding, remodeling, and repair of lung cells and discuss how such events may shape the biologic manifestations of VILI. I refer extensively to the material that Dr. Vlahakis and I have recently presented in a comprehensive review (10).

II. The Histology of VILI Dreyfuss et al. were the ﬁrst to carefully examine the microstructure of the blood–gas barrier of ventilator-injured rat lungs (11). In doing so, they extended earlier observations by John et al. on mechanically ventilated rabbit lungs (12). Electron micrographs of rat lungs, taken after ﬁve minutes of injurious ventilation, showed interstitial edema and endothelial lesions. These consisted of plasma-membrane (PM) blebs and loss of cell contact with the tissue matrix, speciﬁcally the capillary basement membrane. More prolonged exposure to injurious stress produced alveolar epithelial pathology manifesting as inter- and intracellular gap formations, with denuded basement membranes and extensive cell destruction (1,11,12). These changes in the lung-cell ultrastructure may be viewed as evidence for deformationrelated cell remodeling and/or yielding of the cells’ stress-bearing elements. Interestingly, type II alveolar epithelial cells appeared relatively uninjured. This means that these cells are either capable of withstanding larger stresses than other lung cells or that they experience smaller deformations during breathing, on account of their location in alveolar corners. Starting with the report of Webb and Tierney, a recurrent theme of experimental VILI research has been injury associated with the periodic recruitment and derecruitment of peripheral lung units and the protective effects of positive end-expiratory pressure (PEEP) on this injury mechanism (13–17). Muscedere et al. were the ﬁrst to draw attention to the cytopathologic sequelae of the so-called low volume injury, by identifying epithelial lesions in small airways and alveolar ducts of ex vivo ventilated unperfused rat lungs (18). Subsequent investigations have consistently identiﬁed small airway lesions in intact animals after prolonged mechanical ventilation without PEEP (14,19–21). While these studies have established the epithelial cells of conducting airways as important injury targets, there remain many questions about the nature of the injurious stress on a cellular scale (22–25). In a series of studies motivated by the interest in high-altitude physiology, West (26) examined the consequences of capillary pressure on the blood–gas barrier. The blood–gas barrier of rabbit lungs exposed to high capillary pressures revealed not only transcellular epithelial gaps and

Response of Cellular Plasma Membrane to Mechanical Stress

47

endothelial lesions, as had been observed in ventilator-injured lungs, but also basement membrane breaks (27–29). Frank cell necrosis and large alveolar wounds were not seen. On the basis of these ﬁndings, West suggested that under certain conditions, capillary pressures could exceed the structural limit of the basement membrane, which is the primary stressbearing element of the blood–gas barrier. For the most part, the basement membrane is composed of a network of type IV collagen ﬁbers that can withstand considerable tensile stress (26). Physiologic and pathological conditions in which pulmonary capillary stress failure has been observed or suspected include high altitude pulmonary edema (30), congestive heart failure, mitral stenosis, Goodpasture’s syndrome, which is characterized by an immune-mediated weakening of the collagen IV lattice (31), highintensity exercise in race horses (32), and mechanical ventilation in a patient with acute respiratory distress syndrome (ARDS) (33). Even though elite athletes were reported to experience minor changes in pulmonary vascular barrier properties during strenuous exercise, frank capillary stress failure does not occur in normal humans even during extreme activity at sea level (34). West’s work also underscored important mechanical interactions between lung volume, capillary pressure, and the probability of capillary stress failure (35). Although to date, the structural failure of intra-alveolar capillary basement membranes has not been demonstrated in the absence of capillary hypertension, the presence of pulmonary hemorrhage in rat and canine VILI models, which is readily apparent to the naked eye, raises this possibility (11,13,36). On the other hand, light microscopic images of rat lungs subjected to large tidal ventilation localize perivascular hemorrhage to extra-alveolar blood vessels (Fig. 1). As detailed in Chapter 2, interdependence mechanisms generate large stress concentration between relatively inelastic blood vessels and surrounding parenchyma (37). Not all hydrostatic pulmonary edema results in capillary stress failure (38,39). This is because the transmural pressure at which the intra-alveolar capillaries experience yield stress is quite high. West and Mathieu-Costello estimated that normal basement membranes support pressures up to 40 mmHg without yielding (40). In comparison, capillary pressures associated with ultrastructural changes in adherent endothelial and epithelial cells are considerably lower (39). Pressure-regulated endothelial gaps between and within endothelial cells have been demonstrated not only in the lung but also in the systemic capillaries (41–45). These gaps close rapidly upon removal of the deforming stress, thereby restoring normal vascular permeability (46,47). These observations are in keeping with the observed plasticity of the blood–gas barrier in transient pulmonary venous hypertension (46) and after intermittent hyperinﬂation (47). The plasticity of endothelial cells manifest as intracellular gap formation not only is conﬁned to states with high capillary wall stress, but also reﬂects cellular remodeling responses associated with transendothelial diapedesis of neutrophils (48). For an

48

Hubmayr

Figure 1 H&E-stained rat lung section after ex vivo mechanical ventilation at injurious settings. Note the location of red blood cells surrounding a large extraalveolar vessel. Abbreviation: H&E, hematoxylin and eosin.

in-depth discussion of vascular barrier properties in VILI, the reader is referred to Chapters 4 and 5.

III. Cellular Stress Failure in Ventilator-Injured Lungs Several lines of evidence suggest that stress failure and repair of lung cells play a central role in the pathobiology of VILI. Building on earlier reports of the reversibility of vascular barrier lesions, Gajic et al. devised a method for quantifying lung-cell injury and repair in ventilator-injured rat lungs (49). They perfused ex vivo mechanically ventilated rat lungs with solutions containing the membrane impermeant label propidium iodide (PI). When PI enters a cell through a PM defect, it interchelates with DNA, and emits red ﬂuorescence upon excitation with blue light. Gajic et al. used laser confocal microscopy to identify PI-positive cells in optical sections of subpleural airspaces and demonstrated that the number of subpleural cells with membrane defects increases with tidal volume and duration of mechanical ventilation. Moreover, PI staining–based indices of cell injury correlated reasonably well with conventional physiologic and histologic markers of lung injury.

Response of Cellular Plasma Membrane to Mechanical Stress

49

PI is used extensively in ﬂow cytometry to identify dead or dying cells. However, in the context of physical trauma, the presence of PM lesions large enough to allow PI to enter a cell cannot be taken as proof that the injury had been lethal (10,50,51). This is because most nucleated cells are capable of repairing PM wounds through Ca2þ-dependent vesicular trafﬁcking mechanisms (51). To test if cell repair mechanisms are operative in ventilator-injured lungs, Gajic et al. compared the cellular PI uptake in lungs that had been labeled during injurious mechanical ventilation with that of lungs that had been labeled after removal of the injurious stress. They reasoned that in the presence of PI, injured cells would take up the label regardless of their ultimate fate (be it cell repair and survival or failure to repair and cell death). However, when the label is perfused after removal of an injurious stress, injured cells with repaired PM no longer take up PI. As Gajic et al. had postulated, there were consistently fewer PI-positive cells in lungs that had been labeled after injury, than could be identiﬁed in lungs that were labeled while the insult was inﬂicted. In other words, the number of dead and dying cells was considerably smaller than the number of injured cells, indicating that many injured cells heal their PM wounds. Gajic et al. recognized that a rapid recovery of barrier properties reduces the PI concentration in the alveolar space, possibly biasing the results in favor of cell repair. However, PI dosing and barrier function studies suggested that this bias, if present at all, was too small to account for the differences in labeling with timing of label administration (52). It is conceivable that most injured cells are of endothelial origin, which would make alveolar PI concentrations irrelevant. However, this hypothesis is not consistent with the histology of lungs subjected to mechanical ventilation with large tidal volumes (1,11,12). On the basis of their observations, Gajic et al. concluded that in a normocapnic, hyperoxic environment, approximately 60% of injured rat lung cells heal their PM wounds and presumably survive the insult. While data from reduced systems clearly indicate that cell wounding is a proinﬂammatory stimulus (53), the implications of cell survival as opposed to cell death as determinants of innate immune responses in ventilator-injured lungs are incompletely understood at this time. I return to discuss the gene response to wounding and repair in Section VII. The mechanisms responsible for PM wounding in ventilator-injured lungs have been detailed in Chapters 2 and 7, and will be mentioned here only brieﬂy. Overdistention of well-aerated lung units forces the resident cells to assume volumes and shapes that are not compatible with matrix adherence or the maintenance of sublytic PM tensions (10,54). Moreover, liquid in small airways is responsible for the abrasion of the apices of small airway lining cells by interfacial forces (22,24). Both injury mechanisms are ampliﬁed by interdependence mechanisms and surfactant dysfunction. Doerr et al. examined the effects of hypercapnic acidosis on the extent of cell injury and on the probability of cell repair in the isolated perfused rat VILI preparation (52). In recent years, a growing number of

50

Hubmayr

studies have detailed the interactive effects of pH and pCO2 on vascular barrier properties, lung mechanics, gas exchange, and innate immune responses to diverse insults (55–61). The lessons from these studies are complex and, on the surface, contradictory. Seemingly minor variations in experimental design can result in outcomes that change from desirable to undesirable (62). There is a general agreement that carbon dioxide shapes the interactions between the many unstable reactive oxygen and nitrogen species, and that it does so in a dose-, substrate-, environment-, and organ-dependent manner (63,64). Consequently, the biologic responses to hypercapnic acidosis are so nuanced that to classify them as desired, as opposed to undesired, could be misleading. Doerr et al. conﬁrmed earlier observations that CO2 has barrier-protective effects, and that it lessens the accumulation of lung water in injury states (55). At the same time, Doerr et al. observed that mechanical ventilation with CO2 supplementation was associated with an increased number of dead and dying lung cells, and concluded that hypercapnic acidosis interferes with normal PM remodeling and repair (52). This hypothesis was conﬁrmed in a cell injury model in tissue culture, but the responsible molecular mechanisms have not been delineated (see Section VI).

IV. Determinants of PM Tension Normally, the PM carries tensile stress that is at least one order of magnitude lower than that borne by ﬁlamentous actin (65–68). The tension at which the PM fractures is estimated to range between 1 and 25 mN/m, corresponding to membrane strains of only 1% to 3% (69,70). Lytic tensions vary with the composition and organization of the lipid bilayer as well as with the time frame for breakage (68,71). At least in model membranes, lytic tensions are loading-rate dependent, implying a kinetic process that begins with nucleation of a molecular-scale defect, which either resolves spontaneously or grows to become an unsustainable hole. The tension in the plane of the PM has been inferred from retraction force measurements of PM tethers. Sheetz and coworkers pioneered the use of optical tweezers to measure PM mechanics (72,73) and have thereby laid the foundation for the current views on the biophysical determinants of PM remodeling (66). To this end, ligand-coated particles are trapped with laser light and brought into apposition with the cell. After a few seconds, during which ligands bind to corresponding PM proteins but before they associate with the cytoskeleton, the particle is pulled away from the cell, and the PM lipid bilayer lifts off the subcortical cytoskeleton and forms a tether. As the cell seeks to retrieve the tether, the particle is displaced from the center of the optical trap in an inverse proportion to the trap’s stiffness. The nanoscale displacement of the particle (usually a

Response of Cellular Plasma Membrane to Mechanical Stress

51

gamma immune globulin (IGG) or Concavalin A–coated silicone bead) can be measured and the force calculated. Figure 2 shows a primary rat type II alveolar epithelial cell with a lipid tether and a corresponding length/force tracing. The tracing has three informative parts, namely, an initial step change associated with the formation of the tether, a plateau phase during which the retractive force remains more or less constant, and a terminal phase during which the force rises until the trap can no longer hold the particle. Once that happens, the cell rapidly retrieves the tether on account of molecular lipid–protein interactions at the tether base. Experimental observations and theoretical models have established that tether force is determined by adhesive molecular interactions between the polar lipid head groups and cytoskeletal proteins, by the in-plane tension of the PM, by osmotic pressure gradients between the tether and the cell, and to a lesser extent by tether radius and bending stiffness of the lipid bilayer (66). The in-plane membrane tension and the adhesion energy between PM and the subcortical cytoskeleton are tightly regulated. A key second messenger in the regulation of adhesion energy is the PM phosphatidylinositol 4,5-bisphosphate (PIP2), which binds tightly to actin regulatory proteins such as proﬁlin, gelsolin, and coﬁlin and mediates actin cross-linking and focal adhesion contact assembly (74). To the extent to which tether force is a readout of adhesion energy, interventions that decrease PM PIP2, such as phospholipase Cd activation, have been shown to lower tether force and promote endocytosis (75). It is intriguing to consider that deforming stress, which in some systems has been shown to activate phospholipase-dependent signaling pathways (76), could also promote endocytosis by this mechanism.

Figure 2 The insert shows a primary rat type II alveolar epithelial cell with a lipid tether at three o’clock. The typical length tension tracing shows an initial force plateau of 5 pN and terminal exponential force increase. For detailed discussion, see text.

52

Hubmayr

In the cells examined to date, the force required to form a tether is typically between 5 and 15 pN (66). It is substantially lower in blebbed cells (77), suggesting that lateral tension in the plane of the membrane is small compared to the adhesive lipid–protein interactions. The observation that the tether retractive force is constant over a considerable length range (plateau phase in Fig. 2) indicates that some cells have a large reservoir of lipid, which may be recruited into the tether as it is being pulled. Cells are capable of supplying tethers with excess lipid by unfolding PM rufﬂes and/or by trafﬁcking intracellular vesicular organelles to the cell surface, where they can dock and fuse with the PM (78,79). As will be detailed below, there is ample evidence that deforming stress triggers a vesicular trafﬁcking response by an energy-, temperature-, and cytoskeleton-dependent mechanism (50,80). As observed by Raucher and Sheetz, it follows that the maximum tether length in ﬁbroblasts increases with each subsequent tether pull and that tether length is inversely proportional to the rate of tether elongation (81). These observations are in keeping with mechanistic interactions between PM tension and the control of cell volume and surface area (82). Morris considered the evolutionary question of how apparent regulatory feedback loops for cell volume and surface area came into being (83). She concludes that size regulation in the earliest protocells would have been governed by liposome physics, and argues that monitoring and regulation of lipid bilayer tension ultimately determines a cell-size set point. According to that theory, changes in bilayer tension could have altered membrane conductivity for osmolytes and consequently affected volume. Modern cells have evolved more elaborate control mechanisms linked to proteinregulated expenditures of energy and ﬂuxes of materials. Cells sense a multitude of ﬂux rates such as the rates of endocytosis, exocytosis, protein channel or pump activities, and the polymerization rates of cytoskeletal networks. Volume, surface area, and shape are simply consequences of the weighted distributions of the respective rate constants, but are not the sensed quantities themselves. PM tension is central to many of these control loops.

V. Cell Deformation–Associated PM Remodeling Deformation of the connective tissue matrix during breathing imposes a stress on adherent lung cells. Many studies have examined associated mechanotransduction events by stretching lung cells over a large range of rates and amplitudes. Vlahakis et al. were motivated by VILI, and speciﬁcally examined the role of deformation-induced lipid trafﬁcking (DILT) in the prevention of PM stress failure in alveolar epithelial cells (50). They observed that tonic stretch triggers a net exocytic lipid trafﬁcking response that is temperature and energy dependent and requires an intact cytoskeleton. Recent observations have conﬁrmed the importance of DILT over PM

Response of Cellular Plasma Membrane to Mechanical Stress

53

unfolding in the surface area regulation of tonically stretched type II rat alveolar epithelial cells (84). Vlahakis et al. also showed that inhibition of DILT is detrimental for the maintenance of membrane integrity when the cells are stretched in culture. The cell’s susceptibility to stress failure was dependent on DILT and not on the cytoskeletal stiffness as measured with magnetic twisting cytometry (85). This suggests that a cell’s ability to remodel and thereby unload stress-bearing elements is critical for the maintenance of its structural integrity. Consistent with this hypothesis, the probability of stretch-related cell wounding was found to be exquisitely strain-rate dependent. Stretch-induced lipid trafﬁcking is not a purely exocytic event (86,87). Figure 3 shows a representative example of the stretch-induced endocytosis of the ﬂuorescent sphingolipid analogue, Bodipy sphingomyelin (SM), in a human alveolar epithelium–derived cell (A549). SM is essentially restricted to the outer layer of the PM. Therefore, barring cell wounding or a yet to be described stretch effect on membrane thermodynamics or lipid transport proteins (the so-called ‘‘ﬂippases’’), SM can only be internalized by vesicular transport (endocytosis), a process that is shut down at 4 C. Consistent with this reasoning, there is no Bodipy SM in vesicular organelles (ﬂuorescent

Figure 3 Confocal images of Bodipy-SM labeled A549 cells. The left-hand image shows the label conﬁned to the PM. Warming the specimen to 37 C is associated with the internalization of label into punctuate structures (endocytosis). Endocytosis is increased in stretched cells relative to unstretched controls. Label uptake is inhibited at 4 C. Abbreviations: SM, sphingomyelin; PM, plasma membrane. Source: From Ref. 88.

54

Hubmayr

label had been removed from the PM by back exchange) when cells are maintained at 4 C. The situation is quite different when the cells are allowed to warm to room air, because there is label internalization into punctuate structures that is greatly enhanced by stretch. Preliminary experiments on A549 cells suggest that the endocytosis associated with DILT does not involve the classic clathrin-dependent pathway (87). It should be noted that in the freshly isolated primary type II cells from rats, SM targeting to lamellar bodies is not restricted to an endocytic pathway (89). In contrast to lactosyl ceramide, a glycosphingolipid, SM and phosphatidylcholine gain access to lamellar bodies at 4 C by a protein- and energy-dependent transport mechanism. These observations either undermine the relevance of lipid sorting data from cell lines such as A549 cells and/or underscore the molecular speciﬁcity of secretory as opposed to nonsecretory lipid trafﬁcking mechanisms (90,91). In aggregate, experimental evidence suggests that DILT, in addition to its vital role in cell volume and surface area regulation, is also a means of altering PM lipid compositions and (presumably) cell surface protein expressions in response to deforming stress. The PM of eukaryotic cells is enriched with cholesterol and phosphatidylcholine and also contains high levels of sphingolipids (92,93). Lipids and proteins of the PM are hydrophobically matched in order to maintain a low membrane-energy state. Changes in the protein and lipid composition alter the membrane-energy state and, thereby, inﬂuence cell function (94). Sphingolipids play important roles in a wide variety of cell functions, including mechanotransduction (95). Their concentration in cell membranes is tightly regulated in close association with cholesterol, with which they form membrane microdomains (96,97). In endothelial cells, ﬁbroblasts, and some epithelial cells, the sphingolipids, cholesterol, and glycosylphosphatidylinositol-anchored proteins appear to have a preferential association with 50 to 100 nm pits called caveolae, as deﬁned by the marker protein caveolin (98). These structures play an important role in non–clathrin-dependent endocytosis. In contrast to surfactant-secreting type II alveolar epithelial cells, caveolae can be readily identiﬁed in type I alveolar epithelium (99–101). To the same extent to which caveolae are PM invaginations that may unfold when the PM is laterally stressed, they could be important for the mechanotransduction and in addition confer protection against deformation-related injury. It is of note that freshly harvested as well as transformed ATII cells express caveolin 1 and acquire caveolae when they are tonically stretched in culture (Pandey et al., unpublished observation). Several observations on normal and injured lungs raise interest in the molecule and pathway speciﬁcity of deformation-triggered vesicular trafﬁcking. When lungs suffer relatively mild forms of interstitial edema, the lipid microdomains of lung cell–surface membranes undergo a substantial reorganization (102,103). In vitro, such changes are associated with

Response of Cellular Plasma Membrane to Mechanical Stress

55

changes in cell phenotype and, by inference, with changes in the cells’ susceptibility to mechanical injury (103–105). Similar remodeling responses can be effected in vitro by the exposure of cells to cholesterol (106,107). This may be relevant insofar as the alveolar exudate of injured lungs contains cell debris and is cholesterol rich (108). As the progenitor of the type I cell, during alveolar wound healing, type II cells must divide and differentiate in a cholesterol-rich environment. If and how excess cholesterol effects the differentiation of alveolar epithelial type II cells to the ATI phenotype in vivo, and the consequences of this differentiation on DILT and mechanotransduction are not known.

VI. PM Repair The ability to restore membrane integrity following cell wounding is essential for cell survival, and virtually all cells possess the means to do so (Fig. 4) (51). For a long time, the prevailing view held was that injured PMs repaired primarily by ‘‘self-sealing,’’ whereby hydrophobic interactions between phospholipids and water would drive lipid ﬂow toward the free edges of a defect (109,110). Indeed, this mechanism is readily observed in model membranes and red blood cells. However, by itself, it is insufﬁcient for the repair of large wounds and does not account for resealing in nucleated cells (111). In 1994, Steinhardt et al. described wounding responses in sea urchin eggs and provided the ﬁrst clues that repair was governed by Ca2þ-dependent membrane trafﬁcking and fusion events (112). During the subsequent decade, investigators have extended these observations to mammalian cells and have added considerable detail to our understanding of the responsible molecular mechanisms (51). Small disruptions on the order of 1 mm evoke a calcium-dependent exocytosis of vesicles near the wound site, lower PM tension, and thereby facilitate wound closure (113). The generation and trafﬁcking of vesicles involve nonmuscle myosin and are sensitive to disruptions of the actin cytoskeleton (114,115). A rise in cytosolic Ca2þ consequent to the loss of PM integrity also promotes the coalescence of vesicular endomembranes (116–119). These are transported as a ‘‘patch’’ to the site of larger defects and fuse there with the PM. Lysosomes appear to be a ubiquitous source of endomembrane patches in wounded cells, and it has been argued that the release of lysosomal contents following membrane injury could be a primitive defense mechanism against invading pathogens (120–125). However, lysosomes do not appear to be essential for the process of membrane resealing. Molecular approaches to docking and fusion of protein targets suggest that the pool of membrane vesicles for resealing is a heterogenous population as opposed to a speciﬁc subtype (90,126,127).

56

Hubmayr

Figure 4 Schematic of the cellular response to membrane stress failure. Calcium enters the cell through a PM defect. Sustained large elevations in intracellular Ca2þ produce necrosis. Smaller transients in intracellular Ca2þ initiate cell repair responses. Cells repair membrane defects but by several mechanisms (right-hand side). Mechanism 1 involves lateral ﬂow PM lipids–driven free energy (analogous to surface tension) at the wound edge. This mechanism is thought to play a role in the healing of small defects. Mechanism 2 is the fusion of early endosomes with the PM. Mechanism 3 involves the coalescence of vesicular organelles (usually lysosomes) that form a patch and plugs the wound by Ca2þ-induced, site-directed exocytosis. Wounding and repair also trigger the translocation of nuclear transcription factors such as NFjB, leading to the induction of early stress response genes, and thereby initiating proinﬂammatory signaling cascades. Abbreviations: NFjB, nuclear factor jB; PM, plasma membrane. Source: From Ref. 10.

The machinery mediating Ca2þ-dependent exocytosis requires the coordinated interactions between Ca2þ sensors and the assembly of N-ethylmaleimide sensitive factor attachment protein receptors (SNAREs) (91,121,128,129). The SNAREs’ syntaxin and synaptosome-associated protein of 25 kDa form a heterodimer at the PM and serve as targets for the vesicle-associated SNARE, synaptobrevin. Vesicle docking is initiated by the coupling between PM-resident and vesicle-resident SNAREs and is greatly accelerated in the presence of Ca2þ-sensing proteins such as synaptotagmins (129). Once formed, the SNARE complex pulls the membranes

Response of Cellular Plasma Membrane to Mechanical Stress

57

close to each other to destabilize the lipid–water interface and to initiate mixing of the lipids. Synaptotagmins are a large family of at least 15 isoforms that share the structural organization of tandem C2 domains. Nearly all bind the target SNARE heterodimer, and many interact with Ca2þ and phospholipids (91). C2 domains are essential for successful resealing in 3T3 ﬁbroblasts (127). Shen et al. reported average resealing rates of 0.077/sec in 3T3 ﬁbroblasts following PM micropuncture. When cells were wounded twice the resealing rates nearly doubled. The molecular signaling responses vary with the site of the second wound. The facilitated response to wounding at the same site depends on the generation of new vesicles in a protein kinase C- and Golgi-dependent manner (114). The potentiated resealing response to wounding at a different site is cyclic adenosine monophosphate and protein kinase A dependent, and becomes protein synthesis dependent after a few hours, but does not require the formation of new vesicles (127,130). Synaptotagmin VII–deﬁcient mice show defects in cell resealing and develop a form of autoimmune myositis (131). This suggests that defective membrane repair and the consequent release of intracellular contents overwhelm immune tolerance to self-antigens. While the lung morphology and function of synaptotagmin VII–deﬁcient mice have not been characterized to date, it is of note that another syndrome with impaired lysosomal exocytosis, the Hermansky–Pudlak syndrome, is associated with lung pathology (132).

VII. Effects of PM Wounding on Gene Expression and Cell Survival Since Tremblay et al. demonstrated a relationship between ventilator settings and inﬂammatory signaling, the immune and inﬂammatory responses of lungs to mechanical stress have been extensively studied (133–137). Notwithstanding some debate about model, cell, and timing speciﬁc differences in the expression of inﬂammatory mediators, in aggregate, the evidence leaves little doubt that inﬂammation is integral to the pathobiology of the syndrome (138–141). The concomitant impairment in the lung barrier function contributes to the loss of compartmentalization and may account for many of the systemic manifestations of ventilator-associated lung injury (142–148). The inﬂammatory effect of mechanical deformation on uninjured lungs (‘‘one hit’’) compared to that on preinjured lungs (‘‘two hits’’) remains an area of important and continued investigation (149). The signal transduction pathways that link deformation to some gene responses are being characterized in ever increasing detail (see Chapters 1 and 8). Nevertheless, the importance of PM wounding in initiating a widespread proinﬂammatory gene response in ventilator-injured lungs is difﬁcult to discern (141,150). It is clear that not all molecular responses

58

Hubmayr

to mechanical ventilation are associated with lung edema or micron-scale PM lesions (151). Indeed, observations on macrophages and epithelial cells in culture suggest that deforming stress can trigger the release of proinﬂammatory cytokines in the absence of gross cell injury (139,140,151). However, many animal models of VILI in which inﬂammatory mechanisms have been characterized have employed ventilation strategies known to produce cell and PM stress failure (11,49,133). When PM lesions are produced in cell cultures, they invariably cause the translocation of nuclear factor jB (NFjB) followed by the induction of early stress-response genes (53). It is reasonable to think that similar events occur in intact mechanically ventilated lungs. Moreover, the resulting induction of CXC chemokines could be ampliﬁed and transmitted to uninjured cells by cell contact—as well as non–cell contact–dependent pathways (141,152). In that scenario PM wounding becomes the critical mechanosensing event, which initiates and propagates the inﬂammatory response in the whole organ. Inﬂammation may be initiated and sustained by cytokines and growth factors that are either released by wounded cells or are secreted by uninjured cells as part of a coordinated mechanotransduction response. Both the mechanisms have been demonstrated in reduced systems (54,139,153). In this context, Berrios et al. recently examined the effects of strain rate on intercellular adhesion molecule-1 (ICAM-1) gene expressions in A549 cells (154). ICAM-1 is a transmembrane protein expressed constitutively on the apical surface of type I cells, which binds to leukocytes via adhesion proteins alpha (M) beta(2) (Mac-1) and alpha (L) beta(2) (lymphocyte function-associated antigen-1) (155,156). Consistent with the known effects of deforming stress on NFjB kinetics, cyclic stretch induced a transient upregulation of ICAM-1 mRNA, which was followed by increased protein expression. More interesting, however, is the observation that different strain rates produced distinct time courses in ICAM-1 gene response. Rapid stretch (>100%/sec) was associated with a transient doubling in ICAM-1 message at 30 minutes, while the same deformation applied at a low strain rate (<15%/sec) produced a gradual increase in gene response, which was sustained for at least four hours. The notable fact about these differences in gene response is that rapid strain rates produce wounding in A549 cells, while low strain rates do not (50). Understanding the gene proﬁle of cells that wound, compared to those that are able to prevent or repair wounding, may provide potential protein candidates for further study and chemotherapeutic targeting. In the absence of timely repair, cell wounding leads to cell death by necrosis. It is not known if cell survival consequent to membrane repair results in delayed programmed cell death, or if injury followed by repair alters a cell’s susceptibility to other proapoptotic stimuli. A growing number of studies are focusing on apoptotosis as a pathogenetic mechanism in ventilator-injured lungs (157–159). While type II cell hyperplasia and cell

Response of Cellular Plasma Membrane to Mechanical Stress

59

necrosis dominate the acute phase of VILI, tissue specimens from patients with resolving ARDS have revealed type II cell apoptosis (160). The paucity of apoptotic cells in acutely injured lungs of hyperventilated animals contrasts with the abundance of apoptotic cells in the kidney and gastrointestinal tract (148). VIII. Conclusion I have reviewed the morphologic, physiologic, and biochemical evidence that underscores the importance of lung-cell wounding and repair in the pathogenesis of ventilator-associated lung injury. I have emphasized that structural remodeling of stress-bearing elements is a critical response of lung cells to deforming stress, and that the rate at which the cells accomplish this task determines the probability of wounding as well as cell survival. I have highlighted the importance of lipid-trafﬁcking mechanisms in the regulation of cell surface area and PM tension, and have pointed out that environmental conditions such as hypercapnia and acidemia may be powerful modulators of cellular injury and repair responses. I conclude with the hope that emerging insights into lipid sorting and vesicular-trafﬁcking mechanisms will accelerate the search for molecular targets helpful for protecting the lungs against deformation injury and biotrauma. Acknowledgment This work was supported by NHLBI grant HL 63178 and grants from the Brewer and Mayo Foundation. References 1. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157(1):294–323. 2. Pinhu L, Whitehead T, Evans T, Grifﬁths M. Ventilator-associated lung injury. Lancet 2003; 361(9354):332–340. 3. Ricard JD, Dreyfuss D, Saumon G. Ventilator-induced lung injury. Eur Respir J Suppl 2003; 42:2S–9S. 4. Gattinoni L, Carlesso E, Cadringher P, et al. Physical and biological triggers of ventilator-induced lung injury and its prevention. Eur Respir J Suppl 2003; 47:15S–25S. 5. Imanaka H, Shimaoka M, Matsurra N, et al. Ventilator-induced lung injury is associated with neutrophil inﬁltration, macrophage activation, and TGF-beta 1 mRNA upregulation in rat lungs. Anesth Analg 2001; 92(2):428–436. 6. Broccard AF, Shapiro RS, Schmitz LL, et al. Inﬂuence of prone position on the extent and distribution of lung injury in a high tidal volume oleic acid model of acute respiratory distress syndrome. Crit Care Med 1997; 25(1):16–27.

60

Hubmayr

7. Murray JF, Marthay MA, Luce JM, et al. An expanded deﬁnition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988; 138(3):720–723. 8. Bernard GR, Artigas A, Brigham KL, et al. The American-European Consensus Conference on ARDS: deﬁnitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149(3 Pt 1):818–824. 9. Tremblay LN, Slutsky AS. Ventilator-induced injury: from barotrauma to biotrauma. Proc Assoc Am Physicians 1998; 110(6):482–488. 10. Vlahakis NE, Hubmayr RD. Cellular stress failure in ventilator injured lungs. Am J Respir Crit Care Med 2005 (published ahead of print on February 1, 2005 as doi: 10.1164/rccm.200408–1036SO). 11. Dreyfuss D, Basset G, Soler P, et al. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 1985; 132(4):880–884. 12. John E, McDevitt M, Wilbom W, et al. Ultrastructure of the lung after ventilation. Br J Exp Pathol 1982; 63(4):401–407. 13. Webb HH, Tiemey DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures: protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110(5):556–565. 14. Taskar V, John J, Evander E, et al. Surfactant dysfunction makes lungs vulnerable to repetitive collapse and reexpansion. Am J Respir Crit Care Med 1997; 155(1):313–320. 15. Argiras EP, Blakeley CR, Dunnhill MS, et al. High PEEP decreases hyaline membrane formation in surfactant deﬁcient lungs. Br J Anesth 1987; 59(10):1278–1285. 16. Gattinoni L, Pelosi P, Crotti S, et al. Effects of positive end-expiratory pressure on regional distribution of tidal volume and recruitment in adult respiratory distress syndrome. Am J Respir Crit Care Med 1995; 151(6):1807–1814. 17. Lachmann B. Open up the lung and keep the lung open. Int Care Med 1992; 18(6):319–321. 18. Muscedere JG, Mullen JB, Gan K, et al. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med 1994; 149(5): 1327–1334. 19. D’Angelo E, Pecchiari M, Baraggia P, et al. Low-volume ventilation causes peripheral airway injury and increased airway resistance in normal rabbits. J Appl Physiol 2002; 92(3):949–956. 20. Goldstein L, Bughalo MT, Marquette CH, et al. Mechanical ventilationinduced air space enlargement during experimental pneumonia in piglets. Am J Respir Crit Care Med 2001; 163(4):958–964. 21. Farias LL, Faffe DS, Xisto DG, et al. Positive end-expiratory pressure prevents lung mechanical stress caused by recruitment/derecruitment. J Appl Physiol 2005; 98(1):53–61. 22. Kay SS, Bilek AM, Dee KC, et al. Pressure gradient, not exposure duration, determines the extent of epithelial cell damage in a model of pulmonary airway reopening. J Appl Physiol 2004; 97(1):269–276. 23. Fujioka H, Grotberg JB. Steady propagation of a liquid plug in a twodimensional channel. J Biomech Eng 2004; 126(5):567–577.

Response of Cellular Plasma Membrane to Mechanical Stress

61

24. Bilek AM, Dee KC, Gaver DP III. Mechanisms of surface-tension-induced epithelial cell damage in a model of pulmonary airway reopening. J Appl Physiol 2003; 94(2):770–783. 25. Gaver DP III, Kute SM. A theoretical model study of the inﬂuence of ﬂuid stresses on a cell adhering to a microchannel wall. Biophys J 1998; 75(2):721–733. 26. West JB. Thoughts on the pulmonary blood-gas barrier. Am J Lung Cell Mol Physiol 2003; 285(3):L501–L513. 27. West JB, Tsukimoto K, Mathieu-Costello O, et al. Stress failure in pulmonary capillaries. J Appl Physiol 1991; 70(4):1731–1742. 28. Tsukimoto K, Mathieu-Costello O, Prediletto R, et al. Ultrastructural appearances of pulmonary capillaries at high transmural pressures. J Appl Physiol 1991; 71(2):573–582. 29. Costello MX, Mathieu-Costello O, West JB. Stress failure of alveolar epithelial cells studied by scanning electron microscopy. Am Rev Respir Dis 1992; 145(6):1446–1455. 30. Swenson ER, Maggiorini M, Mongovin S, et al. Pathogenesis of high-altitude pulmonary edema: inﬂammation is not an etiologic factor. JAMA 2002; 257(17):2228–2235. 31. Wieslander J, Heinegard D. The involvement of type IV collagen in Goodpasture’s syndrome. Ann NY Acad Sci 1985; 460:363–374. 32. West JB, Mathieu-Costello O, Jones JH, et al. Stress failure of pulmonary capillaries in racehorses with exercise-induced pulmonary hemorrhage. J Appl Physiol 1993; 75(3):1097–1109. 33. Hotchkiss JR, Simonson DA, Marek DJ, et al. Pulmonary microvascular fracture in a patient with acute respiratory distress syndrome. Crit Care Med 2002; 30(10):2368–2370. 34. Hopkins SR, Schoene RB, Martin TR, et al. Intense exercise impairs the integrity of the pulmonary blood gas barrier in elite athletes. Am J Resp Crit Care Med 1997; 155(3):1090–1094. 35. Fu Z, Costello MX, Tsukimoto K, et al. High lung volume increases stress failure in pulmonary capillaries. J Appl Physiol 1992; 73(1):123–133. 36. Broccard A, Shapiro RS, Schmitz LL, et al. Prone positioning attenuates and redistributes ventilator-induced lung injury in dogs. Crit Care Med 2000; 28(2):295–303. 37. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28(5):596–608. 38. Bachofen H, Schu¨rch S, Michel RP, et al. Experimental hydrostatic pulmonary edema in rabbit lungs: morphology. Am Rev Respir Dis 1993; 147(4):989–996. 39. Bachofen H, Schu¨rch S, Weibel ER. Experimental hydrostatic pulmonary edema in rabbit lungs: barrier lesions. Am Rev Respir Dis 1993; 147(4):997–1004. 40. West JB, Mathieu-Costello O. Structure, strength, failure, and remodeling of the pulmonary blood-gas barrier. Annu Rev Physiol 1999; 61:543–572. 41. Neal CR, Michel CC. Transcellular openings through microvascular walls in acutely inﬂamed frog mesentery. Exp Physiol 1992; 77(6):917–920.

62

Hubmayr

42. Neal CR, Michel CC. Transcellular gaps in microvascular walls of frog and rat when permeability is increased by perfusion with the ionophore A23187. J Physiol 1995; 488(Pt 2):427–437. 43. Neal CR, Michel CC. Openings in frog microvascular endothelium induced by high intravascular pressures. J Physiol 1996; 492(Pt 1):39–52. 44. Feng D, Nagy NA, Hipp J, et al. Reinterpretation of endothelial cell gaps induced by vasoactive mediators in guinea-pig, mouse and rat: many are transcellular pores. J Physiol 1997; 504(Pt 3):747–761. 45. Neal CR, Michel CC. Transcellular openings through frog microvascular endothelium. Exp Physiol 1997; 82(2):419–422. 46. Elliott AR, Fu Z, Tsukimoto K, et al. Short-term reversibility of ultrastructural changes in pulmonary capillaries caused by stress failure. J Appl Physiol 1992; 73(3):1150–1158. 47. Dreyfuss D, Soler P, Saumon G. Spontaneous resolution of pulmonary edema caused by short periods of cyclic overinﬂation. J Appl Physiol 1992; 72(6):2081–2089. 48. Carman CV, Springer TA. A transmigratory cup in leukocyte diapedesis both through individual vascular endothelial cells and between them. J Cell Biol 2004; 167(2):377–388. 49. Gajic O, Lee J, Doerr CH, et al. Ventilator-induced cell wounding and repair in the intact lung. Am J Respir Crit Care Med 2003; 167(8):1057–1063. 50. Vlahakis NE, Schroeder MA, Pagano RE, et al. Role of deformation-induced lipid trafﬁcking in the prevention of plasma membrane stress failure. Am J Respir Crit Care Med 2002; 166(2):1282–1289. 51. McNeil PL, Steinhardt RA. Plasma membrane disruption: repair, prevention, adaptation. Annu Rev Cell Dev Biol 2003; 19:697–731. 52. Doerr CH, Gajic O, Berrios JC, et al. Hypercapnic acidosis impairs plasma membrane wound reseating in ventilator injured lungs. Am J Respir Crit Care Med 2005 (published ahead of print on February 1, 2005 as doi: 10.1164/ rccm.200309–1223OC). 53. Grembowicz KP, Sprague D, McNeil PL. Temporary disruption of the plasma membrane is required for c-fos expression in response to mechanical stress. Mol Biol Cell 1999; 10(4):1247–1257.. 54. Tschumperlin DJ, Margulies SS. Equibiaxial deformation-induced injury of alveolar epithelial cells in vitro. Am J Physiol 1998; 275(6 Pt 1):L1173–L1183. 55. Shibata K, Cregg N, Engelberts D, et al. Hypercapnic acidosis may attenuate acute lung injury by inhibition of endogenous xanthine oxidase. Am J Respir Crit Care Med 1999; 158(5 Pt 1):1578–1584. 56. Broccard AF. Protective effects of hypercapnic acidosis on ventilator-induced lung injury. Am J Respir Crit Care Med 2001; 164(5):802–806. 57. Laffey JG, Tanaka M, Engelberts D, et al. Therapeutic hypercapnia reduces pulmonary and systemic injury following in vivo lung reperfusion. Am J Respir Crit Care Med 2000; 162(6):2287–2294. 58. Sinclair SE, Kregenow DA, Lamm WJ, et al. Hypercapnic acidosis is protective in an in vivo model of ventilator-induced lung injury. Am J Respir Crit Care Med 2002; 166(3):403–408.

Response of Cellular Plasma Membrane to Mechanical Stress

63

59. Strand M, Ikegami M, Jobe AH. Effects of high pCO2 on ventilated preterm lamb lungs. Pediatr Res 2003; 53(3):468–472. 60. Laffey JG, Honan D, Hopkins N, et al. Hypercapnic acidosis attenuates endotoxin-induced acute lung injury. Am J Respir Crit Care Med 2004; 169(1):46–56. 61. Lang JD, Figueroa M, Sanders KD, et al. Hypercapnia via reduced rate and tidal volume contributes to lipopolysaccharide-induced lung injury. Am J Respir Crit Care Med 2005; 171(2):147–157. 62. Kavanagh BP. Therapeutic hypercapnia: careful science, better trials. Am J Respir Crit Care Med 2005; 171(2):96–97. 63. Vasela A, Wilhelm J. The role of carbon dioxide in free radical reactions of the organism. Physiol Res 2002; 51(4):335–339. 64. Squadrito GL, Pryor WA. Oxidative chemistry of nitric oxide: the roles of superoxide, peroxinitrite, and carbon dioxide. Free Radic Biol Med 1998; 25(4–5):392–403. 65. Wang N, Naruse K, Stamenovic D, et al. Mechanical behavior in living cells consistent with the tensegrity model. Proc Natl Acad Sci USA 2001; 98(14):7765–7770. 66. Sheetz MP. Cell control by membrane-cytoskeleton adhesion. Nat Rev Mol Cell Biol 2001; 2(5):392–396. 67. Olbrich K, Rawicz W, Needham D, et al. Water permeability and mechanical strength of polyunsaturated lipid bilayers. Biophys J 2000; 79(1):321–327. 68. Waugh RE. Effects of abnormal cytoskeletal structure on erythrocyte membrane mechanical properties. Cell Motil 1983; 3(5–6):609–622. 69. Evans E, Needham D. Physical properties of surfactant bilayer membranes composed of lipids, cholesterol and polypeptides: thermal transitions, elasticity, cohesion, and colloidal interaction. J Phys Chem 1987; 91(16):4219– 4228. 70. Bloom M, Evans E, Mouritsen OG. Physical properties of the ﬂuid bilayer component of cell membranes. Q Rev Biophys 1991; 24(3):293–397. 71. Evans E, Heinrich V, Ludwig F, et al. Dynamic tension spectroscopy and strength of biomembranes. Biophys J 2003; 55(4):2342–2350. 72. Kuo SC, Sheetz MP. Optical tweezers in cell biology. Trends Cell Biol 1992; 2(4):116–118. 73. Dai J, Sheetz MP. Mechanical properties of neuronal growth cone membranes studied by tether formation with laser optical tweezers. Biophys J 1995; 68(3):988–996. 74. Janmey PA. Protein regulation by phosphatidylinositol lipids. Chem Biol 1995; 2(2):61–65. 75. Raucher D, Stauffer T, Chen W, et al. Phosphatidylinositol 4,5-bisphosphate functions as a second messenger that regulates cytoskeleton-plasma membrane adhesion. Cell 2000; 100(2):221–228. 76. Ruwhof C, van Wamel JET, Noordzij LAW, et al. Mechanical stress stimulates phospholipase C activity and intracellular calcium ion levels in neonatal rat cardiomyocytes. Cell Calcium 2001; 29(2):73–83. 77. Dai J, Sheetz MP. Membrane tether formation from blebbing cells. Biophys J 1999; 77(6):3363–3370.

64

Hubmayr

78. Chieregatti E, Meldolesi J. Regulated exocytosis: new organelles for nonsecretory purposes. Nat Rev Mol Cell Biol 2005; 6(2):181–187. 79. Solsona C, Innocenti B, Fernandez JM. Regulation of exocytotic fusion by cell inﬂation. Biophys J 1998; 74(2 Pt 1):1061–1073. 80. Vlahakis NE, Schroeder MA, Pagano RE, et al. Deformation-induced lipid trafﬁcking in alveolar epithelial cells. Am J Physiol Lung Cell Mol Physiol 2001; 280(5):L938–L946. 81. Raucher D, Sheetz MP. Characteristics of a membrane reservoir buffering membrane tension. Biophys J 1999; 77(4):1992–2002. 82. Morris CE, Homann U. Cell surface area regulation and membrane tension. J Membr Biol 2001; 179(2):79–102. 83. Morris CE. How did cells get their size? Anat Rec 2002; 268(3):239–251. 84. Fisher JL, Levitan I, Margulies SS. Plasma membrane surface increases with tonic stretch of alveolar epithelial cells. Am J Respir Cell Mol Biol 2004; 31(2):200–208. 85. Berrios JC, Schroeder MA, Hubmayr RD. Mechanical properties of alveolar epithelial cells in culture. J Appl Physiol 2001; 91(1):65–73. 86. Truschel ST, Wang E, Ruiz WG, et al. Stretch-regulated exocytosis/endocytosis in bladder umbrella cells. Mol Biol Cell 2002; 13(3):830–846. 87. Berrios JC, Hubmayr RD. Deforming stress triggers endocytosis in alveolar epithelial cells [abstr.]. Am J Respir Crit Care Med 2003; 167:A57. 88. Vlahakis NE. Biophysical factors leading to ventilator induced lung injury. In: Brochard L, Slutsky AS, eds. Mechanical Ventilation. Springer Verlag, 2004:213–226. 89. Rajan N, Vlahakis NE, Marks DL, et al. Selective uptake and sorting of sphingomyelin in rat lung. Am J Respir Crit Care Med 2000; 161:A658. 90. Borgonovo B, Cocucci E, Racchetti G, et al. Regulated exocytosis: a novel, widely expressed system. Nat Cell Biol 2002; 4(12):955–962. 91. Chapman ER. Synaptotagmin: a Ca2þ sensor that triggers exocytosis? Nat Rev Mol Cell Biol 72002; 3:1–11. 92. Brown DA, London E. Structure and function of sphingolipid- and cholesterol-rich membrane rafts. J Biol Chem 2000; 275(23):17221–17224. 93. Pagano RE, Watanabe R, Wheatley C, et al. Applications of BODIPYsphingolipid analogs to study lipid trafﬁc and metabolism in cells. Sphingolipid metabolism and cell signaling, PTB. Methods Enzymol 2000; 312:523–534. 94. Mitra K, Ubarretxena-Belandia L, Taguchi T, et al. Modulation of the bilayer thickness of exocytic pathway membranes by membrane proteins rather than cholesterol. Proc Natl Acad Sci USA 2004; 101(12):4083–4088. 95. Volonte D, Galbiati F, Pestell RG, et al. Cellular stress induces the tyrosine phosphorylation of caveolin-1 (Tyr(14)) via activation of p38 mitogenactivated protein kinase and c-Src kinase—evidence for caveolae, the actin cytoskeleton, and focal adhesions as mechanical sensors of osmotic stress. J Biol Chem 2001; 276(11):8094–8103. 96. Simons K, Toomre D. Lipid rafts and signal transduction. Nat Rev Mol Cell Biol 2000; 7(1):31–39. 97. van Meer G, Holthuis JCM. Sphingolipid transport in eukaryotic cells. Biochim Biophys Acta 2000; 1486(1):145–170.

Response of Cellular Plasma Membrane to Mechanical Stress

65

98. Razani B, Woodman SE, Lisanti MP. Caveolae: from cell biology to animal physiology. Pharmacol Rev 2002; 54(3):431–467. 99. Gil J. Number and distribution of plasmalemmal vesicles in the lung. Fed Proc 1983; 42(8):2414–2418. 100. Kasper M, Reimann T, Hempel U, et al. Loss of caveolin expression in type I pneumocytes as an indicator of subcellular alterations during lung ﬁbrogenesis. Histochem Cell Biol 1998; 109(1):41–48. 101. Newman GR, Campbell L, von Ruhland C, et al. Caveolin and its cellular and subcellular immunolocalisation in lung alveolar epithelium: implications for alveolar epithelial type I cell function. Cell Tissue Res 1999; 295(1):111–120. 102. Palestini P, Calvi C, Conforti E, et al. Compositional changes in lipid microdomains of air-blood barrier plasma membranes in pulmonary interstitial edema. J Appl Physiol 2003; 95(4):1446–1452. 103. Daffara R, Botto L, Beretta E, et al. Endothelial cells as early sensors of pulmonary interstitial edema. J Appl Physiol 2004; 97(4):1575–1583 104. Gutierrez JA, Ertsey R, Scavo LM, et al. Mechanical distention modulates alveolar epithelial cell phenotypic expression by transcriptional regulation. Am J Respir Cell Mol Biol 1999; 21(2):223–229. 105. Gutierrez JA, Suzara VV, Dobbs LG. Continuous mechanical contraction modulates expression of alveolar epithelial cell phenotype. Am J Respir Cell Mol Biol 2003; 29(1):81–87. 106. Kolleck L, Guthmann F, Ladhoff AM, et al. Cellular cholesterol stimulates acute uptake of palmitate by redistribution of fatty acid translocase in type II pneumocytes. Biochemistry 2002; 41(20):6369–6375. 107. Kolleck I, Wissel H, Guthmann F, et al. HDL-holoparticle uptake by alveolar type II cells: effect of vitamin E status. Am J Respir Cell Mol Biol 2002; 27(1):57–63. 108. Swendsen CL, Skita V, Thrall RS. Alterations in surfactant neutral lipid composition during the development of bleomycin-induced pulmonary ﬁbrosis. Biochim Biophys Acta 1996; 1301(1–2):90–96. 109. Benz R, Zimmermann U. The resealing process of lipid bilayers after reversible electrical breakdown. Biochim Biophys Acta 1981; 640(1):169–178. 110. Lipowsky R. The conformation of membranes. Nature 1991; 349(6309): 475–481. 111. McNeil PL, Miyake K, Vogel SS. The endomembrane requirement for cell surface repair. Proc Natl Acad Sci USA 2003; 100(8):4592–4597. 112. Steinhardt RA, Bi G, Alderton JM. Cell membrane resealing by a vesicular mechanism similar to neurotransmitter release. Science 1994; 263(5154): 390–393. 113. Togo T, Krasieva TB, Steinhardt RA. A decrease in membrane tension precedes successful cell-membrane repair. Mol Biol Cell 2000; 11(12):4339–4346. 114. Togo T, Alderton JM, Bi GQ, et al. The mechanism of facilitated cell membrane resealing. J Cell Sci 1999; 112(Pt 5):719–731. 115. Togo T, Steinhardt RA. Nonmuscle myosin HA and HB have distinct functions in the exocytosis-dependent process of cell membrane repair. Mol Biol Cell 2004; 15(2):688–695.

66

Hubmayr

116. Terasaki M, Miyake K, McNeil PL. Large plasma membrane disruptions are rapidly resealed by Ca2þ-dependent vesicle-vesicle fusion events. J Cell Biol 1997; 139(1):63–74. 117. McNeil PL, Vogel SS, Miyake K, et al. Patching plasma membrane disruptions with cytoplasmic membrane. J Cell Sci 2000; 113(Pt 11):1891–1902. 118. Bi GQ, Alderton JM, Steinhardt RA. Calcium-regulated exocytosis is required for cell membrane resealing. J Cell Biol 1995; 131(6 Pt 2):1747–1758. 119. Bi GQ, Morris RL, Liao G, et al. Kinesin- and myosin-driven steps of vesicle recruitment for Ca2þ-regulated exocytosis. J Cell Biol 1997; 138(5):999–1008. 120. Reddy A, Caler EV, Andrews NW. Plasma membrane repair is mediated by Ca(2þ)-regulated exocytosis of lysosomes. Cell 2001; 106(2):157–169. 121. Jahn R, Lang T, Sudhof TC. Membrane fusion. Cell 2003; 112(4):519–533. 122. McNeil PL. Repairing a torn cell surface: make way, lysosomes to the rescue. J Cell Sci 2002; 115(Pt 5):873–879. 123. Andrews NW. Regulated secretion of conventional lysosomes. Trends Cell Biol 2000; 10(8):316–321. 124. Rodriguez A, Webster P, Ortego J, et al. Lysosomes behave as Ca2þ-regulated exocytic vesicles in ﬁbroblasts and epithelial cells. J Cell Biol 1997; 137(1):93–104. 125. Rodriguez A, Martinez I, Chung A, et al. cAMP regulates Ca2þ-dependent exocytosis of lysosomes and lysosome-mediated cell invasion by trypanosomes. J Biol Chem 1999; 274(24):16754–16759. 126. Jedd G, Chua NH. A new self-assembled peroxisomal vesicle required for efﬁcient resealing of the plasma membrane. Nat Cell Biol 2000; 2(4):226–231. 127. Shen SS, Tucker WC, Chapman ER, et al. Molecular regulation of membrane resealing in 3T3 ﬁbroblasts. J Biol Chem 2005; 280(2):1652–1660. 128. Rao SK, Huynh C, Proux-Gillardeaux V, et al. Identiﬁcation of SNAREs involved in synaptotagmin VII-regulated lysosomal exocytosis. J Biol Chem 2004; 279(19):20471–20479. 129. Detrait ER, Yoo S, Eddleman CS, et al. Plasmalemmal repair of severed neurites of PC12 cells requires Ca(2þ) and synaptotagmin. J Neurosci Res 2000; 62(4):566–573. 130. Togo T, Alderton JM, Steinhardt RA. Long-term potentiation of exocytosis and cell membrane repair in ﬁbroblasts. Mol Biol Cell 2003; 14(1):93–106. 131. Chakrabarti S, Kobayashi KS, Flavell RA, et al. Impaired membrane resealing and autoimmune myositis in synaptotagmin VII-deﬁcient mice. J Cell Biol 2003; 162(4):543–549. 132. Lyerla TA, Rusiniak ME, Borchers M, et al. Aberrant lung structure, composition, and function in a murine model of Hermansky–Pudlak syndrome. Am J Physiol Lung Cell Mol Physiol 2003; 285(3):L643–L653. 133. Tremblay L, Valenza F, Ribeiro SP, et al. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99(5):944–952. 134. Frank JA, Matthay MA. Science review: mechanisms of ventilator-induced injury. Crit Care 2003; 7(3):233–241. 135. Pugin J. Molecular mechanisms of lung cell activation induced by cyclic stretch. Crit Care Med 2003; 31(suppl 4):S200–S206.

Response of Cellular Plasma Membrane to Mechanical Stress

67

136. Uhligj S. Ventilation-induced lung injury and mechanotransduction: stretching it too far? Am J Physiol Lung Cell Mol Physiol 2002; 282(5):L892–L896. 137. Dos Santos CC, Slutsky AS. Invited review: mechanisms of ventilator-induced lung injury: a perspective. J Appl Physiol 2000; 89(4):1645–1655. 138. Dreyfuss D, Ricard JD, Saumon G. On the physiologic and clinical relevance of lung-borne cytokines during ventilator-induced lung injury. Am J Respir Crit Care Med 2003; 167(11):1467–1471. 139. Pugin J, Dunn I, Jolliet P, et al. Activation of human macrophages by mechanical ventilation in vitro. Am J Physiol 1998; 275(6 Pt 1):L1040–L1050. 140. Vlahakis NE, Schroeder MA, Limper AH, et al. Stretch induces cytokine release by alveolar epithelial cells in vitro. Am J Physiol Lung Cell Mol Physiol 1999; 277(1 Pt 1):L167–L173. 141. Belperio JA, Keane MP, Burdick MD, et al. Critical role for CXCR2 and CXCR2 ligands during the pathogenesis of ventilator-induced lung injury. J Clin Invest 2002; 110(11):1703–1716. 142. Slutsky AS, Tremblay LN. Multiple system organ failure. Is mechanical ventilation a contributing factor? Am J Respir Crit Care Med 1998; 157(6 Pt 1):1721–1725. 143. Murphy DB, Cregg N, Tremblay L, et al. Adverse ventilatory strategy causes pulmonary-to-systemic translocation of endotoxin. Am J Respir Crit Care Med 2000; 162(1):27–33. 144. Chiumello D, Pristine G, Slutsky AS. Mechanical ventilation affects local and systemic cytokines in an animal model of acute respiratory distress syndrome. Am J Respir Crit Care Med 1999; 160(1):109–116. 145. Nahum A, Hoyt J, Schmitz L, et al. Effect of mechanical ventilation strategy on dissemination of intratracheally instilled Escherichia coli in dogs. Crit Care Med 1997; 25(10):1733–1743. 146. Verbrugge SJ, Sorm V, van’t Veen A, et al. Lung overinﬂation without positive end-expiratory pressure promotes bacteremia after experimental Klebsiella pneumoniae inoculation. Intensive Care Med 1998; 24(2):172–177. 147. Haitsma JJ, Uhlig S, Goggel R, et al. Ventilator-induced lung injury leads to loss of alveolar and systemic compartmentalization of tumor necrosis factoralpha. Intensive Care Med 2000; 26(10):1515–1522. 148. Imai Y, Parodo J, Kajikawa O, et al. Injurious mechanical ventilation and endorgan epithelial cell apoptosis and organ dysfunction in an experimental model of acute respiratory distress syndrome. JAMA 2003; 289(16):2104–2112. 149. Bouadma L, Schortgen F, Ricard JD, et al. Ventilation strategy affects cytokine release after mesenteric ischemia-reperfusion in rats. Crit Care Med 2004; 32(7):1563–1569. 150. Copland IB, Kavanagh BP, Engelberts D, et al. Early changes in lung gene expression due to high tidal volume. Am J Respir Crit Care Med 2003; 168(9):1051–1059. 151. Bhattacharya S, Sen N, Yiming MT, et al. High tidal volume ventilation induces proinﬂammatory signaling in rat lung endothelium. Am J Resp Crit Care Med 2003; 28(2):218–224. 152. Boitano S, Safdar Z, Welsh DG, et al. Cell–cell interactions in regulating lung function. Am J Physiol Lung Cell Mol Physiol 2004; 287(3):L455–L459.

68

Hubmayr

153. dos Santos CC, Han B, Andrade CF, et al. DNA microarray analysis of gene expression in alveolar epithelial cells in response to TNFalpha, LPS, and cyclic stretch. Physiol Genomics 2004; 19(3):331–342. 154. Berrios JC, Zolhnan PL, Limper AH, Hubmayr RD. NF-jB mediates stretch induced ICAM-1 upregulation in A549 cells [abstr.]. Am J Respir Crit Care Med 2003; 167:A654. 155. Diamond MS, Staunton DE, de Fougerolles AR, et al. ICAM-1(CD54): a counter receptor for Mao-1(CDI lb/CD18). J Cell Biol 1990; 111(6 Pt 2):3129–3139. 156. Marlin SD, Springer TA. Puriﬁed intercellular adhesion molecule-1(ICAM-1) is a ligand for lymphocyte function-associated antigen (LFA-1). Cell 1987; 51(5):813–819. 157. Martin TR, Nakamura M, Matute-Bello G. The role of apoptosis in acute lung-injury. Crit Care Med 2003; 31(suppl 4):S184–S188. 158. Del Riccio V, van Tuyl M, Post M. Apoptosis in lung development and neonatal lung injury. Pediatr Res 2004; 55(2):183–189. 159. Nanjundan M, Possmayer F. Pulmonary phosphatidic acid phosphatase and lipid phosphate phosphohydrolase. Am J Physiol Lung Cell Mol Physiol 2003; 284(1):L1–L23. 160. Bardales RH, Xie SS, Schaefer RF, et al. Apoptosis is a major pathway responsible for the resolution of type II pneumocytes in acute lung injury. Am J Pathol 1996; 149(3):845–852.

4 Acute Passive and Active Changes in Microvascular Permeability During Lung Distention

JAMES C. PARKER

TAKASHIGE MIYAHARA

Department of Physiology, University of South Alabama College of Medicine Mobile, Alabama, U.S.A.

First Department of Internal Medicine, Shinshu University Matsumoto, Nagano, Japan

MIRCEA ANGHELESCU Department of Pathophysiology, University of Medicine and Pharmacy Timisoara, Timis, Romania

I. Introduction The contributing factors to lung injury that result from lung overdistention remain a topic of intense investigation (1). However, the severity of ventilator-induced lung injury (VILI) has been generally recognized to be both time and pressure dependent in experimental studies as well as in clinical settings (2–4). Yoshikawa et al. (5) recently observed a size-selective alveolar–capillary transport of proteins of different hydrodynamic radii, which was dependent upon peak inﬂation pressures (PIP) and ventilation time. Fluxes of Clara cell secretory protein (CCSP; 1.9 nm) from airway ﬂuids to plasma, and albumin (3.6 nm) and immunoglobin (5.6 nm) from plasma to bronchoalveolar lavage (BAL) ﬂuid exhibited increases that were both time and pressure related (Fig. 1). Progressively higher PIP levels reduced the restriction of successively larger proteins (Fig. 1A), but size restriction was also lost as ventilation time at an injurious PIP (55 cmH2O) was prolonged even though signiﬁcant clearance rates of all three proteins occurred even at 30 minutes (Fig. 1B). Although the time course of the interaction of vascular permeability, signaling molecules, and the inﬂammatory response 69

70

Parker et al.

Figure 1 Increases in plasma and BAL concentrations of CCSP (CCSP, 1.9 nm radius); albumin (3.6 nm radius); and IgG (5.6 nm radius) as a function of PIP (A) and ventilation time at 55 cmH2O (B). Abbreviations: BAL, bronchoalveolar lavage; CCSP, Clara cell secretory protein; IgG, immunoglobin; PIP, peak inﬂation pressure. Source: From Ref. 5.

remains controversial (6), most studies indicate a rapid initial increase in vascular permeability during mechanical strain of the lung. The rapidity of this increase appears to precede development of the cytokine cascade and recruitment of inﬂammatory cells, which may then amplify the lung injury (7,8). Ventilation of isolated perfused lungs of dog, rats, mice, and rabbits with high PIP for periods of only 20 minutes produce signiﬁcant increases in vascular permeability as assessed by ﬁltration coefﬁcients

Acute Passive and Active Changes in Microvascular Permeability

71

(9–12). In intact rats, Dreyfuss et al. (13) reported an increase in lung extravascular albumin space when ventilated with tidal volumes for only ﬁve minutes. An increased BAL protein in mice and rats occurred within 30 minutes during high tidal volume ventilation (5,14,15). These permeability increases generally preceded measurable increases in BAL cytokines and neutrophil recruitment (8,16). A preponderance of evidence indicates that the pathway for ﬂuid and protein movement across the alveolar–capillary barrier after mechanical injury indicates a paracellular route coupled to ﬁltration (17). The discovery of a high-afﬁnity binding protein for albumin and the ability to affect albumin uptake by depletion of this receptor in endothelial monolayers has lent support to the theory of active protein transport via vesicles (18,19). However, the contribution of active transport of proteins to the increased protein ﬂuxes observed during lung vascular injury would be too small to be measured by currently available methods (17,20). Previous lung vascular permeability studies indicate a size-selective protein sieving coupled to ﬁltration (21). In addition, size-selective restricted diffusion of solutes from alveolar ﬂuid to blood ﬁts with a passive paracellular transport route (22,23). The initial vascular leak sites following overdistention of the lung have been localized to the alveolar capillaries and microvessels using methacrylate casts (24). These studies suggest a direct route of entry of ﬂuid and protein into the alveolar spaces and interstitium. Separation of endothelial cells from basement membranes in small vessels and capillaries suggests that loss of focal adhesion and adherens junction integrity are also involved (5,14). However, lung injury due to different ventilatory patterns could produce different patterns of vascular injury because the injury due to repeated opening and collapse of alveoli (atelectotrauma) may occur through different mechanisms than simple overdistention (volutrauma) (1,25,26). Injury of vessels at the small terminal airways could also result in leakage across small airway epithelium as has previously been suggested (27,28).

II. Passive Effects of Lung Distention Hyperinﬂation of the lung increases the strain on tissue load–bearing elements as the lung approaches total lung capacity. Type I collagen ﬁbers provide structural support to the alveoli and alveolar ducts, whereas type IV collagen provide tensile strength to the basement membrane surrounding the capillaries (29). High lung distending pressures would be expected to create high levels of mechanical stress on the alveolar wall capillaries as well as perivascular cuffs surrounding small extra alveolar vessels (30,31). West et al. have proposed the concept of stress failure of lung capillaries subjected to high transmural pressures to explain the hemorrhage and leak of ﬂuid and protein at high vascular and airway pressures (32–34). The high

72

Parker et al.

tensile strength of the basement membrane collagen compared to cellular elements would be the limiting barrier to passive capillary rupture. Small breaks in the cell membranes of epithelial and endothelial cells near cell junctions occurred in proportion to high pulmonary vascular pressures and high airway pressures, and many such breaks traversed the endothelium, epithelium, and basement membranes of the alveolar–capillary barrier (35–37). However, the number of cell breaks decreased with time when vascular or airway pressures were reduced (38). Neal and Michel (39) observed similar transcellular pathways through endothelial cells in systemic frog capillaries exposed to high vascular pressures. While hemorrhage into the alveolar spaces and interstitium implies rupture of capillaries and small vessels, the transient breaks in the cell plasma membrane that do not extend through the basement membrane may not be the major route for the bulk of ﬂuid ﬁltration. Maron et al. (40) did not ﬁnd a close correlation between ﬁltration coefﬁcient and electron microscopic evidence of cell breaks in isolated dog lungs subjected to high vascular pressures, but the transient nature of cell breaks may have obscured such a relationship. In other isolated dog lung studies, increases in ﬁltration coefﬁcient induced by both high vascular pressures and airway pressures tended to return toward control with modest degrees of injury (11,41). The primacy of active endothelial cell control of capillary ﬁltration following mechanical injury was demonstrated by Parker and Ivey (42) who observed that high pulmonary vascular pressure–induced increases in ﬁltration coefﬁcient in isolated rat lungs could be attenuated by isoproterenol pretreatment even in the presence of signiﬁcant hemorrhage. Because isoproterenol enhances endothelial junctional integrity by increases in intracellular cyclic adenosine monophosphate (cAMP), an active role of the endothelial cells in modulating ﬂuid conductance was present even with concomitant capillary rupture. Enhanced intracellular cAMP also prevented high airway pressure–induced increases in ﬁltration coefﬁcient in this isolated rat lung preparation (43). Thus, an active regulation of the endothelial barrier appears to remain the most quantitatively signiﬁcant modulator of ﬂuid ﬁltration short of overwhelming destruction of capillary structural integrity.

III. Active Endothelial Control of Vascular Permeability A. Intracellular Control of Paracellular Permeability

The rapid onset of capillary permeability during lung distention undoubtedly involves both a direct mechanical tensile failure of tissue elements and rapid signal transduction events in endothelial cells. Endothelial permeability is controlled by a balance between the tethering forces of cell–cell and cell– matrix adhesions and the centripetal cell retractile tension on cytoskeletal elements (44–46). The basic structural components of the endothelial

Acute Passive and Active Changes in Microvascular Permeability

73

junctions are shown in (Fig. 2). The major cell–cell tethering proteins thought to regulate paracellular passage of ﬂuid and protein include the proteins that make up the adherens junctions and tight junctions. The cadherins are transmembrane proteins that bind the cells together through homophilic adhesion. While vascular endothelial (VE)-cadherin is predominant in most endothelia, epithelial (E)-cadherin has recently been reported in rat lung microvascular endothelial cells (47). Cadherins are anchored to the actin cytoskeleton through a cytoplasmic complex of proteins consisting of p120, a, b, and c catenins, a-actinin, and protein 4.1 (48). The tight junction proteins include a complex of homophilic occludins with cytoplasmic proteins ZO-1 and -2 and cingulin, which binds to the actin cytoskeleton and spectrin (49). Although details of tight junction control are sketchy, integrity of these junctions appears to be linked to integrity of adherens junction in endothelial cells. The adherens junctions in turn can be controlled by changes in the amount and phosphorylation state of junctional catenins (50). An increase in phosphorylation of b-catenin generally results in a loss of cadherin binding, gap formation, and an increase in endothelial permeability (51). b-Catenin can be phosphorylated by glycogen synthase kinase-3, which is downstream of a phosphorylation cascade that may involve the phosphoinositol-3 kinase (PI3K) target, Akt (protein kinase B), and other kinases (52).

Figure 2 Diagram showing the major proteins of the adherens junction, tight junctions, and focal adhesion plaques, which modulate paracellular permeability in endothelium.

74

Parker et al.

Tethering of cells to the basement membrane and interstitial matrix is a function of the focal adhesions, which are aggregates of integrins that bind to matrix components such as collagen, ﬁbronectin, vitronectin, and laminin (48). The strength of focal adhesions is also dependent upon a complex of cytoplasmic linker proteins that control integrin adhesion and cytoskeletal anchoring, which includes focal adhesion kinase (FAK), paxillin, talin, a-actinin, vinculin, and the tyrosine kinase, pp60src (Src). The phosphorylation state of FAK and other proteins in the focal adhesion complex control the strength of adhesion. Increased phosphorylation of FAK is associated with remodeling and formation of new focal adhesions, but a high state of phosphorylation of FAK generally results in a loss of adhesion and increased endothelial barrier permeability (53,54). The actin cytoskeleton also contributes to endothelial barrier function by exerting centripetal tension on cell junctions and possibly by regulating ion channels. The organization of cortical actin into peripheral bands supports junctional integrity and focal adhesions, but formation of more central stress ﬁbrils results in cell tension and facilitates gap formation (44). The cytoskeleton can affect permeability by a reorganization into stress ﬁbers and an increase in cytoskeletal tension mediated by an active contraction of the actin–myosin complex (55). Figure 3 shows the pathways that control phosphorylation of myosin light chain (MLC) and cytoskeletal tension. MLC kinase (MLCK) can be activated when phosphorylated by Src or calcium calmodulin, which in turn is activated by an increased intracellular calcium (56). Cytoskeletal tension can also be enhanced by Rho kinase

Figure 3 Potential pathways for increasing cytoskeleton tension in endothelial cells by a mechanical strain through increased phosphorylation of nonmuscle myosin. Abbreviations: MLC, myosin light chain; CaM, calmodulin; MLCK, myosin light chain kinase; SACC, stretch-activated cation channels; PLC, phospholipase C.

Acute Passive and Active Changes in Microvascular Permeability

75

inhibition of MLC phosphatase and actin depolymerization (57). There also appears to be a positive feedback function of cytoskeletal tension to increase calcium entry by increasing conductance of nonselective cation channels involved in store-operated and stretch-activated calcium entry (58). B. Potential Signaling Pathways Affecting Permeability During Mechanical Stress

The signal pathways that alter vascular permeability to ﬂuid and protein in the intact lung during mechanical stress are complex and involve mechanogated ion channels, strain-induced signals from matrix-bound integrin and homophilic bound cadherins, cytoskeletal structures, transactivation of growth factor receptors and ligation of purinergic receptors (1,59). The intact lung has several cell types that can interact and endothelial cell phenotypes that differ markedly between vascular regions within the lung and even between branch points and straight portions of individual microvessels (60,61). Increases in intracellular calcium and protein phosphorylation in endothelial cells can alter the integrity of adherens junctions and focal adhesions and increased vascular permeability (62,63). Many of the ionic and phosphorylation events thought to affect vascular permeability have been demonstrated in cultured cell monolayers during mechanical stress. However, relatively few of these pathways have been unequivocally linked to the strain-induced permeability response in intact lungs. The importance of validating in mammalian lungs any hypotheses that are based solely on reductionist cell systems has recently been emphasized by Lindner and Uhlig (64), who reported a lack of correlation in the permeability responses to several proinﬂammatory mediators when cultured endothelial monolayers were compared to isolated perfused lungs. Mechanogated Ion Channels

An increase in intracellular Ca2þ is a necessary component for increased vascular permeability induced by most mediators, and the permeability response to lung overdistention appears to be no exception (48,65). An altered ion channel activity occurs within seconds and is one of the most rapid responses to mechanical strain in both isolated lungs and cultured cell preparations (66,67). Cultured endothelial cells show a rapid increase in intracellular calcium during mechanical strain. Naruse and Sokabe (68) observed a response that was blocked by gadolinium chloride (GdCl3), an inhibitor of stretch-activated cation channels (SACC), and was mediated primarily by calcium entry rather than store release. However, other investigators have implicated a Ca2þ entry pathway in endothelial cells mediated by phospholipase C (PLC) or phospholipase D during cyclic stretch or phospholipase A2 (PLA2) during hypotonic swelling (69,70). Other studies implicate a signiﬁcant store release component, to the initial endothelial

76

Parker et al.

Ca2þ peak after cyclic stretch, in addition to a later GdCl3-sensitive Ca2þ entry that could be blocked (71). Kuebler et al. (67) reported increased intracellular calcium transients induced by high vascular pressures in endothelial cells in situ in venular capillaries of isolated rat lungs (Fig. 4A). The response was proinﬂammatory, causing an increase in P-selectin expression, and was Gd3þ sensitive. Lung distention by high airway pressure also increased epithelial cell calcium in an isolated lung preparation, which was rapidly communicated to adjacent cells (72,73). Parker et al. investigated the role of mechanogated calcium entry in isolated rat lungs on VILI and observed that high peak inﬂation pressure (PIP)–induced increases in ﬁltration coefﬁcient were attenuated using Gd3þ (Fig. 4B)

Figure 4 Isolated rat lung studies showing the increase in endothelial cell calcium produced by an increased pulmonary venous pressure (PLA) (A) and the increase in endothelial ﬂuid conductance (ﬁltration coefﬁcient) induced by high PIP ventilation (B). Note that gadolinium prevented the increase in both calcium levels and permeability induced by mechanical stress. Key: , p < 0.05 versus paired low pressure syndrome; #, p < 0.05 versus control group. Abbreviations: PLA, phospholipase A. Source: From Refs. 65, 67.

Acute Passive and Active Changes in Microvascular Permeability

77

(12,65). These studies conﬁrm that the necessary role of Ca2þ entry in lung distention–induced permeability is similar to the Ca2þ requirement for vascular permeability increases induced by a wide variety of receptor-mediated inﬂammatory mediators (48). Although isolated lung and endothelial and epithelial monolayer experiments provide presumptive evidence that VILI may be initiated by Ca2þ entry through stretch-activated calcium channels, the molecular identity of these channels has not been established. The main route for mechanogated Ca2þ entry is thought to be through nonselective cation channels (74). Recent evidence suggests that these channels may belong to the transient receptor potential (TRP) family of channel proteins (75,76). Both TRPC1 and C4 are present in lung endothelial cells and appear to contribute to both nonselective and selective pathways for Ca2þ entry (77). TRPC4 also contributes to receptor- and PLC-mediated Ca2þ entry, and TRPC1 may respond to shear stress (78). A likely candidate for the mechanogated nonselective cation channel is TRPV4. This protein contributes to channels that are activated by heat, phorbol derivatives, and mechanical stress (79,80). These channels may be activated secondarily by mechanical stimulation of PLA2 activity and subsequent arachidonic acid release. P450 derivatives of arachidonic acid metabolism, in particular the 5,6-epoxyeicosatrienoic acids, strongly activate the TRPV4 channels (81). Recent studies in isolated rat lungs indicate that infusion of 5,6-epoxyeicosatrienoic acid also induces a vascular permeability increase suggestive of a possible role of these channels in permeability (82). In addition, inhibition of PLA2 activity in mice with arachidonyl triﬂuoromethyl ketone reduced lung protein leak induced by high PIP ventilation (Fig. 5) (83). The protein leak was also augmented by the absence of Clara cell secretory protein (CCSP), which also inhibits PLA2 activity. Multiple mediators of lipoxygenase, cyclooxygenase, and P450 pathways may have affected permeability in these experiments. Because lung microvascular and macrovascular endothelial cells differ signiﬁcantly relative to calcium entry via selective and nonselective cation channels, mechanogated Ca2þ entry through multiple channels in regional endothelial cell phenotypes may also be involved (77). In fact, there were marked differences in TRP channel abundances and subtypes even between acutely harvested individual endothelial cells of the same vessel when examined by real-time polymerase chain reaction analysis (84). Because Ca2þ entry into cells may be signiﬁcantly affected by the electrochemical gradient across the cell membrane, activation of Kþ and Cl channels either by mechanical stress or by intracellular Ca2þ ([Ca2þ]I) itself could also affect Ca2þ entry (74). Freshly harvested aortic endothelial cells possessed both calcium activated potassium (KCA) and chloride channels, which tend to hyperpolarize the membrane and maintain a driving gradient for Ca2þ entry during a Ca2þ transient (85). In contrast, inward-rectifying

78

Parker et al.

Figure 5 Lavage albumin concentrations from groups of mice ventilated at a PIP of 10 cmH2O (low PIP) or a PIP of 45 cmH2O (high PIP) showing the effects of high PIP ventilation on wild type and CCSP null mice and the attenuating effect of PLA2 inhibition with arachidonyl triﬂuoromethyl ketone (ATK). Key: , P < 0.05 versus low PIP; #, P < 0.05 versus untreated CCSP-/- group. Abbreviations: CCSP, Clara cell secretory protein; PLA2, phospholipase A2; PIP, peak inﬂation pressure. Source: From Ref. 83.

Kþ channels (Kir), which have been extensively studied in cultured endothelial cells, including those from rat pulmonary artery, were not found in freshly isolated cells (86). The Kir 2.1 channel cloned from aortic endothelial cells was activated by shear stress and blocked by inhibition of tyrosine kinase (87). Chatterjee et al. (88) found that membrane potential and Ca2þ entry were also modulated by mechanically activated nucleotide-gated (KATP) channels in rat pulmonary microvascular endothelial cells during shear stress. These channels were increased during chronic exposure to shear stress but largely lost during static culture. KCA and SACC were also found to be upregulated in endothelial cells after exposure to shear stress (89,90). Shimoda et al. (91) showed that Kir currents in lung macrovascular and microvascular endothelial cells were inhibited by cyclic guanosine monophosphate (cGMP), and Pearse et al. (92) observed an increase in lung cGMP after ventilation of sheep lungs. The volume-regulated anion channels, which produce endothelial chloride currents during cell swelling may also inﬂuence Ca2þ entry because mechanical strain also activates these chloride channels, which then modulate the hyperpolarizing effect of Kþ channel opening in endothelial cells (93). An additional pathway for signiﬁcant Ca2þ entry

Acute Passive and Active Changes in Microvascular Permeability

79

may result from transient plasmalemmal membrane breaks induced by excessive strain and have been reported in cultured lung epithelial and endothelial cells as well as isolated lungs (see Chapter 3) (33,94–96). Mechanically Induced Phosphorylation Signaling

Phosphorylation of endothelial and epithelial cell proteins has a signiﬁcant role in modulating vascular permeability in response to numerous receptor ligands (63). In addition, mechanical stress can directly increase the phosphorylation state of junctional and focal adhesion proteins, and such increases in the phosphorylation state of these proteins are generally associated with an increased transvascular permeability (52). Nonreceptor tyrosine kinases are directly involved in these phosphorylation events. Yano et al. (97,98) showed that cyclical strain of endothelial monolayers increased phosphorylation of the focal adhesion proteins, FAK and paxillin. Protein phosphorylation, cell shape changes, and cell migration were prevented by C3, an inhibitor of the small guanosine triphosphatase (GTPase) protein, rho. Activation of rho is associated with formation of stress ﬁbers and loss of junctional integrity in most studies of cultured endothelial cells, and such studies show an association with E-cadherin and VE-cadherin (99,100). However, inhibition of rho and rho kinase by Adamson et al. (101) did not ablate the transient permeability increase induced by bradykinin and platelet-activating factor in intact microvessels. Another small GTPase, Src, may also have a signiﬁcant role because Src can alter actinomyocin contraction, cytoskeletal organization, and focal adhesion integrity by phosphorylation of MLCK, cortactin, and FAK (52). Both Ca2þ entry and Src activation were required for cell shape changes to occur in endothelial cells during cyclical stretch (102,103). Mechanical activation of PI3K activity may also contribute to overdistention injury because this activity is increased in endothelial cells by shear stress (104). Downstream targets of PI3K include the kinases, Akt, extracellular signal-regulated kinase (ERK) 1/2, and protein kinase C, which may phosphorylate proteins related to cytoskeletal tension and glycogen synthase 3b, the enzyme that can phosphorylate b-catenin, a regulatory protein of the adherens junction (52). Uhlig et al. (105) recently demonstrated that inhibition of PI3K prevented nuclear factor kappa B (NF-jB) activation and cytokine mRNA induction in alveolar macrophages in intact mouse and rat lung after high volume ventilation. In other isolated lung studies, venular endothelial cells produced increased levels of nitric oxide (NO) in response to increased tidal volume or increased venous pressure (106). NO release was blocked by inhibition of PI3K, and involved increased phosphorylation of Akt. Endothelial cells acutely mobilized from isolated rat lungs demonstrated increased phosphorylation of FAK and paxillin after high tidal volume ventilation that could be prevented by the

80

Parker et al.

tyrosine kinase inhibitor, genistein (107). Increased phosphorylation of FAK and b-catenin are generally associated with increased vascular permeability (108). In isolated rat lung studies, inhibition of tyrosine kinase with genistein attenuated high-pressure ventilation–induced increases in ﬁltration coefﬁcient, whereas inhibition of protein tyrosine phosphatase with phenyl arsine oxide (PAO) greatly augmented the increase in capillary ﬁltration coefﬁcient and increased vascular hemorrhage (Fig. 6) (110). Thus, mechanical strain causes phosphorylation of many proteins associated with a decrease in cell adhesion and increased permeability but the precise pathways that quantitatively contribute to VILI have not yet been elucidated. Cytoskeletal Alterations

An increase in cytoskeletal tension can oppose adherens and focal adhesion tethering to produce gap formation and increase permeability of endothelial monolayers exposed to a variety of mediators (44). Cyclical strain of endothelial cells increases phosphorylation of many intracellular proteins that can produce an increase in cell tension and shape change and may contribute to loss of barrier function. Birukov et al. (111) observed a strain-dependent phosphorylation of MLC in cultured endothelial cells. Cytoskeletal remodeling and MLC phosphorylation during shear stress were MLCK- and rho-dependent (112). Other studies of endothelial cells

Figure 6 Filtration coefﬁcients from isolated rat lungs ventilated with low PIP (7 cmH2O) or high PIP (35 cmH2O) showing the effects of tyrosine phosphatase inhibition with PAO or tyrosine kinase inhibition with genistein. Key: , P < 0.05 versus low PIP groups; , P < 0.05 versus all other groups. Abbreviations: PIP, peak inﬂation pressure; PAO, phenyl arsine oxide. Source: From Ref. 109.

Acute Passive and Active Changes in Microvascular Permeability

81

during cyclic strain showed formation of stress ﬁbers, as well as phosphorylation of proteins in mitogen-activated protein (MAP) kinase pathways (113,114). Small GTPase proteins such as Rho and Src have been implicated in the cytoskeletal remodeling and shape change induced by cyclical strain (102,103). The direct contribution of increased cytoskeletal tension to junctional paracellular permeability is uncertain because an increase in cytoskeletal motility could not overcome enhanced junctional stabilization by transfection of endothelial cells with Ca2þ-activated adenylyl cyclase (115). Alternatively, an increased actin–myosin tension can also markedly affect channel conductance and Ca2þ entry in endothelial cells (77,116). In isolated rat lungs, inhibition of MLCK attenuated the high PIP ventilation–induced increase in ﬁltration coefﬁcient (43). Histologic evidence of VILI was decreased and survival was also enhanced in endotoxin-treated MLCK knockout mice compared to wild type controls after mechanical ventilation (Figs. 7 and 8) (117). Thus, an increase in endothelial cytoskeletal tension contributes signiﬁcantly to the vascular permeability response in lungs subjected to high volume ventilation either through retraction of endothelial cell borders or an increased Ca2þ entry. C. Modulators of Permeability Cyclic Nucleotide Modulation of Mechanical Injury

The intracellular level of cAMP is a critical modulator of endothelial junctional permeability, and strategies that increase cAMP by activating adenylyl cyclase or inhibiting phosphodiesterases reverse increases in permeability produced by numerous receptor agonists (48). cAMP increases junctional

Figure 7 Survival data from groups of WT mice and nonmuscle MLCK null mice following endotoxin administration and mechanical ventilation. Abbreviations: LPS, lipopolysaccharide; WT, wild type. Source: From Ref. 117.

82

Parker et al.

Figure 8 Filtration coefﬁcients from isolated rats’ lungs ventilated at low and high PIP showing the attenuating effect of MLCK inhibition using ML-7 or an increased intracellular cyclical AMP induced by treatment with isoproterenol and rolipram. Key: #, P < 0.05 versus low PIP controls and high PIP treated groups. Abbreviations: AMP, adenosine monophosphate; MLCK, myosin light chain kinase; PIP, peak inﬂation pressure. Source: From Ref. 43.

strand number, stabilizes junctions, and inhibits MLCK activity and cell retraction (118–120). The myriad cell functions mediated by cAMP obtain speciﬁcity by localization of adenylyl cyclases and phosphodiesterases on scaffolding A kinase anchoring proteins (121). Recent studies indicate that only increases in cAMP localized to junctional microdomains actually reduce permeability (77). Junctional localization of type VI Ca2þ inhibited adenylyl cyclase and cAMP-speciﬁc phosphodiesterases appear to localize these actions (122). Complexes of A kinase anchoring proteins with cytoskeletal elements and small GTPases may likewise localize cAMP actions on cytoskeletal remodeling and tension (121). Effects of cAMP on Ca2þ entry channels and Ca2þ extrusion pumps also can modulate permeability responses (77). In isolated perfused rat lungs, attenuation by b agonists of vascular pressure stress induced increases in ﬁltration coefﬁcient but not hemorrhage, indicating that cAMP can modulate vascular permeability even in the presence of severe mechanical injury (42). Likewise, increases in ﬁltration coefﬁcient and edema formation after high PIP ventilation were reversed in isolated rat lungs by pretreatment with isoproterenol and rolipram, a phosphodiesterase IV inhibitor (43). Although levels of cAMP are modestly increased in intact lungs during distention and slightly increased in endothelial cells during cyclical stretch (123–125), control of permeability during mechanical stress appears to be dominated by Ca2þ entry rather than cAMP. Inhibition of protein kinase A did not prevent the permeability increase in isolated rat lungs after high

Acute Passive and Active Changes in Microvascular Permeability

83

PIP ventilation, or shape changes of endothelial cells in monolayers during cyclical stretch which indicates dependence on intracellular Ca2þ rather than cAMP levels (43,126). In contrast to the effect of cAMP, cGMP levels produce variable permeability effect, which are vascular bed dependent (48). Unlike cAMP, exogenous cGMP was unable to critically phosphorylate MLCK and reduce its activity in endothelial monolayers and may have different effects on permeability depending upon the intracellular Ca2þ concentration (127,128). However, increases in pulmonary vascular pressures produced increases in NO production by endothelial cells in isolated perfused lungs (106), and inhibition of NO production signiﬁcantly attenuated lung vascular protein leakage in intact rats and increased ﬁltration coefﬁcients in isolated rabbit lungs following high volume ventilation (129,130). Interactions of Mechanical Stress and Growth Factor

Growth factors can acutely impact VILI through a complex interaction with mechanical stress that results from the signal convergence of phosphorylation events induced by growth factor receptors and mechanical stress on integrins through common signaling pathways (131). In particular, both growth factors and mechanical stress can induce phosphorylation of FAK, and mechanical stress can cause transactivation and phosphorylation of epithelial growth factor receptors (131–133). Growth factor receptor kinases and FAK can both interact with Src and rho, which in turn may alter downstream pathways affecting vascular permeability (52). The initiation of mechanical injury can induce endogenous growth factors for subsequent lung repair, which may either increase vascular permeability or protect against injury. Ventilation of rats with high tidal volumes for two hours increased serum levels of vascular endothelial growth factor (VEGF) (130). VEGF is well established as a mediator of increased vascular permeability in many vascular beds where its effect is critically dependent upon calcium entry and PLC activation but not calcium store release (134). VEGF can phosphorylate signal pathways such as Src, PLC, and PI3K, which results in increased phosphorylation of myosin light chains, catenins, and FAK with subsequent reduced cell tethering and increased vascular permeability (52). In contrast, epithelium-speciﬁc growth factors such as keratinocyte growth factor (KGF), ﬁbroblast growth factor, and hepatocye growth factor have impressive protective effects against a variety of injurious stimuli. These growth factors have heparin-binding capability and an increased expression during injury that facilitates wound healing (135). KGF accelerated wounded closure and reduced susceptibility to mechanical deformation injury in airway epithelial monolayers during cyclical stretch (136,137). Repair of epithelial injury by KGF occurs by nonmitogenic and proliferative mechanisms (138). Finally, pretreatment with KGF attenuated lung vascular permeability and edema effects and reduced histologic evidence of injury following high volume ventilation in intact rats (139,140).

84

Parker et al. Activation of Metalloproteinases by Mechanical Strain

Proteases have long been known to affect vascular permeability (141). Neutrophil elastase and caspases have been shown to degrade adherens junction proteins and increase vascular permeability to ﬂuid and protein (142). There is mounting evidence that the zinc-dependent matrix metalloproteinases (MMPs) contributed signiﬁcantly to vascular permeability not only by their ability to degrade interstitial matrix but also by their action to degrade adherens and tight junction proteins such as VE-cadherin, betacatenin, and occludins (143,144). MMP activity is regulated at the gene transcription level for synthesis of inactive zymogens, by proteolytic activation of these zymogens, and by interaction of MMPs with tissue inhibitors of metalloproteinases (145). MMP synthesis and activation can be stimulated by several growth factors and cytokines (146). Inactive MMPs are present in the walls of blood vessel and can be activated by oxidants, vessel distention, or increased blood ﬂow (145,147). Membrane type 1 MMP (MT1-MMP) in endothelial cells can also activate proMMP-2. Cyclical strain of pulmonary microvascular endothelial cells caused an increased release and activation of MMP-2 and MMP-1 through an MT1MMP–dependent mechanism (148). These effects were inhibited by the MMP inhibitor, Prinomastat. Cyclical stretch of vascular smooth muscle cells increased expression of MMP-2 mRNA, which was blunted in cells lacking the nicotinamide adenine dinucleotide phosphate oxidase subunit p47phox (149). FAK was also found to have a pivotal role in MMP-2 and MMP-9 secretion and activation, which further indicates a signiﬁcant link of MMP activation to mechanical stress (150). In contrast, MMP-9 is only weakly expressed in endothelial cells but abundantly produced by macrophages (151). Foda et al. (152) observed a marked upregulation of mRNA for MT-MMP1, extracellular MMP inducer, MMP-2, and MMP9 in rat lungs ventilated at high and low volumes. Pretreatment with the MMP inhibitor, Prinomastat, attenuated the lung injury and MMP activity. Control of MMP activity is highly dependent upon G-proteins and tyrosine phosphorylation events (147), because E-cadherin proteolysis appears to require activation of Rac1 (153). Tyrosine phosphorylation– dependent increases in vascular permeability appear to be MMP dependent because permeability induced by the tyrosine phosphatase inhibitor PAO could be blocked by a MMP-2 inhibitor (144). The interaction of high-pressure ventilation–induced injury and tyrosine phosphorylation has been demonstrated in isolated rat lungs; PAO treatment greatly enhanced protein leak and hemorrhage in high volume ventilated lungs but had no signiﬁcant effect on low volume ventilated lungs (109). MMPs have also been demonstrated to associate with focal adhesion and tight junction proteins such as occludins and tyrosine kinases, the epithelial growth factor receptor, Src and Fer, as well as tyrosine phosphatases (147). Thus, MMP activation

Acute Passive and Active Changes in Microvascular Permeability

85

possibly may be a ﬁnal pathway for the release of cell tethering from the interstitial matrix and adjacent cells, which initiates mechanical stress– induced vascular permeability increases. Interaction of Endothelium with Epithelium, Macrophages and Neutrophils

Overdistention of the lung produces mechanical stress on many cell types including epithelial, endothelial, ﬁbroblast, myoepithelial and smooth muscle cells, in addition to resident alveolar macrophages and marginating neutrophils. The interactions of these cell types determine the overall state of cell activation and results in the observed increase in vascular permeability and cytokine responses. Alveolar type I and II epithelial cells communicate directly via gap junctions or indirectly through nucleotide release and purinergic receptors (154), and mechanical stretch of the alveolus sends calcium waves throughout the alveolus (50). Communication of these Ca2þ waves from type I to type II epithelial cells controlled surfactant exocytosis after alveolar distention (72). Kuebler et al. (73) injected single alveoli with tumor necrosis factor and observed the spread of a Ca2þ signal between epithelial cells and adjacent endothelial cells with activation of PLA2 and increased expression of P-selectin, but communication between epithelial and endothelial cells was not dependent upon transmission through gap junctions. Injurious ventilation is also associated with recruitment and activation of neutrophils (155,156). However, the initial rapid increase in intravascular permeability following VILI may depend upon interactions of endothelial cells with marginating neutrophils prior to signiﬁcant neutrophil recruitment. Bhattacharya et al. (107) observed marked increases in FAK and paxillin tyrosine phosphorylation in lung endothelial cells after high volume ventilation when neutrophils were present compared to their absence, even though the neutrophils were not activated. Activation of alveolar macrophages also appears to have a critical role in VILI because depletion of macrophages markedly attenuated vascular permeability increases following high volume ventilation (15,157). Mechanical stretch of cultured macrophages produced IL-8 and MMP-9 and enhanced their cytokine response to lipopolysaccharide challenge (158,159). High volume lung distention also induced activation of the NF-jB pathways in macrophages and alveolar type II cells, as well as the PI3K pathway in macrophage, epithelial cells, and endothelial cells (105– 107). Inhibition of PI3K reduced cytokine production and the permeability response in isolated perfused mouse lungs (105). Cyclical stretch of ﬁbroblasts has been implicated in hyaluronan and growth factor production, which may also contribute to macrophage and endothelial cell activation (160,161). Neurokinins secreted by nerve endings during high volume mechanical stress are other factors which may contribute to the vessel

86

Parker et al.

permeability affect (162). Thus, the interaction of multiple cell types will determine the ﬁnal pathways that modulate vessel permeability, but the quantitative contribution of each pathway and cell type to the ﬁnal injury remains to be discovered. IV. Conclusion VILI is a multifactorial injury, but pharmacologic targets may be available for prevention. The initiation of the acute permeability response appears to be the most promising phase for intervention. The signal events such as stretch-activation of ion channels, mechanically induced tyrosine phosphorylation events that are matrix adhesion–dependent, or receptor-activated events are potential targets. Inhibition of potential signaling molecules such as PLA2, PI3K, NF-jB, or MAP kinases such as ERK 1/2, p38, or c-Jun N-terminal kinase may prove effective in preventing VILI or the systemic consequences of VILI (1). References 1. Pinhu L, Whitehead T, Evans T, Grifﬁths M. Ventilator-associated lung injury. Lancet 2003; 361:332–340. 2. Dos Santos CC, Slutsky AS. Invited review: mechanisms of ventilator-induced lung injury: a perspective. [see comments]. J Appl Physiol 2000; 89:1645–1655. 3. Dreyfuss D, Saumon G. Ventilator induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 4. Parker JC, Hernandez LA, Peevy K. Mechanisms of ventilator induced injury. Crit Care Med 1993; 21:131–143. 5. Yoshikawa S, King JA, Reynolds SD, Stripp BR, Parker JC. Time and pressure dependence of transvascular Clara cell protein, albumin, and IgG transport during ventilator-induced lung injury in mice. Am J Physiol 2004; 286:L604–L612. 6. Dreyfuss D, Ricard JD, Saumon G. On the physiologic and clinical relevance of lung-borne cytokines during ventilator-induced lung injury. Am J Respir Crit Care Med 2003; 167:1467–1471. 7. Ricard JD, Dreyfuss D, Saumon G. Production of inﬂammatory cytokines in ventilator-induced lung injury: a reappraisal. Am J Respir Crit Care Med 2001; 163:1176–1180. 8. Yoshikawa S, King J, Penton, Eyal FG, Parker JC. Acute ventilator induced vascular permeability and cytokine responses in isolated and in situ mouse lungs. J Appl Physiol 2004; 97:2190–2199. 9. Hernandez LA, Peevy KJ, Moise AA, Parker JC. Chest wall restriction limits high airway pressure-induced lung injury in young rabbits. J Appl Physiol 1989; 66:2364–2368. 10. Parker JC, Gillespie MN, Taylor AE, Martin SL. Capillary ﬁltration coefﬁcient, vascular resistance and compliance in isolated mouse lungs. J Appl Physiol 1999; 87:1421–1427.

Acute Passive and Active Changes in Microvascular Permeability

87

11. Parker JC, Townsley MI, Rippe B, Taylor AE, Thigpen J. Increased microvascular permeability in dog lungs due to high peak airway pressures. J Appl Physiol 1984; 57:1809–1816. 12. Parker JC, Yoshikawa S. Vascular segmental permeabilities at high peak inﬂation pressure in isolated rat lungs. Am J Physiol 2002; 283:L1203–L1209. 13. Dreyfuss D, Soler P, Saumon G. Spontaneous resolution of pulmonary edema caused by short periods of cyclic overinﬂation. J Appl Physiol 1992; 72:2081–2089. 14. Dreyfuss D, Basset G, Soler P, Saumon G. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 1985; 132:880–884. 15. Eyal FG, Hamm CR, Coker-Flowers PJ, Stober M, Parker JC. The neutralization of alveolar macrophages reduces barotrauma induced lung injury [Abstract]. FASEB J 2002; 16(4):A410. 16. Quinn DA, Moufarrej RK, Volokhov A, Hales CA. Interactions of lung stretch, hyperoxia, and MIP-2 production in ventilator-induced lung injury. J Appl Physiol 2002; 93:517–525. 17. Parker JC, Townsley MI. Evaluation of lung injury in rats and mice. Am J Physiol 2004; 286:L231–L246. 18. John TA, Vogel SM, Minshall RD, Ridge K, Tiruppathi C, Malik AB. Evidence for the role of alveolar epithelial gp60 in active transalveolar albumin transport in the rat lung. J Physiol 2001; 533:547–559. 19. Schnitzer JE, Oh P. Albondin-mediated capillary permeability to albumin. Differential role of receptors in endothelial transcytosis and endocytosis of native and modiﬁed albumins. J Biol Chem 1994; 269:6072. 20. Rippe B, Rosengren BI, Carlsson O, Venturoli D. Transendothelial transport: the vesicle controversy. J Vascul Res 2002; 39:375–390. 21. Parker JC, Parker RE, Granger DN, Taylor AE. Vascular permeability and transvascular ﬂuid and protein transport in the dog lung. Circ Res 1981; 48:561. 22. Egan EA, Nelson RM, Oliver RE. Lung inﬂation and alveolar permeability to nonelectrolytes in the adult sheep in vivo. J Physiol (Lond) 1976; 260:409–424. 23. Hermans C, Bernard A. Lung epithelium-speciﬁc proteins. Characteristics and potential applications as markers. Am J Respir Crit Care Med 1999; 159:646–678. 24. Parker JC, Drake DA, King JA, Yoshikawa S. Ventilator induced lung injuryLeak sites identiﬁed using methacrylate vascular casts [Abstract]. Am Rev Respir Crit Care Med 2003; 167(7):A775. 25. Muscedere JG, Mullen JB, Gan K, Slutsky AS. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med 1994; 149:1327–1334. 26. Uhlig S, Uhlig U. Pharmacologic interventions in ventilator induced lung injury. Trends Pharmacol Sci 2004; 25:592–600. 27. Staub NC, Nagano H, Pearce ML. Pulmonary edema in dogs: especially the sequence of ﬂuid accumulation in the lungs. J Appl Physiol 1967; 22:227–240. 28. Yoneda K. Fluid-overload pulmonary oedema in mice: the intercellular junctions of bronchiolar epithelium and arterial endothelium. Br J Exp Pathol 1983; 64:215–224.

88

Parker et al.

29. West JB, Mathieu-Costello O. Strength of the pulmonary blood-gas barrier. Respir Physiol 1992; 88:141–148. 30. Parker JC, Allison RC, Taylor AE. Edema affects intra-alveolar ﬂuid pressures and interdependence in dog lungs. J Appl Physiol Respir Environ Exer Physiol 1981; 51:911–921. 31. Smith JC, Mitzner W. Analysis of pulmonary vascular interdependence in excised dog lobes. J Appl Physiol 1980; 48:450–467. 32. West JB, Mathieu Costello O, Jones JH, et al. Stress failure of pulmonary capillaries in racehorses with exercise-induced pulmonary hemorrhage. J Appl Physiol 1993; 75:1097–1109. 33. West JB. Invited review: pulmonary capillary stress failure. J Appl Physiol 2000; 89:2483–2489. 34. West JB. Thoughts on the pulmonary blood-gas barrier. Am J Physiol Lung Cell Mol Physiol 2003; 285:L501–L513. 35. Costello ML, Mathieu CO, West JB. Stress failure of alveolar epithelial cells studied by scanning electron microscopy. Am Rev Respir Dis 1992; 145:1446–1455. 36. Fu Z, Costello ML, Tsukimoto K, et al. High lung volume increases stress failure in pulmonary capillaries. J Appl Physiol 1992; 73:123–133. 37. Tsukimoto K, Mathieu Costello O, Prediletto R, Elliott AR, West JB. Ultrastructural appearances of pulmonary capillaries at high transmural pressures. J Appl Physiol 1991; 71:573–582. 38. Elliott AR, Fu Z, Tsukimoto K, Prediletto R, Mathieu Costello O, West JB. Short-term reversibility of ultrastructural changes in pulmonary capillaries caused by stress failure. J Appl Physiol 1992; 73:1150–1158. 39. Neal CR, Michel CC. Openings in frog microvascular endothelium induced by high intravascular pressures. J Physiol (Lond) 1996; 492:39–52. 40. Maron MB, Fu Z, Mathieu-Costello O, West JB. Effect of high transcapillary pressures on capillary ultrastructure and permeability coefﬁcients in dog lung. J Appl Physiol 2001; 90:638–648. 41. Rippe B, Townsley M, Thigpen J, Parker JC, Korthuis RJ, Taylor AE. Effects of vascular pressure on the pulmonary microvasculature in isolated dog lungs. J Appl Physiol 1984; 57:233–239. 42. Parker JC, Ivey CL. Isoproterenol attenuates high vascular pressure induced permeability increases in isolated rat lungs. J Appl Physiol 1998; 83:1962–1967. 43. Parker JC. Inhibitors of myosin light chain kinase, phosphodiesterase and calmodulin attenuate ventilator induced lung injury. J Appl Physiol 2000; 89:2241–2248. 44. Dudek SM, Garcia JGN. Cytoskeletal regulation of pulmonary vascular permeability. J Appl Physiol 2001; 91:1487–1500. 45. Garcia JG, Schaphorst KL. Regulation of endothelial cell gap formation and paracellular permeability. J Invest Med 1995; 43:117–126. 46. Schaphorst KL, Pavalko FM, Patterson CE, Garcia JG. Thrombin- mediated focal adhesion plaque reorganization in endothelium: role of protein phosphorylation. Am J Respir Cell Mol Biol 1997; 17:443–455. 47. Safdar Z, Wang P, Ichimura H, Issekutz AC, Quadri S, Bhattacharya J. Hyperosmolarity enhances the lung capillary barrier. J Clin Invest 2003; 112:1541–1549.

Acute Passive and Active Changes in Microvascular Permeability

89

48. Michel CC, Curry FE. Microvascular permeability. Phys Rev 1999; 79:703–761. 49. Mitic LL, Anderson JM. Molecular architecture of tight junctions. Ann Rev Physiol 1998; 60:121–142. 50. Liu M, Tanswell AK, Post M. Mechanical force-induced signal transduction in lung cells. Am J Physiol 1999; 277:L667–L683. 51. Del Maschio A, Zanetti A, Corada M, et al. Polymorphonuclear leukocyte adhesion triggers the disorganization of endothelial cell-to-cell adherens junctions. J Cell Biol 1996; 135:497–510. 52. Bogatcheva NV, Garcia JG, Verin AD. Role of tyrosine kinase signaling in endothelial cell barrier regulation. Vascul Pharmacol 2002; 39:201–212. 53. Bates DO, Harper SJ. Regulation of vascular permeability by vascular endothelial growth factors. Vascul Pharmacol 2002; 39:225–237. 54. Vepa S, Scribner WM, Parinandi NL, English D, Garcia JG, Natarajan V. Hydrogen peroxide stimulates tyrosine phosphorylation of focal adhesion kinase in vascular endothelial cells. Am J Physiol 1999; 277:L150–L158. 55. Garcia JG, Davis HW, Patterson CE. Regulation of endothelial cell gap formation and barrier dysfunction: role of myosin light chain phosphorylation. J Cell Physiol 1995; 163:510–522. 56. Verin AD, Gilbert-McClain LI, Patterson CE, Garcia JG. Biochemical regulation of the nonmuscle myosin light chain kinase isoform in bovine endothelium. Am J Respir Cell Mol Biol 1998; 19:767–776. 57. Garcia JG, Verin AD, Schaphorst K, et al. Regulation of endothelial cell myosin light chain kinase by Rho, cortactin, and p60(src). Am J Physiol 1999; 276:L989–L998. 58. Watanabe H, Takahashi R, Zhang XX, et al. An essential role of myosin lightchain kinase in the regulation of agonist- and ﬂuid ﬂow-stimulated Ca2þ inﬂux in endothelial cells. FASEB J 1998; 12:341–348. 59. Boitano S, Safdar Z, Welsh DG, Bhattacharya J, Koval M. Cell-cell interactions in regulating lung function. Am J Physiol 2004; 287:L455–L459. 60. Kuebler WM, Ying X, Bhattacharya J. Pressure-induced endothelial Ca(2þ) oscillations in lung capillaries. Am J Physiol Lung Cell Mol Physiol 2002; 282:L917–L923. 61. Stevens T, Rosenberg R, Aird W, et al. NHLBI workshop report: endothelial cell phenotypes in heart, lung, and blood diseases. Am J Physiol Cell Physiol 2001; 281:C1422–C1433. 62. Tiruppathi C, Minshall RD, Paria BC, Vogel SM, Malik AB. Role of Ca2þ signaling in the regulation of endothelial permeability. Vascul Pharmacol 2002; 39:173–185. 63. Yuan SY. Protein kinase signaling in the modulation of microvascular permeability. Vascul Pharmacol 2002; 39:213–223. 64. Lindner K, Uhlig S. Differential effects of pro-inﬂammatory mediators on vascular permeability in endothelial monolayers and intact rat lungs [Abstract]. Am J Respir Crit Care Med 2004; 169(7):A156. 65. Parker JC, Ivey C, Tucker A. Gadolinium prevents high airway pressure induced permeability increases in isolated rat lungs. J Appl Physiol 1998; 84:1113–1118. 66. Davies PF. Flow mediated endothelial mechanotransduction. Phys Rev 1995; 75:519–560.

90

Parker et al.

67. Kuebler WM, Ying X, Singh B, Issekutz AC, Bhattacharya J. Pressure is proinﬂammatory in lung venular capillaries. J Clin Invest 1999; 104: 495–502. 68. Naruse K, Sokabe M. Involvement of stretch-activated ion channels in Ca2þ mobilization to mechanical stretch in endothelial cells. Am J Physiol 1993; 264(Pt 1):C1037–C1044. 69. Evans L, Frenkel L, Brophy CM, et al. Activation of diacylglycerol in cultured endothelial cells exposed to cyclic strain. Am J Physiol 1997; 272:C650–C656. 70. Oike M, Droogmans G, Nilius B. Mechanosensitive Ca2þ transients in endothelial cells from human umbilical vein. Proc Natl Acad Sci USA 1994; 91:2940–2944. 71. Rosales OR, Isales CM, Barrett PQ, Brophy C, Sumpio BE. Exposure of endothelial cells to cyclic strain induces elevations of cytosolic Ca2þ concentration through mobilization of intracellular and extracellular pools. Biochem J 1997; 326:385–392. 72. Ashino Y, Ying X, Dobbs LG, Bhattacharya J. [Ca(2þ)](i) oscillations regulate type II cell exocytosis in the pulmonary alveolus [see comments]. Am J Physiol Lung Cell Mol Physiol 2000; 279:L5–L13. 73. Kuebler WM, Parthasarathi K, Wang PM, Bhattacharya J. A novel signaling mechanism between gas and blood compartments of the lung [see comments] J Clin Invest 2000; 105:905–913. [erratum appears in J Clin Invest 2000; 106(4):607]. 74. Nilius B, Droogmans G. Ion channels and their functional role in vascular endothelium. Physiol Rev 2001; 81:1415–1459. 75. Strotmann R, Harteneck C, Nunnenmacher K, Schultz, Plant TD. OTRPC4, a nonselective cation channel that confers sensitivity to extracellular osmolarity. Nature Cell Biol 2000; 2:695–702. 76. Tiruppathi C, Freichel M, Vogel SM, et al. Impairment of store-operated Ca2þ entry in TRPC4(-/-) mice interferes with increase in lung microvascular permeability [see comments]. Circ Res 1912; 91:70–76. 77. Ciofﬁ DL, Wu S, Stevens T. On the endothelial cell I (SOC). Cell Calcium 2003; 33:323–336. 78. Nilius B, Droogmans G, Wondergem R. Transient receptor potential channels in endothelium: solving the calcium entry puzzle? Endothelium 2003; 10:5–15. 79. Watanabe H. Activation of TRPV4 channels (hVRL-2/mTRP12) by phorbol derivatives. J Biol Chem 2002; 277:13569–13577. 80. Watanabe H. Heat-evoked activation of TRPV4 channels in an HEK293 cell expression system and in native mouse aorta endothelial cells. J Biol Chem 2002; 277:47044–47051. 81. Watanabe H, Vriens J, Prenen J, Droogmans G, Voets T, Nilius B. Anandamide and arachidonic acid use epoxyeicosatrienoic acids to activate TRPV4 channels. Nature 1924; 424:434–438. 82. Alvarez DF, Gjerde EA, Townsley MI. Role of EETs in regulation of endothelial permeability in rat lung. Am J Physiol 2004; 286:L445–L451. 83. Yoshikawa S, Miyahara T, Reynolds SD, et al. Clara cell secretory protein and phospholipase A2 activity modulate acute ventilator induced lung injury in mice. J Appl Physiol. 2005; 98:1264–1271.

Acute Passive and Active Changes in Microvascular Permeability

91

84. Kohler R, Brakemeier S, Kuhn M, et al. Expression of ryanodine receptor type 3 and TRP channels in endothelial cells: comparison of in situ and cultured human endothelial cells. Cardiovas Res 2001; 51:160–168. 85. Suh SH, Vennekens R, Manolopoulos VG, et al. Characterisation of explanted endothelial cells from mouse aorta: electrophysiology and Ca2þ signalling. Pﬂugers Arch Eur J Physiol 1999; 438:612–620. 86. Anghelescu M, Tarpey SB, Miyahara K, Miller G, Penton AM, Parker JC. Differential expression of potassium channels in lung endothelial phenotypes [Abstract]. FASEB J 2004. 87. Hoger JH, Ilyin VI, Forsyth S, Hoger A. Shear stress regulates the endothelial Kir2.1 ion channel. Proc Natl Acad Sci 2002; 99:7780–7785. 88. Chatterjee S, Al Mehdi AB, Levitan I, Stevens T, Fisher AB. Shear stress increases expression of a KATP channel in rat and bovine pulmonary vascular endothelial cells. Am J Physiol Cell Physiol 2003; 285:C959–C967. 89. Brakemeier S, Eichler I, Hopp H, Kohler R, Hoyer J. Up-regulation of endothelial stretch-activated cation channels by ﬂuid shear stress [comment]. Cardiovasc Res 2002; 53:209–218. 90. Brakemeier S, Kersten A, Eichler I, et al. Shear stress-induced up-regulation of the intermediate-conductance Ca(2þ)-activated K(þ) channel in human endothelium. Cardiovasc Res 2001; 60:488–496. 91. Shimoda LA, Welsh LE, Pearse DB. Inhibition of inwardly rectifying K(þ) channels by cGMP in pulmonary vascular endothelial cells. Am J Physiol Lung Cell Mol Physiol 2002; 283:L297–L304. 92. Pearse DB, Wagner EM, Permutt S. Effect of ventilation on vascular permeability and cyclic nucleotide concentrations in ischemic sheep lungs. J Appl Physiol 1999; 86:123–132. 93. Nilius B, Droogmans G. Amazing chloride channels: an overview. Acta Physiol Scand 2003; 177:119–147. 94. Gajic O, Lee J, Doerr CH, Berrios JC, Myers JL, Hubmayr RD. Ventilatorinduced cell wounding repair in the intact lung. Am J Respir Crit Care Med 2003; 167:1057. 95. Parker JC, Miller GT, Tarpey SB, Anghelescu M, Penton AM, Adkison JB. Cell wounding modulates junctional proteins in rat pulmonary artery and microvascular endothelial cells [Abstract]. Am J Respir Crit Care Med 2004; 169:A158. 96. Vlahakis NE, Hubmayr RD. Response of alveolar cells to mechanical stress. Curr Opin Crit Care 2003; 9:2–8. 97. Yano Y, Geibel J, Sumpio BE. Tyrosine phosphorylation of pp125FAK and paxillin in aortic endothelial cells induced by mechanical strain. Am J Physiol 1996; 271:C635–C649. 98. Yano Y, Saito Y, Narumiya S, Sumpio BE. Involvement of rho p21 in cyclic strain-induced tyrosine phosphorylation of focal adhesion kinase (pp125FAK), morphological changes and migration of endothelial cells. Biochem Biophys Res Commun 1996; 224:508–515. 99. Braga VM, Del M, Machesky L, Dejana E. Regulation of cadherin function by Rho and Rac: modulation by junction maturation and cellular context. Mol Biol Cell 1999; 10:9–22.

92

Parker et al.

100. Wojciak-Stothard B, Ridley AJ. Rho GTPases and the regulation of endothelial permeability. Vascul Pharmacol 2002; 39:187–199. 101. Adamson RH, Curry FE, Adamson G, et al. Rho and rho kinase modulation of barrier properties: cultured endothelial cells and intact microvessels of rats and mice. J Physiol Online 2002; 539:295–308. 102. Naruse K, Sai X, Yokoyama N, Sokabe M. Uni-axial cyclic stretch induces c-src activation and translocation in human endothelial cells via SA channel activation. FEBS Lett 1998; 441:111–115. 103. Sokabe M, Naruse K, Sai S, et al. Mechanotransduction and intracellular signaling mechanisms of stretch-induced remodeling in endothelial cells. Heart Vessels Suppl 1997:12–13. 104. Li S, Chen BP, Azuma N, et al. Distinct roles for the small GTPases Cdc42 and Rho in endothelial responses to shear stress. J Clin Invest 1999; 103:1141–1150. 105. Uhlig U, Fehrenbach H, Lachmann RA, et al. Phosphoinositide 3-OH kinase inhibition prevents ventilation-induced lung cell activation. Am J Respir Crit Care Med 2004; 169:201–208. 106. Kuebler WM, Uhlig U, Goldmann T, et al. Stretch activates nitric oxide production in pulmonary vascular endothelial cells in situ. Am J Respir Crit Care Med 2001; 168:1391–1398. 107. Bhattacharya S, Sen N, Yiming MT, et al. High tidal volume ventilation induces proinﬂammatory signaling in rat lung endothelium. Am J Respir Cell Mol Biol 2003; 28:218–224. 108. Yuan SY. Signal transduction pathways in enhanced microvascular permeability. Microcirculation 2000; 7:395–403. 109. Parker JC, Ivey CL, Tucker A. Phosphotyrosine phosphatase and tyrosine inhibition modulate airway pressure induced lung injury. J Appl Physiol 1998; 85:1753–1761. 110. Parker JC. Phosphotyrosine phosphatase inhibition increases susceptibility of rat lungs to airway pressure induced injury. Am Rev Respir Crit Care Med 1997; 155:A88. 111. Birukov KG, Jacobson JR, Flores AA, et al. Magnitude-dependent regulation of pulmonary endothelial cell barrier function by cyclic stretch. Am J Physiol 2003; 285:L785–L797. 112. Birukov KG, Birukova AA, Dudek SM, et al. Shear stress-mediated cytoskeletal remodeling and cortactin translocation in pulmonary endothelial cells. Am J Respir Cell Mol Biol 2002; 26:453–464. 113. Iba T, Sumpio BE. Morphological response of human endothelial cells subjected to cyclic strain in vitro. Microvasc Res 1991; 42:245–254. 114. Ikeda M, Takei T, Mills I, Kito H, Sumpio BE. Extracellular signal-regulated kinases 1 and 2 activation in endothelial cells exposed to cyclic strain. Am J Physiol 1999; 276:1–22. 115. Ciofﬁ DL, Moore TM, Schaack J, Creighton JR, Cooper DM, Stevens T. Dominant regulation of interendothelial cell gap formation by calcium-inhibited type 6 adenylyl cyclase. J Cell Biol 1924; 157:1267–1278. 116. Oike M, Schwarz G, Sehrer J, et al. Cytoskeletal modulation of the response to mechanical stimulation in human vascular endothelial cells. Pﬂugers Arch Eur J Physiol 1994; 428:569–576.

Acute Passive and Active Changes in Microvascular Permeability

93

117. Wainwright MS, Rossi J, Schavocky J, et al. Protein kinase involved in lung injury susceptibility: evidence from enzyme isoform genetic knockout and in vivo inhibitor treatment. Proc Natl Acad Sci 2003; 100:6233–6238. 118. Adamson RH, Liu B, Fry GN, Rubin LL, Curry FE. Microvascular permeability and number of tight junctions are modulated by cAMP [Generic]. Am J Physiol Lung Cell Mol Physiol 1998; 274:H1885–H1894. 119. Lampugnani MG, Giorgi M, Gaboli M, Dejana E, Marchisio PC. Endothelial cell motility, integrin receptor clustering, and microﬁlament organization are inhibited by agents that increase intracellular cAMP. Lab Invest 1990; 63:521–531. 120. Moy AB, Bodmer JE, Blackwell K, Shasby S, Shasby DM. cAMP protects endothelial barrier function independent of inhibiting MLC20-dependent tension development. Am J Physiol 1998; 274:L1024–L1029. 121. Tasken K, Aandahl EM. Localized effects of cAMP mediated by distinct routes of protein kinase A. Phys Rev 2004; 84:137–167. 122. Stevens T, Nakahashi Y, Cornﬁeld DN, McMurtry IF, Cooper DM, Rodman DM. Ca(2þ)-inhibitable adenylyl cyclase modulates pulmonary artery endothelial cell cAMP content and barrier function. Proc Natl Acad Sci USA 1995; 92:2696–2700. 123. Cohen CR, Mills I, Du W, Kamal K, Sumpio BE. Activation of the adenylyl cyclase/cyclic AMP/protein kinase A pathway in endothelial cells exposed to cyclic strain. Exp Cell Res 1997; 231:184–189. 124. Iba T, Mills I, Sumpio BE. Intracellular cyclic AMP levels in endothelial cells subjected to cyclic strain in vitro. J Surg Res 1992; 52:625–630. 125. Russo LA, Rannels SR, Laslow KS, Rannels DE. Stretch-related changes in lung cAMP after partial pneumonectomy. Am J Physiol 1989; 257:E261–E268. 126. Yamada T, Naruse K, Sokabe M. Stretch-induced morphological changes of human endothelial cells depend on the intracellular level of Ca2þ rather than of cAMP. Life Sci 2000; 67:2605–2613. 127. Hempel A, Noll T, Muhs A, Piper HM. Functional antagonism between cAMP and cGMP on permeability of coronary endothelial monolayers. Am J Physiol 1996; 270(Pt 2):H1264–H1271. 128. Holschermann H, Noll T, Hempel A, Piper HM. Dual role of cGMP in modulation of macromolecule permeability of aortic endothelial cells. Am J Physiol 1997; 272:H91–H98. 129. Broccard AF, Feihl F, Vannay C, Markert M, Hotchkiss J, Schaller MD. Effects of L-NAME and inhaled nitric oxide on ventilator-induced lung injury in isolated, perfused rabbit lungs. Crit Care Med 2004; 32(9):1872–1878. 130. Choi WI, Quinn DA, Park KM, et al. Systemic microvascular leak in an in vivo rat model of ventilator-induced lung injury. Am J Respir Crit Care Med 2003; 167:1627. 131. Rozengurt E. Convergent signalling in the action of integrins, neuropeptides, growth actors and oncogenes. Cancer Surv 1995; 24:81–96. 132. Correa-Meyer E, Pesce L, Guerrero C, Sznajder JI. Cyclic stretch activates ERK1/2 via G proteins and EGFR in alveolar epithelial cells. Am J Physiol Lung Cell Mol Physiol 2002; 282:L883–L891. 133. Tschumperlin DJ, Dai G, Maly IV, et al. Mechanotransduction through growthfactor shedding into the extracellular space. Nature 2004; 429(6987):83–86.

94

Parker et al.

134. Bates DO, Hillman NJ, Williams B, Neal CR, Pocock TM. Regulation of microvascular permeability by vascular endothelial growth factors. J Anat 2002; 200:581–597. 135. Ware LB, Matthay MA. Keratinocyte and hepatocyte growth factors in the lung: roles in lung development, inﬂammation, and repair. Am J Physiol 2002; 282:L924. 136. Oswari J, Matthay MA, Margulies SS. Keratinocyte growth factor reduces alveolar epithelial susceptibility to in vitro mechanical deformation. Am J Physiol Lung Cell Mol Physiol 2001; 281:L1068–L1077. 137. Waters CM, Savla U. Keratinocyte growth factor accelerates wound closure in airway epithelium during cyclic mechanical strain. J Cell Physiol 1999; 181:424–432. 138. Atabai K, Ishigaki M, Geiser T, Ueki I, Matthay MA, Ware LB. Keratinocyte growth factor can enhance alveolar epithelial repair by nonmitogenic mechanisms. Am J Physiol 2002; 283:L163–L169. 139. Lesur O, Hermans C, Chalifour JF, et al. Mechanical ventilation-induced pneumoprotein CC-16 vascular transfer in rats: effect of KGF pretreatment. Am J Physiol Lung Cell Mol Physiol 2003; 284:L410–L419. 140. Welsh DA, Summer WR, Dobard EP, Nelson S, Mason CM. Keratinocyte growth factor prevents ventilator-induced lung injury in an ex vivo rat model. Am J Respir Crit Care Med 2000; 162:1081–1086. 141. Wang JG, Miyazu M, Matsushita E, Sokabe M, Naruse K. Uniaxial cyclic stretch induces focal adhesion kinase (FAK) tyrosine phosphorylation followed by mitogen-activated protein kinase (MAPK) activation. Biochem Biophys Res Commun 2001; 288:356–361. 142. Bannerman DD, Goldblum SE. Mechanisms of bacterial lipopolysaccharideinduced endothelial apoptosis. Am J Physiol Lung Cell Mol Physiol 2003; 284(6):L899–L914. 143. Herren B, Levkau B, Raines EW, Ross R. Cleavage of beta-catenin and plakoglobin and shedding of VE-cadherin during endothelial apoptosis: evidence for a role for caspases and metalloproteinases. Mol Biol Cell 1998; 9(6):1589–1601. 144. Wachtel M, Frei K, Ehler E, Fontana A, Winterhalter K, Gloor SM. Occludin proteolysis and increased permeability in endothelial cells through tyrosine phosphatase inhibition. J Cell Sci 1999; 112:4347–4356. 145. Galis ZS, Khatri JJ. Matrix metalloproteinases in vascular remodeling and atherogenesis: the good, the bad, and the ugly. Circ Res 2002; 90:251–262. 146. Parks WC, Shapiro SD. Matrix metalloproteinases in lung biology. Respir Res 2001; 2(1):10–19. 147. Alexander JS, Elrod JW. Extracellular matrix, junctional integrity and matrix metalloproteinase interactions in endothelial permeability regulation. J Anat 2002; 200:561–574. 148. Haseneen NA, Vaday GG, Zucker S, Foda HD. Mechanical stretch induces MMP-2 release and activation in lung endothelium: role of EMMPRIN. Am J Physiol 2003; 284:L541–L547. 149. Grote K, Flach I, Luchtefeld M, et al. Mechanical stretch enhances mRNA expression and proenzyme release of matrix metalloproteinase-2 (MMP-2) via NAD(P)H oxidase-derived reactive oxygen species. Circ Res 2003; 92:80e–886.

Acute Passive and Active Changes in Microvascular Permeability

95

150. Sein TT, Thant AA, Hiraiwa Y, et al. A role for FAK in the Concanavalin A-dependent secretion of matrix metalloproteinase-2 and -9. Oncogene 2000:5539–5542. 151. Longo GM, Xiong W, Greiner TC, Zhao Y, Fiotti N, Baxter BT. Matrix metalloproteinases 2 and 9 work in concert to produce aortic aneurysms. J Clin Invest 2002; 110:625–632. 152. Foda HD, Rollo EE, Drews M, et al. Ventilator-induced lung injury upregulates and activates gelatinases and EMMPRIN: attenuation by the synthetic matrix metalloproteinase inhibitor, Prinomastat (AG3340). Am J Respir Cell Mol Biol 2001:717–724. 153. Ito K, Okamoto I, Araki N, et al. Calcium inﬂux triggers the sequential proteolysis of extracellular and cytoplasmic domains of E-cadherin, leading to loss of beta-catenin from cell-cell contacts. Oncogene 1999; 18(50):7080–7090. 154. Koval M. Sharing signals: connecting lung epithelial cells with gap junction channels. Am J Physiol 2002; 283:L875–L893. 155. Imanaka H, Shimaoka M, Matsuura N, Nishimura M, Ohta N, Kiyono H. Ventilator-induced lung injury is associated with neutrophil inﬁltration, macrophage activation, and TGF-beta 1 mRNA upregulation in rat lungs. Anesth Analg 2000; 92:428–436. 156. Matsuoka T, Kawano T, Miyasaka K. Role of high-frequency ventilation in surfactant-depleted lung injury as measured by granulocytes. J Appl Physiol 1994; 76:539–544. 157. Matthay MA, Bhattacharya S, Gaver D, et al. Ventilator-induced lung injury: in vivo and in vitro mechanisms. Am J Physiol 2002; 283:L678–L682. 158. Dunn I, Pugin J. Mechanical ventilation of various human lung cells in vitro: identiﬁcation of the macrophage as the main producer of inﬂammatory mediators. Chest 1999; 116:95S–97S. 159. Pugin J, Dunn I, Jolliet P, et al. Activation of human macrophages by mechanical ventilation in vitro. Am J Physiol 1998; 275:L1040–L1050. 160. McKee CM, Lowenstein CJ, Horton MR, et al. Hyaluronan fragments induce nitric-oxide synthase in murine macrophages through a nuclear factor kappaB-dependent mechanism. J Biol Chem 1997; 272:8013–8018. 161. McKee CM, Penno MB, Cowman M, et al. (HA) fragments induce chemokine gene expression in alveolar macrophages. The role of HA size and CD44. J Clin Invest 1996; 98:2403–2413. 162. Chavolla-Calderon M, Bayer MK, Perez Fontan JJ. Bone marrow transplantation reveals an essential synergy between neuronal and hemopoietic cell neurokinin production in pulmonary inﬂammation. J Clin Invest 2003; 111:973–980.

This page intentionally left blank

5 Hemodynamic Interactions During Ventilator-Induced Lung Injury

JOHN J. MARINI and LEOPOLDO FERRER Division of Pulmonary and Critical Care Medicine, University of Minnesota, Regions Hospital St. Paul, Minnesota, U.S.A.

I. Introduction Numerous experimental studies demonstrate the potential for adverse patterns of mechanical ventilation to initiate acute lung injury (ALI), which is characterized by proteinaceous edema, inﬂammation, and hemorrhage. The great majority of these investigations into ventilator-induced lung injury (VILI) have focused on the characteristics of the individual tidal cycle–tidal volume, inspiratory ﬂow rate, and end-expiratory airway pressure (PEEP). In the largest and, perhaps, the most widely cited clinical trial of ventilator strategy in ALI yet undertaken, smaller tidal volumes were associated with a reduced mortality (1). This result was attributed to the generally lower peak alveolar pressures and reduced mechanical stresses associated with smaller tidal volumes. To date, almost all strategies that have been proposed to limit VILI have altered the ventilatory variables associated with a single tidal cycle. Lower tidal volume, lower plateau pressures, and higher PEEP settings have been used to avoid VILI, decrease length of intensive care unit stay, and curtail the associated mortality rate (2). Although these features 97

98

Marini and Ferrer

are of unquestioned importance, other characteristics of the clinical environment have been shown to modify the intensity and/or nature of the resulting damage. The fragile alveolus serves as the interface between gas and blood; consequently, the intraluminal pressures applied to the airway epithelium also impact the vascular endothelium. Conversely, vascular pressures and ﬂows have the potential to impact the development and/or evolution of VILI. This overview addresses the experimental evidence linking alveolar and vascular events, and the hemodynamic factors in the generation of barrier breakdown and VILI.

II. Effect of Pulmonary Expansion on the Pulmonary Vascular Tree The vascular pathway from the pulmonary artery to the left atrium can be considered as a series of three functional segments: arterial, ‘‘intermediate’’ (which includes alveolar capillaries and contiguous microvessels), and venous (3). Under normal conditions, arterial and venous segments—which are entirely extra-alveolar—contribute most to overall pulmonary vascular resistance (PVR). These compliant arterial and venous segments (extraalveolar vessels) are inﬂuenced primarily by interstitial pressure (crudely estimated by pleural pressure), whereas the compliant intermediate segment (alveolar vessels) is most strongly inﬂuenced by alveolar pressure. As a consequence, the latter undergoes the greatest ‘‘change’’ in the overall vascular resistance that occurs during ventilation. The behaviors of alveolar and extra-alveolar vessels during lung expansion are fundamentally different. The forces of interdependence cause interstitial pressures to fall during inﬂation, even during positive pressure ventilation (4). This reduction of interstitial pressure tends to increase the transmural pressure of the vessels in the immediate environment, thus dilating them. In turn, this increases wall tension, in particular, in the vessels upstream from the alveoli. Something quite different, however, happens at the alveolar level. During inﬂation of a normal, fully aerated (‘‘open’’) lung, the majority of capillaries embedded within the alveolar wall are compressed by the expansion of adjoining alveoli, even as extra-alveolar vessels dilate (Fig. 1). Consequently, alveolar vessels become longer and narrower as the lung expands, and usually decrease their volume, whereas extra-alveolar vessels become longer and wider, thereby increasing their volume. However, because increases in both the length and the diameter of vessels increase blood volume but have opposite effects on resistance, the change in PVR during lung inﬂation cannot always be predicted from changes in blood volume (3). The effects of vessel elongation and alveolar capillary compression

Hemodynamic Interactions During Ventilator-Induced Lung Injury

99

Figure 1 Inﬂuence of lung expansion on alveolar and extra-alveolar vasculature. Inﬂation compresses wall-embedded capillaries but dilates extra-alveolar microvessels.

outweigh the tendency for extra-alveolar vessels to dilate when inﬂation occurs with positive pressure at all lung volumes above the functional residual capacity (FRC), so that PVR rises monotonically as a function of lung volume (5). During negative pressure inﬂation, PVR falls initially and then rises gradually with further volume changes. Sufﬁcient positive pressure promotes the development of West zone I, deﬁned as the sector of the lung in which alveolar pressure exceeds both pulmonary artery and pulmonary venous pressures. As classically described, pulmonary arterio-venous blood ﬂow is not observed in zone I. However, several groups have demonstrated that some blood ﬂow occurs even under this condition (6,7); in fact up to about 15% of the normal resting cardiac output can ﬂow through rabbit lungs that are perfused with Tyrode’s solution maintained completely in zone I (6). In an intact-animal experiment, Lamm et al. found that pulmonary arterio-venous blood ﬂow persisted about 16 cmH2O into zone I and occurred via alveolar vessels (8). Such studies support the uniqueness of the so-called ‘‘corner’’ vessels, which are located at the junctions of three or more alveolar septae. Such vessels are simultaneously inﬂuenced by competing stresses arising from alveolar and interstitial pressures and therefore do not behave as the wall-embedded capillaries do. Alveolar corner vessel recruitment depends both on transpulmonary pressure (alveolar corner vessels are distended by lung inﬂation, as are extra-alveolar vessels) and on pulmonary artery pressure. Functionally, they behave like extra-alveolar vessels; indeed, they may serve as conduits for some blood to ﬂow through the intermediate segment, even when alveolar pressure exceeds pulmonary arterial pressure (8). With reference to the vascular contribution to VILI, it

100

Marini and Ferrer

is important to consider that even for the normal lung, inﬂation imposes competing vascular stresses on different classes of microvessels. As discussed later, these competing forces are ampliﬁed by the heterogeneity of ALI.

III. Response of the Endothelial Cell to Shear Forces Pulmonary endothelial cells (ECs) form a semiselective barrier to macromolecule transport into the lung, which is routinely exposed to biophysical forces in the form of shear stress (SS) and mechanical strain imposed by the blood ﬂow and the respiratory cycle. Pulmonary EC barrier function in vivo and in vitro is compromised by vasoactive substances such as thrombin, histamine, and tumor necrosis factor (9). These events result in lifethreatening edema. However, in contrast to the well-known effects of these bioactive peptides, the role of mechanical factors in pulmonary EC signaling and in disrupting the barrier function is not as completely understood. Increased SS is known to modulate EC function by initiating a wide range of responses, including activation of ﬂow-sensitive ion channels (10), changes in the expression of various gene products (11), G-protein–coupled signaling, and tyrosine kinase and focal adhesion kinase activation—effects with the potential to trigger early cytoskeletal reorganization (12). Lung ECs experience different ﬂow patterns in vivo under physiologic and pathologic conditions that are dictated by the unique features of the pulmonary circulation. Analysis of the cellular mechanisms of endothelial adaptation to ﬂow suggests time-dependent activation of intracellular signaling pathways, which culminates in cytoskeletal rearrangement. SS is widely accepted as a regulator of cytoskeletal organization and of cellular components such as G-protein–coupled receptors, caveolae, integrins, focal adhesion kinases, and mitogen-activated receptors (13). Moreover, SS is known to induce the transcription of some endothelial genes, including platelet-derived growth factor-B, monocyte chemostatic protein-1, and intercellular adhesion molecule-1. Morphologic and morphodynamic studies suggest three distinct phases of EC adaptation to ﬂow. The initial phase of this response is characterized by compensatory enhancement of EC cytoskeleton, characterized by increased stress ﬁber formation, thicker intercellular junctions, and more apical actin ﬁlaments. In the second phase, EC exhibit characteristics of motility and remodeling of their intercellular junctions, whereas the ﬁnal phase of SS-induced EC monolayer remodeling is characterized by EC orientation in the direction of ﬂow, and reestablishment of the intercellular contacts as well as the monolayer integrity (14,15). Although the intermediate-term effects of SS (several hours) on cytoskeletal reorientation, focal adhesion, and adherens junctions have been well described, little is known about molecular mechanisms triggering the initial phase of SS-activated cytoskeletal remodeling.

Hemodynamic Interactions During Ventilator-Induced Lung Injury

101

Birukov et al. demonstrated that pulmonary ECs exhibit a rapid cytoskeletal response to ﬂow, which may play an adaptive role in the maintenance of EC barrier properties under varying hemodynamic conditions (16). This observation suggests that the potential barrier-disrupting action of biophysical forces during the early phase of ﬂow readjustment may be counteracted by enhancement of the endothelial cortical cytoskeleton via cortactin and Rac-dependent mechanisms that promote the structural stability of the cell (16). Nitric oxide (NO) may play an important role in the development of hyperpermeability. Tangential forces generated by ﬂow across the EC surface trigger the generation of NO in the EC (17). Large tidal volumes repeatedly increase mean intrathoracic pressure during inspiration (which falls during exhalation), undoubtedly decreasing and increasing venous ﬂow and amplifying shearing stresses on the vessels. SS can increase NO production not only by increasing endothelial NO synthase (eNOS) expression, but also by phosphorylation of eNOS. This process may occur independently of changes in the intracellular calcium levels. SS-related NO production can occur within a few minutes of force application (18). The pathway by which NO increases vascular permeability may be through the extracellular signal-regulated kinases 1 and 2, and the vascular endothelial growth factor (VEGF). Inhibition of extracellullar signal-regulated kinases 1 and 2 in human umbilical vein ECs blocks cyclic guanosine monophosphate and VEGF-induced hyperpermeability (19). NO modulates VEFG-induced vascular permeability, and eNOS predominantly mediates this process. Cyclic stretch also upregulates the transforming growth factor B1 production (20), and the latter appears to directly increase the alveolar epithelial permeability by depletion of intracellular glutathione in an ALI model (21). Transforming growth factor B1 is also able to stimulate VEGF release in an epithelial cell line (22). Therefore, the production and interaction of multiple cytokines elaborated from the endothelium play an important role in promoting the vascular injury associated with mechanical stress. In addition to increasing the amount of cytokines in the lung, it is proposed that overinﬂation during mechanical ventilation may encourage the release of cytokines into the blood, potentially giving a generative role for mechanical ventilation in multiorgan dysfunction (23).

IV. Interactions Between Airway and Pulmonary Vascular Pressures The normal lung exhibits up to three perfusion zones, depending on the relationship between alveolar pressure and pulmonary arterial venous pressure. According to the familiar conceptual model popularized by West et al. (24),

102

Marini and Ferrer

gas pressures within the aerated alveoli are everywhere equivalent under static conditions, whereas vascular pressures are inﬂuenced by gravity. Unfortunately, the application of these useful concepts to the problem of the acutely injured lung is not straightforward. Indeed, their validity for this circumstance—in which collapsed, edematous, inﬂamed, and even ﬁbrotic lung units may coexist in the same microenvironment—can rightfully be questioned. In the setting of ALI, both alveolar and pulmonary arterial pressures are considerably greater than under normal conditions. Moreover, a variety of perfusion states are likely to exist, even along the same isogravitational plane. Filling of the small airways and the alveolar and interstitial spaces with cells and ﬂuid alters the normal relationships among the pressures and ﬂows of gases and blood. Independent of any variations of local pathology, increased lung tissue density and accentuated pleural pressure gradients tend to collapse dependent lung units, developing shunts and/or extending zone II conditions to more caudal levels as the interstitial pressures surrounding the microvasculature rise. It is also worth noting that the ventilation mode may affect the hemodynamics of the microvascular environment. The pressure drop occurring across the pulmonary circulation is greater for a positive pressure than it is for a negative pressure breath. Large tidal volume ventilations have been found capable of detaching ECs from their basement membranes within ﬁve minutes of exposure (25). During mechanical stress, the endothelium also has been shown to respond by forming paracellular gaps, resulting in increased permeability (9). Under the high permeability conditions of the ﬁrst stage of ALI, even minor increases in pulmonary microvascular pressure will increase edema formation dramatically. Moreover, unlike healthy tissues whose blood–gas barrier is intact, there is no clear pressure threshold for edema formation (26). The physiologic consequences of pulmonary edema are well known; alveolar edema compromises gas exchange, some blood proteins reduce the effectiveness of alveolar surfactants and thus increase the surface tension of the alveolar lining layer (27), and edematous airways impede airﬂow and secretion clearance. From the standpoint of VILI, however, alveolar ﬂooding may produce competing effects. A well-known simplistic model of interdependence proposed by Mead et al. suggests that collapsed alveoli are subjected to shearing forces that are proportional to the disparity in alveolar dimensions between the collapsed alveolus and its distended neighbors (28). Therefore, completely ﬂuid-ﬁlled (ﬂooded) alveoli, theoretically, are subjected to lower shearing stresses than atelectatic units, as the gas–liquid interface is eliminated and the alveolar dimensions increase. Edema-caused elimination of surface tension also allows capillaries that are fully embedded in the alveolar walls to bulge further into the interior, encouraging their rupture (27). Thus the normal surface tension of the alveolar lining layer protects the capillaries to some extent, and this

Hemodynamic Interactions During Ventilator-Induced Lung Injury

103

protective action is lost when the alveoli are ﬁlled with ﬂuid (28). On the other hand, the increased weight of the edematous lung may encourage small airway compression and accentuate the tendency for tidal opening and closure to occur. Which of these competing effects predominates cannot be said with certainty. Thus, while the inﬂuence of preformed edema on lung mechanics and gas exchange is reasonably well described, the importance of the microvasculature in the generation of VILI is less understood. The remainder of this brief review will focus on what is currently known regarding the interactions between airway pressures and vascular pressures in the generation and maintenance of VILI.

V. Mechanisms Disrupting the Blood–Gas Barrier Clinicians have long been aware that certain inﬂammatory conditions of the lung produce tissue hemorrhage in the absence of ventilatory stress. These vessel-disrupting inﬂammatory injuries may originate from either the alveolar side (e.g., pneumonia, abscess, etc.) or from the vascular side of the blood–gas interface. During normal breathing, the alveolar walls simply unfold rather than show an elastic deformation; therefore they undergo little stress under these conditions (29). But the blood–gas barrier is extremely thin and fails when it is exposed to high transmural pressures. The spectrum of injury (deﬁned by electron microscopy on perfusion-ﬁxed tissue specimens) potentially includes endothelial and epithelial membrane blebs, transcellular and intercellular gaps, and overt breaks of the basement membrane (29). These lesions have been described in small as well as large animals after vascular perfusion with high pressures or inﬂation to volumes exceeding the total lung capacity. Although clinicians recognize alveolar barrier damage radiographically as air that leaks into the interstitial spaces to cause intrapulmonary gas cysts, mediastinal emphysema, pneumothorax, and systemic gas embolism, it is equally clear that the vasculature can lose its integrity in advance of epithelial fragmentation (30). Postmortem examination of tissues from patients with acute respiratory distress syndrome (ARDS) often reveals areas of interstitial and alveolar hemorrhage, ﬁndings that generally have been attributed to the underlying inﬂammatory process. Yet, in both small and large animal models, the application of adverse ventilatory patterns to previously healthy lungs not only causes the formation of proteinaceous edema, but also stimulates neutrophil aggregation and hemorrhage (25,31). Although inﬂammation is obviously of potential importance in the breakdown of the lung’s structural architecture, simply elevating transmural pulmonary vascular pressure to high levels may cause vascular rents or tears. Perhaps the clearest clinical example in this category occurs in severe

104

Marini and Ferrer

mitral stenosis, a condition in which the pulmonary venous and capillary pressures can exceed 35 to 40 mmHg. Acute edema that forms in this setting is typically blood tinged, and the presence of hemosiderin-laden macrophages in expectorated or lavaged samples strongly suggests that this process originates at the alveolar level, from the pulmonary circulation (rather than from bronchial). Another circumstance under which elevating transmural vascular pressures may cause hemoptysis in the absence of preexisting inﬂammation occurs during extreme exertion, when blood ﬂows through the lung are extremely high and excursions of alveolar pressure are unusually large. Postexertional lung hemorrhage is a well-described occurrence in racehorses (32), and hemoptysis has been reported after heavy exertion in elite human athletes, as well (33). Finally, forceful inspiratory efforts made during upper airway obstruction may produce transvascular pressures of sufﬁcient magnitude to cause hemorrhagic pulmonary edema (34). It is well demonstrated that the walls of capillaries exposed to increased capillary transmural pressures show an ultrastructural change. This change consists of disruption of the capillary endothelium, the alveolar epithelium, or sometimes all layers of the wall, and is known as stress failure. It may be useful to consider the three different forces acting on the capillary wall, which promote stress failure: (i) circumferential (or hoop) stress caused by the transmural pressure, (ii) longitudinal tension in the alveolar wall associated with lung inﬂation, which is partly transmitted to the capillary wall, and (iii) surface tension of the alveolar lining layer (27). In experiments undertaken in the laboratories of West et al., electron microscopy has demonstrated the potential for mechanical disruption of the microvasculature—‘‘capillary stress fracture’’—to occur when microvascular pressures are elevated to very high levels relative to their usual operating conditions (35–38). The pressures necessary to cause capillary stress fracture vary among species, with disruptions being observed in healthy small animal lungs (e.g., rabbits), at pressures as low as 40 mmHg. Larger animals, such as dogs, withstand much higher microvascular pressures without losing the structural integrity of the capillary network (37). Although the range of microvascular pressures applied in these studies might seem to preclude their physiological relevance, much lower vascular pressures might be required if the framework of the lung were degraded by inﬂammation. Moreover, there is excellent reason to believe that regional transmural vascular forces may be dramatically different when mechanically heterogeneous lungs are ventilated with adverse ventilatory patterns. The increased longitudinal tension in the alveolar walls as a result of the high state of lung inﬂation is partially transmitted to the capillary walls, making them more vulnerable to stress failure, particularly if there is any concomitant increase in the capillary transmural pressure. When edema ﬂuid moves into the alveoli, abolition of the normally compressive action

Hemodynamic Interactions During Ventilator-Induced Lung Injury

105

of surface tension on the wall-embedded capillaries could exacerbate microvascular stress failure. Although unproven, this could be a vicious cycle. If the capillary transmural pressure rises transiently, stress failure and leakage of ﬂuid into the alveoli could abolish the normal protective mechanism of the alveolar lining layer, and the result is that more ﬂuid would extravasate (38). Flooding, however, would tend to negate the ampliﬁcation of tension at the borders of previously open alveoli.

VI. Behavior of Airway and Vascular Pressures in Heterogeneous Areas Studies conducted in our laboratory strongly indicate that in the supine position, hemorrhagic edema forms preferentially in dependent areas (31,39). This proclivity is not subtle, and has been corroborated by the work of other investigators using different injury models (40). The tendency for hemorrhage to occur preferentially in the most dependent regions of the lung may have several explanations. One compelling reason to expect microvascular disruption to occur in this area is that the mechanical stresses applied by the tidal inﬂation cycle are greatly ampliﬁed at the interface of opened and closed lung tissues (heterogeneous areas). More than three decades ago Mead et al. described a simpliﬁed model of alveolar mechanics in which they proposed that an alveolus attempting to close in an environment in which it was surrounded by inﬂated tissue would experience traction forces that are ampliﬁed in a nonlinear proportion to the alveolar pressures existing in the open units (28). By their reasoning, the coefﬁcient linking effective pressure (Peff) to that actually applied (Papp) is the ratio between the alveolar volume that corresponds to the Papp (V) and the volume of the collapsed alveolus (V0), raised to the 2/3 power: Peff ¼ Papp (V/V0)2/3. Their admittedly oversimpliﬁed geometrical argument suggested that at 30 cmH2O alveolar pressure, for example, the effective stress applied at the junction of closed and open tissues might approximate a value 4.5 times as great as that experienced within the free walls of open alveoli. Whatever its quantitative accuracy, a similar line of reasoning might be applied when tissues are already atelectatic and the lung is exposed to high ventilating pressures, as in ARDS. Extrapolating from the Mead equation, the traction forces applied to junctional tissues when the alveolar pressure is 30 cmH2O could approximate 140 cmH2O, or approximately 100 mmHg. Thus, transvascular microvascular forces during tidal ventilation could be in the range that West et al. suggested as necessary for a stress fracture to occur in large animals (dogs), and far exceed those for a small animal (e.g., rat or rabbit) (37). However, it does seem reasonable to assume that mechanical shearing forces experienced in ‘‘junctional’’ tissues are likely to exceed those experienced elsewhere in the lung. Moreover, even within

106

Marini and Ferrer

fully inﬂated regions, the competing forces of capillary compression and extra-alveolar vessel dilation/elongation are expected to be ampliﬁed when both the lung volumes and the vascular pressures are high. It is not difﬁcult, therefore, to envision vascular rupture from ventilatory pressure under the pathological conditions of ALI. Another intriguing possibility to explain disproportionate vascular disruption in dependent lung regions is that dorsally situated tissues receive a majority of the lung’s total blood ﬂow and are subjected to greater hydrostatic pressures in the supine position. These higher intraluminal vascular pressures or ﬂows might amplify the tensile forces external to the microvessels or give rise to shearing stresses within the vascular endothelium, which initiate inﬂammation-mediated tissue breakdown (mentioned previously). There are hints in the early experimental literature of VILI that vascular pressure could play an important—if not pivotal—role in VILI development or expression. Dreyfuss and Saumon, for example, found that ventilation with negative pressures caused more severe damage than that caused by positive pressures, implicating the role of increased blood ﬂow to ventilationrelated damage (41). These same investigators provided further support for this hypothesis by showing that rats given dopamine to increase their cardiac output suffered increased albumin leak when ventilated with a high pressure, and ascribed a major portion of PEEP’s protective effect in the setting of high pressure ventilation to its reduction of pulmonary perfusion (42).

VII. Role of Vascular Pressure and Flow on Genesis of VILI Our group has also explored the vascular contribution to VILI in a series of experiments using isolated, ventilated and perfused (IVP) rodent lungs (43–46). The IVP system offers numerous advantages for the investigation of the interactions between alveolar and vascular pressures. The progress of edema formation can be monitored by continuously weighing the heart/lung block suspended from a strain gauge. Breakdown of the alveolar capillary barrier can be inferred from the ﬁltration constant (KF) derived from the weight–time relationship. In-ﬂow and out-ﬂow vascular pressures and/or the perfusion rate can be precisely measured and/or regulated. Finally, the composition and physical properties of the perfusate can be adjusted. In our ﬁrst experiment, we exposed isolated rabbit lungs to perfusion levels that were equivalent to, and approximately 50% greater or less than the normal resting blood ﬂow of that animal species (43). All lungs were ventilated identically with airway pressures that proved damaging in vivo. In this model of VILI, we demonstrated that perfusion amplitude contributed to the reduced lung compliance resulting from an adverse ventilatory pattern and promoted both lung edema and hemorrhage.

Hemodynamic Interactions During Ventilator-Induced Lung Injury

107

We also found a strong correlation between the indices of lung injury and the vascular pressure changes resulting from the interaction between ventilation and perfusion (43). Although data from this experiment strongly suggested the primacy of perfusion pressure, it was not possible in the initial experiment to deﬁnitively determine which of these two variables was more important in modulating VILI, because vascular pressure increased in parallel with ﬂow. In a second IVP lung experiment designed to address that question, we varied airway pressure proﬁles to allow the arterial pressure to vary while blood ﬂow was held constant (44). Our results indicated that the mean airway pressure was a higher impact variable than the tidal excursion amplitude in determining the severity of the lung hemorrhage and lung permeability alterations resulting from an adverse pattern of mechanical ventilation. Histologic injury scores were virtually identical for large and small tidal volumes when high mean airway pressures were achieved, whether by lengthening inspiratory time or by increasing PEEP, respectively. The results of those experiments emphasized the potential for deleterious interactions to occur between lung volumes and pulmonary hemodynamics. Taken together, our initial two studies demonstrated that modiﬁcations of vascular pressure within and upstream from the intermediate segment could inﬂuence the severity of VILI inﬂicted by an unchanging adverse pattern of ventilation.

VIII. Effect of Respiratory Rate and Flow on Expression of VILI Although ventilation is the product of tidal volume and frequency, surprisingly little attention has been directed to the role of the latter in the generation of VILI. Therefore, having concluded that upstream microvascular pressure might be an important cofactor in the development of VILI, we next addressed the question of how the number of ventilatory cycles occurring over a timed interval inﬂuences the rate of edema formation or the severity of histologic alterations when maximum, minimum, and mean airway pressures are held identical. Almost 15 years ago, Bshouty and Younes reported that for the same minute ventilation target, raising tidal volume at a constant ventilating rate and raising frequency at a constant tidal volume produced similar degrees of edema in canine lobes perfused in situ at elevated hydrostatic pressures (47). We used our IVP model in experiments that tested the hypothesis that cumulative damage occurs as a function of the number of stress cycles as well as stress magnitude. In these experiments, the pressure-controlled mode with a peak pressure of 30 cmH2O and a PEEP of 3 cmH2O was used in each preparation (45). Our main ﬁndings were that lungs ventilated at

108

Marini and Ferrer

low frequencies and high peak pulmonary artery pressures formed less edema and displayed markedly less perivascular hemorrhage than did those ventilated at higher frequencies but identical peak and mean pulmonary arterial pressures. In addition, lungs ventilated with high peak pulmonary artery pressures and ﬂows demonstrated more extensive histologic alterations and edema formation than did those subjected to the same ventilatory pattern, but at lower peak vascular pressures and ﬂows (45). Only a very small fraction of this difference was attributable to differences in the mean hydraulic pressure. Several mechanisms come to mind to explain the diminution of lung edema formation and perivascular hemorrhage that we observed by decreasing the respiratory frequency. A higher ventilatory frequency could have depleted surfactant more efﬁciently, thereby increasing alveolar surface tension, lowering end-expiratory extravascular pressure, and promoting alveolar ﬂooding. Upstream, the increased transvascular pressure gradient across extra-alveolar vessels would also favor ﬂuid transudation, vessel disruption, and perivascular hemorrhage. Conceivably, the larger number of imposed stress cycles could have induced cumulative damage in a fashion similar to that experienced in a variety of biomaterials that are subjected to sufﬁcient repeated stress (48). Overt stress fractures similar to those found by West et al. have been demonstrated by scanning electron microscopy in our laboratory and in a recently reported human patient (Fig. 2A and B microphotography) (49). A type of ‘‘materials failure’’ of structural elements seems an attractive explanation, in that we found that reducing the stress application frequency (respiratory rate) as well as the stress amplitude (pulmonary artery peak pressure) effectively limited VILI. Whether our results are relevant to intact animals should be questioned. For example, in contrast to our ex-vivo studies, Rich et al. found that reducing the respiratory rate did not reduce the development of injury in intact sheep. At reduced respiratory rates, the development of lung injury was independent of the inspiratory time during high-pressure, large-volume mechanical ventilations. But reducing the inspiratory ﬂow was protective against the development of VILI, as measured by all indices of injury. Limiting inspiratory ﬂow and altering its pattern of delivery to a constant rate signiﬁcantly reduced lung injury, as measured physiologically and histologically (50). Douillet and coworkers, working with another model of VILI in rats, showed that increasing the respiratory rate at low tidal volumes had little impact on injury parameters, but respiratory rate reduction under VILI-promoting conditions signiﬁcantly limited lung injury (51). Taking account of these seemingly contradictory data, we believe that the unifying hypothesis might be that frequency assumes importance only if the alveolar microvascular barrier is sufﬁciently stressed.

Hemodynamic Interactions During Ventilator-Induced Lung Injury

109

Figure 2 Electron micrographs of rabbit lung (A) and patient lung (B) injured by high inﬂation pressures, low PEEP, and elevated vascular inﬂow pressures. The previously healthy rabbit lung was injured by the application of purely mechanical forces, whereas the patient had been ventilated for ARDS and had received vigorous intravascular volume resuscitation. Large, nonanatomical disruptions are evident in both. Abbreviations: PEEP, end-expiratory airway pressure; ARDS, acute respiratory distress syndrome.

Such data strongly indicate that not only are the characteristics of the tidal cycle and vascular pressures of fundamental importance to VILI, but also that minute ventilation, reﬂecting the number of stress cycles of a potentially damaging magnitude per unit time (or their cumulative number), might be as well. If cumulative damage is important, providing a lower

110

Marini and Ferrer

frequency and/or lower pulmonary vascular pressure would both be expected to reduce the tendency for material stress failure. Moreover, it is interesting to consider that a low frequency may allow sufﬁcient time between adjacent cycles for reparative processes to operate. Surprisingly little time appears to be needed to reseal small disruptions in tissue barriers (52,53). Therefore, the contribution of mechanical ventilation in the genesis of lung injury resides not only in the magnitude of the alveolar inﬂation or in the frequency at which this is applied, but also in the rate of gas delivery. This complex interaction makes more difﬁcult the control of all mechanical variables that effect the development of VILI.

IX. Cyclic Effect on the Microvascular Environment Induced by Mechanical Ventilation Because rigorous limitation of pulmonary vascular pressures signiﬁcantly attenuated the damage in lungs exposed to a ﬁxed ventilatory pattern, the work outlined above suggests that elevations of pulmonary vascular pressure arising from interactions between lung volume, PVR, and pulmonary vascular ﬂow could worsen ventilator-associated lung injury. Our redirected attention toward the vascular side of the alveolar capillary barrier stimulated us to ask this question: Is periodic inﬂation a necessary component of the vascular injury incurred during VILI? Knowing that the frequency of ventilation was an important determinant of VILI when high airway pressures were in use, we reasoned that a lung exposed to pulsatile vascular pressure, ‘‘but not ventilated,’’ might experience signiﬁcant injury, even without ﬂuctuations of airway pressure. In an experiment designed to test this question, we applied a damaging pattern of airway pressure (plateau 30 cmH2O, PEEP 5 cmH2O) to one of three sets of rabbit lung preparations and allowed others to remain motionless (46). In the ventilated group, peak pulmonary artery pressure was allowed to rise to 35 mmHg and in the other two groups a vascular pump applied pulsatile pulmonary artery pressure to motionless lungs at frequencies of 3 or 20 pulses/min. Our main ﬁndings were that lungs exposed to cyclic elevations of pulmonary artery pressure in the absence of ventilation, formed less edema and displayed less perivascular and alveolar hemorrhage than ventilated lungs exposed to similar peak and mean pulmonary artery pressures and mean airway pressure (46). In addition, lungs ventilated with high peak pulmonary artery pressures and ﬂows exhibited more extensive histologic alterations and edema formation than did those subjected to the same ventilatory pattern but at lower peak vascular pressures and ﬂows. Interestingly, under static continuous positive airway pressure conditions, the higher pulsing frequency was associated with a greater degree of perivascular hemorrhage, indicating that the pulsatile frequency of vascular pressure

Hemodynamic Interactions During Ventilator-Induced Lung Injury

111

did indeed contribute to VILI. Thus, the effects of respiratory frequency and vascular pressures on VILI are not mediated primarily by pulsatile vascular pressure per se, but rather by a phenomenon related to the cyclical modulation of the vascular microenvironment induced by ventilation. Because alveolar and extra-alveolar microvessels are stressed differently by lung expansion, these experiments focused our attention on the extra-alveolar microvasculature, suggesting that the cyclic changes in perivascular pressures surrounding extra-alveolar, juxta-capillary microvessels might be important in the genesis of VILI.

X. Effect of Postalveolar Vascular Pressure on the Development of VILI Given that elevation of pulmonary vascular inﬂow pressure accentuated VILI, it seems logical that reducing postalveolar vascular pressure would be protective. Merit from reducing left atrial pressures might be expected for at least two reasons. The edematous lungs tend to collapse under their own weight and develop dependent atelectasis, which could lead to cyclic opening and collapse, intensiﬁed shearing stresses (SSs), and a tendency for VILI in dependent areas. Moreover, exudation of protein-rich ﬂuids has the potential to inactivate surfactant, further altering membrane permeability by increasing both surface tension and radial traction on pulmonary microvessels. On the other hand, increased left atrial pressure might help to limit VILI by ﬂooding the alveoli of dependent regions, thereby reducing regional mechanical stresses. Moreover, reducing capillary pressure could promote cyclical vascular recruitment and derecruitment as the lungs transition from West’s zone III to zone II condition during the course of the positive pressure inﬂation/deﬂation cycle. Hydrodynamic forces may well be accentuated by the higher velocities and surface SSs that occur along the vascular endothelium under such conditions. But reducing the outﬂow pressure increases the gradient of pressure appearing across the alveoli, and consequently, the energy dissipated across the intermediate segment. For these reasons, the impact of selectively reducing the pulmonary venous pressure during ventilation with high airway pressures cannot easily be predicted. Broccard et al. compared isolated perfused rabbit lungs ventilated with moderately high peak alveolar pressures with normal and low left atrial pressures, and demonstrated a striking difference in favor of the normal vascular pressure subset (Fig. 3) (54). This rather surprising result suggests that the cyclical opening and closure of stressed microvessels could be important in the generation of VILI. Alternatively, decreasing outﬂow pressure might amplify microvascular stresses at or near the alveolar level, presumably acting through

112

Marini and Ferrer

Figure 3 Effect of reducing outﬂow pressure on edema formation and vascular permeability when ventilating isolated perfused rabbit lungs with high tidal inﬂation pressure and low PEEP. HPN and HPL refer to pulmonary venous pressures: normal and low. Note that little change can be detected in any condition until alveolar pressure rises to its highest value (30 cmH2O). When peak alveolar pressure is 30 cmH2O, lungs perfused with a high inﬂow pressure and a low outﬂow pressure sustain more edema and altered permeability than do lungs whose vascular outﬂow pressures are reduced. Abbreviation: PEEP, end-expiratory airway pressure. Source: From Ref. 34.

interdependence of the pulmonary vascular network. The reduction of postcapillary pressure allows an increased amount of energy to dissipate across the ‘‘middle segment.’’ We speculate that direct mechanotransduction of inﬂammatory signals, increased transalveolar energy dissipation, or materials failure at the stressed boundary could be important linking mechanisms.

XI. Potential Clinical Implications The interactions between the vascular pressure and ventilation outlined in this review suggest strongly that closer attention should be paid to interventions that impact vascular pressures, ﬂows, and resistances when high inﬂation pressures are in use. Because microvascular stresses appear to be a potent cofactor in the development of pulmonary edema as well as in the lung damage resulting from an injurious pattern of ventilation, the clinician managing ALI must reconcile the competing objectives of ensuring adequate oxygen delivery and minimizing adverse effects. For example, an increase in cardiac output is generally held to be a beneﬁcial consequence of management; however, increases in cardiac output are associated with an increased prealveolar microvascular pressure and a higher vascular pressure gradient across the lung. Because an increased

Hemodynamic Interactions During Ventilator-Induced Lung Injury

113

prealveolar microvascular pressure accentuates a tendency for VILI, attempts to raise the cardiac output may have unintended consequences. On the other hand, taking steps to reduce oxygen consumption demands could beneﬁt the lung by reducing the pressure gradient developed across the microvasculature. Similarly, a reduction in the left atrial pressure with maintained cardiac output is generally believed to beneﬁt lung function, and this perception is almost certainly accurate with respect to hydrostatic edema formation. But excessive reduction of left atrial pressures could amplify the tendency for VILI when high airway pressures are in use (54). Therefore, because reducing ventilation frequency decreases the number of stress cycles and reducing respiratory ﬂow rate may diminish shear stresses, a reduction in minute ventilation effected either by a decrease in tidal volume or a decrease in ventilatory frequency or limiting respiratory ﬂow might have a salutary effect in reducing the tendency for VILI. It follows that a reduced need for the delivery of blood ﬂow and oxygen supply to tissues might reduce the VILI risk. In fact, the available literature suggests that reducing the demands for cardiac output and ventilation could dramatically reduce the tendency for VILI to occur, even when using patterns that generate similar values for peak and end-expiratory alveolar pressures. Whether these intriguing possibilities are relevant to the clinical setting will require extensive and careful additional study.

XII. Conclusions To summarize, during positive pressure ventilation, the majority of capillaries embedded within the alveolar walls of open alveoli are compressed by the expansion of adjoining lung units. At the same time, lung expansion decreases interstitial pressure, which increases the transmural pressure of the vessels in the intermediate segment. Raising precapillary and/or reducing postcapillary microvascular pressures simultaneously, increases the pressure gradient and energy dissipated across the middle segment of the pulmonary microvasculature. These actions appear to worsen edema and/or accentuate barrier failure when airway mechanical stresses are sufﬁciently high. On the other hand, cyclic opening and closure of the microvessels may amplify shear forces and stretching of the vascular endothelium (similar to that undergone by the alveolar wall), with the potential to initiate inﬂammation-mediated tissue breakdown. Thus, by promoting alveolar vascular collapse and amplifying extra-alveolar vascular stress, excessive reductions in the lung volume and microvascular pressure have the potential to contribute to VILI (Fig. 4). Although more studies are clearly necessary to determine the exact relationship between vascular pressure and lung injury, it would seem

114

Marini and Ferrer

Figure 4 Hypothetical interactions between airspace and pulmonary vascular pressures in the causation of ventilator-induced lung injury.

prudent to diminish unnecessary demands for ventilation and cardiac output; the direct clinical implication is that conditions of agitation, high fever, pain, and work of breathing should be avoided when potentially damaging alveolar pressures are approached.

References 1. Brower RG, Matthay M, Schoenfeld D. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials. Am J Respir Crit Care Care Med 2002; 166:1515–1517. 2. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338:347–354. 3. Hakim TS, Michel RP, Chang HK. Effect of lung inﬂation on pulmonary vascular resistance by arterial and venous occlusion. J Appl Physiol 1982; 53:1110–1115. 4. Lai-Fook SJ. Perivascular interstitial pressure measured by micropipettes in isolated dog lung. J Appl Physiol 1982; 52:9–15. 5. Fishman AP. Pulmonary circulation. In: Fishman AP, Fisher AB, Geiger SR, eds. Handbook of Physiology-Section 3: The Respiratory System. American Physiological Society, 1987:93–97. 6. Koyama S, Lamn J, Hildebrandt J, Albert RK. Flow characteristics of open vessels in zone 1 rabbit lungs. J Appl Physiol 1989; 15:241–245. 7. Bruderman LK, Somers K, Hamilton W, Tooley W, Butler J. Effects of surface tension on circulation in excised lungs of dogs. J Appl Physiol 1964; 19:707–712.

Hemodynamic Interactions During Ventilator-Induced Lung Injury

115

8. Lamm W, Kirk K, Hanson W, Wagner W, Albert R. Flow through zone 1 lungs utilizes alveolar corner vessels. J Appl Physiol 1991; 70(4):1518–1523. 9. Dudek SM, Garcia JG. Cytoskeletal regulation of pulmonary vascular permeability. J Appl Physiol 2001; 91:1487–1500. 10. Barakat AI, Leaver EV, Pappone PA, Davies PF. A ﬂow-activated chloride selective membrane current in vascular endothelial cells. Int J Mol Med 1999; 4:323–332. 11. Gudi S, Nolan JP, Fragos JA. Modulation of GTPase activity of G proteins by ﬂuid shear stress and phospholipid composition. Proc Natl Acad Sci USA 1998; 95:2515–2519. 12. Garcia JG, Shaphorst AD, Verin S, et al. Diperoxavanadate alters endothelial cell focal contacts and barrier function: role of tyrosine phosphorylation. J Appl Physiol 2000; 89:2333–2343. 13. Fisher AB, Chien S, Barakat AI, Nerem RM. Endothelial cellular response to altered shear stress. Cell Mol Physiol 2001; 281:L529–L533. 14. Noria SD, Cowan DB, Gotlieb AI, Langile BL. Transient and steady-state effects of shear stress on endothelial cell adherens junctions. Cir Res 1999; 85:504–514. 15. Davies PF. Flow-mediated endothelial mechanotransduction. Physiol Rev 1995; 75:519–560. 16. Birukov K, Birukova A, Dudek S, et al. Shear stress-mediated cytoskeletal remodeling and cortactin translocation in pulmonary endothelial cells. Am J Respir Cell Mol Biol 2002; 26:453–462. 17. Quinn TP, Schlueter M, Soifer SJ, Gutierrez JA. Cyclic mechanical stretch induces VEGF and FGF-2 expression in pulmonary vascular smooth muscle cells. Physiol Lung Cell Mol Physiol 2002; 282:L897–L903. 18. Buga GM, Gold ME, Fukuto JM, Ignarro LJ. Shear stress-induced release of nitric oxide from endothelial cells grown on beads. Hypertension 1991; 17:187–193. 19. Varma S, Bresin JW, Lai BK, Pappas PJ, Hobson RW II, Duran WN. P42/ 44(MAPK) regulates baseline permeability and cGMP-induced hyperpermeability in endothelial cells. Microvasc Res 2002; 63:172–178. 20. Yamamoto H, Teramoto H, Uetani K, Igawa K, Shimizu E. Cyclic stretch upregulates interleukin-8 and transforming growth factor-beta 1 production through a protein kinase C-dependent pathway in alveolar epithelial cells. Respirology 2002; 7:103–109. 21. Pittet JF, Grifﬁths MJ, Geiser T, et al. TGF-beta is a critical mediator of acute lung injury. J Clin Invest 2001; 107:1537–1544. 22. Boussat S, Eddahibi S, Coste A, et al. Expression and regulation of vascular endothelial growth factor in human pulmonary epithelial cells. Am J Physiol Lung Cell Mol Physiol 2000; 279:L371–L378. 23. Pugin J. Is the ventilator responsible for lung and systemic inﬂammation? Intensive Care Med 2002; 28:817–819. 24. West JB, Dollery CT, Naimark A. Distribution of blood ﬂow in isolated lung; relation to vascular and alveolar pressures. J Appl Physiol 1964; 19:713–724. 25. Dreyfuss D, Saumon G. Ventilator induced lung injury: lessons from experimental studies. Am J Resp Care Med 1998; 157:294–323.

116

Marini and Ferrer

26. Brigham KL, Woolverton WC, Blake LH, Staub NC. Increased sheep lung vascular permeability caused by pseudomonas bacteremia. J Clin Invest 1974; 54(4):792–804. 27. Namba Y, Kurdak S, Fu Z, Mathieu-Costello O, West J. Effect of reducing alveolar surface tension on stress failure in pulmonary capillaries. J Appl Physiol 1995; 79(6):2114–2121. 28. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28:218–233. 29. Vlahakis N, Hubmayr R. Cellular responses to mechanical stress. Invited review: plasma membrane stress failure in alveolar epithelial cells. J Appl Physiol 2000; 89:2490–2496. 30. Amato MB, Marini JJ. Barotrauma, volutrauma, and the ventilation of acute lung injury. In: Marini JJ, Slutsky AS, eds. Physiological Basis of Ventilatory Support. New York: Marcel Dekker, 1998:1187–1245. 31. Broccard A, Shapiro R, Schmitz L, Adams AB, Nahum A, Marini J. Prone positioning attenuates and redistributes ventilator-induced lung injury in dogs. Crit Care Med 2000; 28:2295–2303. 32. West JB, Mathieu-Costello O, Jones HJ, et al. Stress failure of pulmonary capillaries in racehorses with exercise-induced pulmonary hemorrhage. J Appl Physiol 1993; 75:1097–1109. 33. Hopkins SR, Schoene RB, Martin TR, Henderson WR, Spragg RG, West JB. Intense exercise impairs the integrity of the pulmonary blood-gas barrier in elite athletes. Am J Respir Crit Care Med 1997; 155:1090–1094. 34. Broccard AF, Liaudet L, Aubert JD, Schnyder P, Schaller MD. Negative pressure post-tracheal extubation alveolar hemorrhage. Anesth Analg 2001; 92(1):273–275. 35. Costello ML, Mathieu-Costello OM, West JB. Stress failure of alveolar epithelial cells studied by scanning electron microscopy. Am Rev Respir Dis 1992; 145:1446–1455. 36. West JB, Tsukimoto K, Mathieu-Costello O, Prediletto R. Stress failure in pulmonary capillaries. J Appl Physiol 1991; 70:1731–1742. 37. Mathieu-Costello O, Willford DC, Fu Z, Garden RM, West JB. Pulmonary capillaries are more resistant to stress failure in dogs than in rabbits. J Appl Physiol 1995; 79:908–917. 38. Fu Z, Costello ML, Tsukimoto K, et al. High lung volume increases stress failure in pulmonary capillaries. J Appl Physiol 1992; 73:123–133. 39. Broccard AF, Shapiro RS, Schmitz LL, Ravenscraft SA, Marini JJ. Inﬂuence of prone position on the extent and distribution of lung injury in a high tidal volume oleic acid model of acute respiratory distress syndrome. Crit Care Med 1997; 25(1):16–27. 40. Hirschl RB, Tooley R, Parent A, Johnson K, Bartlett RH. Evaluation of gas exchange, pulmonary compliance, and lung injury during total and partial liquid ventilation in the acute respiratory distress syndrome. Crit Care Med 1996; 24(6):1001–1008. 41. Dreyfuss D, Saumon G. Role of tidal volume, FRC, and end-inspiratory volume in the development of pulmonary edema following mechanical ventilation. Am Rev Respir Dis 1993; 148:1194–1203.

Hemodynamic Interactions During Ventilator-Induced Lung Injury

117

42. Dreyfuss D, Soler P, Basset G, Saumon G. High inﬂation pressure pulmonary edema. Respective effects of high airway pressure, high tidal volume, and positive end-expiratory pressure. Am Rev Respir Dis 1988; 137(5):1159–1164. 43. Broccard AF, Hothchkiss JR, Kuwayama N, et al. Consequences of vascular ﬂow on lung injury induced by mechanical ventilation. Am J Respir Crit Care Med 1998; 157:1935–1942. 44. Broccard AF, Hotchkiss JR, Suzuki S, Olson D, Marini JJ. Effects of mean airway pressure and tidal excursion on lung injury induced by mechanical ventilation in an isolated perfused rabbit lung model. Crit Care Med 1999; 27:1533–1541. 45. Hotchkiss JR, Blanch LL, Murias G, et al. Effects of decreased respiratory frequency on ventilator induced lung injury. Am J Respir Crit Care Med 2000; 161:463–468. 46. Hotchkiss JR, Blanch LL, Naviera A, Adams AB, Olson D, Marini JJ. Relative roles of vascular and airspace pressures in ventilator induced lung injury. Crit Care Med 2001; 29(8):1593–1598. 47. Bshouty Z, Younes M. Effect of breathing pattern and level of ventilation on pulmonary ﬂuid ﬁltration in dog lung. Am Rev Respir Dis 1992; 145:3672–3676. 48. Hashin Z, Rotem A. A cumulative damage theory of fatigue failure. Mater Sci Eng 1978; 34:147–160. 49. Hotchkiss JR, Simonson DA, Marek DJ, Marini JJ, Dries DJ. Pulmonary microvascular fracture in a patient with acute respiratory distress syndrome. Crit Care Med 2002; 30(10):2368–2370. 50. Rich P, Reickert C, Sawada S, et al. Effect of rate and inspiratory ﬂow on ventilator-induced lung injury. J Trauma 2000; 49:903–911. 51. Rich PB, Douillet CD, Hurd H, Boucher RC. Effect of ventilatory rate on airway cytokine levels and lung injury. J Surg Res 2003; 113(1):139–145. 52. Dreyfuss D, Soler P, Saumon G. Spontaneous resolution of pulmonary edema caused by short periods of cyclic overinﬂation. J Appl Physiol 1992; 72(6): 2081–2089. 53. Vlahakis NE, Hubmayr RD. Invited review: plasma membrane stress failure in alveolar epithelial cells. J Appl Physiol 2000; 89(6):2490–2496. 54. Broccard A, Vannay C, Feihl F, Schaller MD. Impact of low pulmonary vascular pressure on ventilator-induced lung injury. Crit Care Med 2002; 30:2183–2190.

This page intentionally left blank

6 Lung Mechanics and Pathological Features During Ventilation-Induced Lung Injury

DIDIER DREYFUSS

JEAN-DAMIEN RICARD

Paris 7-Denis Diderot Medical School Paris, France Service de Re´animation Me´dicale, Assistance Publique-Hoˆpitaux de Paris, Hoˆpital Louis Mourier Colombes, France

Service de Re´animation Me´dicale, Assistance Publique-Hoˆpitaux de Paris, Hoˆpital Louis Mourier Colombes, France Inserm U 722, Paris 7-Denis Diderot Medical School Paris, France

GEORGES SAUMON EA 3512, IFR 02 Claude Bernard, Paris 7-Denis Diderot Medical School Paris, France

I. Introduction Few simple experimental concepts have helped in improving the care of critically ill patients the way the concept of ventilator-induced lung injury (VILI) did. The safety of mechanical ventilation for the treatment of patients with acute respiratory failure was questioned soon after its introduction into medical practice. Several studies showed that ventilation with high airway pressure may result in abnormal lung function. Whereas ventilation of goats with a peak airway pressure of 13 cmH2O for two weeks proved safe provided the fraction of inspired oxygen remained low (1), atelectasis and increased surface tension of lung extracts were seen after ventilation of dogs with a peak airway pressure of 26 to 32 cmH2O (2). The very concept of VILI arose from the study by Webb and Tierney (3) who showed that ventilation of rats with high peak airway pressures [30–45 cmH2O peak inspiratory pressure (PIP)] resulted in pulmonary edema, the severity of which depended on peak airway pressure magnitude and was attenuated by the use of positive end-expiratory pressure (PEEP). Later studies showed that this edema 119

120

Dreyfuss et al.

was due to microvascular injury and underlined the role of increased lung distension and thus high lung tissue stress in VILI development (4). If VILI had a clinical counterpart (termed VALI, ventilation-associated lung injury), easing the stress applied to diseased lungs would improve acute (adult) respiratory distress syndrome (ARDS) prognosis. This has been unambiguously demonstrated by the results of the ARDS network study that showed a 22% lessening of mortality in patients who received 6 mL/kg instead of the more conventional 12 mL/kg tidal volume (VT) (5). However, the precise reasons for this improvement are not yet completely clariﬁed because VILI is composite. Different phenomena occur depending on the nature and level of mechanical stress and interplay to produce complete VILI features. Very high mechanical stretch produces lung ‘‘overinﬂation’’ (this term is discussed later) and acute physiological and pathological alterations. It is also the case when ventilating damaged lungs at low volume, because of the repetitive opening and closure of distal airways, whether this phenomenon is due to cyclic expansion of atelectatic areas or movement of edema foam in airways. This kind of injury may be experimentally lessened by application of a sufﬁcient level of PEEP (6–8). However, these ﬁndings have not received a deﬁnitive clinical demonstration (9). Mechanical alterations may eventually induce an inﬂammatory and repair process, as is the case for all tissues that have been wounded. Inﬂammation may apparently worsen lung damage in the short term because it is accompanied by an increase in capillary permeability that tends to aggravate pulmonary edema. But it has also been suggested that ventilating the lungs under less extreme mechanical conditions may create a subacute inﬂammatory reaction without any evidence of initial cellular lesions, because stretched lung cells release proinﬂammatory mediators that recruit and activate leukocytes. This concept has been termed ‘‘biotrauma’’ and is discussed in Chapters 9 and 10 of this book. This chapter presents the physiological concepts that underlie acute VILI, the pathological consequences of high lung stretch, and how respiratory mechanics might inform us on the risk of overinﬂation.

II. Acute Pulmonary Edema Consecutive to High-Lung-Volume Ventilation A. Increased Filtration Pressure

The concept of VILI emerged from the observation that pulmonary edema rapidly develops when animals are ventilated at high lung volume (3,4,10). This edema severity is due to the conjunction of two phenomena: an increased driving pressure for ﬂuid ﬁltration and permeability alterations of alveolar–capillary barrier. The contribution and mechanisms of transmural vascular pressure and permeability changes are described in detail

Lung Mechanics and Pathological Features During VILI

121

in Chapters 3 and 4 of this book and are therefore only brieﬂy presented. The role of hydrostatic alterations was put forward in early publications (3,11). Although vascular pressure increase was estimated to be about 10 cmH2O during ventilation of closed-chest lambs at 58 cmH2O peak inspiratory pressure (12) or of open-chest dogs at 64 cmH2O peak inspiratory pressure (11), even such modest changes may have important consequences when microvascular permeability is abnormal. Indeed, in the face of a microvascular barrier with altered sieving properties, any increase in driving pressure will accelerate edema formation (13–15). Fluid leakage (whether it occurs by pure ﬁltration or also because of permeability changes) from microvessels takes place at both alveolar and extra-alveolar levels during VILI (16,17). Lung inﬂation decreases interstitial pressure (thus, increases transmural pressure) because of the interdependence phenomenon and surfactant inactivation. During lung inﬂation, pressure in the perivascular space surrounding extra-alveolar vessels decreases, and this in turn increases transmural pressure. This effect of lung volume on pulmonary vessels is well documented (18). Pressure–volume (PV) characteristics of the vascular bed have been determined in isolated lung lobes in dogs (19). Inﬂating lungs dilates extra-alveolar vessels. During inﬂation from a low transpulmonary pressure, the increase in vessel diameter is such that an effective outward-acting pressure in excess of pleural pressure (1 to 2 cmH2O for each centimeter of water increase in transpulmonary pressure) expands these vessels (20). The potential importance of ﬂuid leakage through extra-alveolar vessels has been established in both excised lungs (21) and in situ lungs in open-chest animals (22). Ventilation of excised lungs alters pulmonary PV curves and increases surface tension of lung extracts commensurate with the magnitude of tidal volume and duration of ventilation (23–25). These anomalies are the result of surfactant inactivation (25–28) because of lower amounts of organized lipid–protein structures (29), or a loss of surface-active material in the airways (30). Surfactant alterations are less when volume excursions are reduced by applying PEEP (23,24,26,30), which may help explain the protective effect of PEEP during VILI (see below). This increase in alveolar surface tension due to surfactant inactivation indeed increases ﬁltration through microvessels (31–33). B. Alterations of Alveolar–Capillary Barrier

Alveolar epithelial permeability alterations in response to increased lung volume were reported in both animals and humans. Increasing functional residual capacity (FRC) by increasing the level of PEEP during mechanical ventilation (34) or during spontaneous ventilation (35) in sheep was associated with an increase in aerosolized diethylenetriaminepentaacetic acid (DTPA) clearance. Similarly, DTPA clearance increased regardless of

122

Dreyfuss et al.

whether lung inﬂation was obtained by positive pressure breathing or voluntary hyperinﬂation in humans (36). However, these increases in permeability were modest under physiologic conditions. By contrast, static lung inﬂation beyond physiological volume resulted in considerable increases in epithelial permeability of ﬂuid-ﬁlled lobes of sheep in situ (37,38). Equivalent pore radii increased from about 1 nm at 20 cmH2O inﬂating pressure to 5 nm at 40 cmH2O alveolar pressure. Free diffusion of albumin across the epithelium was observed in some instances, suggesting the presence of large leaks. The magnitude of this permeability alteration was related to the amount of lung distension. When lung volume was increased 6- to 12-fold from FRC, the epithelium became permeable to large molecules such as albumin (39). Epithelial alveolar permeability alterations are present in varying degrees during VILI. The decrease of the concentration of secretory Clara cell protein in bronchoalveolar lavage ﬂuid in rats (40) and its appearance in the plasma of mice (41) during mechanical ventilation at high lung volume attested to these epithelial permeability alterations. Lung distension also affects microvascular endothelial permeability, in both isolated lung preparations and intact animals. Ventilation of isolated perfused lobes of dogs for 20 minutes with varying PIPs was associated with increases in the capillary ﬁltration coefﬁcient when airway pressure reached 45 to 65 cmH2O (42). Protein reﬂection coefﬁcient decreased only at the highest airway pressure. Major microvascular permeability alterations were also observed in intact rats subjected to 45 cmH2O PIP ventilation (4). Fulminating and lethal pulmonary edema with massive tracheal ﬂooding occurred after only 20 to 30 minutes of ventilation. At autopsy, the lungs were markedly enlarged and congestive (Fig. 1). The severity of the microvascular permeability defect was attested to by the slope of the relationship observed between dry lung weight and extravascular lung water, which indicated that protein concentration in extravasated ﬂuid was close to that of plasma, suggesting the presence of numerous large capillary leaks. Similar ﬁndings relating the duration of high PIP ventilation and alterations in capillary permeability have been made in mice (41). This increase in endothelium permeability occurs both at the alveolar and extra-alveolar level (43). The time course of edema development depends on the size of the species: in animals such as rats, high PIP ventilation for as little as two minutes is sufﬁcient to produce permeability edema (10). In larger animals, longer durations (several hours) of similar high PIP ventilation are required to produce signiﬁcant alterations (12,44). C. Ultrastructural Correlates of Alveolar–Capillary Barrier Permeability Alterations

Both endothelial and epithelial cellular abnormalities have been found after ventilation at high lung volume (4,10,41,45). Some endothelial cells were

Lung Mechanics and Pathological Features During VILI

123

Figure 1 Macroscopic aspect of rat lungs after mechanical ventilation at 45 cmH2O peak airway pressure. (Left): normal lungs; (middle): after ﬁve minutes of high airway pressure mechanical ventilation. Note zones of atelectasis (arrow). (Right): after 20 minutes, the lungs were markedly enlarged and congestive; edema ﬂuid ﬁlls the tracheal cannula. Source: From Ref. 17.

detached from their basement membrane, resulting in the formation of intracapillary blebs ﬁlled with plasma-like material and focal disruptions after 5 to 10 minutes of 45 cmH2O peak airway pressure ventilation (Fig. 2). Bleb formation has been reported in experimental high-permeability edema, regardless of the nature of the causative agent, and in ARDS (46–49) but not in experimental hydrostatic pulmonary edema (46,47,50). Pathologic studies showed that the severe edema that occurred after longer durations (20 minutes) of high peak pressure ventilation in rats was accompanied by diffuse damage to the alveolar–capillary barrier. Severity of alterations varied: whereas in some sites the epithelial lining appeared intact, in many areas ﬁndings included discontinuities (Fig. 3) and sometimes almost complete destruction of type I cells, leaving a denuded basement membrane (Fig. 4). In contrast, type II cells appeared preserved. Hyaline membranes ﬁlled the alveolar spaces in many of the sections examined (Fig. 5). Similar to the endothelial abnormalities, these lesions are not speciﬁc and occur in toxic injuries as well as in ARDS (46,49). They are similar to those observed during increased endothelial cell NO production. Studies using both transmission and scanning electron microscopy have shown breaks in endothelial and epithelial cells when capillary

124

Dreyfuss et al.

Figure 2 Ultrastructural aspect of the blood–air barrier after ﬁve minutes of 45 cmH2O peak airway pressure in a closed-chest rat. The most striking change is the formation of an endothelial bleb due to the disruption and detachment of the thin part of the En from the basement membrane. This bleb (Bleb) is ﬁlled with a material slightly less electron-dense than plasma. At this stage, Ep appear intact. Abbreviations: En, endothelial cell; Ep, epithelial type I cell; AS, alveolar space; RBC, red blood cells. Source: Courtesy of Dr. Paul Soler.

transmural pressure was raised to 40 mmHg or more (51–53). These breaks are the morphological correlates of the increased microvascular permeability reported with very high microvascular pressures (54,55) and described as the stretched pore phenomenon (56,57). Similarities between capillary stress failure and acute VILI are quite remarkable. First, the electron microscopic appearance of endothelial and epithelial cell lesions bear similarities (Fig. 6) (51,52). Second, both types of damage are partly reversible. Albumin leakage from capillaries during high-volume ventilation ceased almost immediately after discontinuation of ventilation (10). When capillary pressures were lowered to normal after elevation to a level causing stress failure, the number of endothelial and epithelial breaks fell compared with control experiments in which pressure remained elevated (58). This reversibility suggests that vascular endothelium response to mechanical stress may be more than purely passive. In vitro studies have shown that cell plasticity and deformation-induced lipid trafﬁcking protect against strain injury (59) and are detailed elsewhere in this book. Besides, as explained in Chapter 4 of this book, blocking nonselective stretch-activated cation channels by gadolinium annulled the increase in microvascular permeability induced by high peak airway

Lung Mechanics and Pathological Features During VILI

125

Figure 3 Ultrastructural aspect of the blood–air barrier after 20 minutes of 45 cmH2O peak airway pressure in a closed-chest rat. Type I cells show numerous gaps (arrows). Abbreviation: AS, alveolar space. Source: From Ref. 17.

pressure in an isolated perfused rat lung model (60). This increase in permeability may thus be initiated by the cellular entry of Ca2þ and the activation of downstream pathways (61,62). Taken together, these results suggest that the increase in microvascular permeability below the cell rupture point, which occurs under extreme stretch conditions [‘‘stress failure’’ (51,63)] is not a passive physical phenomenon, but the result of biochemical reactions. Importantly, capillary stress failure is inﬂuenced by lung inﬂation: at a capillary transmural pressure of 32.5 cmH2O, increasing lung volume from a transpulmonary pressure of 5 to 20 cmH2O resulted in a signiﬁcant increase in the number of capillary endothelium and alveolar epithelium breaks (Fig. 6) (63). Thus, vascular pressures that are too low to affect microvascular permeability when lung volume is normal may produce permeability alterations when combined with a sufﬁcient increase in lung volume. Furthermore, the increase in the number of breaks was roughly the same for comparable rises in transpulmonary pressure on the one hand and capillary transmural pressure on the other, suggesting that ‘‘increases in transpulmonary pressure and capillary transmural pressure are approximately equivalent in terms of

126

Dreyfuss et al.

Figure 4 Epithelial and endothelial cells are almost completely destroyed, resulting in denuded basement membranes (black arrows). A polymorphonuclear neutrophil (Leu) inside the capillary lumen exhibits cytoplasmic processes (white arrows) protruding through gaps in the capillary endothelium. Abbreviation: AS, alveolar space. Source: From Ref. 17.

their effect on capillary wall stress’’ (63). Increased permeability was found at the alveolar (capillary) and extra-alveolar (larger vessels) level during VILI by Parker and Yoshikawa (43). These authors demonstrated that high peak inﬂation pressure of 45 cmH2O increased ﬁltration coefﬁcient in arteries, veins, and capillaries about 4-, 6-, and 10-fold, respectively. A purely mechanistic explanation would rest on calculations like those made by West et al. who proposed the concept of capillary stress failure. They calculated that this capillary stress failure occurs when capillary wall radial stress is roughly equivalent to that of the aorta in the presence of a 100 mmHg transmural pressure (51). Such high transmural pressures are more likely to occur in heterogeneously damaged lungs. D. Inﬂuence of Lung Inhomogeneity on Tissue Stress Level

Mead et al. (64) drew attention to the fact that local tissue pressures might greatly differ from transpulmonary pressure in unevenly expanded lungs. They calculated that the pressure that would expand 10-fold an atelectatic region surrounded by fully aerated lung would be approximately 140 cm H2O at a transpulmonary pressure of 30 cmH2O because of the amplifying effect of the convergence of tissue matrix (Fig. 7). Most of above-mentioned studies were conducted in animals with healthy lungs. The uneven distribution of

Lung Mechanics and Pathological Features During VILI

127

Figure 5 Ultrastructural aspect of the blood–air barrier after 20 minutes of 45 cmH2O peak airway pressure in a closed-chest rat. Complete lysis of the epithelial layer results in denudation of the basement membrane (arrows). AS is occupied by HM composed of cell debris and ﬁbrin (f). Abbreviations: AS, alveolar space; HM, hyaline membranes; En, endothelial cells; In, interstitial edema. Source: From Ref. 4.

Figure 6 Electron micrographs of the blood–gas barrier in newborn rabbit lungs with perfusion ﬁxed at a capillary transmural pressure of 15 cmH2O. (A) disruption of the capillary endothelium (closed arrows) and (B) epithelium (open arrow)—a, alveolus; c, capillary.

128

Dreyfuss et al.

Figure 7 Photographs of a hexagonal array of identical springs subjected to uniform tension at the margins. (A) The central hexagon has the same size as the surrounding hexagons; the tension (which is proportional to length) is the same in all springs. (B) The central hexagon is reduced in size; tension is higher in all springs directly attached to it and progressively decreases in distant ones. Source: From Ref. 64.

alterations consecutive to high-volume ventilation (65) may result in ventilation distribution inhomogeneity and patchy distribution of edema owing to the different behavior of peripheral airspaces. Uneven distribution of ventilation may be exacerbated by high ﬂow rates, and as a matter of fact, it has been shown that high inspiratory ﬂow rates, on their own, have deleterious effects (66). Rabbits ventilated with high VT had signiﬁcantly more gas exchange deterioration, fall in respiratory system compliance (Crs), and macroscopic and microscopic lung injury when peak inspiratory ﬂow rate was higher (67). Conversely, reduction of inspiratory ﬂow resulted in less lung injury (68) in a model of high peak pressure (50 cmH2O) ventilation in sheep. Inhomogeneity is even more pronounced in already diseased lungs that are thus expected to be more sensitive than healthy lungs to the deleterious effects of mechanical ventilation. Lung susceptibility to the obnoxious effects of ventilation with a high tidal volume is thus related to the lung mechanical properties, and these may widely differ from one subject (animal or human being) to another. Thus, there is an increased risk of important local tissue stress during mechanical ventilation of inhomogenous lungs, and it is important to try to delineate this risk with simple measurements such as those easily obtained at the bedside.

III. Respiratory Mechanics and Severity of VILI A. The Respiratory System PV Curve

The monitoring of mechanical ventilation usually relies on airway pressure. This is why the deleterious effects of mechanical ventilation of the lung have often been termed ‘‘barotrauma.’’ In fact, high airway pressure without

Lung Mechanics and Pathological Features During VILI

129

concomitant increase in lung volume (in other words when transalveolar pressure is not increased) does not produce VILI (12,45,69). This shows that the monitoring of respiratory system mechanics should be considered with some caution because of the role played by the thoracoabdominal wall. As a matter of fact, changes in respiratory system PV curve shape have sometimes been attributed to changes in wall compliance (70). Indeed, it would be better to directly measure lung mechanical properties; but there are no means in routine clinical or experimental practice to determine precisely the overall state of lung inﬂation and, more importantly, whether global or regional overdistension occurs. Monitoring static airway pressure may, however, help in detecting overinﬂation because there is a direct correspondence between end-inspiratory (quasistatic) airway pressure (and not peak pressure) and maximal transalveolar pressure in small animals with high chest wall compliance. In such conditions, the higher the end-inspiratory airway pressure, the higher the likelihood of lung overdistension. This correspondence between endinspiratory pressure and maximal transalveolar pressure also holds for patients when the wall compliance is not too far from that of a normal population, for the above-mentioned reasons. In such a case, end-inspiratory pressure value may warn about the risk of excessive tissue stress (Fig. 8). It is thus important to understand which pertinent information can be obtained from a respiratory system PV curve and how it is modiﬁed by acute lung injury and by VILI, in order to recognize when a particular ventilation modality may aggravate (create) lung injury. The respiratory system PV curve usually displays an S shape, with the lower horizontal arm of the S more or less pronounced depending on the presence of injury or of ﬂuid in airspaces (Fig. 8). Several mathematical models (72–74) have been used to describe this curve, all of which aimed at quantifying three parameters describing curve shape: position (pressure or volume) at which the lower change of slope is observed (called ‘‘lower inﬂection point’’ or LIP), position at which the upper change of slope is observed (called ‘‘upper inﬂection point’’ or UIP), and the slope of the linear part of the curve between these points (the Crs).

IV. Respiratory System PV Curve Changes During Lung Injury Acute lung injury is characterized by excess ﬂuid in lung interstitium and distal airspaces, inactivation of surfactant by plasma proteins in airspace ﬂuid, and airway obstruction by edema foam. Water accumulation in interstitium per se has little inﬂuence on lung elasticity (75,76). Changes in lung and thus respiratory system PV curve have been described during

130

Dreyfuss et al.

Figure 8 Lung PV curves of patients with pulmonary ﬁbrosis (which produces a ‘‘baby lung’’ syndrome: decreased FRC and vital capacity because of loss of aerated lung volume), drawn using lung volume as a percentage of predicted total lung capacity. Note that inﬂating lungs from FRC with the same VT of 10% TLC (which is basically the same as delivering VT tailored on predicted BW) might result in very different end-inspiratory pleural pressures (arrows), and thus tissue stress. Normal values lie within the shaded area. Abbreviations: BW, body weight; FRC, functional residual capacity; TLC, total lung capacity; PV, pressure–volume; VT, tidal volume. Source: From Ref. 71.

experimental pulmonary edema of the permeability type, produced either by injection of toxics like oleic acid or a-naphthylthiourea (ANTU), or during VILI development.

A. Effect of Lung Injury on the Crs and Position of the UIP

The decrease in lung compliance observed during pulmonary edema (Fig. 9) is not the result of a change in lung tissue elastic properties that are mainly due to the presence of elastic and collagen ﬁbers, but of the loss of ventilatable units (76). This decrease in ventilatable lung volume has been called ‘‘baby lung’’ in the context of ARDS (77) or ‘‘shrunken lung’’ in that of pulmonary ﬁbrosis (71). The hypothesis that pulmonary edema development reduces ventilatable lung volume essentially because of distal airway obstruction was examined in rats (74). When distal airways of rats were obstructed by instilling a viscous liquid, the changes in the shape of the PV curve (gradual decrease in compliance and in the volume at which

Lung Mechanics and Pathological Features During VILI

131

Figure 9 Changes in the respiratory system PV curve during pulmonary edema development. The curve on the left was obtained in a normal rat. During edema development, PV curves are shifted to the right. An LIP progressively appears whereas the slope (Crs) of the line drawn on the linear part of the curve (dotted line) progressively decreases. The UIP is the change in slope seen at high lung volume. Abbreviations: Crs, respiratory system compliance; LIP, lower inﬂection point; PV, pressure–volume; UIP, upper inﬂection point. Source: From Ref. 74.

the UIP was seen, and progressive increase in end-inspiratory pressure) were very similar to those seen during the development of pulmonary edema. The decrease in Crs and in the UIP volume (but not UIP pressure that remained essentially the same) were similarly correlated during edema development and airway obstruction, suggesting that both reﬂect the amount of ventilatable lung volume (Fig. 10).

B. Explanations for the LIP

The LIP seen in atelectasis-prone lungs (isolated lungs, open-chest, and surfactant depletion models) may be the consequence of re-expansion of atelectatic parenchyma. Using in vivo video microscopy, Schiller and colleagues (78–82) directly observed and quantiﬁed the dynamic changes in alveolar size throughout the ventilatory cycle during tidal ventilation in normal lungs and in surfactant-deactivated lungs. In normal lungs, alveoli never collapsed, in accordance with former ﬁndings from Wilson and Bachofen that showed that alveoli do not change volume appreciably during ventilation (83). Collapse and reopening and increase in the alveolar size at end-inspiration were observed in surfactant-deactivated lungs. In a

132

Dreyfuss et al.

Figure 10 (A) Instillation of increasing amounts of a viscous liquid in airways progressively decreased both Crs and the volume of the UIP in rats (each animal is represented by closed circles connected by a line), (B) the same decreases and relationship was observed during pulmonary edema (permeability type) development. The dotted polygon represents the envelope of values obtained with liquid instillation. Abbreviations: Crs, respiratory system compliance; UIP, upper inﬂection point. Source: From Ref. 74.

Lung Mechanics and Pathological Features During VILI

133

subsequent study, the same authors (82) found that application of PEEP to a surfactant-deactivated lung decreased the changes in alveolar area from inspiration to expiration, reﬂecting stabilization of terminal lung units. In contrast with these ﬁndings in surfactant-depleted lungs, using a parenchymal marker technique Martynowicz et al. (84) showed that oleic acid injury did not affect FRC (even in dependant lung regions) in dogs and that lung volume distribution between dependant and nondependant regions did not change between FRC (below the LIP) and higher lung volumes, which suggested that increasing lung volume did not recruit atelectatic zones in this model. Wilson et al. (85) suggested that the LIP in edematous lungs might reﬂect the mechanical properties of lungs with partially ﬂuid-ﬁlled alveoli and constant surface tension instead of being the result of abrupt opening of airways or recruitment of atelectatic parenchyma. Obstructing airways of a rat by instillation of increasing amounts of liquid results in changes in the respiratory system PV curve that are depicted in Figure 11. The curve ﬂattens, the volume of the UIP decreases and another inﬂection, with an upward concavity, appears initially at a high lung volume above the UIP. When a sufﬁciently high number of airways are obstructed, the volume at which this upward concavity occurs decreases and the change in slope appears to be quite similar to the LIP, as described in Figure 9. A possible explanation for these changes is presented in Figure 12. C. Respiratory System PV Curve and Susceptibility to VILI During Ventilation at High Lung Volume

It is usual, both in experimental studies and in clinical trials (this point will be further discussed below), to tailor VT based on animal or patient weight. This does not account for the variability of lung size. For example, there is about 20% variability in lung size (compared to 3% for the brain) in a normal Sprague-Dawley rat population (86). It is thus conceivable that delivering the same high VT in mL/kg BW would result in different VILI courses depending on actual lung size. As a matter of fact, it has been observed that the accumulation rate of plasma proteins in the lungs (traced with 111Intransferrin) widely differed between animals that were given the same 38 mL/kg ventilation (87). The higher risk for VILI could be predicted from the respiratory system PV curve shape. The higher was the compliance and the volume of the UIP before ventilation (and thus the higher was lung size), the lesser was 111In-transferrin accumulation and thus capillary permeability alterations (Fig. 13). It is conceivable that edema foam in airways of injured lungs reduces the number of alveoli that receive the tidal volume, exposing them to overinﬂation and rendering them more susceptible to injury, further

134

Dreyfuss et al.

Figure 11 Inﬂation PV curves during the successive instillation of 50- to 100-mL aliquots of a viscous solution into airways. Before instillation (left), the PV curve has a normal shape. The UIP (small arrow) is determined by ﬁtting the curve with joined linear segments. The instillation of the solution progressively distorted the shape of the curve, which was shifted to the right. The usual PV shape is shrunken but remains recognizable at low pressure below the LIP, which is easily identiﬁable in the lowest curve (open arrow). A second UIP is found above the LIP, at high airway pressures (large arrow). Abbreviations: LIP, lower inﬂection point; PV, pressure– volume; UIP, upper inﬂection point. Source: From Ref. 74.

reducing aerated lung volume and resulting in positive feedback. To explore this possibility, alveolar ﬂooding was produced by instilling saline into the trachea of rats that were immediately ventilated with tidal volumes of up to 33 mL/kg (88). Flooding by itself did not affect microvascular permeability, but capillary permeability alterations were signiﬁcantly more pronounced in ﬂooded than in intact animals during ventilation with high tidal volume. The effect of high-volume ventilation on injured lungs was investigated by comparing different degrees of lung distention in rats that received ANTU (65). ANTU infusion alone caused moderate interstitial pulmonary edema of the permeability type. It was possible to calculate how much mechanical ventilation would theoretically injure lungs diseased by ANTU by summing up the separate effects of mechanical ventilation alone or ANTU alone on edema severity (Fig. 14). Animals injured by ANTU and ventilated at high VT (45 mL/kg BW) had more severe permeability pulmonary edema than predicted, indicating synergy between the two insults.

Lung Mechanics and Pathological Features During VILI

135

Figure 12 Diagram showing possible consecutive lung states during inﬂation depending on the amount of edema. Airway pressures are the same for each of the successive states (numbered 1–4) in (A) and (B). For the sake of simplicity, territories excluded from ventilation are depicted by airway closure. This does not reﬂect the precise site or the mechanism of exclusion from ventilation (alveolar spaces ﬁlled with liquid, foam in airways, bronchial edema, etc.). (A) shows how a limited volume of excluded (but potentially recruitable) lung results in a shrunken PV curve, with the ‘‘usual’’ shape at low airway pressure. 1–2: Inﬂation of ventilatable areas; 3: the distensibility of these zones is decreased because of overinﬂation; 4: recruitment of excluded territories at airway pressures above the opening pressure. (B) shows how a large volume of excluded but potentially recruitable lung results in a PV curve displaying a LIP. 1–3: Progressive inﬂation of a small number of open units; 4: simultaneous opening of excluded volumes resulting in a steep increase in the PV curve slope when airway pressure exceeds the opening pressure. Abbreviations: LIP, lower inﬂection point; PV, pressure–volume. Source: From Ref. 74.

The extent to which aerated lung volume was reduced prior to ventilation was a key factor in this synergy. The amount of pulmonary edema produced by high-volume mechanical ventilation in animals given ANTU was inversely proportional to the Crs (65,74) or the volume of the UIP (74) measured at the very beginning of mechanical ventilation. The airway pressure–time curve during mechanical ventilation with constant ﬂow is another way of depicting respiratory system mechanical properties. An upward concavity of this curve reﬂects that compliance decreases as tidal volume is delivered, and thus that end-inspiratory volume

136

Dreyfuss et al.

Figure 13 Accumulation of a systemic protein tracer in the lungs (111Intransferrin) after a period of high-volume ventilation in rats was correlated with the volume at which the UIP was found on the respiratory system PV curve before beginning high-volume ventilation. Abbreviations: PV, pressure–volume; UIP, upper inﬂection point. Source: From Ref. 87.

is situated above the UIP. Such a ventilation pattern was associated with histological lung injury in isolated, nonperfused, lavaged lungs (77) and in a model of surfactant depletion in the pig (89). Taken together, these results conﬁrm that the position of the UIP is a marker of the amount of ventilatable lung volume and is both inﬂuenced by and predictive of the development of edema during mechanical ventilation. Lung instability, produced by surfactant inactivation (for example in the saline lavage model), promotes atelectasis that, besides reducing aerated lung volume, will locally shrink the lung and expose lung tissue to increased risk of high stress (Fig. 7). Spontaneous breathing during prolonged anesthesia results in degradation of surfactant activity and promotes focal atelectasis. Even such minor alterations (65) were sufﬁcient to increase the harmful effects of high-volume ventilation (Fig. 15), providing indirect validation of the concept of interdependence (64). D. Is There a Threshold for the Susceptibility of Lungs to High-Volume VILI?

The UIP on the respiratory system inspiratory PV curve in patients with ARDS has been interpreted as reﬂecting the beginning of overinﬂation (90,91), but may also reﬂect the end of recruitment (92,93) in edematous lungs. Whereas it has been shown that the level above the UIP at which end-inspiration occurs is correlated to the risk of acute permeability edema (74), whether ventilation below the UIP would be safe is unsettled.

Lung Mechanics and Pathological Features During VILI

137

Figure 14 Interaction between previous lung injury and mechanical ventilation on pulmonary edema: Effect of previous toxic lung injury. Qwl after mechanical ventilation in normal rats (open circles) and in rats with mild lung injury produced by ANTU (closed circles). Tidal volume varied from 7 to 45 mL/kg BW. The solid line represents the Qwl value expected for the aggravating effect of ANTU on edema caused by ventilation, assuming additivity. ANTU did not potentiate the effect of ventilation with VT up to 33 mL/kg BW. In contrast, VT at 45 mL/kg BW produced an increase in edema that greatly exceeded additivity, indicating synergy between the two insults. Abbreviations: ANTU, a-naphthylthiourea; BW, body weight; Qwl, extravascular lung water; VT, tidal volume. Source: From Ref. 65.

Overinﬂation (extraphysiological inﬂation) may be deﬁned as the volume above which ventilating lungs is unsafe. It is unclear whether there is a true threshold for VILI, particularly in previously diseased lungs. Initial studies on VILI were only concerned with the effects of acute distension and the changes in capillary permeability produced by high lung stretch. Later, it was shown that ventilating lungs with lower distending pressures for longer times produced different lesions, mainly of the inﬂammatory type. The airway pressure below which ventilating lungs, and particularly inhomogenously diseased lungs, is safe is still unknown because it would require a simple means to monitor regional volume changes. Thus, there certainly is a lung level above which overinﬂation takes place, but this level still remains unknown and is probably variable as a function of the severity of previous lung injury and of individual susceptibility. Considering endinspiratory pressure rather than tidal volume scaled by predicted BW when setting ventilation parameters is a clinically relevant issue and questions the value of setting ventilation parameters only on the basis of a theoretical

138

Dreyfuss et al.

Figure 15 Effect of lung functional alteration by prolonged anesthesia. Intact rats were anesthetized and breathed spontaneously for 30 to 120 minutes before ventilation with 7 mL/kg BW (open bars) or 45 mL/kg BW (shaded bars). Qwl of animals ventilated with a high VT was signiﬁcantly higher than in animals ventilated with a normal VT. Qwl was not affected by the duration of anesthesia in animals ventilated with a normal VT. In contrast, 120 minutes of anesthesia before high VT ventilation resulted in a larger increase in Qwl than did 30 minutes of anesthesia (**p < 0.01). Abbreviations: BW, body weight; VT, tidal volume; Qwl, extravascular lung water. Source: From Ref. 65.

phenotype value. This point will be discussed in the section devoted to the clinical relevance of the concept of VILI. A threshold volume for capillary permeability changes as the lung is ‘‘overexpanded’’ was suggested by Carlton et al. (12). PIP (and therefore tidal volume) was increased in three successive steps (each lasting four hours) from 16 to 61 cmH2O in lambs. No change in lymph ﬂow or protein composition was observed until the highest pressure was reached (corresponding to a tidal volume of 57 mL/kg VT). However, the albumin–globulin ratio in lymph versus plasma decreased before this maximum pressure was reached, suggesting altered protein sieving. Although not speciﬁcally discussed in the article, this ﬁnding may indicate either that the threshold was lower than the maximum value studied or that there was no threshold. In smaller species such as rats mechanical ventilation– induced lung damage develops more rapidly. Indeed, edema was observed after 30 minutes of mechanical ventilation with 30 cmH2O peak airway pressure in this species (3), whereas Tsuno and coworkers (94) found that more than 40 hours were required to observe an increase in lung weight and leukocyte recruitment when sheep were ventilated with this same PIP. This inﬂammation may be the consequence of the cellular lesions

Lung Mechanics and Pathological Features During VILI

139

produced by the repetition of microtrauma (either due to local overinﬂation, or to ventilation at low lung volume as explained below) or by the release of inﬂammatory mediators by lung cells as a consequence of ‘‘extraphysiological’’ mechanical deformation (95–98), or both. Indeed, leukocyte sequestration may occur early in the course of VILI (99). However, there is no evidence (16) that the early permeability defect seen during high-volume ventilation is due to an inﬂammatory process. There has, however, been no study speciﬁcally designed to clarify whether there is a true threshold below which ventilation of injured lungs might be safe. E. VILI During Ventilation at Low Lung Volume

Ventilation of edematous lungs at low volume may produce airway epithelium injury, probably, in part, because of the increased shear produced by the movement of foam in airway lumen. These aspects are considered in Chapter 7 of this book. Another potential harmful effect of ventilation at FRC is cyclic collapse and re-expansion of atelectasis-prone lungs, as it is the case, for example, in surfactant-depleted lungs. Increasing FRC by PEEP is the usual means to avoid these harmful events. Several studies have investigated the effects of conventional mechanical ventilation on acutely injured lungs and have shown that high PEEP levels (high enough to keep small airways ‘‘open’’ during the whole ventilatory cycle) might lessen deterioration. Most of the available data on this issue were generated by Sykes and coworkers (6,7), who studied the effect of ventilation on surfactant-depleted lungs. Rabbits were ventilated with a PIP ranging from 15 mmHg at the beginning of the experiment to 25 mmHg at the end (ﬁve hours later) because of the fall in lung compliance (VT was not stated), and a PEEP adjusted to set FRC either above or below the LIP. Mortality rate was not inﬂuenced by the PEEP level, but arterial oxygen tension (PaO2) was better preserved in the high PEEP group (6,7). There was less HM formation in animals ventilated with a high PEEP than in those ventilated with a low PEEP. This lessening of pathological alterations was observed even with ventilatory settings responsible for identical mean airway pressure in the low and high PEEP groups (7). Comparable results were reported in isolated, nonperfused, lavaged rabbit lungs ventilated with a low tidal volume and with a level of PEEP set either below or above the inﬂection point (8). PEEP may prevent diffuse lung damage during prolonged ventilation by stabilizing distal lung units (100). It is noteworthy that Sykes and colleagues could not replicate the same ﬁndings in hydrochloric acid–injured rabbit lungs using the same ventilatory settings (101). Thus, it is conceivable that the protective effect of PEEP set above the LIP of the PV curve is observed only in the very special context of surfactant deﬁciency (atelectasis-prone lung) but not during pulmonary edema because lung instability and airspace collapse are more

140

Dreyfuss et al.

prominent during the former. The reality of the repetitive opening and closure of terminal units and the signiﬁcance of the LIP on the PV curve in this context have been recently challenged by Martynowicz and coworkers. Using the parenchymal marker technique, these authors observed that sinusoidal oscillation of lung volume at FRC of oleic acid–injured lungs did not suggest cyclic reopening and collapse in dependant zones (Fig. 16). Their ﬁndings thus do not support the hypothesis that a superimposed gravitational pressure gradient during pulmonary edema produces compression atelectasis of dependent lung that in turn produces shear injury from cyclic recruitment and collapse (84). The protective effect of PEEP is however not restricted to the peculiar situation of surfactant depletion because it was also observed during VILI (3,4,102). The beneﬁcial effect of PEEP is however variable in different

Figure 16 Effects of PEEP on volume and ventilation of a dependent region of a dog lung injured with oleic acid (the volume of the region is expressed as a percentage of TLC). Oleic acid edema resulted in the exclusion of this region from ventilation (derecruitment) in the absence of PEEP. Application of 7.5 cmH2O PEEP for 20 minutes was associated with the reappearance of low-amplitude volume oscillations. Application of 15 cmH2O PEEP for 20 minutes restored normal volume oscillations. Abbreviations: PEEP, positive end-expiratory pressure; TLC, total lung capacity. Source: From Ref. 84.

Lung Mechanics and Pathological Features During VILI

141

models of lung injury, PEEP being less effective in models of lung inﬂammation such as after intratracheal endotoxin instillation (103). As explained above, occurrence of a LIP may represent the transition from the liquid-ﬁlled state to the air-ﬁlled state in edematous lungs (85). There would thus not be an abrupt expansion of distal lung units when the LIP is reached. This casts some doubt on the usefulness of LIP determination for the prevention of VILI (104). The difﬁculty in proving whether or not PEEP exerts a protective effect on VILI may be explained by taking into account the distinction between atelectatic and ﬂuid-ﬁlled distal airways in the interpretation of the LIP. Indeed, in the case of diffuse ﬁlling of distal airways with liquid, VILI may occur through overdistension of already aerated zones rather than because of shear stress due to opening of collapsed zones (105). If so, it would be difﬁcult to expect any reduction of injury by PEEP in this context. F. Effect of PEEP on High-Volume VILI Development

For the same level of overall lung inﬂation, increasing FRC by PEEP (and thus reducing tidal volume) results in less severe alterations, the reasons for which may be reduction of VT at the same respiratory rate, and hence of peak inspiratory ﬂow, or stabilization of terminal units. Another factor may be a decrease in microvascular ﬁltration pressure because of hemodynamic alteration, as will be discussed in the next section. In perfused canine lobes, Bshouty et al. (106) found that for equivalent perfusion ﬂow rates and microvascular hydrostatic pressures, the rate of hydrostatic edema formation increased with tidal volume. When an identical increase in mean airway pressure was achieved either by applying PEEP or by increasing tidal volume, edema was less marked under the former condition, suggesting that large cyclic changes in lung volume promote the development of edema. This explanation was also put forward by Corbridge et al. (107), who observed in hydrochloric acid–injured dog lungs that ventilation with a large tidal volume and a low PEEP resulted in more severe edema than ventilation with a small tidal volume and a high PEEP. Webb and Tierney showed that for a given level of teleinspiratory pressure (i.e., overall level of pressure), edema was less severe when a 10 cmH2O PEEP was applied (3). It was subsequently shown (45) that although the amount of edema ﬂuid was smaller with PEEP and no alveolar ﬂooding was observed on light microscopic examination (3,45), capillary permeability alterations were similar to those observed with zero end-expiratory pressure (ZEEP). However, whereas diffuse alveolar damage was present in the animals ventilated with ZEEP, no epithelial cell alterations were observed on electron microscopic examination of the lungs of animals ventilated with PEEP. The only ultrastructural alterations consisted of endothelial blebbing.

142

Dreyfuss et al.

This preservation of the alveolar epithelial layer has received no satisfactory explanation. It may be that PEEP eliminated repetitive opening and closing of terminal airways, thereby decreasing shear stress at this level (the uncertainties on this potential mechanism were explained above). Alternatively, avoidance of alveolar ﬂooding by PEEP (108,109) may have protected the epithelial lining from injury through the effects of as yet unidentiﬁed humoral mediators. Finally, the potential role of hemodynamic alterations during PEEP ventilation should also be considered. For instance, rats ventilated with 45 cmH2O peak airway pressure and 10 cmH2O PEEP had more edema when PEEP-induced hemodynamic alterations were corrected by dopamine administration (110). Moreover, arterial blood pressure was found to be signiﬁcantly correlated with the amount of pulmonary edema under such conditions. The reason why use of PEEP during ventilation with high peak airway pressure is associated with reductions in both the amount of edema and the severity of cell damage may be a combination of hemodynamic alterations, shear stress reduction, and surfactant modiﬁcations.

G. Adverse Effects of Too High PEEP

It should be borne in mind that this beneﬁcial effect of PEEP during overinﬂation edema contrasts with the usual lack of reduction or even increase in edema reported with PEEP in most forms of experimental pulmonary edema (111). Moreover, prophylactic application of expiratory positive airway pressure (a form of PEEP) did not reduce edema formation during oleic acid injury (112). It would be inappropriate to conclude from the data summarized above that tidal volume, which governs opening and closure of terminal units, is the sole determinant of VILI. On the contrary, the overall degree of lung distension (i.e., teleinspiratory volume) is probably the crucial factor, as explained in a previous section of this chapter. Hence, rats ventilated with a tidal volume within the physiologic range at two levels of PEEP (10 and 15 cmH2O) developed pulmonary edema only at the higher level of PEEP (Fig. 17) (110). Similarly, doubling tidal volume had no effect in animals ventilated with ZEEP but resulted in pulmonary edema when 10 cmH2O PEEP was used (110). Thus, the safety of small tidal volumes depends on whether or not FRC is increased. Raising end-inspiratory volume by increasing FRC may cause lung injury independently of tidal volume (110). It is now clear that mechanical ventilation–induced lung injury occurs whenever a certain degree of lung overinﬂation is reached, by whatever means. For a given level of teleinspiratory pressure (and volume), adjunction of PEEP seems to slow the development or diminish the severity of alterations (102,110) but does not prevent the occurrence of permeability pulmonary edema (45,110).

Lung Mechanics and Pathological Features During VILI

143

Figure 17 Effect of increasing PEEP from 0 to 15 cmH2O during ventilation with two VT values (7 mL/kg of BW ¼ Lo VT; 14 mL/kg BW of BW ¼ Med VT). Increasing PEEP resulted in pulmonary edema (as evaluated by extravascular lung water). The level of PEEP required to produce edema varied with tidal volume. No edema was observed until a level of 15 cmH2O PEEP was reached during ventilation with a low VT but occurred at 10 cmH2O PEEP during ventilation with a moderately increased VT. Key: , p < 0.05; , p < 0.01 vs. ZEEP and the same VT. Abbreviations: BW, body weight; PEEP, positive end-expiratory pressure; VT, tidal volume; ZEEP, zero end-expiratory pressure. Source: From Ref. 110.

V. Improvement of Lung Mechanical Properties and Protection from VILI A. Partial Liquid Ventilation

Partial liquid ventilation (113) with perﬂuorocarbons has been developed during the last decade as an alternative to conventional gas mechanical ventilation for the treatment of acute respiratory failure because it may improve gas exchange, lung mechanics, and lung histology during different models of acute respiratory failure (surfactant depletion, oleic acid injury, hydrochloric acid injury, and prematurity) (114). A dose-dependent improvement in gas exchange has been observed when Perﬂubron (LiquiVent1) was instilled in the trachea of rabbits with surfactant-depleted lungs (115). But large volumes of perﬂuorocarbon combined with high PEEP or

144

Dreyfuss et al.

increased tidal volume resulted in barotrauma (namely pneumothorax) (116). Perﬂuorocarbon may reduce the mechanical nonuniformity of diseased lungs by suppressing air–liquid interfaces and allowing reopening of collapsed or liquid-ﬁlled areas. This would render lungs less susceptible to VILI. As a matter of fact, the severity of hyperinﬂation-induced lung injury during alveolar ﬂooding (this situation was mimicked by instilling saline into airways, as explained above) in rats was lessened by perﬂuorocarbon instillation (88). Tracheal instillation of a low dose (3.3 mL/kg) of Perﬂubron (LiquiVent1) in these ﬂooded lungs considerably reduced VILI and decreased permeability alterations. Whereas saline instillation alone raised LIP pressure to values as high as 25 cmH2O and produced a signiﬁcant increase in the end-inspiratory pressure, administration of Perﬂubron signiﬁcantly reduced this pressure and normalized end-inspiratory pressure. To further investigate the effect of partial liquid ventilation on VILI with respect to the dosage of perﬂuorocarbon, intact animals received increasing doses of Perﬂubron (LiquiVent1) (from 6 to 20 mL/kg) (117). Hyperinﬂation-induced pulmonary edema tended to decrease in animals that were administered doses of Perﬂubron lower than l0 mL/kg as compared to animals not given Perﬂubron. By contrast, ventilator-induced pulmonary edema was aggravated in animals given 13 and 16 mL/kg, and even more in animals given 20 mL/kg Perﬂubron. End-inspiratory pressure was signiﬁcantly correlated with the capillary permeability alterations. This conﬁrms that administration of small doses of Perﬂubron (< 10 mL/kg) tends to decrease hyperinﬂation edema and thus protects lungs against volutrauma (88). The worsening of VILI by large doses of Perﬂubron was due to its effects on FRC. Indeed, such doses resulted in an increase in FRC, thus in end-inspiratory volume for the same VT, and favored gas trapping in the distal lung. Gas trapping was indeed evidenced by computed tomography (CT) imaging (117). These observations suggest that monitoring the endinspiratory pressure could help detect the risk of volutrauma during partial liquid ventilation. B. Prone Positioning

Prone position lessened the deleterious effect of high PIP ventilation in dogs with previous oleic acid lung injury (118), in rabbits (119), and in rats (120). Prone position resulted in a more homogenous distribution of lung injury (121), an observation that was not conﬁrmed by the other studies in rabbits or rats. These differences are probably due to difference in lung sizes, larger lung size increasing the inﬂuence of gravity. Indeed, CT scans in rats did not disclose a gradient of lung inﬂation at end-expiration as in humans (120). In rats, the beneﬁcial effect of prone position was ascribed to a more homogenous distribution of VT and thus of strain, because of the downward displacement of the diaphragm (120).

Lung Mechanics and Pathological Features During VILI

145

VI. Clinical Considerations The most important feature of VILI is that it is not the consequence of barotrauma but rather of volutrauma (45) and that end-inspiratory volume (the overall lung distension) seems to be the major determinant of VILI rather than tidal volume or FRC (which is dependent on the level of PEEP) (110). For many years, clinicians were only concerned with improving gas exchange during ARDS, without taking into account the possibility that mechanical ventilation may promote further injury. This is why high tidal volumes were applied, ranging from 12 to 24 mL/kg BW (122–124). However, the awareness of the experimental concept of VILI ﬁnally resulted in profound changes in the goals of mechanical ventilation. The term ventilator-associated lung injury (VALI) (125) formed the rationale behind all lung protective strategies that aimed to reduce the risk of lung overinﬂation during ventilation of patients with ARDS. Interestingly, the validity of this concept could not be directly demonstrated in patients but the physiological rationale was sound enough to support the progressive reduction of tidal volume over the years. For instance, a recent international survey showed that the mean tidal volume was less than 9 mL/kg BW during ARDS (126). The heterogeneity of lungs during ARDS was well documented (77,127), resulting in the coexistence of healthy tissue, recruitable tissue, and diseased tissue unresponsive to pressure changes. Healthy units may represent as little as 20% to 30% of total units, forming a ‘‘baby lung’’ (77). If a ‘‘normal’’ (10–12 mL/kg BW) tidal volume is applied to such lungs, it may result in severe overdistension of these healthy units, which may receive the equivalent of 40 mL/kg BW, precisely the value that consistently results in severe VALI in animals. The simplest way to avoid VALI was a reduction of tidal volume with resulting permissive hypercapnia (128). The publication of the ARDS Network study (5) has unambiguously demonstrated the clinical relevance of available experimental data. Indeed, mortality decreased from 40% in patients ventilated with a tidal volume of 12 mL/kg of ideal BW to 31% in those patients ventilated with a tidal volume of 6 mL/kg. This was associated with a marked reduction in plateau pressure, lending credence to the hypothesis that reduction of lung stress was responsible for this improvement in prognosis (129). But this study also raises interesting questions: the same reduction in tidal volume was employed in all patients allocated to the low tidal volume group in the ARDS Network trial, without taking into account the individual lung mechanical properties, which were not necessarily identical in all patients (130). Indeed, the pressure and volume considered safe for some patients with ARDS may cause lung overdistension in others (77,90,91,131). It may then be that unnecessary reduction in tidal volume was applied to some patients, leading to unwanted respiratory acidosis and the need for deeper sedation and paralysis. These uncertainties underline two important facts.

146

Dreyfuss et al.

First, better tailoring of tidal volume could be obtained from the analysis of the inspiratory PV curve of the respiratory system. Indeed, the risk of overinﬂation would probably be better appreciated if the signiﬁcance of its particular characteristics (LIP, UIP) is better understood (74). Second, similar to the difﬁculty in determining whether or not a threshold tidal volume exists below which experimental VILI does not occur in previously injured lungs (this point was previously discussed in this chapter), the ARDS Network study (5) was not designed to answer this question directly but reported the interesting ﬁnding that the beneﬁt of ventilation with a lower tidal volume was independent of the static compliance of the respiratory system at baseline. This ﬁnding would suggest the absence of threshold. However, this interpretation was debated (130). Depending on the actual existence of such a threshold, tidal volume may be reduced only to the value that ensures a plateau pressure within the so-called safe limits (125,129) or should be drastically reduced in all patients. Whatever the response to this uncertainty, this clinical study (5) strongly suggests that end-inspiratory lung volume is the main determinant of volutrauma (as contended at the beginning of this section) because reduction of mortality was achieved by simply reducing end-inspiratory lung stress. Moreover, this contention is supported by the recent report that higher tidal volume may be associated with the onset of acute lung injury (132) or ARDS (133) in patients ventilated for other causes of respiratory failure. Another important piece of information was derived from the recently published result of a multicenter study that compared low and high PEEP levels in patients with ARDS ventilated with 6 mL/kg BW (134) and showed no difference of mortality between groups. This suggests that, contrary to the experimental and clinical demonstration of the importance of reducing lung stretch by reducing tidal volume, the clinical relevance of the concept of low lung volume injury (discussed above) is less obvious. Another possibility for reducing the risk of VALI is to improve lung mechanical properties independently of the implementation of lung protective strategies. Unfortunately, treatments aimed at improving lung mechanical properties, namely surfactant administration and partial liquid ventilation, have not resulted in improved prognosis in the majority of trials. This failure could have been predicted on physiological and experimental grounds. Surfactant treatment did not improve mortality in two studies in adults with ARDS, but this treatment did not markedly improve oxygenation or lung mechanics, suggesting their lack of physiological efﬁcacy (135,136). A recent study of a novel natural surfactant containing high levels of surfactantspeciﬁc protein B in a pediatric population with acute lung injury (premature infants were not included) was associated with increased survival (137). But this was not associated with an increase in ventilator-free days despite the fact that the oxygenation index (which takes into account both lung mechanical properties and oxygenation) was markedly improved.

Lung Mechanics and Pathological Features During VILI

147

The clinical relevance of VILI was highlighted by the recent failure of a multicenter clinical trial of partial liquid ventilation with Perﬂubron (a perﬂuorocarbon) during mechanical ventilation of adult patients with acute lung injury (138). Markedly higher inspiratory pressures were observed in patients receiving Perﬂubron because the amount of administered liquid was high and the protocol promoted the use of a high level of PEEP ( >13 cmH2O). As a result of this very high distending pressure, the incidence of macroscopic barotrauma was very high, more than double that in the controls (17% vs. 6%, p < 0.05). More worrisomely, mortality was higher (although not signiﬁcantly) in patients receiving liquid ventilation. Careful analysis of experimental studies easily explains (and could predict) these results. Indeed, increased incidence of barotrauma (116) and worsening of VILI (88,117) were observed in experimental studies when both the amount of instilled perﬂuorocarbon and the pressures delivered by the respirator were high. In conclusion, the experimental concept of VILI allowed fundamental breakthroughs in the treatment of ARDS patients. A better comprehension of the mechanisms underlying VILI and of the role of lung mechanics monitoring would promote reﬁnement of ventilator strategies to allow further progress in the management of this still deadly disease. References 1. Nash G, Bowen JA, Langlinais PC. Respirator lung: a misnomer. Arch Path 1971; 21:234–240. 2. Greenﬁeld LJ, Ebert PA, Benson DW. Effect of positive pressure ventilation on surface tension properties of lung extracts. Anesthesiology 1964; 25:312–316. 3. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110:556–565. 4. Dreyfuss D, Basset G, Soler P, Saumon G. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 1985; 132:880–884. 5. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342:1301–1308. 6. Argiras EP, Blakeley CR, Dunnill MS, Otremski S, Sykes MK. High peep decreases hyaline membrane formation in surfactant deﬁcient lungs. Br J Anaesth 1987; 59:1278–1285. 7. Sandhar BK, Niblett DJ, Argiras EP, Dunnill MS, Sykes MK. Effects of positive end-expiratory pressure on hyaline membrane formation in a rabbit model of the neonatal respiratory distress syndrome. Intensive Care Med 1988; 14:538–546.

148

Dreyfuss et al.

8. Muscedere JG, Mullen JBM, Gan K, Bryan AC, Slutsky AS. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med 1994; 149:1327–1334. 9. Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351:327–336. 10. Dreyfuss D, Soler P, Saumon G. Spontaneous resolution of pulmonary edema caused by short periods of cyclic overinﬂation. J Appl Physiol 1992; 72: 2081–2089. 11. Parker JC, Hernandez LA, Longenecker GL, Peevy K, Johnson W. Lung edema caused by high peak inspiratory pressures in dogs. Role of increased microvascular ﬁltration pressure and permeability. Am Rev Respir Dis 1990; 142:321–328. 12. Carlton DP, Cummings JJ, Scheerer RG, Poulain FR, Bland RD. Lung overexpansion increases pulmonary microvascular protein permeability in young lambs. J Appl Physiol 1990; 69:577–583. 13. Prewitt RM, McCarthy J, Wood LDH. Treatment of acute low pressure pulmonary edema in dogs. Relative effects of hydrostatic and oncotic pressure, nitroprusside and positive end-expiratory pressure. J Clin Invest 1981; 67:409–418. 14. Huchon GJ, Hopewell PC, Murray JF. Interactions between permeability and hydrostatic pressure in perfused dogs’ lungs. J Appl Physiol 1981; 50: 905–911. 15. Guyton AC, Lindsey AW. Effect of elevated left atrial pressure and decreased plasma protein concentration on the development of pulmonary edema. Circ Res 1959; 7:649–653. 16. Yoshikawa S, King JA, Lausch RN, Penton AM, Eyal FG, Parker JC. Acute ventilator-induced vascular permeability and cytokine responses in isolated and in situ mouse lungs. J Appl Physiol 2004; 97:2190–2199. 17. Dreyfuss D, Saumon G. Ventilator-induced lung injury: Lessons from experimental studies (State of the Art). Am J Respir Crit Care Med 1998; 157:1–30. 18. Permutt S. Mechanical inﬂuences on water accumulation in the lungs. In: Fishman, Renkin, eds. Pulmonary Edema. Am Physiol Soc, Clinical Physiology Series. Baltimore: Williams and Wilkins, 1979:175–193. 19. Howell JBL, Permutt S, Proctor DF, Riley RL. Effect of inﬂation of the lung on different parts of pulmonary vascular bed. J Appl Physiol 1961; 16: 71–76. 20. Benjamin JJ, Murtagh PS, Proctor DF, Menkes HA, Permutt S. Pulmonary vascular interdependence in excised dog lobes. J Appl Physiol 1974; 37: 887–894. 21. Hiff LD. Extra-alveolar vessels and edema development in excised dog lungs. Circ Res 1971; 28:524–532. 22. Albert RK, Lakshminarayan AS, Huang TW, Butler J. Fluid leaks from extraalveolar vessels in living dog lungs. J Appl Physiol 1978; 44:759–762. 23. Faridy EE, Permutt S, Riley RL. Effect of ventilation on surface forces in excised dogs’ lungs. J Appl Physiol 1966; 21:1453–1462. 24. McClenahan JB, Urtnowski A. Effect of ventilation on surfactant, and its turnover rate. J Appl Physiol 1967; 23:215–220.

Lung Mechanics and Pathological Features During VILI

149

25. Veldhuizen RA, Tremblay LN, Govindarajan A, van Rozendaal BA, Haagsman HP, Slutsky AS. Pulmonary surfactant is altered during mechanical ventilation of isolated rat lung. Crit Care Med 2000; 28:2545–2551. 26. Wyszogrodski I, Kyei Aboagye K, Tauecsh HW Jr, Avery ME. Surfactant inactivation by hyperventilation: conservation by end-expiratory pressure. J Appl Physiol 1975; 38:461–466. 27. Oyarzun MJ, Clements JA. Control of lung surfactant by ventilation, adrenergic mediators, and prostaglandins in the rabbit. Am Rev Respir Dis 1978; 117:879–891. 28. Nicholas TE, Barr HA. The release of surfactant in rat lung by brief periods of hyperventilation. Resp Physiol 1983; 52:69–83. 29. Veldhuizen RA, Welk B, Harbottle R, et al. Mechanical ventilation of isolated rat lungs changes the structure and biophysical properties of surfactant. J Appl Physiol 2002; 92:1169–1175. 30. Faridy EE. Effect of ventilation on movement of surfactant in airways. Respir Physiol 1976; 27:323–334. 31. Pattle RE. Properties, function and origin of the alveolar lining layer. Nature (Lond) 1955; 175:1125–1126. 32. Clements JA. Pulmonary edema and permeability of alveolar membranes. Arch Environ Health 1961; 2:280–283. 33. Albert RK, Lakshminarayan S, Hildebrandt J, Kirk W, Butler J. Increased surface tension favors pulmonary edema formation in anesthetized dogs’ lungs. J Clin Invest 1979; 63:1015–1018. 34. Cooper JA, Van Der Zee H, Line BR, Malik AB. Relationship of endexpiratory pressure, lung volume, and 99mTc-DTPA clearance. J Appl Physiol 1987; 63:1586–1590. 35. O’Brodovich H, Coates G, Marrin M. Effect of inspiratory resistance and PEEP on 99mTc-DTPA clearance. J Appl Physiol 1986; 60:1461–1465. 36. Marks ID, Luce JM, Lazar NM, Wu IN, Lipavsky A, Murray JF. Effect of increases in lung volume on clearance of aerosolized solute from human lungs. J Appl Physiol 1985; 59:1242–1248. 37. Egan EA, Nelson RM, Olver RE. Lung inﬂation and alveolar permeability to non-electrolytes in the adult sheep in vivo. J Physiol (Lond) 1976; 260:409–424. 38. Egan EA. Response of alveolar epithelial solute permeability to changes in lung inﬂation. J Appl Physiol 1980; 49:1032–1036. 39. Egan EA. Lung inﬂation, lung solute permeability, and alveolar edema. J Appl Physiol 1982; 53:121–125. 40. Lesur O, Hermans C, Chalifour JF, et al. Mechanical ventilation-induced pneumoprotein CC-16 vascular transfer in rats: effect of KGF pretreatment. Am J Physiol Lung Cell Mol Physiol 2003; 284:L410–L419. 41. Yoshikawa S, King JA, Reynolds SD, Stripp BR, Parker JC. Time and pressure dependence of transvascular Clara cell protein, albumin, and IgG transport during ventilator-induced lung injury in mice. Am J Physiol Lung Cell Mol Physiol 2004; 286:L604–L612. 42. Parker JC, Townsley MI, Rippe B, Taylor AE, Thigpen J. Increased microvascular permeability in dog lungs due to high airway pressures. J Appl Physiol 1984; 57:1809–1816.

150

Dreyfuss et al.

43. Parker JC, Yoshikawa S. Vascular segmental permeabilities at high peak inﬂation pressure in isolated rat lungs. Am J Physiol Lung Cell Mol Physiol 2002; 283:L1203–L1209. 44. Borelli M, Kolobow T, Spatola R, Prato P, Tsuno K. Severe acute respiratory failure managed with continuous positive airway pressure and partial extracorporeal carbon dioxide removal by an artiﬁcial membrane lung. Am Rev Respir Dis 1988; 138:1480–1487. 45. Dreyfuss D, Soler P, Basset G, Saumon G. High inﬂation pressure pulmonary edema. Respective effects of high airway pressure, high tidal volume, and positive end-expiratory pressure. Am Rev Respir Dis 1988; 137:1159–1164. 46. Cottrell TS, Levine OR, Senior RM, Wiener J, Spiro D, Fishman AP. Electron microscopic alterations at the alveolar level in pulmonary edema. Circ Res 1967; 21:783–797. 47. Teplitz C. Pulmonary cellular and interstitial edema. In: Fishman AP, Renkin EM, eds. Pulmonary Edema. Clinical Physiology Series. Bethesda: American Physiological Society, 1979:97–111. 48. Hurley TV. Types of pulmonary microvascular injury. Ann NY Acad Sci 1982; 384:269–286. 49. Bachofen M, Weibel ER. Structural alterations of lung parenchyma in the adult respiratory distress syndrome. In: Bone RC, ed. Clinics in Chest Medicine. Vol 3. Philadelphia: WB Saunders, 1982:35–56. 50. DeFouw DO, Berendsen PB. Morphological changes in isolated perfused dog lungs after acute hydrostatic edema. Circ Res 1978; 43:72–82. 51. West JB, Tsukimoto K, Mathieu-Costello O, Prediletto R. Stress failure in pulmonary capillaries. J Appl Physiol 1991; 70:1731–1742. 52. Tsukimoto K, Mathieu-Costello O, Prediletto R, Elliott AR, West JB. Ultrastructural appearances of pulmonary capillaries at high transmural pressures. J Appl Physiol 1991; 71:573–582. 53. Costello ML, Mathieu-Costello O, West JB. Stress failure of alveolar epithelial cells studied by scanning electron microscopy. Am Rev Respir Dis 1992; 145:1446–1455. 54. Nycolaysen G, Waaler BA, Aarseth P. On the existence of stretchable pores in the exchange vessels of the isolated rabbit lung preparation. Lymphology 1979; 12:201–207. 55. Rippe B, Townsley M, Thigpen J, Parker JC, Korthuis RJ, Taylor AE. Effects of vascular pressure on the pulmonary microvasculature in isolated dog lungs. J Appl Physiol 1984; 57:233–239. 56. Shirley HH, Wolfram CG, Wasserman K, Mayerson HS. Capillary permeability to macromolecules: stretched pore phenomenon. Am J Physiol 1957; 190: 189–193. 57. Fishman AP, Pietra GG. Hemodynamic pulmonary edema. In: Fishman and Renkin, eds. Pulmonary Edema. Am Physiol Soc 1979:79–96. 58. Elliott AR, Fu Z, Tsukimoto K, Prediletto R, Mathieu-Costello O, West JB. Short-term reversibility of ultrastructural changes in pulmonary capillaries caused by stress failure. J Appl Physiol 1992; 73:1150–1158. 59. Vlahakis NE, Hubmayr RD. Response of alveolar cells to mechanical stress. Curr Opin Crit Care 2003; 9:2–8.

Lung Mechanics and Pathological Features During VILI

151

60. Parker JC, Ivey CL, Tucker JA. Gadolinium prevents high airway pressureinduced permeability increases in isolated rat lungs. J Appl Physiol 1998; 84:1113–1118. 61. Parker JC. Phosphotyrosine phosphatase inhibition increases susceptibility of rat lungs to airway pressure-induced injury [abstract]. Am J Respir Crit Care Med 1997; 155:A88. 62. Parker JC. Inhibitors of myosin light chain kinase and phosphodiesterase reduce ventilator-induced lung injury. J Appl Physiol 2000; 89: 2241–2248. 63. Fu Z, Costello ML, Tsukimoto K, et al. High lung volume increases stress failure in pulmonary capillaries. J Appl Physiol 1992; 73:123–133. 64. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28:596–608. 65. Dreyfuss D, Soler P, Saumon G. Mechanical ventilation-induced pulmonary edema. Interaction with previous lung alterations. Am J Respir Crit Care Med 1995; 151:1568–1575. 66. Peevy KJ, Hernandez LA, Moise AA, Parker JC. Barotrauma and microvascular injury in lungs of nonadult rabbits: effect of ventilation pattern. Crit Care Med 1990; 18:634–637. 67. Maeda Y, Fujino Y, Uchiyama A, Matsuura N, Mashimo T, Nishimura M. Effects of peak inspiratory ﬂow on development of ventilator-induced lung injury in rabbits. Anesthesiology 2004; 101:722–728. 68. Rich PB, Reickert CA, Sawada S, et al. Effect of rate and inspiratory ﬂow on ventilator-induced lung injury. J Trauma 2000; 49:903–911. 69. Hernandez LA, Peevy KJ, Moise AA, Parker JC. Chest wall restriction limits high airway pressure-induced lung injury in young rabbits. J Appl Physiol 1989; 66:2364–2368. 70. Ranieri VM, Brienza N, Santostasi S, et al. Impairment of lung and chest wall mechanics in patients with acute respiratory distress syndrome: role of abdominal distension. Am J Respir Crit Care Med 1997; 156:1082–1091. 71. Gibson GJ, Pride NB. Pulmonary mechanics in ﬁbrosing alveolitis. Am Rev Respir Dis 1977; 116:637–647. 72. Ranieri VM, Zhang H, Mascia L, et al. Pressure-time curve predicts minimally injurious ventilatory strategy in an isolated rat lung model. Anesthe siology 2000; 93:1320–1328. 73. Venegas JG, Harris RS, Simon BA. A comprehensive equation for the pulmonary pressure–volume curve. J Appl Physiol 1998; 84:389–395. 74. Martin-Lefevre L, Ricard JD, Roupie E, Dreyfuss D, Saumon G. Signiﬁcance of the changes in the respiratory system pressure-volume curve during acute lung injury in rats. Am J Respir Crit Care Med 2001; 164:627–632. 75. Hauge A, Bo G, Waaler BA. Interrelations between pulmonary liquid volumes and lung compliance. J Appl Physiol 1975; 38:608–614. 76. Grossman RF, Jones JG, Murray JF. Effects of oleic acid-induced pulmonary edema on lung mechanics. J Appl Physiol 1980; 48:1045–1051. 77. Gattinoni L, Pesanti A, Avalli L, Rossi F, Bombino M. Pressure-volume curves of total respiratory system in acute respiratory failure. Computed tomographic scan study. Am Rev Respir Dis 1987; 136:730–736.

152

Dreyfuss et al.

78. Steinberg JM, Schiller HJ, Halter JM, et al. Alveolar instability causes early ventilator-induced lung injury independent of neutrophils. Am J Respir Crit Care Med 2004; 169:57–63. 79. Schiller HJ, McCann UG 2nd, Carney DE, Gatto LA, Steinberg JM, Nieman GF. Altered alveolar mechanics in the acutely injured lung. Crit Care Med 2001; 29:1049–1055. 80. Halter JM, Steinberg JM, Schiller HJ, et al. Positive end-expiratory pressure after a recruitment maneuver prevents both alveolar collapse and recruitment/derecruitment. Am J Respir Crit Care Med 2003; 167:1620–1626. 81. Steinberg J, Schiller HJ, Halter JM, et al. Tidal volume increases do not affect alveolar mechanics in normal lung but cause alveolar overdistension and exacerbate alveolar instability after surfactant deactivation. Crit Care Med 2002; 30:2675–2683. 82. McCann UG 2nd, Schiller HJ, Carney DE, Gatto LA, Steinberg JM, Nieman GF. Visual validation of the mechanical stabilizing effects of positive end-expiratory pressure at the alveolar level. J Surg Res 2001; 99:335–342. 83. Wilson TA, Bachofen H. A model for mechanical structure of the alveolar duct. J Appl Physiol 1982; 52:1064–1070. 84. Martynowicz MA, Minor TA, Walters BJ, Hubmayr RD. Regional expansion of oleic acid-injured lungs. Am J Respir Crit Care Med 1999; 160:250–258. 85. Wilson TA, Anaﬁ RC, Hubmayr RD. Mechanics of edematous lungs. J Appl Physiol 2001; 90:2088–2093. 86. Reichling TD, German RZ. Bones, muscles and visceral organs of proteinmalnourished rats (Rattus norvegicus) grow more slowly but for longer durations to reach normal ﬁnal size. J Nutr 2000; 130:2326–2332. 87. Bouvet F, Dreyfuss D, Lebtahi R, Martet G, Le Guludec D, Saumon G. Noninvasive evaluation of acute capillary permeability changes during highvolume ventilation in rats with and without hypercapnic acidosis. Crit Care Med 2005; 33:155–160; discussion 250–252. 88. Dreyfuss D, Martin-Lefevre L, Saumon G. Hyperinﬂation-induced lung injury during alveolar ﬂooding in rats: effect of perﬂuorocarbon instillation. Am J Respir Crit Care Med 1999; 159:1752–1757. 89. Grasso S, Terragni P, Mascia L, et al. Airway pressure-time curve proﬁle (stress index) detects tidal recruitment/hyperinﬂation in experimental acute lung injury. Crit Care Med 2004; 32:1018–1027. 90. Roupie E, Dambrosio M, Servillo G, et al. Titration of tidal volume and induced hypercapnia in acute respiratory distress syndrome. Am J Respir Crit Care Med 1995; 152:121–128. 91. Dambrosio M, Roupie E, Mollet J-J, et al. Effects of PEEP and different tidal volumes on alveolar recruitment and hyperinﬂation. Anesthesiology 1997; 87:497–503. 92. Hickling KG. The pressure-volume curve is greatly modiﬁed by recruitment. A mathematical model of ARDS lungs. Am J Respir Crit Care Med 1998; 158:194–202. 93. Jonson B, Richard JC, Straus C, Mancebo J, Lemaire F, Brochard L. Pressure-volume curves and compliance in acute lung injury: evidence of

Lung Mechanics and Pathological Features During VILI

94. 95.

96. 97.

98. 99.

100.

101.

102.

103.

104.

105.

106.

107.

108.

153

recruitment above the lower inﬂection point. Am J Respir Crit Care Med 1999; 159:1172–1178. Tsuno K, Prato P, Kolobow T. Acute lung injury from mechanical ventilation at moderately high airway pressures. J Appl Physiol 1990; 69:956–961. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99:944–952. Pugin J, Dunn L Jolliet P, et al. Activation of human macrophages by mechanical ventilation in vitro. Am J Physiol 1998; 275:L1040–L1050. Vlahakis NE, Schroeder MA, Limper AH, Hubmayr RD. Stretch induces cytokine release by alveolar epithelial cells in vitro. Am J Physiol 1999; 277:L167–L173. Pugin J, Jolliet P. Ventilator-induced lung injury: an inﬂammatory disease? Clin Pulm Med 1998; 5:290–299. Choudhury S, Wilson MR, Goddard ME, O’Dea KP, Takata M. Mechanisms of early pulmonary neutrophil sequestration in ventilator-induced lung injury in mice. Am J Physiol Lung Cell Mol Physiol 2004; 287:L902–L910. D’Angelo E, Pecchiari M, Baraggia P, Saetta M, Balestro E, Milic-Emili J. Low-volume ventilation causes peripheral airway injury and increased airway resistance in normal rabbits. J Appl Physiol 2002; 92:949–956. Sohma A, Brampton WJ, Dunnill MS, Sykes MK. Effect of ventilation with positive end-expiratory pressure on the development of lung damage in experimental acid aspiration pneumonia in the rabbit. Intensive Care Med 1992; 18:112–117. Valenza F, Guglielmi M, Irace M, Porro GA, Sibilla S, Gattinoni L. Positive end-expiratory pressure delays the progression of lung injury during ventilator strategies involving high airway pressure and lung overdistention. Crit Care Med 2003; 31:1993–1998. Kloot TE, Blanch L, Melynne Youngblood A, et al. Recruitment maneuvers in three experimental models of acute lung injury. Effect on lung volume and gas exchange. Am J Respir Crit Care Med 2000; 161:1485–1494. Dreyfuss D, Saumon G. Pressure-volume curves: searching for the grail or laying patients with adult respiratory distress syndrome on procrustes’ bed? Am J Respir Crit Care Med 2001; 163:2–3. Hubmayr RD. Perspective on lung injury and recruitment: a skeptical look at the opening and collapse story. Am J Respir Crit Care Med 2002; 165: 1647–1653. Bshouty Z, Ali J, Younes M. Effect of tidal volume and PEEP on rate of edema formation in in situ perfused canine lobes. J Appl Physiol 1988; 64:1900–1907. Corbridge TC, Wood LDH, Crawford GP, Chudoba MJ, Yanos J, Sznadjer JI. Adverse effects of large tidal volume and low PEEP in canine acid aspiration. Am Rev Respir Dis 1990; 142:311–315. Malo J, Ali J, Wood LDH. How does positive end-expiratory pressure reduce intrapulmonary shunt in canine pulmonary edema? J Appl Physiol 1984; 57:1002–1010.

154

Dreyfuss et al.

109. Pare´ PD, Warriner B, Baile EM, Hogg JC. Redistribution of pulmonary extravascular water with positive end-expiratory pressure in canine pulmonary edema. Am Rev Respir Dis 1983; 127:590–593. 110. Dreyfuss D, Saumon G. Role of tidal volume, FRC and end-inspiratory volume in the development of pulmonary edema following mechanical ventilation. Am Rev Respir Dis 1993; 148:1194–1203. 111. Rizk NW, Murray JF. PEEP and pulmonary edema. Am J Med 1982; 72: 381–383. 112. Luce JM, Huang TW, Robertson HT, et al. The effects of prophylactic expiratory positive airway pressure on the resolution of oleic acid-induced lung injury in dogs. Ann Surg 1983; 197:327–336. 113. Fuhrman BP, Paczan PR, DeFrancisis M. Perﬁuorocarbon-associated gas exchange. Crit Care Med 1991; 19:712–722. 114. Lachmann B, Fraterman A, Verbrugge SJC. Liquid Ventilation. In: Physiological basis of mechanical ventilation. New York: Marcel Dekker, 1998. 115. Tu¨tu¨ncu¨ AS, Akpir K, Mulder P, Erdmann W, Lachmann B. Intratracheal perﬂuorocarbon administration as an aid in the ventilatory management of respiratory distress syndrome. Anesthesiology 1993; 79:1083–1093. 116. Cox PN, Frndova H, Tan PS, et al. Concealed air leak associated with large tidal volumes in partial liquid ventilation. Am J Respir Crit Care Med 1997; 156:992–997. 117. Ricard JD, Dreyfuss D, Laissy JP, Saumon G. Dose–response effect of perﬂuorocarbon administration on lung microvascular permeability in rats. Am J Respir Crit Care Med 2003; 168:1378–1382. 118. Broccard AF, Shapiro RS, Schmitz LL, Ravenscraft SA, Marini JJ. Inﬂuence of prone position on the extent and distribution of lung injury in a high tidal volume oleic acid model of acute respiratory distress syndrome. Crit Care Med 1997; 25:16–27. 119. Nishimura M, Honda O, Tomiyama N, Johkoh T, Kagawa K, Nishida T. Body position does not inﬂuence the location of ventilator-induced lung injury. Intensive Care Med 2000; 26:1664–1669. 120. Valenza F, Guglielmi M, Mafﬁoletti M, et al. Prone position delays the progression of ventilator-induced lung injury in rats: does lung strain distribution play a role? Crit Care Med 2005; 33:361–367. 121. Broccard A, Shapiro RS, Schmitz LL, Adams AB, Nahum A, Marini JJ. Prone positioning attenuates and redistributes ventilator-induced lung injury in dogs. Crit Care Med 2000; 28:295–303. 122. Jardin F, Farcot JC, Boisante L, Curien N, Margairaz A, Bourdarias JP. Inﬂuence of positive end-expiratory pressure on left ventricular performance. N Engl J Med 1981; 304:387–392. 123. Suter PM, Fairley B, Isenberg MD. Optimum end-expiratory airway pressure in patients with acute pulmonary failure. N Engl J Med 1975; 292: 284–289. 124. Falke KJ, Pontoppidan H, Kumar A, Leith DE, Gefﬁn B, Laver MB. Ventilation with end-expiratory pressure in acute lung disease. J Clin Invest 1972; 51:2315–2323.

Lung Mechanics and Pathological Features During VILI

155

125. International consensus conferences in intensive care medicine: Ventilatorassociated Lung injury in ARDS. Am J Respir Crit Care Med 1999; 160: 2118–2124. 126. Esteban A, Anzueto A, Frutos F, et al. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study. JAMA 2002; 287:345–355. 127. Maunder RJ, Shuman WP, McHugh JW, Marglin SI, Butler J. Preservation of normal lung regions in the adult respiratory distress syndrome. Analysis by computed tomography. JAMA 1986; 255:2463–2465. 128. Hickling KG, Henderson SJ, Jackson R. Low mortality associated with low volume pressure limited ventilation with permissive hypercapnia in severe adult respiratory distress syndrome. Intensive Care Med 1990; 16:372–377. 129. Tobin MJ. Culmination of an era in research on the acute respiratory distress syndrome [editorial; comment]. N Engl J Med 2000; 342:1360–1361. 130. Deans KJ, Minneci PC, Cui X, Banks SM, Natanson C, Eichacker PQ. Mechanical ventilation in ARDS: One size does not ﬁt all. Crit Care Med 2005; 33:1141–1143. 131. Gattinoni L, Pelosi P, Crotti S, Valenza F. Effets of positive end-expiratory presure on regional distribution of tidal volume and recruitment in adult respiratory distress syndrome. Am J Respir Crit Care Med 1995; 151: 1807–1814. 132. Gajic O, Dara SI, Mendez JL, et al. Ventilator-associated lung injury in patients without acute lung injury at the onset of mechanical ventilation. Crit Care Med 2004; 32:1817–1824. 133. Gajic O, Frutos-Vivar F, Esteban A, Hubmayr RD, Anzueto A. Ventilator settings as a risk factor for acute respiratory distress syndrome in mechanically ventilated patients. Intensive Care Med 2005; 26:26. 134. Brower RG, Lanken PN, Macntyre N, et al. Higher versus lower positive endexpiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351:327–336. 135. Anzuetto A, Baughman R, Guntupalli KK, et al. Aerosolized surfactant in adults with sepsis-induced acutre respiratory distress syndrome. N Engl J Med 1996; 334:1417–1421. 136. Spragg RG, Lewis JF, Walmrath HD, et al. Effect of recombinant surfactant protein C-based surfactant on the acute respiratory distress syndrome. N Engl J Med 2004; 351:884–892. 137. Willson DF, Thomas NJ, Markovitz BP, et al. Effect of exogenous surfactant (calfactant) in pediatric acute lung injury: a randomized controlled trial. JAMA 2005; 293:470–476. 138. Kacmarek RM, Wiedemann HP, Lavin PT, Wedel MK, Tutuncu AS, Slutsky AS. Partial liquid ventilation in adult patients with the acute respiratory distress syndrome. Am J Respir Crit Care Med 2005. In Press.

This page intentionally left blank

7 The Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume Ventilator-Induced Lung Injurya

DONALD P. GAVER III and ANNE-MARIE JACOB Department of Biomedical Engineering, Tulane University New Orleans, Louisiana, U.S.A.

ANASTACIA M. BILEK Center for Devices and Radiological Health, Food and Drug Administration Rockville, Maryland, U.S.A.

KAY C DEE Department of Applied Biology and Biomedical Engineering, Rose-Hulman Institute of Technology Terre Haute, Indiana, U.S.A.

I. Introduction In this chapter, we describe how pulmonary ﬂuid–structure interactions, lining ﬂuid physics, and surfactant biophysical properties interrelate to inﬂuence the lung’s micromechanical environment during low-volume ventilation. In conditions such as acute respiratory distress syndrome (ARDS) and respiratory distress syndrome (RDS), the lining ﬂuid and/or surfactant systems may be abnormal, and the pulmonary tissue is particularly susceptible to mechanical trauma. In these cases, low-volume ventilation strategies may result in VILI. This chapter describes our present understanding of the ﬂuid-structure and physicochemical interactions that are related to low-volume VILI, and potential techniques for reducing the likelihood or severity of this injury. For a more complete description of pulmonary lining ﬂows and their relationship to pulmonary disease, the reader is referred to Ref. (1). a

The opinions expressed in this chapter are those of the author and do not necessarily represent the ofﬁcial policies of the Food and Drug Administration.

157

158

Gaver et al. II. Background

The structure of lungs is optimized to provide a large respiratory surface area with a very thin tissue barrier for the rapid diffusion and equilibration of respiratory gases. The airway surfaces are lined with epithelial cells that vary in structure and function in different generations. At the level of the respiratory bronchioles (approximately, generation 17), alveoli (the sites of gas exchange) begin to appear in the walls of the airways, and the airways terminate in alveolar sacs. The alveoli are thoroughly enveloped by a pulmonary capillary network that is lined with a single layer of endothelial cells covering a large surface area. The alveolar–capillary membrane consists of the alveolar lining ﬂuid, the alveolar epithelium, a network of connective tissue, and the pulmonary capillary endothelium (2). The delicate structure of this portion of the lung makes it particularly susceptible to mechanical injury (3,4). Damage by acute lung injury (ALI), is ‘‘a syndrome of inﬂammation and increased permeability [of the airspaces in the mature lung] . . . associated most often with sepsis syndrome, aspiration, primary pneumonia, or multiple trauma’’ (5). In severe cases, the resulting altered respiratory mechanics can develop rapidly into the characteristic hypoxemia and stiff lungs of the ARDS (6). Both disorders are initiated by damage to the lung’s epithelial and endothelial cell layers, which diminishes the blood–gas barrier and permits the inﬂux of proteinaceous edema ﬂuid and inﬂammatory cells (neutrophils, macrophages, monocytes, and lymphocytes) into the airways and alveoli. The reduced number of intact, functioning alveolar type II cells is no longer sufﬁcient for the adequate synthesis or turnover of pulmonary surfactant. This, compounded by the inactivation of surfactant proteins SP-B and SP-C by plasma proteins in edema ﬂuid, increases the surface tension of the ﬂuid plugging the airspaces and the pressures required to reopen them (7). In addition, plasma protein–rich hyaline membranes form on the exposed basement membrane of the denuded airways and alveoli, which fosters the development of endstage ﬁbrosis during ARDS (8). Finally, because the pulmonary vasculature communicates directly with the heart, the diffuse and intense inﬂammatory response generated by these diseases—including the release of cytokines and chemokines by cells, the alteration of blood plasma (complement system, coagulation, ﬁbrinolysis, and kinin systems), and the induction of protein synthesis (5)—can spread quickly throughout the body’s systemic circulation, where activated inﬂammatory cells and their mediators continually threaten to provoke the catastrophic multiorgan dysfunction syndrome or system failure (MODS or MOSF, respectively). Interestingly, nonsurvivors of ALI and ARDS tend to have three times as much residual protein from pulmonary edema in their lungs than survivors (8). Thus, the associated diminished blood–air barrier function and enhanced epithelial and endothelial permeabilities are very likely to

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

159

play a pivotal role in the ﬁnal prognosis of these patients. This is no surprise, because the inﬂux of proteinaceous edema into the lower airways and alveoli carries with it a wide range of effects on respiratory mechanics. For example, edema ﬂuid inactivates and washes away functioning pulmonary surfactant proteins (7). In addition, the excess serum proteins in edema ﬂuid cause the precipitation and modiﬁcation of lung proteins, the accumulation of hyaline membranes, and the development of end stage ﬁbrosis. The resulting increase in surface tension and decrease in the lung tissue elasticity severely diminish pulmonary compliance and further contribute to alveolar ﬂooding and collapse (8). As we will show below, the increased surface tension can increase the mechanical stresses on airways and alveoli during recruitment. This, in turn, can compromise the integrity of the pulmonary epithelial cell monolayer, which further reduces surfactant production by type II cells, increases the permeability of the epithelial cell layer, and impairs the removal of edema ﬂuid from the airspaces (9), resulting in a self-perpetuating feedback loop. Mechanical ventilation, though inarguably an indispensable therapeutic modality regularly used to treat RDS, ALI, and ARDS, exerts a wide range of excessive, irregular mechanical stresses and strains on the delicate tissues that make up the airspaces of the lung, further contributing to the disease-related damage mechanisms already in progress. Optimal ventilation protocols generally rely upon a strategy of preventing the collapse and reopening of compliant airways and alveoli, while simultaneously avoiding the overdistension of patent regions of the lung. In 1970, Mead et al. (10) ﬁrst suggested that cyclic collapse and reopening of lung units, as would occur with ventilation at low lung volumes and pressures, could generate large, potentially damaging stresses in the surrounding tissues. When a surfactant is either insufﬁcient in quantity (as in RDS) or ineffective (e.g., in the presence of pulmonary edema ﬂuid during ARDS) or when the lung becomes more compliant (e.g., due to emphysema or aging), airways and alveoli are rendered increasingly unstable and prone to damage from this mode of injury at low lung volumes. For this reason, Argiras et al. (11) proposed that positive end-expiratory pressure (PEEP) be administered during the mechanical ventilation of surfactant-deﬁcient lungs to prevent the repetitive collapse and reopening of the distal airways and alveoli. While low-volume ventilation may damage the lung due to airway and alveolar collapse and recruitment, the early work of Webb and Tierney (12) demonstrated that mechanical ventilation with large tidal volumes and peak airway pressures rapidly produced an ALI with severe pulmonary edema in rats. Although slower to develop, a similar injury has been shown in larger animals such as sheep (13). These studies illustrated that high-volume mechanical ventilation may cause injury de novo in otherwise healthy lungs. Dreyfuss and Saumon (14) established that large tidal volumes (not high airway pressures) result in excessive lung distension that causes air leaks,

160

Gaver et al.

acute respiratory failure, pulmonary edema, and alveolar cell dysfunction. In human studies, increased pulmonary edema and mortality occur with excessive tidal volumes (15). Analysis of data from a recent ARDS Network study of patients with ALI or ARDS indicates that reducing the imposed tidal volume yielded a concomitant reduction in mortality, from 40% to 31%, regardless of the pulmonary compliance or the original etiology of the ALI or ARDS. These researchers observed that the bulk of deaths resulted from withdrawal of care, sepsis, and MODS or MOSF, and not from hypoxemia (16). It is now generally accepted that the use of excessively high or low volumes may cause VILI. It is common to assume that pressure above the lower inﬂection point (LIP) of the pressure–volume (PV) curve for patients with ARDS reﬂects the region where recruitment of closed airways occurs. It has been hypothesized that PEEP should be applied to ventilate the lung between the LIP and upper inﬂection point (UIP) of the PV curve to prevent volutrauma while minimizing the recruitment or derecruitment damage (Fig. 1). Amato et al. (17) demonstrated that ventilating patients with ARDS using a strategy that speciﬁcally sought to stay between these two inﬂection points reduced mortality. It is therefore essential to establish

Figure 1 PV relationship for the lung. Protective ventilation strategies generally ventilate between LIP and UIP to prevent recruitment damage and tissue overdistension. Abbreviations: PV, pressure–volume; LIP, lower inﬂection point; UIP, upper inﬂection point.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

161

better lung protective strategies for mechanical ventilation that prevent alveolar overdistension, via lower tidal volumes, and simultaneously maintain sufﬁcient functional residual capacity to avoid the repetitive opening and closure of atelectatic lung units, via PEEP (18). However, in practice, this task remains a difﬁcult one, especially in the heterogeneously damaged lungs of individuals with ALI and ARDS, where regions of ﬂooded and collapsed airways and alveoli cause the mechanically ventilated air to preferentially distribute to the limited airspace (7). In addition, studies by Brown and Mitzner (19) have shown that in methacholine-challenged animals, airway closure still occurs even at the highest level of PEEP, suggesting that PEEP alone cannot protect the lung from airway closure. In addition to the direct injury of the pulmonary tissue caused by the physical forces associated with mechanical ventilation, increasing evidence suggests that mechanical stresses may initiate and regulate inﬂammatory processes that, as described above, play a signiﬁcant role in ALI. The initiation or exacerbation of tissue injury as a consequence of mechanical stress–induced inﬂammation has recently been termed ‘‘biotrauma,’’ and can elicit a host of responses in the lung, including tissue remodeling (20,21), the release of cytokines (22), the upregulation of surfactant production and release (23), and apoptosis (24,25). If the inﬂammatory response remains improperly regulated, these mediators can rapidly overwhelm and severely harm the lung and elicit the uncontrolled immune system activation underlying the MODS or MOSF (26). A. Summary

It is our premise that the mechanical stimuli responsible for epithelial damage during low-volume VILI result from macro- and microscale ﬂuid–structure interactions between the lining ﬂuid that coats the interior surfaces of the lung and sensitive pulmonary tissue. These interactions occur primarily during the cyclic derecruitment and recruitment of airways and alveoli, which can induce large pressures on the entire organ and microscale stresses and stress gradients on the tissues surrounding these lung units. As will be shown in the following text, it is also evident that surfactant biophysical properties can protect the lung from biotrauma. Figure 2 provides a synopsis of the sections of this chapter. In section III, we explore the ﬂuid–structure interactions (Fig. 2B and C) that occur during airway collapse and recruitment. In section IV, we investigate the micromechanical stresses that exist during airway recruitment (Fig. 2D) and identify the putative mechanism for epithelial lesions during low-volume ventilation. In section V, we explore the effects of pulmonary surfactant on this system (Fig. 2C). We anticipate that an understanding of the interrelationships between lining ﬂuid ﬂows, surfactant biophysical behavior, and tissue responses to ﬂuid–structure interactions may lead to the development

162

Gaver et al.

Figure 2 Synopsis of the components related to this project. (A) The PV relationship for ventilation. This project addresses issues related to the low-volume portion, where airway closure or reopening can occur; (B) an occluded airway, with a ﬁnger of air that progresses to clear the airway. The mechanical properties depend upon (C) the local ﬂuid ﬂow and surfactant interactions (‘‘physicochemical hydrodynamics’’) that inﬂuence the reopening behavior and determine the (D) mechanical stresses on airway epithelial cells. Abbreviation: PV, pressure–volume.

of ventilator waveforms that reduce the incidence and severity of low-volume VILI, as described in section VI.

III. Introduction to Pulmonary Fluid–Structure Interactions A. Organ-Level Fluid–Structure Interactions

Pulmonary ﬂuid–structure interactions related to VILI occur on both the macroscale (i.e., organ-level) and microscale (i.e., scales of tissues and cells) levels. Under normal conditions, macroscale ﬂuid–structure interactions are clearly evident when one examines the mechanical behavior of the whole lung—in particular, its ability to expand and contract in response to changes in pressure during the respiratory cycle, which depends in large part on the ﬂuid and chemical properties of the liquid ﬁlm that coats the interior surfaces of the lung. The importance of ﬂuid–structure interactions

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

163

to pulmonary mechanics was ﬁrst demonstrated by Von Neergaard (27), who recognized that surface tension forces were largely responsible for determining the PV behavior of the lung (Fig. 1). Surface tension is the membrane-like quality of an interface, due to differences in molecular attraction that occur between any two phases (e.g., liquid and air) separated by the interface. Physically, this can be demonstrated by considering the Young–Laplace relationship for the pressure drop across a spherical bubble of radius R (a crude model of an alveolus): DP ¼

2c ; R

ð1Þ

where c is the surface tension. This relationship clearly demonstrates that bubble (i.e., alveolar) pressure is directly related to the surface tension, and inversely related to the radius. In the healthy lung, pulmonary surfactant adsorption and desorption causes variations in c, providing the hysteresis loop behavior observed in Figure 1. In fact, when a lung is instead inﬂated with saline (hence removing the air–liquid interface), it is much more compliant, and the hysteresis area diminishes. This indicates that the surface tension of the lining ﬂuid is integral to determining the mechanical properties of the lung, and that surfactant biophysical properties are responsible for the hysteresis. If the surface tension of the lining ﬂuid were to remain constant, this could lead to alveolar instability. The mechanism for instability can be understood from Figure 3. If the surface tension is equal on the left and right units, the pressure in the smaller unit would exceed the pressure in the larger unit, resulting in airﬂow from the small to large unit. This mechanism would result in the collapse of all the small alveoli (until the tissue stretch compensates). Pulmonary surfactant stabilizes this system by reducing the surface tension of the lining ﬂuid in direct relation to the interfacial concentration. This adds to the stability of the lung because initiation of alveolar collapse would result in a decrease in surface tension, due to the compression of the surfactant at the air–liquid interface. This simplistic example (which, among other things, ignores the stabilizing effects of parenchymal tethering) demonstrates the essential role played by the pulmonary surfactant in maintaining normal respiratory mechanics. Surfactant insufﬁciency is a signiﬁcant contributor to pulmonary disease. In 1959, Avery and Mead (28) showed that surface-active material is diminished or absent in the lungs of infants with hyaline membrane disease, now known as RDS. RDS results from lung immaturity at birth, which produces a high lining ﬂuid–surface tension and results in a propensity for airway closure, atelectasis of portions of the lung, and inhomogeneous ventilation. Other diseases, such as acute or adult respiratory distress syndrome (ARDS), though caused by other processes, can result in surfactant insufﬁciency that exacerbates the disease. Studies have also

164

Gaver et al.

Figure 3 Description of surface tension–driven instability between two alveoli, when surface tension is constant. From the Young–Laplace relationship, the pressure in a small alveolus exceeds the pressure in the large alveolus, driving the ﬂow towards the large alveolus. With surfactant, the collapse of an alveolus increases the surfactant concentration at the air–liquid interface. This reduces the surface tension on the collapsing alveolus, reduces the pressure differential, and hence stabilizes the system.

suggested that surfactant deﬁciency can play a role in asthma (29,30), though the evidence for this is inconclusive. In section V, we will provide a more detailed description of how surfactant biophysical properties interact with lining ﬂows to protect the lung from low-volume VILI.

B. Airway Closure

Airway closure and reopening are examples of ﬂuid–structure interactions that occur at the microscale, but can result in macroscale responses. Effective pulmonary ventilation depends upon the availability of free passageways between the mouth and the alveoli. Ventilation can be signiﬁcantly hindered by closure of the small airways, which prevents gas exchange with

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

165

peripheral respiratory units. Closure can occur when ﬂuid accumulates in the lung or when the lung is highly compliant, as in diseases such as emphysema. Because the closure process results in the derecruitment of portions of the lung, it is critically important to our understanding of low-volume VILI. Airway closure can occur through two mechanisms, ‘‘meniscus formation’’ (Fig. 4A) and ‘‘compliant collapse’’ (Fig. 4B) (31). In meniscus formation, the airway lining ﬂuid becomes unstable and a ﬂuid plug forms, which spans the airway. This behavior was reported, by Macklem et al. (32), as existing in cats’ airways with diameters of approximately 0.05 cm. In compliant collapse, the walls of the airway buckle inward and the lining ﬂuid adheres to the walls, as has been observed in histological investigations (33). Here, the liquid occlusion induces a transmural pressure on the airway due to surface tension forces, which buckles the walls and holds them in apposition by the adhesive properties (surface tension and viscosity) of

Figure 4 Schematic description of airway closure (1). (A) Meniscus occlusion. (B) Compliant collapse.

166

Gaver et al.

the lining ﬂuid (31). In both cases, airway closure leads to an occlusion that prevents air exchange. Ventilation inhomogeneity and ventilation or perfusion mismatch can occur if an airway remains closed for a signiﬁcant portion of a tidal ventilation cycle. Kamm and Schroter (34) tested Macklem’s (35) hypothesis that airway closure arises through a surface tension–induced instability of the liquid lining, by experimentally simulating small airway closure using liquid-lined rigid tubes. That study showed that a minimum volume of lining ﬂuid was necessary to induce meniscus formation, and the ratio of liquid volume to airway diameter needed to induce instability is relatively independent of the airway length. Below this critical volume, the ﬁlm readjusted to form stable unduloids that did not occlude the tube. Dynamic unduloid formation by a thin viscous ﬁlm coating the interior of a cylindrical tube of circular cross-section was described by Hammond (36), while meniscus formation by a thicker ﬂuid layer was predicted by Gauglitz and Radke (37) and Johnson et al. (38) (accounting also for ﬂuid inertia). These studies identiﬁed a critical ﬁlm thickness, hc 0.14R, for meniscus formation in a cylindrical tube of radius R; below this critical thickness, the ﬁlm evolves into stable unduloids, and above it, the ﬁlm becomes unstable and forms liquid bridges. Timescales for the generation of liquid bridges were calculated to be in the order of 65 msec, using realistic airway parameters (38). According to these models, meniscus occlusion in airways is most likely to occur at the end of expiration, when R is smallest and the liquid lining thickness is most likely to exceed hc. It is reasonable to expect that compliant collapse will increase peripherally, because the compliance of airways increases with airway generation (31,33). Airway collapsibility has therefore been included in subsequent theoretical models of airway closure (38–43). For example, Halpern and Grotberg (39) theoretically modeled the coupled effects of wall compliance, liquid lining viscosity and surface tension, and demonstrated that wall ﬂexibility enhances closure and that hc decreases with increasing wall ﬂexibility. Compliant collapse (Fig. 4B) was predicted for sufﬁciently ﬂoppy tubes or for high surface tension, and collapse was found to occur within a fraction of a millisecond. The studies of Heil (44,45) demonstrated the importance of nonaxisymmetric wall buckling by showing that the minimum volume required for the formation of a static liquid plug in a compliant tube can be as much as 10 times smaller than the volume needed to block an axisymmetric tube. Thus a ﬁlm coating a compliant tube that does not contain enough ﬂuid to form an axisymmetric plug can nevertheless undergo a nonaxisymmetric instability, leading to meniscus occlusion. Though airway closure is a local event, it can result in macroscale responses. For example, once an airway closes, the surface tension of the lining ﬂuid can result in regions of atelectasis that are much longer than the original

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

167

region of closure. Furthermore, all subtended airways and alveoli are then subject to hypoinﬂation that results in ventilation or perfusion mismatch. C. Airway Reopening

While it is still unclear which mechanism is responsible for airway closure in ALI (46), several investigators have observed that once collapsed, the airways take on a ﬂat ‘‘ribbon-like’’ conﬁguration and reopening separates the walls in a peeling motion (47,48). Macklem et al. (32) were the ﬁrst to describe the (re)opening of airways as ‘‘gradually peeling apart the opposing walls, and the liquid remaining in situ presumably lining that part of the bronchiole that had opened.’’ A schematic of a reopening airway is shown in Figure 5. Here, the airway is open on the left, with walls held in apposition on the right by the viscous liquid–lining ﬂuid that ‘‘glues’’ the airway shut. To open the airway, a long bubble must propagate through the collapsed airway and separate the walls. Estimates of the upstream pressure necessary to inﬂate collapsed airways were initially conducted by Gaver et al. (49). These studies indicate that a ‘‘yield’’ pressure (Pyield 8c/R) must be exceeded in order for airways to be reopened. This estimate is consistent with data from experiments in an ex vivo lung preparation (Fig. 6) (47). Subsequent benchtop experimental investigations have identiﬁed the reopening characteristics of more complex systems. Two-dimensional experimental models of reopening demonstrated an approximately 50% decrease in the yield pressure (50), indicating that the three-dimensional meniscus curvature increases the yield pressure, and further demonstrated the possibility of unstable reopening at low reopening velocities. NonNewtonian ﬂuids that mimic mucus were studied by Hsu et al. (51–53).

Figure 5 Schematic of airway reopening. Semi-inﬁnite bubble separates the airway walls and penetrates viscous occlusion.

168

Gaver et al.

Figure 6 Airway diameter–pressure relationship for reopening rat airways. (A) Tantalum bronchograms of selected airways just before reopening, immediately after reopening at higher inﬂation. (B) Yield pressure versus diameter relationship from experiments, indicating consistency with predictions by Gaver et al. (49). Source: From Ref. 47.

These studies show that ﬂuid elasticity can signiﬁcantly inﬂuence the PV relationship for reopening by inducing ﬂow instabilities at large reopening velocity due to a sol–gel transition in the viscoelastic properties. In ex vivo studies, Yap et al. (48) demonstrated the inﬂuence of parenchymal tethering on airway reopening by modiﬁcation of pleural pressure. This study showed that the reopening behavior can exhibit unstable ﬂuttering at low pleural pressure, stable-peeling reopening at intermediate parenchymal tethering, and rapid ‘‘popping’’ open with large parenchymal tethering. Subsequent benchtop model studies (54) indicate that tethering stresses will reduce

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

169

the airway pressure necessary to inﬂate a compliant airway. Thus, a reduction of tethering (e.g., emphysema) could greatly increase the airway pressure necessary to maintain airway patency. Of particular importance to low-volume VILI are the predictions of reopening pressures emanating from these studies. Using an approximation of Pyield 8c/R, the estimate for yield pressure for obstructed respiratory bronchioles for adults with normal surfactant function is Pyield 5 cmH2O; a pressure small enough to prevent biotrauma or volutrauma. Alternatively, for adults with ARDS resulting in surfactant deﬁciency, Pyield 15 to 20 cmH2O is required. Estimates for premature infants with surfactant deﬁciency are Pyield 50 cmH2O. It is most important to recognize that pressures of this magnitude are exposed to all airways and alveoli that are in direct communication with the atelectic region. Thus, even though airway closure is a local (or microscale) event, the pressures necessary to recruit collapsed airways can result in macroscale responses through the hyperinﬂation of patent airways and alveoli (Fig. 7).

Figure 7 Closure of regions of the lung can lead to inhomogeneity of ventilation. The high resistance associated with airway closure can lead to volutrauma of open portions of the lung, even when protective ventilation strategies are followed.

170

Gaver et al. IV. Microscale Fluid–Structure Interactions Leading to VILI

In the sections above, the focus had been on the airway and/or alveolar obstruction and recruitment as being responsible for lung injury on the macroscale (or organ level). Additionally, the recruitment of airways and alveoli can have a microscale (tissue-level) response. To describe the signiﬁcance of these phenomena in the creation of biotrauma at the microscale, we ﬁrst outline the general concept of mechanical stress and the potential responses of epithelial cells to different modes of stress. Subsequently, we will investigate how lining ﬂows and surfactant interactions can develop and modulate these mechanical stresses.

A. Mechanical Stress Deﬁnitions

In this subsection, we describe the concept of a mechanical stress ﬁeld and the resulting behavior that this stress ﬁeld can elicit from a structure. While the analysis of the coupled system can be quite complex, the basic behavior is not difﬁcult to comprehend. That is: 1.

2.

3.

4.

The mechanical stress ﬁeld describes the stresses on orthogonal surfaces, as shown schematically in Figure 8A. A mechanical stress ﬁeld refers to the distribution of internal tractions that balance a set of external tractions and/or body forces. Mechanically, an external traction Tij (a stress tensor) is a force intensity (force per unit area) with indices representing the direction of the outward facing normal vector (i) and the direction of the stress ( j), with the index values 1, 2, 3 representing the x, y, z directions, respectively; The inﬂuence of Tij on a surface is determined by the orientation of the surface. This is described by the unit normal ^n and tangential ^t vectors, as shown in Figure 8B; The stresses on the surface are determined by the vector product of ^ n and Tij. This results in normal (sn ¼ ^n T ^n) and tangential n T ^t) components of stress, as shown in Figure 8C. These (ss ¼ ^ stresses are considered as ‘‘bound vectors,’’ because, in general, they cannot slide to a different location and have the same magnitude, for the surface normal and/or the stress magnitudes will change with position. This occurs because the mechanical stresses are functions of both the local ﬂow ﬁeld and the domain; The local stress ﬁeld can redistribute the surface. This is shown schematically in Figure 8C by the dashed line, and can be due to tissue remodeling (slow time scale), or the mechanical redistribution of the surface due to compression stretching or wear

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

171

Figure 8 Stress ﬁeld description and its implication on ﬂuid stress interactions. (A) A representation of the stress tensor that describes the force intensity (force or area) on orthogonal surfaces. For ﬂuid ﬂow, the magnitude of each term depends upon the velocity ﬁeld and a constitutive equation that relates ﬂow characteristics to stress. (B) The geometrical description of a structural surface; (C) the stress vector along the surface, which depends upon the orientation of the surface. The stress vector is shown decomposed into normal and tangential components, each of which can inﬂuence the surface topography and biological responses of the cells. Here a dashed line (– – –) represents a hypothetical redistribution of the surface due to the applied stress ﬁeld. Motion of the surface would, in turn, modify the ﬂow ﬁeld and stresses experienced by the surface.

172

Gaver et al. (fast timescale). Each of these processes can inﬂuence the biologic functioning of the cells that make up the surface, as described in section IV-C. B. Mechanical Stress Field from Airway Reopening

Here we focus on external stresses in the lung as the stimulus for biotrauma at the microscale. In the case of airway reopening, the external traction on an airway is imposed by the ﬂuid stresses that exist at the boundary between the lining ﬂuid and tissue, which are generated as the ﬁnger of air propagates through a liquid-occluded airway and separates the airway walls. As described above, it is most convenient to consider two types of stresses—normal stresses (sn), which act perpendicular to a surface, and tangential (or shear) stresses (ss) that are directed along a surface, with s deﬁned as a surface coordinate. In the process of airway reopening, the airway walls are separated in a peeling motion by the bubble progression, as analyzed by Gaver et al. (55) and indirectly observed experimentally (47,48). An example of the ﬂuid domain and the stresses acting on an airway during airway reopening is shown in Figure 9 (for the deﬁnition of Ca see Section IV. E, below). Reopening induces large and rapid changes in normal and shear stress along the airway walls. Combining the stress relationships with those in Figure 10 provides a graphical representation of the mechanical environment of the airway epithelium as a bubble of air progresses across the surface. For a given epithelial cell, the time-dependent nature of the stresses may be quite signiﬁcant. Far downstream of the bubble, the cell is not notably stressed by any force. As the bubble approaches that cell, a moderate shear stress, directed towards the bubble, is experienced in addition to a large inward-directed normal stress due to the bending of the airway wall. As the bubble propagates directly over the cell, a large shear stress and a step-jump in pressure is imparted for a brief instant. Finally, after the bubble has passed by, the airway wall experiences an outward normal stress from the pressure that is required to push the bubble forward. Tensions applied to the cells as a result of the walls bending during reopening and the hoop-stress caused by the distending pressure are not represented in Figure 10. The spatial and temporal gradients of these stresses exert dynamic, large, and potentially damaging forces on the airway epithelium that are not typically seen in one-phase steady ﬂow conditions (55–57). C. Biological Responses to Micromechanical Stresses from Airway Reopening

The fundamental issue related to low-volume VILI at the micromechanical level is the determination of the biological response to the stress ﬁeld near the migrating ﬁnger of air during airway reopening. The associated stress

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

173

Figure 9 The effect of the capillary number (Ca) on the (A) bubble/airway geometry (B) normal stress, and (C) shear stress during the steady propagation of a bubble through a ﬂexible-walled channel. Source: From Ref. 55.

cycle can be separated into four potentially injurious components—shear stress, the gradient of shear stress, pressure, and the gradient of pressure. These stresses, and the gradients of the stresses, may be responsible for microscale atelectrauma, as graphically described in Figure 11. The potential effects are as described in the following section.

174

Gaver et al.

Figure 10 Schematic representation of stresses on the airway wall from the steady propagation of a bubble propagating through a ﬂexible-walled channel representation of an obstructed pulmonary airway.

Shear Stress

Shear stress by the ﬂowing ﬂuid over the surface of a cell might induce deformations of the cell causing plasma membrane disruptions. Additionally, shear stresses can be translated from the surrounding ﬂuid to the nontethered components of the cell membrane. The cell membrane may be ‘‘rareﬁed’’ in regions due to shifting of nontethered membrane components, where it would become more susceptible to tearing (58).

Figure 11

Potential effects of stress and stress gradients on a cell.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

175

Shear Stress Gradient

Shear stress gradients cause force imbalances within the plane of the cell membrane, directly increasing the tension of the cell membrane. Cell membranes can carry only a very small tension (58) and may rupture under gradients of shear stress. Pressure

Maximum pressure occurs in patent regions of the airway, and exerts an evenly distributed normal stress on the cells. While studies have shown that a uniform pressure stimulus alters the behavior of some cells (59,60), it is unlikely that this stimulus can directly compromise the cell membrane, because transmural pressure will equilibrate at the transmission speed of an acoustic wave. Pressure Gradient

Pressure gradients create normal stress imbalances on the cell membrane over the length of the cell. For a low proﬁle, predominately ﬂat region of a cell, the nonuniformly distributed load may depress and stretch small regions of a cell’s membrane. In addition, we speculate that the normalstress difference could induce transient internal ﬂows within the cell that could exert hydrodynamic stresses on the extracellular surface of the cell membrane, which might injure the membrane by the same mechanisms as extracellular stresses. For high proﬁle cells or regions of a cell, for example, at the protruding cell nucleus, net normal forces will act on either side of the region in opposition; thus, a pressure gradient will pinch that region. The pinching could tear the membrane at the base of the protrusion or force the ﬂuid upward, rupturing the top surface of the cell. D. Model Studies

To our knowledge, the only investigations of micromechanical damage due to ﬂuid–structure interactions during airway reopening are those given in Refs. 61 and 62, which are synopsized here. In these studies, a rigid parallel-plate chamber was used as a model of an occluded airway (Fig. 12). Using parallel plates simpliﬁes the geometry of the model to an essentially two-dimensional system from which the air–liquid interfacial shape and the ﬂuid dynamical behavior can be computationally evaluated. The upper and lower walls of the chamber were formed by two glass microscope slides, with the pulmonary epithelial cells cultured to the top wall. The dimensions of the ﬂow chamber were chosen to mimic those of occluded airways. The diameter of closed airways have been measured in normal lungs to be as small as 0.025 cm and as large as 0.2–0.4 cm (33,48,63,64). The channel

176

Gaver et al.

Figure 12 Schematic of ﬂow chamber and resulting stresses for the evaluation of reopening on cell behavior.

height of the experimental model was 0.17 cm, which corresponds to this range. In addition, in injured or surfactant-deﬁcient lungs, closure may occur in larger airways. Two cell lines were investigated, a fetal rat pulmonary epithelial cell line (CCL-149) and a human pulmonary epithelial cell line (A549), each from the American Type Culture Collection (Manassas, Virginia). Each cell line was cultured to conﬂuence on a small (1 cm2), square region of the top glass microscope slide, prior to experiments. The investigations studied different reopening scenarios to 1. 2.

identify the component of mechanical stress that damages the epithelial cells during reopening, and determine whether the duration of stress exposure plays a key role in the damage.

To do so, phosphate-buffered saline including 0.1 mg/mL CaCl2 and MgSO4 (PBS) was used to model a surfactant-deﬁcient (high surface tension) airway lining ﬂuid with a low viscosity of 8.0 103 g/cm sec; and PBS supplemented with 14.1 wt.% clinical grade dextran (Sigma, average molecular weight 68,800) with a viscosity of 8.0 102 g/cm sec was used as a high-viscosity reopening ﬂuid.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

177

To determine the stress component most likely to cause damage, two reopening velocities (0.27 and 2.7 cm/sec) were investigated, both with the low-viscosity ﬂuid. To determine the effect of stress duration, a ﬁxed reopening velocity (0.34 cm/sec) was investigated for two occlusion ﬂuids (low and high viscosity). In that study, the period of time an individual cell experiences the traveling stress-wave is very short and nearly constant (5 102 sec). Following removal from the apparatus, a ‘‘Live or Dead’’ assay was used to differentiate ‘‘live’’ from ‘‘dead’’ cells. If injury or death compromises a cell membrane, Eth-1 enters the cell and binds to DNA, producing a outlined ﬂuorescent nucleus. Uninjured cells are marked by the calcein AM binding to active intracellular esterases, producing dark ﬂuorescence at the cell membrane (Fig. 13). To assess the magnitude of damage, the numbers of injured (outlined) cells (Eth-1 stained) in each of the ﬁve random ﬁelds were counted manually using ﬂuorescence micrographs, with the average number of injured cells expressed either as ‘‘injured cells’’ or cells/cm2 of slide surface area. The data are reported as mean standard error of the mean for ﬁve slides per condition. Statistical signiﬁcance was set at p < 0.01, and differences between means were statistically evaluated usingDuncan’s multiple range test, after model adequacy checking veriﬁed the

Figure 13 Fluorescent micrograph of ethidium homodimer-1 (Eth-1) (outlined) and calcein AM (dark) stained cells after exposure to reopening stresses. Outlined cells indicate damage to the cell membrane with potentially deleterious effects.

178

Gaver et al.

Figure 14 Effect of airway reopening on epithelial cell damage using a ﬂowchamber model of reopening. (A) Number of injured cells versus velocity for surfactant-free and surfactant conditions. Key: , signiﬁcantly greater than control, p < 0.01; #, signiﬁcantly greater than Infasurf-occluded channels for the same velocity. p < 0.01. (B) Number of injured cells vs. occlusion ﬂuid for identical reopening speeds. Key: , signiﬁcantly greater than control, p < 0.01; #, signiﬁcantly greater than the PBS/dextran ﬂuid for the same bubble velocity, p < 0.01. Abbreviation: PBS, phosphate buffered saline.

normal distribution of the data. Figure 14A shows the data for the variable velocity studies. In this case, it is evident that the slower reopening results in a marked increase in the damage to the airway epithelial cells, outlined in white. Figure 14B identiﬁes the relative damage from the studies used to investigate the effect of stress duration. Here, it is clear that the lowviscosity–lining ﬂuid resulted in far more damage than that exhibited by

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

179

high-viscosity–lining ﬂuid. The physiologic implications of these results are discussed in the following section. E. Analysis

To understand the stimulus/response mechanisms related to recruitment damage, it is essential to have a quantitative estimate of the mechanical stresses experienced by airway tissue. Correlation between the stresses and cell responses can then be used to test the hypotheses concerning the likely source of airway damage. The results shown in Figure 14 demonstrate that increasing velocity or occlusion viscosity serves to protect the epithelial cells from damage due to bubble progression in a parallel-plate ﬂow chamber. Intuitively, these results alone would suggest that the shear–stress mechanism is not the likely source of damage. To quantify this understanding, we approximate the magnitudes of mechanical stress and the exposure duration, and relate these to the observed cell membrane damage. Stress Magnitudes

To investigate the stress magnitudes in this system, we use the regression formulas provided by Bilek et al. (61) to calculate the maximum shear stress (ss), shear-stress gradient (dss/dx), and pressure gradient (dP/dx) that the cells experience. These relationships were calculated in a dimensionless form, which exploits the fact that the fundamental physical interactions depend on the ratio of viscous to surface tension forces. The dimensionless velocity, also known as the capillary number, Ca ¼ lU = c;

ð2Þ

represents the ratio of viscous to surface tension effects and determines the dynamic response of the system. Here U is the bubble velocity, l the ﬂuid viscosity, and c is surface tension. The stress relationships are as follows: shear stress, ðlUÞ0:36 c0:64 ; H shear stress gradient,

ð3Þ

ðss Þmax ¼ 0:69

0:22c ðlUÞ0:75 c0:25 þ 1:2 ; H2 H2 and pressure gradient ðdsS =dxÞmax ¼

ðdP=dxÞ max ¼ 0:34

c1:29 ðlUÞ0:29 H 2

;

ð4Þ

ð5Þ

180

Gaver et al.

where H is the half height of the channel. From these relationships, it is clear that all stress magnitudes depend directly on the product (lU ); thus, an increase in viscosity (l) has precisely the same effect on the stress magnitude as an increase in the velocity (U ). It is this relationship that allows the experimental variation of stress magnitudes, using viscosity rather than velocity, yielding additional implications for the duration of stress exposure, as will be explained below. There, we will refer to an increase in Ca as an increase in velocity; however, this can also be thought of as an increase in viscosity. The relationships provided in Eqs. (3) to (5) are potentially counterintuitive, and thus it is important to understand the physical processes that cause this behavior. To understand this behavior, consider Figures 15 and 16. Figure 15A presents representative interfacial domain shapes for Ca ¼ 5 102, 5 103. A thin ﬁlm region exists far upstream, and the thickness of this ﬁlm becomes extremely small with decreasing velocity. As velocity decreases, the bubble cap becomes semicircular. Figure 15B presents corresponding shear stress proﬁles along the wall. In the thin ﬁlm, the shear stress is very small, and downstream (x > 0), the shear stress is caused by Poiseuille (or steady parabolic) ﬂow ahead of the bubble. In contrast, the shear stress in the region of the bubble cap is increased as ﬂuid is deposited into the thin ﬁlm after squeezing past the bubble cap. Figure 15C demonstrates the wall normal-stress. In this ﬁgure, P ¼ 0 is the pressure inside the bubble, which was also the maximum pressure in the system. Far upstream of the bubble tip, the pressure is approximately uniform. In contrast, the downstream pressure (x > 0) decreases linearly with distance. For a small velocity, this decrease is very small, and represents the pressure gradient necessary to drive the ﬂow ahead of the bubble. Within the transition region between the thin stagnant ﬁlm and the semispherical cap, a large change in pressure occurs with the largest pressure gradient at x H. As velocity is reduced, the pressure proﬁle approaches that of a static bubble (Ca ¼ 0), where the spherical cap meets the wall at a contact point at x ¼ H, resulting in a pressure discontinuity of magnitude DP ¼ c/H. To further clarify this behavior, consider Figure 16A, which shows a schematic representation of the interface propagating through the ﬂow chamber, with Figure 16B and C representing the magniﬁed view of the domain and pressure ﬁeld, respectively. Figure 16B shows that a decrease in velocity causes the liquid ﬁlm between the bubble and wall to thin. In the limit as velocity ! 0, the bubble approaches the wall as a contact line, spanwise across the channel. The pressure drop between the interior (air) and exterior (liquid) is approximated by the Young–Laplace equation [Eq. (1)] as DPtot ¼ c/H. Therefore, as velocity ! 0, a step-jump in pressure occurs at the contact line (Fig. 16C). Because DPtot is established over an inﬁnitesimal region, dP/dx ! 1. As velocity increases, the bubble leaves

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

181

a minuscule layer of ﬂuid (‘‘lubrication ﬁlm’’) along the wall (Fig. 16B) that grows in depth with increasing velocity, and reduces the magnitude of the pressure gradient (Fig. 16C). So, while the pressure gradient remains large at small Ca, it is reduced by an increase in Ca (or, relatedly, l or U). In contrast, an increase in Ca increases the shear stress because a greater volume of ﬂuid is squeezed over the cell surface in the lubrication ﬁlm. For this reason, an increase in Ca (i.e., velocity) decreases the pressure gradient and increases the shear stress. Stress Exposure Duration

As demonstrated above, a change in the reopening velocity results in a modiﬁcation in the slope of the pressure wave that travels across the cell, and directly relates to the exposure time. To determine the exposure time for a cell as the stress traveling-wave of length Lwave sweeps over the cell surface, consider the representation of the system provided in Figure 16. We approximate DPtot ¼ c/H, because the majority of the pressure drop is due to surface tension, not viscosity (i.e., Ca ¼ lU/c << 1). Using the relationship for dP/dx [Eq. (5)], the extent of the traveling-wave region is approximately 0:29 lU Lwave ¼ 2:94 H Ca0:29 ¼ 2:94 H ð6Þ c The exposure time is the length of time required for the entire width of the traveling-wave region to propagate past a point on the wall occupied by a cell Dtexposure ¼

Lwave HCa0:29 Hl0:29 ¼ 2:94 ¼ 2:94 0:71 0:29 ; U U U c

ð7Þ

This calculation indicates that even though U is held constant in the variable viscosity experiments, the increase in l extends the length of the traveling wave, and thus increases Dtexposure. In contrast, the experiments given in Ref. 61 modiﬁed Ca by increasing U such that an increase in Ca simultaneously decreased Dtexposure. F. Damage Mechanism

Combining the experimental observations of Refs. 61 and 62 with the analysis above, we are in the position to identify the stimulus that induces cell damage. Figure 14A shows that damage is increased with reduced velocity. In contrast, Figure 14B shows that an increase in l (and thus, the exposure time) with no change in U caused a reduction of the cell damage. In both the low- and high-viscosity cases, the damage was signiﬁcantly greater than the control.

182

Figure 15

Gaver et al.

(Caption on facing page)

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

183

Figure 16 (A) Schematic of the bubble propagating through the ﬂow chamber. (B) A magniﬁed view of the lubrication ﬁlm near the bubble tip. (C) The pressure ﬁeld near the bubble. This ﬁgure demonstrates that the increase of the capillary number (Ca) reduces the magnitude of the pressure gradient by increasing the lubrication ﬁlm thickness. From Ref. 62.

Table 1 provides a synopsis of the experimental observations and trends from Refs. 61 and 62. This table shows that when the viscosity is varied, the low Ca (lower l and shorter duration) experiments demonstrate increased damage with a larger pressure-gradient magnitude relative to

Figure 15 (Facing page) Effect of capillary number (Ca) on (A) the domain geometry, (B) the shear stress, and (C) pressure. Source: From Ref. 61.

184

Gaver et al.

Table 1 An Overview of the Trends from Bilek et al. (61) and Kay et al. (62) Constant velocity (U) Kay et al. current study Low Ca (high cellular damage)

High Ca (low cellular damage)

Constant viscosity (l) Bilek et al. study

"

dP dx # Dtexp # ss

"

dP dx " Dtexp " ss

#

#

dP dx " Dtexp # ss dP dx # Dtexp " ss

Note: In all cases cellular damage is increased with decreased Ca, independent of exposure time. The pressure-gradient magnitude is the common factor that explains the cellular damage. Source: From Ref. 62.

the large Ca experiments. On the other hand, when velocity is varied, the low Ca experiments also reveal increased damage with a larger pressuregradient, even though the duration was longer for the lower values of Ca. Because, in both cases, the damage is increased with decreasing Ca (and so is the pressure gradient), this provides compelling evidence that the magnitude of the pressure gradient on the cell (not the exposure durations) is the factor that induces membrane damage. We hypothesize that the pressure gradient may create an imbalance in the pressure over the length of the cell (Fig. 11). This would induce a fore– aft pressure difference on the cell body, approximated as DPcell ¼ (dP/dx) Lcell, where Lcell 40 mm is the approximate cell length. Table 1 shows that DPcell increases with reduced Ca, which is consistent with the observed damage pattern (Fig. 14). The pressure imbalance could result in nonuniform cell compression, leading to ‘‘pinching’’ of the cell and rupturing of the cell membrane, as theorized earlier in this section. From Refs. 61 and 62, signiﬁcant cell membrane damage occurs when DPcell300 dynes/cm2, which is reduced when DPcell120 dynes/cm2. Little membrane rupture was observed for DPcell80 dynes/cm2. Clearly, the propensity for membrane disruption decreases with decreasing DPcell, however, it is not yet evident that a speciﬁc critical level that initiates damage exists. G. Limitations

It should be noted that several stress components that exist in the compliant reopening model (Fig. 10) are missing in this rigid parallel plate model of reopening (Fig. 12B). For example, in the rigid reopening model, a negative

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

185

pressure pulls the cells inward as the bubble approaches. In addition, a substrate tension exists in this region due to wall bending, and potentially signiﬁcant hoop stress exists upstream due to the distending pressure. Nevertheless, the rigid-walled representation is suitable for analysis of the clearance of liquid from an uncollapsed airway, and thus directly relates to issues of recruitment and lung injury that have been raised by Hubmayr (46). In addition, this simpliﬁcation may represent the tip region of a compliantly collapsed airway, because for very low reopening velocity, the walls in the region of the bubble tip are nearly parallel (57). Hubmayr (46) suggests that this ﬂexibility will focus stresses if compliant collapse occurs. Thus, ﬂexibility might induce greater damage in vivo compared to that observed in the rigid-walled experiments. Alternatively, with a rigid substrate, all deformations caused by the applied stimulus are constrained to the cells. Therefore, it is possible that wall ﬂexibility in vivo could protect cells by allowing the stresses to be distributed more fully to surrounding structures. Clearly, though, issues of airway ﬂexibility should be investigated to determine the importance of this aspect of ﬂuid–structure interactions. In addition, the inﬂuence of topological variation has not been investigated. The migration of a ﬁnger of air through convoluted airways may result in more damage to particular portions of airways, akin to the increased likelihood of atherosclerotic plaque near vascular bifurcations. In addition, ﬂuid ﬂow near the perimeter of the bubble cap is restricted to the narrow space between the air–liquid interface and the wall of the channel. If the ﬁlm thickness near the perimeter of the bubble cap is equivalent to a cell protrusion, the ﬂow resistance and stresses may be further ampliﬁed, as discussed in Refs. 65 and 66. Recent theoretical investigations have suggested that that the topological variation may amplify pressure gradients near the migrating ﬁnger of air, and thus may be even more detrimental to the airway’s cellular epithelium than the studies above would suggest (67).

V. The Protective Effect of Pulmonary Surfactant The analyses above have largely ignored the protection afforded by pulmonary surfactant. As discussed in section III, normal lung surfactant (LS) serves to stabilize the airways so that airway collapse and reopening does not occur routinely during tidal breathing. Lungs in animal models with a normal surfactant system can be exposed to repeated cycles of airway collapse and reopening using negative end-expiratory pressure without injury (68,69), indicating that the surfactant can protect the lung from biotrauma. However, even mild surfactant dysfunction leads to severe lung injury, including epithelial destruction, under the same ventilation protocols (69). In situations where the pulmonary surfactant system of the lung is severely compromised (e.g., during RDS and ALI), lung damage occurs with

186

Gaver et al.

spontaneous breathing or conventional mechanical ventilation (70–72). In this section, we ﬁrst brieﬂy describe the biophysical (physicochemical) behavior of pulmonary surfactant because it relates to lining-ﬂuid dynamics. We then explore the concept of a critical concentration for protection from low-volume VILI, and speculate on ventilation processes that might be used to reduce the critical concentration so as to protect the lung when surfactant deﬁciency exists. A. Foundations

Before discussing the protective effects of pulmonary surfactant, it is instructive to provide a brief primer on surface-tension effects. Many more details are provided in Ref. (1). Surface tension originates at a molecular level because liquid molecules attract one another more strongly than air molecules; liquid near an air–liquid interface experiences a net force perpendicular to the interface that acts to minimize the interfacial area. A convenient way in which to characterize this force is to endow the air–liquid interface with an effective surface tension c. Just as the tension in the skin of a balloon balances the difference in pressure between the air inside and outside the balloon, so too does surface tension balance the pressure jump across an air–liquid interface. Statically, this balance is expressed by the general form of the Young–Laplace equation Pl Pg ¼ cj

ð8Þ

Here Pl is the liquid pressure, Pg the gas pressure, and j is the curvature of the interface, so that j¼

1 1 R1 R2

ð9Þ

where R1 and R2 are the radii of curvatures in orthogonal planes. Note that if the interface is convex relative to the liquid, j > 0; if it is concave, j < 0. For a ﬂat interface, j ¼ 0; for a cylindrical interface of radius R, j ¼ 1/R, and for a spherical interface of radius R, j ¼ 2/R. In general, the curvature of an interface is more complex. When the air–liquid interface assumes a shape with nonuniform j, Equation (8) implies that the pressure in the liquid will also be nonuniform. A surface tension–driven ﬂow will then drive the liquid from regions of high to low pressure, redistributing the liquid until j is uniform (at least locally). This behavior is demonstrated in a two-dimensional example in Figure 17A. When a surfactant adsorbs to an air–liquid interface, the intermolecular forces are modiﬁed by the surfactant’s hydrophilic head groups, which reduce the intermolecular force acting perpendicular to the interface, and lowers c by an amount dependent on the local instantaneous surfactant surface concentration, C. Thus if C is spatially nonuniform, then so is c.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

187

Figure 17 Descriptions of surface tension–driven ﬂow. (A) Flow induced by local variation of curvature. P1 < P2, which drives a ﬁlling ﬂow. (B) Flow driven by surface tension variation along the interface. Here the surfactant concentration Cb is greater than the neighboring concentration (Cb > Ca). This causes a local reduction of surface tension (cb < ca), which causes a tangential (Marangoni) stress, sM, that drags the top layer of ﬂuid towards regions of higher surface tension.

A small region of the air–liquid interface in which a variation of surfactant exists will then experience higher c on the side where C is lower and vice versa. The difference in surface tension across the element exerts a net stress (called a Marangoni stress, sM) that is tangential to the interface, directed towards the region of higher c (and lower C). Thus, this tension imbalance causes the surface element to drag the viscous liquid beneath it through factional (viscous) effects. The resulting ﬂow, called a Marangoni ﬂow, in which viscous drag balances surface-tension gradients, leads to the transport of both the liquid and the surfactant adsorbed to its surface from regions of higher to lower concentrations, as illustrated in Figure 17B. This type of ﬂow can be simply observed by the spreading of a drop of dish detergent on the surface of an oil-covered pan. In summary, a surfactant can lower the surface tension and also create spatial variations in the surface tension that drives interfacial ﬂows. In general, the reduction of surface tension will reduce the pressure in the system [from Eq. (8)] but can also induce a ‘‘rigidiﬁcation’’ of the interface by Marangoni stresses, which can have a detrimental effect during airway reopening. This is likely to occur because surfactant that exists within the

188

Gaver et al.

Figure 18 Streamlines surrounding bubble as it steadily propagates through a liquid-ﬁlled channel or tube. Surfactant molecules are convected with the ﬂow, and develop a nonuniform distribution on the air–liquid interface. Surface-tension gradients on the surface can retard the ﬂow through Marangoni stresses that rigidify the interface, sm.

lining ﬂuid (with concentration C) is transported to and from the interface (with concentration C), thereby directly modifying the interfacial surface. Pulmonary surfactant is highly insoluble, so adsorption rates are far greater than desorption rates (73–75). So, during the motion of a bubble, the interface can assume a nonuniform surface tension that will inﬂuence the mechanical behavior of the system. To conceptualize this interaction, consider the ﬂow ﬁeld surrounding a bubble ﬂowing down a liquid-ﬁlled channel or tube as shown in Figure 18. Streamlines are drawn in a bubble-ﬁxed reference frame in which ﬂow enters from the right and exits to the left in the thin ﬁlm. A circulating region near the bubble tip occurs at low velocities [note, this region is a closed recirculation region, when the walls are ﬂexible (55)]. As a result, the rate of interfacial expansion or compression will vary with interfacial position. Variation of the dynamic surface tension, c, can alter the pressure required to push the bubble along the tube through a modiﬁcation of the pressure drop across the air–liquid interface following the Young–Laplace law [Eq. (8)]. The elevation of the pressure drop over that which would exist if the surface tension were in equilibrium is partially due to dynamic surface tension effects that increase the local normal stress. This has been referred to as a ‘‘nonequilibrium normal stress.’’ In addition, variation of the surface tension along the air–liquid interface creates a surface-tension gradient, which allows the interface to support a shear stress sM that is directed from low-to-high surface tension (Fig. 17B). In steady-state situations, the Marangoni stress retards bubble motion, because it creates a stress in a direction opposite to the ﬂow that would exist on a surfactant-free interface. For this reason, Marangoni stresses are referred to as stresses that

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

189

‘‘rigidify’’ an interface. So, if sorption is slow, these nonequilibrium stresses increase both the pressure drop (56,76,77) and the deleterious mechanical stresses on the epithelial cells. In unsteady ﬂow, the Marangoni stress mechanism may be useful in propagating the bubble forward. B. Surfactant Properties

Pulmonary surfactant is a lipid–protein complex formed in the type II alveolar cells. While predominantly comprising lipids (90%), the surfactant proteins (10%) are necessary for normal functioning (78,79). Approximately 80% of the phospholipid content is dipalmitoylphospatidylcholine (DPPC), which is responsible for attaining ultralow surface tensions (<5 dyne/cm) (80–82). Also, phosphatidylglycerol aids in spreading the surfactant (80,81,83,84). The surfactant-associated proteins (SP-A, B, C, and D) inﬂuence physicochemical properties. The hydrophobic proteins, SP-B and SP-C, aid the adsorption and respreading of the surfactant monolayer upon compression to ultralow surface tensions (85–88). Theoretical models of LS function are largely based upon a ‘‘squeezeout’’ theory. This theory was developed from the analysis of individual lipid components of LS from their surface tension lowering and adsorption properties. The squeeze-out theory states that upon compression of a multiconstituent monolayer to high concentrations, the surface tension is reduced to ultralow values, and a collapse phase occurs, wherein the rapidly adsorbed components are selectively squeezed from the interface to the subphase (the bulk region in contact with the interface). Surface balance and morphological studies by Takamoto et al. (89) conﬁrm this behavior with lipid-only LS analogues. Otis et al. (73,74) have developed a mathematical model of LS behavior that is predicated upon this squeeze-out model. Physicochemical properties of LS replacement were estimated from measurements of Surfactant TA (Abbott Laboratories) using a pulsating bubble surfactometer (PBS), with the theory based upon an oscillating bubble system. While speciﬁc lipid components were not modeled, a key aspect of their model was the inclusion of a squeeze-out regime that caused the surfactant to be rejected from the interface and enter the bulk phase. Recent improvements of this model (90) account for loss of surfactant into the bulk phase, and allow the surfactant to adsorb from the highly concentrated subphase to the interface. This was the ﬁrst analytical model to replicate many of the qualitative features of hysteresis loops from oscillating bubble experiments and to estimate adsorption-rate parameters of an LS analogue. Recent studies [Lipp et al. (91)] have indicated that, upon compression of a model LS monolayer with surfactant apoproteins, squeeze-out may not exist. Studies by Zasadzinski and coworkers (91–93) demonstrate that SP-B and SP-C can induce a reversible two- to three-dimensional folding transition at monolayer collapse (Fig. 19). Speciﬁcally, monolayers of

190

Gaver et al.

Figure 19 Schematic of multilayer collapse process. Compression of the monolayer causes the surfactant to buckle and create a multilayer of surfactant. Upon expansion of the interface, the secondary layer can reincorporate into the primary layer if surfactant apoproteins are present.

the pulmonary surfactant containing those proteins in addition to lipids adsorbed to an air–liquid interface will collapse under compression to form subsurface multilayers (89,91,94–96) that will respread to the primary interface when the interfacial surface area expands (97). Thus, the multilayer can provide a reservoir for reincorporation into the monolayer during interfacial expansion. It is now recognized that the kinetics of multilayer development and reinsertion are critical to adequate surfactant function (98–100) and are key mechanisms responsible for the surface tension hysteresis observed during the cycling of a surfactant-doped air–liquid interface with high bulk concentrations (75). This collapse mechanism has signiﬁcant implications for the dynamic surface tension of LS analogues that include SP-B or SP-C, because it indicates that the reintroduction of surfactant to an expanding interface is not due solely to sorption kinetics. When SP-C alone is included the surfactant, the surfactant folds into a bilayer, while SP-B can create multiple folds. As will be described below, these biophysical properties may be useful in protecting the lung from recruitment/derecruitment damage. Studies to investigate the physicochemical nature of an exogenous surfactant replacement, Infasurf (ONY Inc., Buffalo, New York, U.S.A.) with surfactant apoproteins (and hence monolayer or multilayer dynamics)

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

191

were conducted by Krueger (101). Experimental and theoretical studies of dynamic surface tension–area (S–A) behavior were compared to discern the relevant physicochemical parameters in an oscillating bubble system. Experiments were conducted using a PBS (Electronetics, Buffalo, New York, U.S.A.), which provides dynamic surfactant S–A responses by volume-oscillating a small (1 mm) bubble surrounded by a surfactant-doped ﬂuid. Surfactant transport between the hypophase surrounding the bubble and the interface modiﬁes the surface surfactant concentration C, and thus the resulting surface tension c. A representative hysteresis loop from high concentration studies is shown in Figure 20. This ﬁgure shows that the sorption is intimately related to the surfactant surface–structure at the interface. In this model, surfactant can reside on either the primary layer or the collapsed secondary layer. A maximum equilibrium interfacial packing state can exist when C1 ¼ C1 (point A). However, the monolayer compression

Figure 20 Surface tension versus area hysteresis loop for Infasurf. Compression of the surfactant creates a metastable low surface tension region during surfactant collapse to a multilayer. Upon reexpansion, the multilayer reincorporates to the primary layer.

192

Gaver et al.

(as would occur near recirculation at a bubble tip) to C1 > C1 reduces the surface tension well below equilibrium (ceq). When C1 > Cmax (point B), the surface tension is near zero (73), and continuing compression will cause a local buckling in the monolayer (Fig. 19), extending the surfactant into the subphase and ultimately collapsing on the interface and creating the secondary layer, C2 (point B to C). This secondary layer blocks desorption of the primary layer, and may either desorb to the subphase or become available for respreading. The surfactant that remains attached to the interface respreads when the interface expands, if C1 > Cmls (point D to A). From this example, it is clear that the dynamic surface tension of pulmonary surfactant ranges, in magnitude, to values signiﬁcantly greater or less than the equilibrium value. C. Critical Surfactant Concentration

Two studies have sought to identify the critical surfactant necessary to protect the lung. The ﬁrst investigated surfactant adsorption aspects in a physical system, where the pressure required to push a semi-inﬁnite bubble of air down a ﬂuid-ﬁlled cylindrical capillary of radius R was measured (102). The ionic surfactant sodium dodecyl sulfate (SDS) and pulmonary surfactant analogues DPPC and Infasurf (ONY Inc.) were examined. This study showed that the nonequilibrium adsorption of surfactant can create a large nonequilibrium normal stress and a surface shear stress (Marangoni stress) that increases the bubble pressure. As shown in Figure 21, the nonphysiological surfactant, SDS, is capable of ‘‘remobilizing’’ the interface by eliminating the nonequilibrium normal and Marangoni stresses when the concentration exceeds the critical bulk concentration, CCBC, as described

Figure 21 The inﬂuence of nonequilibrium normal stresses on reopening pressures. Key: , SDS; H, DPPC; &, Infasurf. Source: From Ref. 102.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

193

Figure 22 Effect of Infasurf concentration on cell damage using a parallel-plate model of reopening. U ¼ 0.25 cm/sec, CCBC ¼ 1 mg/mL.

in Refs. 103 and 104. CCBC represents the surfactant concentration that will, under static conditions, result in a minimum equilibrium surface tension. In contrast, DPPC is not capable of reducing either stresses, demonstrating slow adsorption properties. The clinically relevant surfactant, Infasurf, has intermediate adsorption properties, but does not fully remobilize the interface. These studies indicate that, for steady reopening, even high concentrations of Infasurf can result in surface tensions that are much higher than static equilibrium values (102). Experiments to assess the critical Infasurf concentration necessary to protect epithelial cells from damage have been performed using the ﬂow chamber described in section IV (105). These studies indicate that, in order to protect the lung, the surfactant concentration must exceed CCBC, even during very low-speed reopening (Fig. 22). Thus, during steady-state reopening, it appears that surfactant-deﬁcient airways are highly susceptible to damage due to high surface tensions that exist from nonequilibrium surfactant transport effects.

VI. Future Directions We have thus far provided evidence that the recruitment of collapsed airways can cause ﬂuid–structure interactions that may damage airway epithelial

194

Gaver et al.

cells. Furthermore, our analyses strongly suggest that this damage is due to a time-dependent normal-stress gradient that sweeps across the airway wall as the airway is recruited. This damage is reduced if surface tensions remain low due to high concentrations of pulmonary surfactant. Unfortunately, surfactant deﬁciency is likely to occur in ALI or ARDS and can result in an elevated nonequilibrium surface tension. Not only does this destabilize the pulmonary airways, but it also increases the mechanical stresses on the airway walls, and hence increases the likelihood of airway damage. An important question that remains is whether ventilation waveforms can be designed that will take advantage of dynamic surface-tension effects to reduce mechanical stresses during the recruitment of collapsed airways under conditions where surfactant deﬁciency exists.

Figure 23 Examples of meniscus-frame streamlines for the four main types of interfacial ﬂow during pulsatile ﬂow in a channel (half-plane): (A) tip streamline convergence (þ) with a divergent transition region stagnation point (–), (B) tip streamline divergence only, (C) tip streamline divergence with a convergent transition region stagnation point, and (D) tip streamline convergence only. Source: From Ref. 111.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

195

To address this issue, we hypothesize that surfactant multilayer formation and expansion could be used to encourage airway reopening with minimal damage. This could occur in a dynamic system if a phase shift between bubble motion and surfactant accumulation/respreading can be developed. Already, evidence suggests that variable tidal volume ventilation can improve the ventilation efﬁcacy in models of ARDS (106,107) and may result in enhanced endogenous surfactant release (108). In addition, highfrequency ventilation has been shown to beneﬁt patients with ARDS (109,110). To predict whether a phase shift in surfactant accumulation and respreading can be accomplished, computational ﬂuid dynamic simulations of airway recruitment have explored only the transport of surfaceassociated contaminants to and/or from a pulsatile bubble of air over a variety of reopening scenarios (111). These studies demonstrate that the pulsatile ﬂow ﬁeld can substantially modify the transport of the surfaceassociated contaminants to an air–liquid interface. The different types of ﬂow behavior are depicted in Figure 23, which show that the ﬂow ﬁeld changes throughout the cycle to direct contaminant towards converging stagnation points (þ)—where the surfactant is expected to accumulate— and away from diverging stagnation points ()—where surfactant will become depleted—on the air–liquid interface. These simulations reveal a

Figure 24 Predictions of cycle spatial–averaged concentrations of passive surfactant. Concentrations are ampliﬁed with oscillation and depend upon amplitude of the oscillation, A, and frequency X.

196

Gaver et al.

large increase of the average concentration of surface-associated contaminant at speciﬁc frequencies (Fig. 24). Experiments to identify the capacity of pulsatile ﬂow to enhance the transport of surfactant to a migrating bubble in a cylindrical tube have been completed with purely sinusoidal motion superimposed with a constant ﬂow. These studies used Infasurf at 1 mg/mL (10 CCBC) and set the oscillation amplitude, A, to induce a bubble oscillation of 0.5 or 2.5 mm with frequencies of 0.1, 0.5, 0.7, and 1.0 Hz. A constant ﬂow syringe pump provided forward ﬂow at rates of 0.025, 0.1, and 0.4 mL/min. Figure 25 shows DPcap for pulsatile ﬂow in comparison to the DPcap for steady migration, and clearly demonstrates that oscillation reduces the pressure required to clear a viscous ﬂuid occlusion. This behavior is qualitatively similar to the above theoretical predictions and suggests that ventilation might be ‘‘tuned’’ to reopen airways with minimal damage. However, to fully understand these systems, it will be necessary to more completely analyze this system from a multiscale perspective. This includes investigations at the molecular scale that determines dynamic surfactant behavior at the air–liquid interface, the continuum scale that describes the liquid ﬂow and mechanical stress ﬁeld within airways, and the tissue- and organ-level mechanical and biological responses.

Figure 25 Capillary pressure drops for a bubble oscillation of 0.5 mm (A ¼ 1) with f ¼ 0.75 Hz. C ¼ 1 mg/mL.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

197

References 1. Gaver DP III, Halpern D, Jensen OE. Surfactant and airway liquid ﬂows. In: Nag K, ed. Molecular Mechanisms in Lung Surfactant (Dys)function. New York: Marcel Dekker, 2005:187–223. 2. Levitzky MG. Pulmonary Physiology. 4th ed. New York: McGraw-Hill, Inc., 1995:299. 3. West JB, Mathieu-Costello O. Stress failure of pulmonary capillaries. In: Crystal RG, West JB, et al., eds. The Lung: Scientiﬁc Foundations. Philadelphia: Lippincott-Raven Publishers, 1997:1493–1501. 4. West JB, Mathieu-Costello O. Structure, strength, failure, and remodeling of the pulmonary blood–gas barrier. Annu Rev Physiol 1999; 61:543–572. 5. Bernard GR, Artigas A, Brigham KL, et al. Report of the American– European Consensus conference on acute respiratory distress syndrome: deﬁnitions, mechanisms, relevant outcomes, and clinical trial coordination. Consensus Committee. J Crit Care 1994; 9(9):72–81. 6. Ranieri VM, Suter PM, Tortorella C, et al. Effect of mechanical ventilation on inﬂammatory mediators in patients with acute respiratory distress syndrome: a randomized controlled trial. J Am Med Assoc 1999; 282(1):54–61. 7. Acute respiratory distress syndrome. In: Matthay MA, ed. Lung Biology in Health and Disease. Vol. 179. New York: Marcel Dekker, 2003:657. 8. Kim KJ, Malik AB. Protein transport across the lung epithelial barrier. Am J Physiol 2003; 284(2):L247–L259. 9. Ware LB, Matthay MA. Medical progress: the acute respiratory distress syndrome. N Engl J Med 2000; 342(18):1334–1349. 10. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28(5):596–608. 11. Argiras EP, Blakeley CR, Dunnill MS, et al. High PEEP decreases hyaline membrane formation in surfactant deﬁcient lungs. Br J Anaesth 1987; 59:1278–1285. 12. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110:556–565. 13. Tsuno K, Prato K, Kolobow T. Acute lung injury from mechanical ventilation at moderately high airway pressures. J Appl Physiol 1990; 69(3):956–961. 14. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 15. Slutsky AS. Lung injury caused by mechanical ventilation. Chest 1999; 116:9S–15S. 16. Eisner MD, Thompson T, Hudson LD, et al. Efﬁcacy of low tidal volume ventilation in patients with different clinical risk factors for acute lung injury and the acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 164(2):231–236. 17. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338(6):347–354.

198

Gaver et al.

18. Martynowicz MA, Minor TA, Walters BJ, Hubmayr RD. Regional expansion of oleic acid-injured lungs. Am J Respir Crit Care Med 1999; 160(1):250–258. 19. Brown RH, Mitzner W. Airway closure with high PEEP in vivo. J Appl Physiol 2000; 89(3):956–960. 20. Wiggs BR, Hrousis CA, Drazen JM, Kamm RD. On the mechanism of mucosal folding in normal and asthmatic airways. J Appl Physiol 1997; 83(6): 1814–1821. 21. Swartz MA, Tschumperlin DJ, Kamm RD, Drazen JM. Mechanical stress is communicated between different cell types to elicit matrix remodeling. Proc Natl Acad Sci USA 2001; 98(11):6180–6185. 22. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99(5):944–952. 23. Torday JS, Rehan VK. Stretch-stimulated surfactant synthesis is coordinated by the paracrine actions of PTHrP and leptin. Am J Physiol Lung Cell Mol Physiol 2002; 283(1):L130–L135. 24. Edwards YS, Sutherland LM, Murray AW. NO protects alveolar type II cells from stretch-induced apoptosis. A novel role for macrophages in the lung. Am J Physiol Lung Cell Mol Physiol 2000; 279(6):L1236–L1242. 25. Sanchez-Esteban J, Wang Y, Cicchiello LA, Rubin LP. Cyclic mechanical stretch inhibits cell proliferation and induces apoptosis in fetal rat lung ﬁbroblasts. Am J Physiol Lung Cell Mol Physiol 2002; 282(3):L448–L456. 26. Slutsky AS, Tremblay LN. Multiple system organ failure. Is mechanical ventilation a contributing factor?. Am J Respir Crit Care Med 1998; 157(6.1): 1721–1725. 27. Von Neergaard K. Neue auffassungen u¨ber einen grundbegriff der atemmechanik. Die retraktionskraft der lunge, abha¨ngig von der oberﬂa¨chenspannung in den alveolen. Z Gesamte Exp Med 1929; 66:373–394. 28. Avery ME, Mead J. Surface properties in relation to atelectasis and hyaline membrane disease. AMA J Dis Child 1959; 97:517–523. 29. Liu M, Wang L, Enhorning G. Surfactant dysfunction develops when the immunized guinea-pig is challenged with ovalbumin aerosol. Clin Exp Allergy 1995; 25(11):1053–1060. 30. Cheng G et al. Compositional and functional changes of pulmonary surfactant in a guinea-pig model of chronic asthma. Respir Med 2001; 95(3): 180–186. 31. Greaves LA, Hildebrandt J, Frederic G, Hoppin J. Micromechanics of the lung. In: Fishman AP, ed. Handbook of Physiology: The Respiratory System III. Bethesda, Maryland: American Physiological Society, 1986:217–231. 32. Macklem PT, Proctor DF, Hogg JC. Stability of peripheral airways. Respir Physiol 1970; 8(2):191–203. 33. Hughes JMB, Rosenzweig DY, Kivitz PB. Site of airway closure in excised dog lungs: histological demonstration. J Appl Physiol 1970; 29(3): 340–344. 34. Kamm RD, Schroter RC. Is airway closure caused by a liquid ﬁlm instability? Respir Physiol 1989; 75:141–156.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

199

35. Macklem PT. Airway obstruction and collateral ventilation. Physiol Rev 1971; 51(2):368–436. 36. Hammond P. Nonlinear adjustment of a thin annular ﬁlm of viscous-ﬂuid surrounding a thread of another within a circular cylindrical pipe. J Fluid Mech 1983; 137:363–384. 37. Gauglitz PA, Radke CJ. An extended evolution equation for liquid-ﬁlm break up in cylindrical capillaries. Chem Eng Sci 1988; 43(7):1457–1465. 38. Johnson M et al. The nonlinear growth of surface-tension-driven instabilities of a thin annular ﬁlm. J Fluid Mech 1991; 233:141–156. 39. Halpern D, Grotberg JB. Fluid-elastic instabilities of liquid-lined ﬂexible tubes. J Fluid Mech 1992; 244:615–632. 40. Halpern D, Grotberg JB. Surfactant effects on ﬂuid-elastic instabilities of liquid-lined ﬂexible tubes: a model of airway closure. J Biomech Eng 1993; 115(3):271–277. 41. Halpern D, Grotberg JB. Oscillatory shear stress induced stabilization of thin ﬁlm instabilities. In: Chang HC, ed. IUTAM Symposium on Non-Linear Waves in Multi-Phase Flow. Dordrecht, The Netherlands: Kluwer Academic, 2000:33–43. 42. Halpern D, Moriarty JA, Grotberg JB. Capillary-elastic instabilities with an oscillatory forcing function. In: Durban D, Pearson JRA, eds. IUTAM Symposium on Non-Linear Singularities in Deformation and Flow. Dordrecht, The Netherlands: Kluwer Academic, 1999:243–255. 43. Otis DR Jr, Johnson M, Pedley TJ, Kamm RD. Role of pulmonary surfactant in airway closure: a computational study. J Appl Physiol 1993; 75(3): 1323–1333. 44. Heil M. Minimal liquid bridges in non-axisymmetrically buckled elastic tubes. J Fluid Mech 1999; 380:309–337. 45. Heil M. Airway closure: occluding liquid bridges in strongly buckled elastic tubes. J Biomech Eng-T Asme 1999; 121(5):487–493. 46. Hubmayr RD. Perspective on lung injury and recruitment: a skeptical look at the opening and collapse story. Am J Respir Crit Care Med 2002; 165(12):1647–1653. 47. Naureckas ET et al. Airway reopening pressure in isolated rat lungs. J Appl Physiol 1994; 76(3):1372–1377. 48. Yap DY, Liebkemann WD, Solway J, Gaver DP III. Inﬂuences of parenchymal tethering on the reopening of closed pulmonary airways. J Appl Physiol 1994; 76(5):2095–2105. 49. Gaver DP III, Samsel RW, Solway J. Effects of surface tension and viscosity on airway reopening. J Appl Physiol 1990; 69(1):74–85. 50. Perun, ML, Gaver DP III. An experimental model investigation of the opening of a collapsed untethered pulmonary airway. J Biomech Eng 1995; 117:1–9. 51. Hsu S-H, Strohl KP, Jamieson AM. Role of viscoelasticity in the tube model of airway reopening. I. Nonnewtonian sols. J Appl Physiol 1994; 76(6): 2481–2489. 52. Hsu SH, Strohl KP, Haxhiu MA, Jamieson AM. Role of viscoelasticity in the tube model of airway reopening. II. Non-Newtonian gels and airway simulation. J Appl Physiol 1996; 80(5):1649–1659.

200

Gaver et al.

53. Hsu SH, Hou CM. Air–liquid interfacial movement in models simulating airway reopening. Med Eng Phys 1998; 20(8):558–564. 54. Perun ML, Gaver DP III. Interaction between airway lining ﬂuid forces and parenchymal tethering during pulmonary airway reopening. J Appl Physiol 1995; 79(5):1717–1728. 55. Gaver DP III, Halpern D, Jensen OE, Grotberg JB. The steady motion of a semi-inﬁnite bubble through a ﬂexible-walled channel. J Fluid Mech 1996; 319:25–65. 56. Yap DYK, Gaver DP III. The inﬂuence of surfactant on two-phase ﬂow in a ﬂexible-walled channel under bulk equilibrium conditions. Phys Fluids 1998; 10(8):1846–1863. 57. Jensen OE, Horsburgh MK, Halpern D, Gaver DP. The steady propagation of a bubble in a ﬂexible-walled channel: asymptotic and computational models. Phys Fluids 2002; 14(2):443–457. 58. Vlahakis NE, Hubmayr RD. Invited review: plasma membrane stress failure in alveolar epithelial cells. J Appl Physiol 2000; 89(6):2490–2497. 59. Acevedo AD, Bowser SS, Gerritsen ME, Bizios R. Morphological and proliferative responses of endothelial cells to hydrostatic pressure: role of ﬁbroblast growth factor. J Cell Physiol 1993; 157:603–614. 60. Roelfcen J, Klein-Nulend J, Burger EH. Mechanical stimulation by intermittent hydrostatic compression promotes bone-speciﬁc gene expression in vitro. J Biomech 1995; 28(12):1493–1503. 61. Bilek AM, Dee KC, Gaver DP III. Mechanisms of surface-tension-induced epithelial cell damage in a model of pulmonary airway reopening. J Appl Physiol 2003; 94(2):770–783. 62. Kay SS, Bilek AM, Dee KC, Gaver DP. Pressure gradient, not exposure duration, determines the extent of epithelial cell damage in a model of pulmonary airway reopening. J Appl Physiol 2004; 97:269–276. 63. Burger EJ, Macklem P. Airway closure: demonstration by breathing 100% O2 at low lung volumes and by N2 washout. J Appl Physiol 1968; 25: 139–148. 64. Otis DR Jr et al. Airway closure and reopening assessed by the alveolar capsule oscillation technique. J Appl Physiol 1996; 80(6):2077–2084. 65. Gaver DP III, Kute SM. A theoretical model study of the inﬂuence of ﬂuid stresses on a cell adhering to a microchannel wall. Biophys J 1998; 75(2): 721–733. 66. Davies PF. Flow-mediated endothelial mechanotransduction. Physiol Rev 1995; 75(3):519–560. 67. Jacob AM, Gaver DP III. An investigation of the inﬂuence of cell topography on epithelial mechanical stresses during pulmonary airway reopening. Phys Fluids 2005; (17). 68. Taskar V et al. Healthy lung tolerates repetitive collapse and reexpansion. Acta Anaesthesiol Scand 1995; 39:370–376. 69. Taskar V et al. Surfactant dysfunction makes lungs vulnerable to repetitive collapse and reexpansion. Am J Respir Crit Care Med 1997; 155:313–320. 70. Lachmann B, Jonson B, Lindroth M, Robertson B. Modes of artiﬁcial ventilation in severe respiratory distress syndrome. Lung function and morphology

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

71.

72.

73.

74. 75.

76.

77. 78.

79. 80. 81.

82.

83.

84. 85.

201

in rabbits after wash-out of alveolar surfactant. Crit Care Med 1982; 10: 724–731. Nilsson R et al. Surfactant treatment and ventilation by high frequency oscillation in premature newborn rabbits: effect on survival, lung aeration, and bronchiolar epithelial lesions. Pediatr Res 1985; 19(1):143–147. Muscedere JG, Mullen JBM, Gan K, Slutsky AS. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med 1994; 149:1327–1334. Otis DR Jr, Ingenito EP, Kamm RD, Johnson M. Dynamic surface tension of surfactant TA: experiments and theory. J Appl Physiol 1994; 77(6): 2681–2688. Ingenito EP et al. Biophysical characterization and modeling of lung surfactant components. J Appl Physiol 1999; 86(5):1702–1714. Krueger MA, Gaver DP III. A theoretical model of pulmonary surfactant multilayer collapse under oscillating area conditions. J Colloid Interface Sci 2000; 229(2):353–364. Ghadiali SN, Gaver DP III. The inﬂuence of non-equilibrium surfactant dynamics on the ﬂow of a semi-inﬁnite bubble in a rigid cylindrical tube. J Fluid Mech 2003; 478:165–196. Ratulowski J, Chang H-C. Marangoni effects of trace impurities on the motion of long gas bubbles in capillaries. J Fluid Mech 1990; 210:303–328. Yu S-H, Possmayer F. Adsorption, compression and stability of surface ﬁlms from natural, lipid extract, and reconstituted pulmonary surfactants. Biochim Biophys Acta 1993; 1167(3):264–271. Hall SB et al. Importance of hydrophobic apoproteins as constituents of clinical exogenous surfactants. Am Rev Respir Dis 1992; 145(1):24–30. Klaus MH, Clements JA, Havel RJ. Composition of surface-active material isolated from beef lung. Proc Natl Acad Sci USA 1961; 47(11):1858–1859. Tchoreloff PA et al. A structural study of interfacial phospholipid and lung surfactant layers by transmission electron microscopy after Blodgett sampling: inﬂuence of surface pressure and temperature. Chem Phys Lipids 1991; 59:151–165. Hawco MW, Coolbear KP, Davis PJ, Keough KMW. Exclusion of ﬂuid lipid during compression of monolayers of mixtures of dipalmitoylphosphatidylcholine with some other phosphatidylcholines. Biochim Biophys Acta 1981; 646(1):185–187. Belbrgey O, Tchoreloff P, Benattar JJ, Proust JE. An x-ray reﬂectivity investigation of a deposited layer of the natural lung surfactant. J Colloid Interface Sci 1991; 146(2):373–381. Bangham AD, Morley CJ, Phillips MC. The physical properties of an effective lung surfactant. Biochim Biophys Acta 1979; 573:552–556. Yu S-H, Possmayer F. Effect of pulmonary surfactant protein B (SP-S) and calcium on phospholipid adsorption and squeeze-out of phosphatidylglycerol from binary phospholipid monolayers containing dipalmitoylphosphatidylcholine. Biochim Biophys Acta 1992; 1126(1):26–34.

202

Gaver et al.

86. Yu SH, Possmayer F. Role of bovine pulmonary surfactant-associated proteins in the surface-active property of phospholipid mixtures. Biochim Biophys Acta 1990; 1046(3):233–241. 87. Wang Z, Hall SB, Notter RH. Dynamic surface activity of ﬁlms of lung surfactant phospholipids, hydrophobic proteins, and neutral lipids. J Lipid Res 1995; 36(6):1283–1293. 88. Wang Z, Hall SB, Notter RH. Roles of different hydrophobic constituents in the adsorption of pulmonary surfactant. J Lipid Res 1996; 37(4):790–798. 89. Takamoto DY et al. Interaction of lung surfactant proteins with anionic phospholipids. Biophys J 2001; 81(1):153–169. 90. Morris J et al. Dynamic behavior of lung surfactant. J Biomech Eng-T Asme 2001; 123(1):106–113. 91. Lipp MM et al. Coexistence of buckled and ﬂat monolayers. Phys Rev Lett 1998; 81(8):1650–1653. 92. Zasadzinski JA. Round-up at the bilayer corral [see comments]. Biophys J 1996; 71(5):2243–2244. 93. Zasadzinski JA et al. The physics and physiology of lung surfactants. Curr Opin Colloid Interface Sci 2001; 6(5–6):506–513. 94. Lu W et al. Folding langmuir monolayers. Phys Rev Lett 2002; 89(14):146107. 95. Diamant H, Witten TA, Gopal A, Lee KYC. Unstable topography of biphasic surfactant monolayers. Europhys Lett 2000; 52(2):171–177. 96. Crane JM, Hall SB. Rapid compression transforms interfacial monolayers of pulmonary surfactant. Biophys J 2001; 80(4):1863–1872. 97. Ding JQ et al. Effects of lung surfactant proteins, SP-B and SP-C, and palmitic acid on monolayer stability. Biophys J 2001; 80(5):2262–2272. 98. Ross M, Krol S, Janshoff A, Galla HJ. Kinetics of phospholipid insertion into monolayers containing the lung surfactant proteins SP-B or SP-C. Eur Biophys J 2002; 31(1):52–61. 99. Perez-Gil J. Molecular interactions in pulmonary surfactant ﬁlms. Biol Neonate 2002; 81(suppl 1):6–15. 100. Piknova B, Schram V, Hall SB. Pulmonary surfactant: phase behavior and function. Curr Opin Struct Biol 2002; 12(4):487–494. 101. Krueger MA. Evaluation of physicochemical properties of a pulmonary surfactant analogue. In: In: Biomedical Engineering. New Orleans, Louisiana: Tulane University, 1999:229. 102. Ghadiali SN, Gaver DP III. An investigation of pulmonary surfactant physicochemical behavior under airway reopening conditions. J Appl Physiol 2000; 88(2):493–506. 103. Stebe KJ, Lin S-Y, Maldarelli C. Remobilizing surfactant retarded ﬂuid particle interfaces. I. Stress-free conditions at the interfaces of micellar solutions of surfactants with fast sorption kinetics. Phys Fluids A 1991; 3(1):3–20. 104. Stebe KJ, Maldarelli C. Remobilizing surfactant retarded ﬂuid particle interfaces. II. Controlling the surface mobility at interfaces of solution containing surface active components. J Colloid Interface Sci 1994; 163:177–189. 105. Bilek AM. Investigations of distal airway epithelial cell responses to mechanical stresses in relation to acute lung injury. In: Biomedical Engineering. New Orleans, Louisiana: Tulane University, 2003; 291.

Signiﬁcance of Air–Liquid Interfacial Stresses on Low-Volume VILI

203

106. Arold SP et al. Variable tidal volume ventilation improves lung mechanics and gas exchange in a rodent model of acute lung injury. Am J Respir Crit Care Med 2002; 165(3):366–371. 107. Suki B et al. Life-support system beneﬁts from noise. Nature 1998; 393(6681):127–128. 108. Arold SP et al. Variable ventilation induces endogenous surfactant release in normal guinea pigs. Am J Physiol Lung Cell Mol Physiol 2003; 285(2): L370–L375. 109. Krishnan JA, Brower RG. High-frequency ventilation for acute lung injury and ARDS. Chest 2000; 118(3):795–807. 110. Singh JM, Stewart TE. High-frequency mechanical ventilation principles and practices in the era of lung-protective ventilation strategies. Respir Care Clin N Am 2002; 8(2):247–260. 111. Zimmer ME IV, Williams HAR, Gaver DP III. The pulsatile motion of a semi-inﬁnite bubble in a channel: ﬂow ﬁeld, and transport of an inactive surface-associated contaminant. J Fluid Mech 2005; 537:1–33.

This page intentionally left blank

8 Cellular and Molecular Basis for Ventilator-Induced Lung Injury

ˆ ME PUGIN and SE´VERINE OUDIN JE´RO Division of Medical Intensive Care, University Hospital of Geneva Geneva, Switzerland

I. Introduction Evidence has accumulated over the past decade that although mechanical ventilation has helped many patients with respiratory failure, it can also cause damage to the lungs, particularly during the course of the acute respiratory distress syndrome (ARDS). The mortality rate of patients with this syndrome remains high, generally exceeding 30% to 40%. A recent study by the ARDS Network (1) has demonstrated that patients subjected to low tidal volume ventilation associated with positive end expiratory pressure (PEEP) had a signiﬁcantly lower mortality than patients receiving higher tidal volumes. This indicated that the ventilatory strategy signiﬁcantly inﬂuenced mortality. In addition to direct lung injury and air leaks (2), mechanical ventilation is responsible for worsening acute lung injury by triggering lung and systemic inﬂammation (3–6). This process is now widely known as ventilator-induced lung injury (VILI) (4,7,8). Various groups have modeled the deleterious effects of mechanical ventilation in animals. Nearly 20 years ago, Dreyfuss et al. (9) reported that edema formed rapidly in the lungs of rats ventilated with large tidal 205

206

Pugin and Oudin

volumes, and proposed the concept of volutrauma. Later, Kawano et al. (10) reported that polymorphonuclear neutrophils (PMN) were indispensable for the development of VILI in a surfactant-depletion rabbit model. One of the ﬁrst studies to directly address the issue of whether various ventilatory regimens could modulate the production of inﬂammatory mediators was performed by the group of Slutsky et al. (11). In an ex vivo nonperfused ventilated rat lung model, these authors showed that an injurious ventilatory regimen was responsible for a massive lung production of inﬂammatory mediators. They reported a marked increase in the bronchoalveolar lavage ﬂuid concentration of tumor necrosis factor (TNF)-a, interleukin (IL)-1b, macrophage inﬂammatory protein (MIP)-2, IL-6, interferon gamma (IFNc) and IL-10 (11) when lungs were ventilated with large tidal volumes in the absence of PEEP. Other investigators were, however, unable to demonstrate the upregulation of TNF-a using the same model (12). Nevertheless, all studies that have measured inﬂammatory mediators in the airway of ventilated animals have invariably shown an upregulation of chemokines (12,13). These mediators are crucial chemoattractants for circulating phagocytes responsible for the transmigration of these cells into the airways. PMNs and their chemoattractants from the CXCR2 family, such as MIP-2 in rodents and IL-8 in higher animal species and humans, were identiﬁed as key players in the pathogenesis of VILI. A recent study in mice showed the critical role of CXCR2 receptor and CXCR2 ligands during the course of VILI (14). The term ‘‘biotrauma’’ has been coined by Slutsky et al. to describe ventilator-induced lung inﬂammation (6). It is proposed that this biological response is due to the mechanical forces applied to lung cells during ventilation (15), and may only be apparent if the lung has been ‘‘primed’’ or pre-injured (4). Lung cells may be submitted to a variety of mechanical forces during positive pressure mechanical ventilation. These include unusual stretching of the lung parenchyma following overinﬂation, shear stress of bronchial cells due to turbulent ﬂuxes of gases, and shear stress of alveolar cells resulting from cyclic opening and closing of alveoli during tidal ventilation (16). Only recently have researchers been interested in deﬁning the cellular and molecular pathways governing cell responses to mechanical forces, and the interplay between the primary inﬂammatory process in the lung and the superimposed effect of mechanical ventilation.

II. Ventilator-Induced Lung Inﬂammation For a long period of time, patients with respiratory failure were submitted to ventilatory regimens set to normalize blood gases. These regimens frequently used large tidal volumes and high airway pressures. It has only

Cellular and Molecular Basis for Ventilator-Induced Lung Injury

207

recently been recognized that overinﬂation of the lung, or parts of the lungs, was deleterious and that it may induce lung inﬂammation or synergize with the primary pulmonary inﬂammatory process. It is now widely accepted, based on a series of animal and human studies, that clinicians should ventilate their patients with lower tidal volumes, and keep the lung open by increasing the end-expiratory volume. It has also been recognized that conventional mechanical ventilation does not signiﬁcantly affect lung functions. For example, patients ventilated for drug overdose or a neuromuscular disease usually do not suffer from prolonged mechanical ventilation in the absence of a pulmonary superinfection or gastric ﬂuid aspiration. The situation is different when mechanical ventilation superimposes on a primary or secondary lung inﬂammatory process. In this case, evidence is accumulating showing that mechanical ventilation worsens lung and possibly systemic inﬂammation. In animal studies, lung cell stretching synergizes with local inﬂammation and signiﬁcantly worsens lung inﬂammation and function. This could be modeled in vitro by showing a synergy between cell stretching and pro-inﬂammatory mediators such as lipopolysaccharide (LPS) or TNF-a in the induction of IL-8, for example (17,18). It is also conceivable that mechanical ventilation induces a subtle lung cell activation that is not clinically apparent, but which primes the lung for a subsequent inﬂammatory injury. It has, for example, been shown that conventional mechanical ventilation of rabbits with healthy lungs induced an increase in monocyte chemotactic protein (MCP)-1, the recruitment of alveolar macrophages, and increased lung mRNA for TNF-a and IL-1b (19). The addition of a noxious stimulus such as intravenous endotoxin in this model was associated with the development of lung inﬂammation to levels that are much higher than those found in nonventilated rabbits injected only with LPS (20). Our current understanding of the pathogenesis of VILI is depicted in Figure 1. Macrophages represent the typical sentinel cells in the airways, sensing the presence of noxious stimuli. The encounter with bacteria and bacterial products, for example, will trigger these cells to secrete locally IL-1b and TNF-a. IL-1b has been shown to be a very important bioactive pro-inﬂammatory cytokine in the lungs from ARDS patients (22,23). These cytokines will stimulate neighboring cells to secrete IL-8, the major chemoatractant for neutrophils (24). Airway neutrophils will in turn augment lung inﬂammation and tissue injury. It is believed that mechanical ventilation acts synergistically with the inﬂammatory process and increases neutrophil recruitment. There is now evidence that the local inﬂammatory process may spread outside the lung and that mechanical ventilation may play a role in the translocation of bacteria, bacterial products and cytokines to the systemic circulation (21,25,26). However a clear relationship between this phenomenon and end-organ dysfunction such as that observed in ARDS patients remains to be established (27).

208

Pugin and Oudin

Figure 1 Cellular and molecular pathogenesis of ventilator-induced lung injury. Schematic view of a bronchiolar duct and an alveolus submitted to cyclic stretch. (A) Alveolar and bronchiolar cells involved in the inﬂammatory response during acute lung injury. (B) Synergy between bacteria, bacterial products (LPS), and cell activation due to cyclic cell stretch (large arrows) for the production of pro-inﬂammatory cytokines (IL)-1b and TNF-a by myeloid cells in the alveolar space, and for the secretion of the neutrophil chemoattractant IL-8 by alveolar phagocytes, type I and type II epithelial cells. Abbreviations: IL, interleukin; LPS, lipopolysaccharide; TNF, tumor necrosis factor. Source: From Ref. 24.

III. Cells Submitted to Mechanical Stress Little is known about the lung cell types in vivo that are sensitive to mechanical forces during mechanical ventilation, as well as the degree of distension of airway cells or cells from the interstitium. Animal studies were performed to visualize lung structure movement during mechanical ventilation directly and indirectly. In vivo microscopy studies of ventilated dogs and pigs showed that the major component of lung volume change during mechanical ventilation was due to alveolar recruitment and de-recruitment (28,29). Morphometric measurements were performed on rabbit lungs ﬁxed after ex vivo ventilation, and showed an increase of 30% in the alveolar surface during inﬂation (30). In ventilated piglets with experimental pneumonia, Goldstein et al. (31) showed an emphysema-like increase of the alveolar surface in nondependent, well-aerated regions of the lung. In regions with pneumonia, these authors observed a distention of the bronchioles with a mean increase of the surface section of 100%, corresponding to a

Cellular and Molecular Basis for Ventilator-Induced Lung Injury

209

circumferential cell elongation of 40%. It can be concluded from these studies that cell stretching occurs during positive pressure mechanical ventilation both in the alveolar and in the bronchiolar compartments, and that cell stretching up to 40% may be measured. IV. What Happens to Cells When They Are Submitted to Cyclic Stretch? Lung cells are not the only cells in the body submitted to stretch. The response to stretch has been studied in a variety of cell types, including osteoblasts (32), chondrocytes (33), mesangial cells, myocytes (34), arterial endothelial cells (35), and vascular smooth muscle cells (36). Of the cell types relevant to the lung, macrophages and monocytic cells (18), alveolar and bronchial epithelial cells (37–43), ﬁbroblasts (44), bronchial and vascular smooth muscle cells (45), and microvascular endothelial cells (46) were tested using various in vitro models generating cell cyclic stretch. The cell responses varied depending on the cell types studied. Alveolar macrophages secreted IL-8 and matrix metalloproteinase (MMP)-9 upon cyclic stretch (18), ﬁbroblasts increased type I collagen expression (44), endothelial cells produced MMPs (46), and vascular smooth muscle cells secreted growth factors (45). In addition to cellular activation, evidence has accumulated that cells submitted to 20% to 30% cyclic elongation underwent structural changes including plasma membrane breaks and cytoskeleton rearrangement. In rat lungs mechanically ventilated ex vivo, Gajic et al. (47) showed that lung cells experienced reversible plasma membrane stress failure. The number of injured cells was signiﬁcantly greater in lungs ventilated with large tidal volumes and zero end-expiratory pressure than in lungs ventilated with small tidal volumes and PEEP. In elegant in vitro studies using ﬂuorescent dyes, Vlahakis et al. (48,49) showed that plasma membrane from alveolar epithelial cells ruptured upon cell stretching. These authors observed that one population of cells was able to reseal despite an extensive increase in cell permeability, whereas another population never resealed and died (48–50). Alveolar epithelial cells were found to exocytose intracellular lipid vesicles to the plasma membrane, not only to prevent cell breaks but also to reseal these breaks (50). This phenomenon can be considered as a wound repair mechanism but also as a cytoprotective mechanism against plasma membrane stress failure. The relationship between cell activation and plasma membrane stress failure remains unclear. V. Mechanosensing The cellular molecules that sense cellular deformation upon cyclic stretch and transform a mechanical strain into a biological sequel remain to be

210

Pugin and Oudin

determined in lung cells. By analogy with other mechanical stresses, such as shear stress, there are several candidate molecules and cell structures that are likely to participate in this proximal event. Based on the relative conservation of these pathways between various cell types, and on recent reports with lung epithelial cells, it can be anticipated that integrins (51), stretchactivated ion channels (52), signaling molecules associated with the focal adhesion plaque, and the cytoskeleton itself play a role in ‘‘mechanosensing’’ (16,53–55). The role of the actin cytoskeleton in the mechanosensing apparatus is potentially important for at least two reasons: (i) it has previously been shown that endothelial cells respond to shear stress as well as to cyclic stretch by reorienting their actin cytoskeleton obliquely to the direction of the mechanical strain, and by reorganizing the actin network in heavy structures known as ‘‘stress ﬁbers’’ (56,57); (ii) stress ﬁbers are linked to the extracellular matrix via specialized plasma membrane platforms, the so-called ‘‘focal adhesion plaques.’’ The focal adhesion plaques are organized plasma membrane structures in which actin ﬁbers anchor to the intracellular domain of integrins. There are several proteins that segregate in these plaques and adapt actin ﬁlaments to integrins, including paxillin, talin, tensin, Src kinase and the focal adhesion kinase (FAK). Some of these proteins also function as signal transduction molecules that can be activated (phosphorylated) upon cell migration or mechanical cell stresses (16,34,53–55,58–62). FAK segregates into focal adhesion plaques, and has been shown to be activated in a variety of cell types by mechanical stresses, such as ﬁbroblast (60,61), cardiac myocytes (34), and endothelial cells (62). Importantly, once activated, FAK turns on signaling pathways such as the Ras-Raf-ERK mitogen-activated protein kinases (MAPK) cascade, a pathway known to be activated by mechanical stresses, as demonstrated in ﬁbroblasts (60), in cardiac myocytes (34), and in endothelial cells submitted to shear stress (54). A viable model for the mechanosensing apparatus of lung cells submitted to cyclic stretch could be the following: integrins, through their anchoring in the extracellular matrix at the level of focal adhesion plaque, would sense the cell deformation, and transmit an ‘‘outside-in’’ signal through the plasma membrane. This would activate adaptor proteins sitting in the intracellular region of the plaque, most likely FAK and paxillin. The activation of these proteins would then induce cytoskeleton rearrangement and turn on signaling pathways leading to the transcription of target genes. In unpublished work, we have shown that cytochalasin D, an inhibitor of actin polymerization, completely blocked the cyclic stretch-dependent activation of IL-8 in BEAS-2B epithelial cells, whereas this compound had no effect on the production of IL-8 by these cells in response to TNF-a. Further studies are needed to unravel a possible link between cytoskeleton rearrangement and gene transcription in cells submitted to mechanical stresses.

Cellular and Molecular Basis for Ventilator-Induced Lung Injury

211

VI. Cyclic Stretch of Lung Epithelial Cells Of all lung cell types, lung epithelial cells were the most studied with regard to responses to cyclic stretch. Cyclic stretch modulates key functions of alveolar epithelial cells. For example, rat alveolar type II cells submitted to cyclic stretch regulate surfactant homeostasis. Cyclic stretch induces the exocytosis of lamellar bodies, a phenomenon dependent on stretchinduced cytoplasmic calcium mobilization (63). The mechanism by which mechanical stimulation of alveolar cells inﬂuences the cell cycle remains poorly understood. Stretch also modulates the proliferation of alveolar epithelial cells depending on their type I or type II phenotype (40,64), and whether they originate from fetal or mature lungs (65,66). Cyclic stretch also activates apoptotic pathways in type II cells (64,67). Others have reported that cyclic stretch induced protective mechanisms against apoptosis in type II cells via mechanisms implicating nitric oxide (68). VII. Cyclic Stretch–Induced Cell Activation One important feature of lung cells submitted to cyclic stretch is the release of inﬂammatory mediators. Rat type II cells produce the IL-8 equivalent in rodents, the MIP-2 (17), whereas human type II cells secrete IL-8 (42,43, 69,70), prostacyclin (71), hepatocyte growth factor (HGF), and transforming growth factor (TGF)-b1 (38,70). There is also evidence that cyclic stretch activates gadolinium-sensitive ion channels in type II cells (52,60), as well as Naþ–Kþ-ATPase pumps (72). Cyclic stretch also induces the reorganization of a5-b1 integrins and the formation of new adhesion plaques (51). Bronchial epithelial cells submitted to a mechanical load increase their production of endothelin-1 and -2 and TGF-b2 (73), and of heparin-binding epidermal growth factor (HB-EGF) via the activation of ERK1/2 MAPK (74). Several protein kinases are also stimulated by cyclic stretch in alveolar cells. Protein kinase C (PKC) is activated in response to stretch in human A549 type II-like cells and participates in the pathway leading to the secretion of HGF, IL-8 and TGF-b1 (38,70). MAPK are also activated by cyclic stretch in alveolar cells. ERK1/2 MAPK are activated in type II cells via an unusual pathway implicating G proteins and epidermal growth factor receptor (EGFR) in stretched A549 cells (39). Stress-activated protein kinase/jun N-terminal kinase (SAPK/JNK) and p38 MAPK are also activated, leading to the secretion of IL-8 (43,75). There is evidence that the IL-8 secretion of type II cells is due to the activation of nuclear factor (NF)-jB and activated protein (AP)-1 responding elements on the IL-8 gene, downstream of PKC, NF-jB-inducing kinase, and SAPK/JNK MAPK (76). Human bronchial epithelial cells (BEAS-2B cells) also increase their secretion of IL-8 in response to cyclic stretch (75). This secretion is

212

Pugin and Oudin

dependent on the activation of MAPK, in particular p38 MAPK. Whether p38 is responsible for increased IL-8 gene transcription or IL-8 mRNA stabilization (76,77,79–82) has not been fully established. In work submitted for publication, we show that NF-jB is a ‘‘stretch responding element’’ in the IL-8 promoter of BEAS-2B cells, and that AP-1 is required for high level production of the IL-8 protein. Interestingly, NF-jB has also been shown to be a ‘‘shear stress responding element’’ in endothelial cells (83–85), and to be activated by cyclic stretch in other cell types, such as ﬁbroblasts (86), skeletal (87) and smooth muscle cells (88,89), and macrophages (18). IL-8 gene transcription is directly increased by cyclic stretch, as demonstrated by the upregulation of nascent IL-8 mRNA transcripts. Actinomycin D, a transcription inhibitor, and cyclohexamide, an inhibitor of protein translation both blocked the production of IL-8 by cyclic stretch. Interestingly, the CEB/P-NF-IL6 site adjacent to the NF-jB site is required for synergistic effects between stretch and inﬂammatory mediators. Taken together, these ﬁndings suggest that cyclic stretch of airway epithelial cells induces a de novo, NF-jB–dependent production of IL-8 transcripts. Kinases of the MAPK family, and p38 MAPK in particular, are responsible for a further increase of the IL-8 protein by a post-transcriptional mechanism through the stabilization of IL-8 mRNA. A summary of the signaling of lung epithelial cells submitted to cyclic stretch (mechanotransduction) is shown schematically in Figure 2.

VIII. Synergy Between Cyclic Stretch and Inﬂammatory Stimuli In the pathogenesis of VILI, injured or infected lungs are more sensitive to cyclic stretch-induced inﬂammation. This phenomenon has been modeled in vitro by combining the effect of stretch with that of a pro-inﬂammatory stimulus in rat type II cells (17). In human bronchial epithelial cells (BEAS-2B cells), we have observed that cyclic stretch combined with pro-inﬂammatory stimuli such as TNF-a and IL-1b, induced a massive secretion of IL-8, revealing a synergistic effect between these two stimuli. The synergy was not present when stretched cells were co-stimulated with bacterial products, where only an additive effect was observed. This synergistic effect between stretch and pro-inﬂammatory cytokines certainly represents an important aspect of the pathogenesis of VILI, and adds to the notion that healthy lungs are relatively insensitive to stretch, but that diseased or inﬂamed lungs are hypersensitive to excessive bronchial and alveolar distension (18). In unpublished studies, we have found that the cooperation of NFs binding to NF-jB and C/EBP-NF-IL6 responding elements represents the molecular basis for the observed synergy between stretch and proinﬂammatory mediators (17,18). C/EBPb, also known as NF-IL6, can bind and activate the promoters of IL-6 and -8 (90). The activity of this

Cellular and Molecular Basis for Ventilator-Induced Lung Injury

213

Figure 2 Schematic view of mechanotransduction in lung epithelial cells. A putative model of IL-8 secretion induced by cyclic stretch. Integrins anchored to the extracellular matrix sense the cellular deformation induced by stretch and send an ‘‘outside-in’’ signal to the interior of cells at the level of focal adhesion plaques. Proteins adapting the intracytoplasmic tail of integrins to the actin cytoskeleton such as FAK, paxillin, talin, tensin and Src kinases get activated and stimulate cytoskeleton rearrangement into ‘‘stress ﬁbers.’’ The MAPK and the NF-jB pathways are activated. NF-jB translocates to the nucleus and activates the IL-8 gene. p38 MAPK is phosphorylated and participates in IL-8 mRNA stabilization. Increased cytoplasmic levels of IL-8 mRNA are responsible for an increased translation, the production and the secretion of mature IL-8 protein. Activation through the C/EBP responding element is responsible for the synergy between cyclic stretch and pro-inﬂammatory mediators, whereas the occupation of the AP-1 site is necessary for a high throughput production of IL-8 mRNA. Abbreviations: AP, activated protein; IL, interleukin; FAK, focal adhesion kinase; MAPK, mitogen-activated protein kinases; NF, nuclear factor.

transcription factor is regulated by phosphorylation by members of the MAP kinase family, i.e., ERK1/2, SAPK/JNK and p38 MAPK (91,92). The mutation of C/EBP responding element in the promoter region of the IL-8 gene abrogated the synergy observed between TNF-a and cyclic stretch on the IL-8 production. Cooperation at the level of promoters of target genes may therefore represent the molecular basis for the observed synergy between cyclic stretch and pro-inﬂammatory cytokines (90,93,94). IX. Genes Activated by Cyclic Stretch A variety of mediators are produced by cells submitted in vitro to cyclic stretch and in the lungs of mechanically ventilated animals or humans.

214

Pugin and Oudin

Since it appears that cyclic stretch increases the transcriptional activity of the IL-8 gene, it would be worthwhile to determine the array of genes that are turned on by cylic stretch, and also genes that are downregulated by this mechanical strain. In a differential gene expression approach using complementary DNA (cDNA) expression arrays (Atlas2, Clontech), we compared the expression proﬁle of human lung epithelial cells (BEAS-2B cells) grown in static conditions versus cells submitted to four hours of cyclic stretch. cDNAs from these cells were hybridized with probes corresponding to 588 cDNAs, representative of six categories of genes important for various cell functions plus housekeeping genes. Among the 588 genes tested were a selection of: oncogenes, tumor suppressors, regulators of cell cycle, signaling molecules, transcription factors, DNA binding proteins, proteins of the stress response, mediators of apoptosis, various receptors, cell surface antigens and receptors, adhesion molecules, growth factors, cytokines, chemokines, and hormones. In three independent experiments, no mRNA was induced more than twofold in a reproducible manner. However, 10 mRNAs were consistently induced at lower levels in all three experiments (Table 1). Interestingly, six of these transcripts were already reported to be induced by mechanical stretch in other cell systems. Importantly, IL-8 was among the genes that were consistently found to be upregulated in these array experiments. Quantitative reverse transcriptase polymerase chain reaction (RT-PCR) was performed with speciﬁc primers for all 10 upregulated

Table 1 Upregulated cDNAs in Human Bronchial Epithelial Cells Submitted to Four Hours of Cyclic Stretch In Vitro, Compared with Cells Grown in Static Conditions (Mean of Three Independent Microarray Experiments) mRNA IL-8a MCP-1a GADD-45a FRA-1a ICAM-1 TGF-b2a EGFR IGFBP-3 ATF-4 HSP-90

Mean fold induction

Induced by stretch in other systems

1.5 1.4 1.8 1.7 1.5 1.4 1.5 1.7 1.2 1.1

Yes (18,38,42,75) Yes (95–97) N.D. Yes (98) Yes (99) Yes (73) N.D. N.D. N.D. Yes (100)

a Upregulation was conﬁrmed by quantitative RT-PCR. Abbreviations: IL, interleukin; MCP, monocyte chemotactic protein; GADD-45, growth arrest and DNA-damage-inducible gene-45; TGF, tumor growth factor; HSP, heat shock protein; RTPCR, reverse transcriptase polymerase chain reaction; ICAM, intracellular adhesion molecule; FRA-1, Fos-related antigen-1; IGFBP, insulin like growth factor binding protein; ATF-4, activating transcription factor-4; EGFR, epidermal growth factor receptor.

Cellular and Molecular Basis for Ventilator-Induced Lung Injury

215

transcripts. Only ﬁve transcripts showed signiﬁcant mRNA upregulation by quantitative RT-PCR in BEAS-2B cells submitted to four hours of cyclic stretch as compared to cells grown in static conditions: IL-8, MCP-1, growth arrest and DNA-damage-inducible gene-45 (GADD-45), FRA-1 and TGF-b2. Whereas IL-8, MCP-1 and TGF-b2 are well recognized mediators of lung inﬂammation and repair and could be expected to be upregulated by stretch, the upregulation of GADD-45 and FRA-1 was unexpected. GADD-45 is a p53 regulated protein implicated in the regulation of cell cycle, and activates p38 and SAPK/JNK MAPK (101–103). FRA-1 is a member of the c-fos oncogene family and can participate in AP-1 transcription (104). c-fos and AP-1 have been previously shown to participate in gene transcription induced by stretch (11,75). A differential gene expression based on microarrays has limitations, however (105). It allows one to test only a subset of genes, a single time point, a ﬁxed condition of cell stretching, and the level of detection might be low, as compared with other techniques such as the quantitative RT-PCR (105). Nevertheless, these experiments show that cyclic stretch does not per se turn on a vast array of genes, but is rather selective in the genes that are upregulated by this mechanical strain. It is not understood at this point why some genes that possess a NF-jB responding element—a putative ‘‘stretch responding element’’—in their promoter, such as IL-8 and MCP-1, are induced by stretch, whereas other genes with NF-jB responding elements, such as TNF-a, IL-1b, or IL-6, are not. Tandem repeats of NF-jB binding sites can be activated by TNF-a, but not by cyclic stretch in lung epithelial cells (75). This points out the importance of the DNA context of a transcription factor binding site inside a promoter region, and forms the basis for the speciﬁcity of transcription (106). Further studies are needed to determine the array of genes that are upregulated and downregulated by stretch to better understand the pathogenesis of VILI.

X. Conclusions and Perspectives Cyclic stretch of lung cells induces a selective activation of genes, mainly inﬂammatory genes and related transcription factors. This occurs through an increase in the rate of transcription of these genes, but also through a stabilization of their mRNA, via a p38-dependent pathway, as demonstrated for IL-8. The chemokines IL-8 and MCP-1 are induced by cyclic stretch in macrophages and epithelial cells, and certainly play an important role in recruiting phagocytes to the airways. NF-jB is an important stretch responding element in the IL-8 promoter. Cyclic stretch synergizes with pro-inﬂammatory cytokines for the production of IL-8. The molecular basis for this phenomenon might reside in the cooperation between NF-jB and C/EBPb at the level of the IL-8 promoter. Further studies are needed to

216

Pugin and Oudin

determine the array of genes that are turned on by cyclic stretch. The relationship between stretch and ﬁbrotic reactions, particularly relevant in the context of ARDS, should also be studied. The study of the effects of stretch on the immune function of cells of the innate immunity—not only the inﬂammatory response—would also be important in the future. It is indeed possible that mechanical ventilation directly affects immune functions of lung cells. Finally, to test the clinical relevance of these mediators of VILI, inhibitors of key cytokines, chemokines or signaling molecules should be tested in animal models to prevent VILI. NF-jB and p38 inhibitors were, for example, tested in a model of chronic obstructive pulmonary disease (107). This approach may lead to the identiﬁcation of pharmacological compounds able to dampen lung inﬂammation and injury related to mechanical ventilation.

Acknowledgment This work was supported in part by a grant from the Stanley Thomas Johnson Foundation to J. P. References 1. ARDS Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342:1301–1308. 2. Rouby JJ, et al. Histologic aspects of pulmonary barotrauma in critically ill patients with acute respiratory failure. Intensive Care Med 1993; 19:383–389. 3. Haitsma JJ, Uhlig S, Goggel R, Verbrugge SJ, Lachmann U, Lachmann B. Ventilator-induced lung injury leads to loss of alveolar and systemic compartmentalization of tumor necrosis factor-alpha. Intensive Care Med 2000; 26:1515–1522. 4. Pugin J, Jolliet P. Ventilator-induced lung injury: an inﬂammatory disease? Clin Pulm Med 1998; 5:290–299. 5. Slutsky AS, Tremblay LN. Multiple system organ failure. Is mechanical ventilation a contributing factor? Am J Respir Crit Care Med 1998; 157:1721–1725. 6. Tremblay LN, Slutsky AS. Ventilator-induced injury: from barotrauma to biotrauma. Proc Assoc Am Phys 1998; 110:482–488. 7. Dos Santos CC, Slutsky AS. Invited review: mechanisms of ventilator-induced lung injury: a perspective. J Appl Physiol 2000; 89:1645–1655. 8. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 9. Dreyfuss D, Basset G, Soler P, Saumon G. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 1985; 132:880–884. 10. Kawano T, et al. Effect of granulocyte depletion in a ventilated surfactantdepleted lung. J Appl Physiol 1987; 62:27–33.

Cellular and Molecular Basis for Ventilator-Induced Lung Injury

217

11. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99:944–952. 12. Held HD, Boettcher S, Hamann L, Uhlig S. Ventilation-induced chemokine and cytokine release is associated with activation of nuclear factor-kappaB and is blocked by steroids. Am J Respir Crit Care Med 2001; 163:711–716. 13. Ricard JD, Dreyfuss D, Saumon G. Production of inﬂammatory cytokines in ventilator-induced lung injury: a reappraisal. [comment]. Am J Respir Crit Care Med 2001; 163:1176–1180. 14. Belperio JA, et al. Critical role for CXCR2 and CXCR2 ligands during the pathogenesis of ventilator-induced lung injury. J Clin Invest 2002; 110:1703–1716. 15. Liu M, Tanswell AK, Post M. Mechanical force-induced signal transduction in lung cells. Am J Physiol 1999; 277:L667–L683. 16. Pugin J. Molecular mechanisms of lung cell activation induced by cyclic stretch. Crit Care Med 2003; 31:S200–S206. 17. Mourgeon E, Isowa N, Keshavjee S, Zhang X, Slutsky AS, Liu M. Mechanical stretch stimulates macrophage inﬂammatory protein-2 secretion from fetal rat lung cells. Am J Physiol Lung Cell Mol Physiol 2000; 279:L699–L706. 18. Pugin J, et al. Activation of human macrophages by mechanical ventilation in vitro. Am J Physiol 1998; 275:L1040–L1050. 19. Bregeon F, et al. Conventional mechanical ventilation of healthy lungs induced pro-inﬂammatory cytokine gene transcription. Respir Physiol Neurobiol 2002; 132:191–203. 20. Altemeier WA, et al. Mechanical ventilation with moderate tidal volumes synergistically increases lung cytokine response to systemic endotoxin. Am J Physiol Lung Cell Mol Physiol 2004; 287:L533–L542. 21. Ranieri VM, Giunta F, Suter PM, Slutsky AS. Mechanical ventilation as a mediator of multisystem organ failure in acute respiratory distress syndrome. JAMA 2000; 284:43–44. 22. Pugin J, Ricou B, Steinberg KP, Suter PM, Martin TR. Proinﬂammatory activity in bronchoalveolar lavage ﬂuids from patients with ARDS, a prominent role for interleukin-1. Am J Respir Crit Care Med 1996; 153:1850–1856. 23. Pugin J, Verghese G, Widmer MC, Matthay MA. The alveolar space is the site of intense inﬂammatory and proﬁbrotic reactions in the early phase of acute respiratory distress syndrome. Crit Care Med 1999; 27:304–312. 24. Goodman RB, Pugin J, Lee JS, Matthay MA. Cytokine-mediated inﬂammation in acute lung injury. Cytokine Growth Factor Rev 2003; 14:523–535. 25. Ranieri VM, et al. Effect of mechanical ventilation on inﬂammatory mediators in patients with acute respiratory distress syndrome: a randomized controlled trial. [comment]. JAMA 1999; 282:54–61. 26. Stuber F, et al. Kinetic and reversibility of mechanical ventilation-associated pulmonary and systemic inﬂammatory response in patients with acute lung injury. Intensive Care Med 2002; 28:834–841. 27. Pugin J. Is the ventilator responsible for lung and systemic inﬂammation? Intensive Care Med 2002; 28:817–819. 28. Carney DE, et al. The mechanism of lung volume change during mechanical ventilation. Am J Respir Crit Care Med 1999; 160:1697–1702.

218

Pugin and Oudin

29. McCann UG II, Schiller HJ, Carney DE, Gatto LA, Steinberg JM, Nieman GF. Visual validation of the mechanical stabilizing effects of positive endexpiratory pressure at the alveolar level. J Surg Res 2001; 99:335–342. 30. Gil J, Bachofen H, Gehr P, Weibel ER. Alveolar volume-surface area relation in air- and saline-ﬁlled lungs ﬁxed by vascular perfusion. J Appl Physiol 1979; 47:990–1001. 31. Goldstein I, Bughalo MT, Marquette CH, Lenaour G, Lu Q, Rouby JJ. Mechanical ventilation-induced air-space enlargement during experimental pneumonia in piglets. Am J Respir Crit Care Med 2001; 163:958–964. 32. Granet C, Boutahar N, Vico L, Alexandre C, Lafage-Proust MH. MAPK and SRC-kinases control EGR-1 and NF-kappa B inductions by changes in mechanical environment in osteoblasts. Biochem Biophys Res Commun 2001; 284:622–631. 33. Hung CT, et al. Mitogen-activated protein kinase signaling in bovine articular chondrocytes in response to ﬂuid ﬂow does not require calcium mobilization. J Biomech 2000; 33:73–80. 34. Seko Y, Takahashi N, Tobe K, Kadowaki T, Yazaki Y. Pulsatile stretch activates mitogen-activated protein kinase (MAPK) family members and focal adhesion kinase (p125(FAK)) in cultured rat cardiac myocytes. Biochem Biophys Res Commun 1999; 259:8–14. 35. Chien S, Li S, Shyy YJ. Effects of mechanical forces on signal transduction and gene expression in endothelial cells. Hypertension 1998; 31:162–169. 36. Zou Y, Hu Y, Metzler B, Xu Q. Signal transduction in arteriosclerosis: mechanical stress-activated MAP kinases in vascular smooth muscle cells. Int J Mol Med 1998; 1:827–834. 37. Cavanaugh KJ Jr, Oswari J, Margulies SS. Role of stretch on tight junction structure in alveolar epithelial cells. Am J Respir Cell Mol Biol 2001; 25:584–591. 38. Yamamoto H, Teramoto H, Uetani K, Igawa K, Shimizu E. Cyclic stretch upregulates interleukin-8 and transforming growth factor-beta1 production through a protein kinase C-dependent pathway in alveolar epithelial cells. Respirology 2002; 7:103–109. 39. Correa-Meyer E, Pesce L, Guerrero C, Sznajder JI. Cyclic stretch activates ERK1/2 via G proteins and EGFR in alveolar epithelial cells. Am J Physiol Lung Cell Mol Physiol 2002; 282:L883–L891. 40. Chess PR, Toia L, Finkelstein JN. Mechanical strain-induced proliferation and signaling in pulmonary epithelial H441 cells. Am J Physiol Lung Cell Mol Physiol 2000; 279:L43–L51. 41. Savla U, Sporn PH, Waters CM. Cyclic stretch of airway epithelium inhibits prostanoid synthesis. Am J Physiol 1997; 273:L1013–L1019. 42. Vlahakis NE, Schroeder MA, Limper AH, Hubmayr RD. Stretch induces cytokine release by alveolar epithelial cells in vitro. Am J Phyiol 1999; 277:L167–L173. 43. Quinn D, Tager A, Joseph PM, Bonventre JV, Force T, Hales CA. Stretchinduced mitogen-activated protein kinase activation and interleukin-8 production in type II alveolar cells. Chest 1999; 116:89S–90S.

Cellular and Molecular Basis for Ventilator-Induced Lung Injury

219

44. Breen EC, Fu Z, Normand H. Calcyclin gene expression is increased by mechanical strain in ﬁbroblasts and lung. Am J Respir Cell Mol Biol 1999; 21:746–752. 45. Quinn TP, Schlueter M, Soifer SJ, Gutierrez JA. Cyclic mechanical stretch induces VEGF and FGF-2 expression in pulmonary vascular smooth muscle cells. Am J Physiol Lung Cell Mol Physiol 2002; 282:L897–L903. 46. Haseneen NA, Vaday GG, Zucker S, Foda HD. Mechanical stretch induces MMP-2 release and activation in lung endothelium: role of EMMPRIN. Am J Physiol Lung Cell Mol Physiol 2003; 284:L541–L547. 47. Gajic O, Lee J, Doerr CH, Berrios JC, Myers JL, Hubmayr RD. Ventilatorinduced cell wounding and repair in the intact lung. [comment]. Am J Respir Crit Care Med 2003; 167:1057–1063. 48. Vlahakis NE, Schroeder MA, Pagano RE, Hubmayr RD. Deformationinduced lipid trafﬁcking in alveolar epithelial cells. Am J Physiol Lung Cell Mol Physiol 2001; 280:L938–L946. 49. Vlahakis NE, Schroeder MA, Pagano RE, Hubmayr RD. Role of deformationinduced lipid trafﬁcking in the prevention of plasma membrane stress failure. Am J Respir Crit Care Med 2002; 166:1282–1289. 50. Vlahakis NE, Hubmayr RD. Response of alveolar cells to mechanical stress. Curr Opin Crit Care 2003; 9:2–8. 51. Zhang HJ, Cai SX, Lu X, Wang YL, Wang HB, Huang QP. Effects of mechanical strain on the proliferation of pulmonary epithelial cells and reorganization of integrins. Acta Biochem Biophys Sin 2002; 34:748–752. 52. Liu M, Xu J, Tanswell AK, Post M. Inhibition of mechanical strain-induced fetal rat lung cell proliferation by gadolinium, a stretch-activated channel blocker. J Cell Physiol 1994; 161:501–507. 53. Chen KD, et al. Mechanotransduction in response to shear stress. Roles of receptor tyrosine kinases, integrins, and Shc. J Biol Chem 1999; 274: 18393–18400. 54. Li S, et al. Fluid shear stress activation of focal adhesion kinase. Linking to mitogen-activated protein kinases. J Biol Chem 1997; 272:30455–30462. 55. Shyy JY, Chien S. Role of integrins in endothelial mechanosensing of shear stress. Circ Res 2002; 91:769–775. 56. Birukov KG, et al. Shear stress-mediated cytoskeletal remodeling and cortactin translocation in pulmonary endothelial cells. Am J Respir Cell Mol Biol 2002; 26:453–464. 57. Hayakawa K, Sato N, Obinata T. Dynamic reorientation of cultured cells and stress ﬁbers under mechanical stress from periodic stretching. Exp Cell Res 2001; 268:104–114. 58. Turner CE. Paxillin and focal adhesion signalling. Nat Cell Biol 2000; 2:E231–E236. 59. Turner CE. Paxillin interactions. J Cell Sci 2000; 113(Pt 23):4139–4140. 60. Wang HB, Dembo M, Hanks SK, Wang Y. Focal adhesion kinase is involved in mechanosensing during ﬁbroblast migration. Proc Natl Acad Sci USA 2001; 98:11295–11300. 61. Wang JG, Miyazu M, Matsushita E, Sokabe M, Naruse K. Uniaxial cyclic stretch induces focal adhesion kinase (FAK) tyrosine phosphorylation

220

62.

63. 64. 65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

Pugin and Oudin followed by mitogen-activated protein kinase (MAPK) activation. Biochem Biophys Res Commun 2001; 288:356–361. Yano Y, Saito Y, Narumiya S, Sumpio BE. Involvement of rho p21 in cyclic strain-induced tyrosine phosphorylation of focal adhesion kinase (pp125FAK), morphological changes and migration of endothelial cells. Biochem Biophys Res Commun 1996; 224:508–515. Wirtz HR, Dobbs LG. Calcium mobilization and exocytosis after one mechanical stretch of lung epithelial cells. Science 1990; 250:1266–1269. Tschumperlin DJ, Margulies SS. Equibiaxial deformation-induced injury of alveolar epithelial cells in vitro. Am J Physiol 1998; 275:L1173–L1183. Gutierrez JA, Gonzalez RF, Dobbs LG. Mechanical distension modulates pulmonary alveolar epithelial phenotypic expression in vitro. Am J Physiol Lung Cell Mol Physiol 1998; 274:L196–L202. Liu M, Skinner SJ, Xu J, Han RN, Tanswell AK, Post M. Stimulation of fetal rat lung cell proliferation in vitro by mechanical stretch. Am J Physiol 1992; 263:L376–L383. Edwards YS, Sutherland LM, Power JH, Nicholas TE, Murray AW. Cyclic stretch induces both apoptosis and secretion in rat alveolar type II cells. FEBS Lett 1999; 448:127–130. Edwards YS, Sutherland LM, Murray AW. NO protects alveolar type II cells from stretch-induced apoptosis. A novel role for macrophages in the lung. Am J Physiol Lung Cell Mol Physiol 2000; 279:L1236–L1242. Tsuda A, Stringer BK, Mijailovich SM, Rogers RA, Hamada K, Gray ML. Alveolar cell stretching in the presence of ﬁbrous particles induces interleukin-8 responses. Am J Respir Cell Mol Biol 1999; 21:455–462. Yamamoto H, Teramoto H, Uetani K, Igawa K, Shimizu E. Stretch induces a growth factor in alveolar cells via protein kinase. Respir Physiol 2001; 127:105–111. Skinner SJ, Somervell CE, Olson DM. The effects of mechanical stretching on fetal rat lung cell prostacyclin production. Prostaglandins 1992; 43: 413–433. Fisher JL, Margulies SS. Na(þ)-K(þ)-ATPase activity in alveolar epithelial cells increases with cyclic stretch. Am J Physiol Lung Cell Mol Physiol 2002; 283:L737–L746. Tschumperlin DJ, Shively JD, Kikuchi T, Drazen JM. Mechanical stress triggers selective release of ﬁbrotic mediators from bronchial epithelium. Am J Respir Cell Mol Biol 2003; 28:142–149. Tschumperlin DJ, et al. Bronchial epithelial compression regulates MAP kinase signaling and HB-EGF-like growth factor expression. Am J Physiol Lung Cell Mol Physiol 2002; 282:L904–L911. Oudin S, Pugin J. Role of MAP kinase activation in interleukin-8 production by human BEAS-2B bronchial epithelial cells submitted to cyclic stretch. Am J Respir Cell Mol Biol 2002; 27:107–114. Li L-F, et al. Stretch-induced IL-8 depends on c-Jun NH2-terminal and nuclear factor-{kappa}B-inducing kinases. Am J Physiol Lung Cell Mol Physiol 2003; 285:L464–L475.

Cellular and Molecular Basis for Ventilator-Induced Lung Injury

221

77. Chaudhary LR, Avioli LV. Regulation of interleukin-8 gene expression by interleukin-1beta, osteotropic hormones, and protein kinase inhibitors in normal human bone marrow stromal cells. J Biol Chem 1996; 271:16591–16596. 78. Kasahara T, Mukaida N, Yamashita K, Yagisawa H, Akahoshi T, Matsushima K. IL-1 and TNF-alpha induction of IL-8 and monocyte chemotactic and activating factor (MCAF) mRNA expression in a human astrocytoma cell line. Immunology 1991; 74:60–67. 79. Kowalski J, Denhardt DT. Regulation of the mRNA for monocyte-derived neutrophil-activating peptide in differentiating HL60 promyelocytes. Mol Cell Biol 1989; 9:1946–1957. 80. Stoeckle MY. Post-transcriptional regulation of gro alpha, beta, gamma, and IL-8 mRNAs by IL-1 beta. Nucl Acids Res 1991; 19:917–920. 81. Villarete LH, Remick DG. Transcriptional and post-transcriptional regulation of interleukin-8. Am J Pathol 1996; 149:1685–1693. 82. Winzen R, et al. The p38 MAP kinase pathway signals for cytokine-induced mRNA stabilization via MAP kinase-activated protein kinase 2 and an AUrich region-targeted mechanism. EMBO J 1999; 18:4969–4980. 83. Chen H, Wu L, Liu X, Chen Y, Wang B. Effects of laminar shear stress on IL-8 mRNA expression in endothelial cells. Biorheology 2003; 40:53–58. 84. Du W, Mills I, Sumpio BE. Cyclic strain causes heterogeneous induction of transcription factors, AP-1, CRE binding protein and NF-jB, in endothelial cells: species and vascular bed diversity. J Biomech 1995; 28:1485–1491. 85. Khachigian LM, Resnick N, Gimbrone MA Jr, Collins T. Nuclear factorkappa B interacts functionally with the platelet-derived growth factor B-chain shear-stress response element in vascular endothelial cells exposed to ﬂuid shear stress. J Clin Invest 1995; 96:1169–1175. 86. Inoh H, et al. Uni-axial cyclic stretch induces the activation of transcription factor nuclear factor kappaB in human ﬁbroblast cells. Faseb J 2002; 16: 405–407. 87. Kumar A, Boriek AM. Mechanical stress activates the nuclear factor-kappaB pathway in skeletal muscle ﬁbers: a possible role in Duchenne muscular dystrophy. Faseb J 2003; 17:386–396. 88. Chaqour B, Howard PS, Richards CF, Macarak EJ. Mechanical stretch induces platelet-activating factor receptor gene expression through the NFkappaB transcription factor. J Mol Cell Cardiol 1999; 31:1345–1355. 89. Hishikawa JC, Oemar BS, Yang Z, Luscher TF. Pulsatile stretch stimulates superoxide production and activates nuclear factor-kappa B in human coronary smooth muscle. Circ Res 1997; 81:797–803. 90. Kunsch C, Lang RK, Rosen CA, Shannon MF. Synergistic transcriptional activation of the IL-8 gene by NF-kappa B p65 (RelA) and NF-IL-6. J Immunol 1994; 153:153–164. 91. Karin M, Liu Z, Zandi E. AP-1 function and regulation. Curr Opin Cell Biol 1997; 9:240–246. 92. Murphy L, Smith S, Chen R, Fingar D, Blenis J. Molecular interpretation of ERK signal duration by immediate early gene products. Nat Cell Biol 2002; 4:556–564.

222

Pugin and Oudin

93. Holtmann H, et al. Induction of interleukin-8 synthesis integrates effects on transcription and mRNA degradation from at least three different cytokineor stress-activated signal transduction pathways. Mol Cell Biol 1999; 19: 6742–6753. 94. Mukaida N, Mane Y, Matsushima K. Cooperative interaction of nuclear factor-kappa B- and cis-regulatory enhancer binding protein-like factor binding elements in activating the interleukin-8 gene by pro-inﬂammatory cytokines. J Biol Chem 1990; 265:21128–21133. 95. Okada M, et al. Cyclic stretch upregulates production of interleukin-8 and monocyte chemotactic and activating factor/monocyte chemoattractant protein-1 in human endothelial ceils. Arterioscler Thromb Vasc Biol 1998; 18:894–901. 96. Suda T, et al. Pressure-induced expression of monocyte chemoattractant protein-1 through activation of MAP kinase. Kidney Int 2001; 60:1705–1715. 97. Wung BS, Cheng JJ, Chao YJ, Lin J, Shyy YJ, Wang DL. Cyclical strain increases monocyte chemotactic protein-1 secretion in human endothelial cells. Am J Physiol 1996; 270:H1462–H1468. 98. Mitchell JA, Lye SJ. Differential expression of activator protein-1 transcription factors in pregnant rat myometrium. Biol Reprod 2002; 67:240–246. 99. Riser BL, Varani J, Cortes P, Yee J, Dame M, Sharba AK. Cyclic stretching of mesangial cells up-regulates intercellular adhesion molecule-1 and leukocyte adherence: a possible new mechanism for glomerulosclerosis. Am J Pathol 2001; 158:11–17. 100. Kippenberger S, Bernd A, Loitsch S, Muller J, Guschel M, Kaufmann R. Cyclic stretch up-regulates proliferation and heat shock protein 90 expression in human melanocytes. Pigment Cell Res 1999; 12:246–251. 101. Jin S, et al. GADD45-induced cell cycle G2-M arrest associates with altered subcellular distribution of cyclin B1 and is independent of p38 kinase activity. Oncogene 2002; 21:8696–8704. 102. Sheikh MS, Hollander MC, Fornance AJ Jr. Roie of Gadd45 in apoptosis. Biochem Pharmacol 2000; 59:43–45. 103. Takekawa M, Saito H. A family of stress-inducible GADD45-like proteins mediate activation of the stress-responsive MTK1/MEKK4 MAPKKK. Cell 1998; 95:521–530. 104. Tulchinsky E. Fos family members: regulation, structure and role in oncogenic transformation. Histol Histopathol 2000; 15:921–928. 105. Kaminski N, Friedman N. Practical approaches to analyzing results of microarray experiments. Am J Respir Cell Mol Biol 2002; 27:125–132. 106. Lomvardas S, Thanos D. Modifying gene expression programs by altering core promoter chromatin architecture. Cell 2002; 26:261–271. 107. Underwood DC, et al. SB 239063, a p38 MAPK inhibitor, reduces neutrophilia, inﬂammatory cytokines, MMP-9, and ﬁbrosis in lung. Am J Physiol Lung Cell Mol Physiol 2000; 279:L895–L902.

Part II: SUBACUTE VILI

9 The Role of Cytokines During the Pathogenesis of Ventilator-Associated and Ventilator-Induced Lung Injury

JOHN A. BELPERIO and MICHAEL P. KEANE Division of Pulmonary, Critical Care, and Hospitalists, David Geffen School of Medicine at UCLA Los Angeles, California, U.S.A.

ROBERT M. STRIETER Division of Pulmonary, Critical Care, and Hospitalists and Pathology and Pediatrics, Department of Medicine, David Geffen School of Medicine at UCLA Los Angeles, California, U.S.A.

I. Introduction Acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) are clinical syndromes (hypoxemia, bilateral pulmonary inﬁltrates, and noncardiogenic pulmonary edema) having multifactorial etiologies either from direct or indirect injury to the lung (1,2). Histopathologically, there is an initial acute exudative phase involving an alveolar–capillary leak in conjunction with leukocyte extravasation. This is followed by a ﬁbroproliferative phase involving the precipitation of alveolar proteins with hyaline membrane formation, persistent inﬂammation, and proliferation of alveolar epithelia and mesenchymal cells. Finally, there is a ﬁbrotic phase in which inﬂammation results in dysregulated repair with denudation of the basement membrane, excessive matrix deposition, and parenchymal ﬁbrosis (1,2). Clinically, these patients develop an increased physiological dead space, progressive shunt with hypoxemia, decreased compliance, and pulmonary artery vasculopathy resulting in a high minute ventilation requiring the need for mechanical ventilation (1). Management has consisted of aggressive treatment of the inciting cause, vigilant supportive care while 223

224

Belperio et al.

on the ventilator, and the prevention of nosocomial infections. However, the mortality rate from ALI/ARDS is approximately 35% to 65% (1–5). Unfortunately, over the last 30 years there has not been a signiﬁcant change in this mortality rate. However, a recent multicentered randomized controlled trial compared traditional ventilation strategy (tidal volume of 12 mL/kg ideal body weight) to a lung-protective strategy (tidal volume of 6 mL/kg ideal body weight). The study consisted of 861 patients and demonstrated that the mortality rate in the lung-protective group was 22% lower than in the traditional ventilation group (6). This sentinel study has changed the standard of care for ventilator management of patients with ALI/ ARDS. However, this study has raised questions with regard to possible mechanism(s) by which the lung-protective strategy reduces mortality. Surprisingly, only a small percentage of patients with ALI/ARDS actually die of respiratory failure (7,8). One simple explanation is that patients with ALI/ARDS are critically ill, are relatively immunosuppressed, and succumb to overwhelming infection/endotoxemia with multiple organ dysfunction syndrome (MODS) and death. Alternatively, the lung injury of ALI/ARDS leads to leak of inﬂammatory mediators/endotoxin/ microbes predisposing the patient to the following sequence: a continuous systemic inﬂammatory response syndrome (SIRS) ! sepsis ! severe sepsis ! ultimately culminating in MODS and death. Moreover, mechanical ventilation used to support the injured failing lung during ALI/ARDS may serve as an initiator or propagator of the inﬂammatory/ﬁbroproliferative response occurring during the pathogenesis of ALI/ARDS. Moreover, this injured lung, now being mechanically stretched and stressed augments alveolar–capillary permeability allowing for increased translocation of inﬂammatory mediators/endotoxin/microbes to the systemic circulation. Multiple organs are then exposed to these mediators/endotoxin/microbes allowing for known and unknown end organ physiologic derangements ultimately contributing to MODS and death.

II. Mechanical Ventilation of the ALI/ARDS Lung Computed tomography (CT) scans of the chest have demonstrated that patients with ALI/ARDS have regions of nondependent lung that are continuously open to ventilation and regions of dependent consolidated/ atelectatic lung not open to ventilation. There are intermediate regions in which alveoli/airways are collapsed or partially collapsed due to proteinaceous exudates. However, these intermediate regions can still be aerated and recruited/derecruited depending on the ventilation strategy (9–11). Hence, patients with ALI/ARDS have a heterogeneous distribution of relatively normal lung regions, intermediate and fully consolidated/atelectatic lung regions producing an overall small ‘‘baby-lung’’ volume available for

The Role of Cytokines During the Pathogenesis of VALI/VILI

225

ventilation (9–11). Thus, when dealing with ventilator patients with ALI/ ARDS, it is important to limit volume and pressure. Studies have predicted that nonuniformly inﬂated ALI/ARDS lungs will have exuberant forces placed upon them. In the normal lung, inﬂation is simple due to the interdependence of alveoli (i.e., open alveoli support the opening of other alveoli because they share the same walls) (12–16). However, the heterogeneous distribution of consolidated/atelectatic lung during ALI/ARDS changes the normal alveoli distending forces. Alveoli still share the same walls; however, on one side of the wall is a normal alveoli while on the other side of the wall is a consolidated/atelectatic alveoli. This will cause a change in local distending forces in the normal alveolus. The distending force in the normal alveoli will be increased to oppose the collapsing force on its shared wall with the consolidated/atelectatic alveoli. Thus the normal alveoli will become overdistended. In fact, it has been hypothesized that at a transpulmonary pressure (difference between the alveoli and pleural pressure) of 30 cmH2O, the pressure required to expand consolidated/atelectatic regions surrounded by fully expanded normal lung would be greater than 100 cmH2O (17). This exempliﬁes the tremendous shear forces occurring on alveoli and airways during mechanical ventilation of ALI/ARDS lungs (17). Consequently, mechanical ventilation can lead to overdistension of the alveoli by a regional high transpulmonary pressure and cyclic recruitment/derecruitment of alveoli/small airways, both of which can propagate lung injury. III. Mechanotransduction Leads to Lung Injury Injurious mechanical ventilation allows for ‘‘mechanotransduction,’’ the conversion of a mechanical stimulus (i.e., cell deformation due to stress) into cellular biochemical signals causing lung injury (18–23). Another example is the physical forces such as cell stretch/deformation that lead to activation of downstream signals critical to lung growth, development, and surfactant production (18–22). Similarly, in vivo and in vitro studies have found that both the degree and the pattern of mechanical stretch (mechanotransduction ! lung injury) are important in determining cellular responses, giving credence to the hypothesis that different strategies of mechanical ventilation can alter cellular gene expression capable of regulating lung injury (19,23). IV. Cytokines and the Pathogenesis of VALI/VILI Multiple animal models of ventilator-induced lung injury (VILI) have demonstrated that overdistension/stretching or cyclical recruitment/derecruitment of alveoli/small airways leads to signiﬁcant leukocyte sequestration and

226

Belperio et al.

lung injury (4,24,25). Cytokines (cytokines and chemokines) are all thought to have speciﬁc/interacting roles during this process; however the intricate details of these mechanisms remain to be elucidated. Moreover, the speciﬁc mediators that orchestrate the extravasation, activation, and recruitment of leukocytes and presumably nonleukocytes into the lung during ventilatorassociated and ventilator-induced lung injury (VALI/VILI), which may perpetuate ALI/ARDS, have not been fully elucidated. Lastly, the concept of VALI/VILI augmenting the production of cytokines and translocation of cytokines/endotoxin/microbes into the systemic circulation, directly causing or perpetuating MODS during ALI/ARDS, also remains to be determined. This chapter reviews the current literature on the role of cytokines during the pathogenesis of VALI/VILI. A. The Role of Type I and II Cytokines During the Pathogenesis of VALI/VILI

Critical to lung repair during ALI/ARDS is a delicate balance between proinﬂammatory and anti-inﬂammatory cytokines. Changes in this balance by VALI/VILI can inﬂuence lung tissue remodeling during lung injury. The speciﬁc mechanisms by which injurious mechanical ventilation can initiate and propagate ALI/ARDS may involve the interactions between Type 1 and Type 2 inﬂammatory cells/cytokines. Naive CD4þ T cells differentiate into two distinct T cell subsets (Th1 or Th2 cells), which have distinct cytokine proﬁles/ functions. Th1 cells are mainly involved in cell-mediated immunity, whereas Th2 cells are associated with humoral immunity. Similarly, mononuclear phagocytes have also been found to polarize toward a Type I (macrophage 1/M1) or Type II (macrophage 2/M2) response (26,27). The nature of the lung injury is the most important factor dictating whether the inﬂammatory response is directed toward a Type I [i.e., interleukin-1 beta (IL-lb), tumor necrosis factor-alpha (TNF-a), IL-12, IL-23, and interferon-gamma (IFN-c)] ] or Type II [i.e., IL-10, transforming growth factor-beta (TGF-b), IL-4, IL-5, and IL-6] response. In addition, Th1/M1 and Th2/M2 cells can cross-regulate each other through their respective cytokine responses (26–29). Type I cytokines are considered the predominate regulators of innate immunity and early inﬂammation by promoting cytotoxic T cell responses and delayed type hypersensitivity (26,27,30,31). Type II cytokines are considered the predominate regulators of the humoral and ﬁbroproliferative responses, which usually end in ﬁbroplasia. Directing the host response from one type of inﬂammatory response to the other may be helpful during ALI/ ARDS; however, the effect of VALI/VILI on this response remains to be elucidated. While many studies have demonstrated that both Type I (i.e., TNF-a and IL-1b) and Type II (i.e., IL-6 and IL-10) inﬂammatory responses are elevated during VALI/VILI (2,32), it remains to be determined whether VALI/VILI tips the balance toward a Type I or II inﬂammatory response.

The Role of Cytokines During the Pathogenesis of VALI/VILI

227

Figure 1 Mechanical ventilation–associated and –induced lung injury may lead to a Type II inﬂammatory response promoting the ﬁbroplasia of ALI/ARDS. ALI/ ARDS patients are supported by mechanical ventilation. The ARDS Network study has demonstrated lower levels of circulating IL-6 from patients randomized to a noninjurious strategy of mechanical ventilation. This suggests that injurious strategies of mechanical ventilation may tip the balance toward a Type II cytokine response. With a Type II inﬂammatory response these patients may be at higher risk for a deﬁcient innate immunity (i.e., prone to more infections) as well as at an increased risk of ﬁbroplasia, perpetuating ALI/ARDS. Abbreviations: ALI, acute lung injury; ARDS, acute respiratory distress syndrome; IL, interleukin.

With the ARDS Network trial demonstrating that protective ventilator strategies lower mortality and IL-6 levels, this may have given us a hint that protective ventilator strategies may drive the inﬂammatory response toward an overall Type I cytokine response, causing the attenuation of the ﬁbroproliferative response involved in the pathogenesis of ALI/ARDS (Fig. 1). V. The Role of TNF-a During the Pathogenesis of VALI/VILI A. Brief Overview of TNF-a Biology

It is assumed that the primary function of TNF-a is in activating the innate immune response, which is beneﬁcial to the host during active infections.

228

Belperio et al.

However, it must be understood that inappropriate production of TNF-a leads to an inﬂammatory response with unwanted tissue injury (33). TNF-a is synthesized as a membrane-associated protein with biological activity. The membrane-associated protein is enzymatically cleaved by TNF-a–converting enzyme (TACE) to a soluble protein that readily homotrimerizes (33–35). Binding of homotrimeric TNF-a (either soluble or cell associated) to its receptors Type I (TNF-a-RI; p60 or p55; and CD12a) and Type II (TNF-a-RII; p80 or p75; and CD120b) induces oligomerization of the receptors and signal transduction. Both Type I and II receptors are present on all cell types (33–35), Type I and II receptors are both subject to proteolytic cleavage by members of the matrix metalloproteinase family and are shed from the surface of cells in response to inﬂammatory signals (33–35). The shed extracellular domains of both receptors retain their ability to bind TNF-a and may act as natural inhibitors of TNF-a bioactivity or as delivery peptides depending on their relative concentrations (33–35). Lastly, naturally occurring TNF-a inhibitors, consisting of the fulllength four-domain or truncated forms of the extracellular region of TNF-aRI are referred to as TNF-a–binding proteins or soluble TNF-a receptors (TNF-a-R), which have been found in tissue and serum (33–35). Thus, TNF-a/receptor biology is very complex, making it critical to determine TNF-a biological activity during the pathogenesis of disease processes. TNF-a is produced by both immune and nonimmune cells; however, monocytes and tissue macrophages are the primary cell sources for TNF-a synthesis during most inﬂammatory responses. TNF-a is a proximal proinﬂammatory cytokine with numerous effects on multiple inﬂammatory and immunologic responses, including enhanced cytolytic activity of natural killer (NK) cells, upregulation of major histocompatibility complex (MHC) class II antigen and IL-2 receptors, and induction of T cell proliferation (36,37). In addition, it plays an important role in the regulation of the Type I immune response by inducing IL-12 and IL-18, which induce IFN-c expression. Furthermore, TNF-a has been shown to be important in the activation of apoptosis or programmed cell death. All of these biological functions may be germane to ALI/ARDS and VALI/VILI. B. The Role of TNF-a During the Pathogenesis of ALI/ARDS

Elevated levels of TNF-a have been associated with ARDS and importantly were found to be biologically active (38–41). Furthermore, in a recent prospective study, TNF-a levels measured in both the serum and bronchoalveolar lavage ﬂuid (BALF) correlated with patients’ acute physiology and chronic health evaluation (APACHE) II scores and outcome (42). In addition, lower TNF-a levels in both the serum and BALF from patients at risk for ARDS exhibited a good negative predictive value for ARDS development (42). Furthermore, TNF-a has been shown to play a key role in the

The Role of Cytokines During the Pathogenesis of VALI/VILI

229

ﬁbroproliferative response during bleomycin-induced pulmonary ﬁbrosis and hepatic ﬁbrosis in rodent models (43,44). Taken together these studies demonstrate that elevated levels of TNF-a are associated with ﬁbroproliferative diseases such as ALI/ARDS and overall outcomes in critically ill patients with ARDS. However, none of these studies determined the effects of different strategies of mechanical ventilation on TNF-a expression during ALI/ARDS. C. The Role of TNF-a During the Pathogenesis of Human VALI

Ranieri et al. set out to examine the inﬂuence of different strategies of mechanical ventilation on lung-derived and systemically released cytokines from patients with ARDS (2). They performed a two-center randomized control trial that included 37 patients with ARDS (predominantly from sepsis and trauma). Volume–pressure curves and BALF and plasma samples were collected from all patients at entry into the study and 24 to 30 and 36 to 40 hours after randomization. Patients were randomized to either a control group of conventional mechanical ventilation (CMV) (n ¼ 19), in which VT were set to obtain normal values of PaCO2 without exceeding a plateau pressure of 35 cmH2O, and positive end-expiratory pressure (PEEP) was set based on the greatest improvement in SaO2 without causing hemodynamic derangement. The lung-protective strategy (n ¼ 18) consisted of VT and PEEP based on the upper and lower inﬂection points of the volume– pressure curves, respectively. Speciﬁcally, the VT was set to obtain a value of plateau pressure based on the upper inﬂection point regardless of PaCO2, while PEEP was set at 2 to 3 cmH2O higher than the pressure determined at the lower inﬂection point. Patients’ physiological characteristics and TNF-a levels were similar in both groups at the start of the study. Signiﬁcant differences between the control and protective ventilator groups were a higher VT, end inspiratory pressure, lower PEEP, and lower PaCO2 in the control group. The patients in the control group had an increase in both plasma and BALF concentrations of TNF-a and soluble TNF-a receptor (Type I and II) over 36 hours. More importantly, patients in the lung-protective strategy group had a reduction in BALF TNF-a, s TNF-a-R55 and in both plasma and BALF concentration of s TNF-a-R75. Moreover, the BALF concentration of TNF-a and its receptors were all decreased in the protected, as compared to the control group at the end of 36 to 40 hours of ventilation. This human study suggests that mechanical ventilation can induce a cytokine response in the lung, which can lead to a similar response in the systemic circulation. Importantly, changing the strategy of ventilation to one that minimizes overdistension and recruitment/derecruitment of the lung can attenuate an aberrant cytokine response. This study gives us some potential insight into the importance of cytokines release during VALI and may explain why the ARDS Network study demonstrated a reduction in mortality when a protective ventilator strategy was used (i.e., decreased cytokine

230

Belperio et al.

release) (6). In addition, these results may partially explain the association between VALI/VILI perpetuating ALI/ARDS leading to MODS (i.e., translocation of cytokines from the lung to the systemic circulation) (6,45–47). The only signiﬁcant confounding factor found in this study was the fact that the protective ventilator strategy had signiﬁcantly higher PaCO2 levels, which has been shown to inhibit cytokine release and to be protective in some models of lung injury including ischemia-reperfusion injury and VILI (45,48–50). However, this human study demonstrates that mechanical ventilation, the therapeutic modality that is used to support ALI/ARDS, is by itself a potential cause for an increased release of cytokines possibly perpetuating ALI/ARDS. D. The Role of TNF-a During Low-Lung-Volume VALI/VILI

While the ARDS Network and the above studies have demonstrated that overdistension of the lung during mechanical ventilation leads to increased cytokine release and lung injury, others have demonstrated that lung injury can occur with low lung volume mechanical ventilation (51). For instance, ex vivo experiments using surfactant-depleted rat lungs demonstrated that mechanical ventilation with (low VT 7 mL/kg with 0 cmH2O PEEP) can result in marked lung injury, as compared to (low VT 7 mL/kg with a PEEP chosen higher than the lower inﬂection point found on the volume–pressure curve) (51). Thus the proposed mechanism for low lung volume ventilatory injury is cyclic recruitment/derecruitment (atelectrauma) occurring when PEEP is not optimized. However, another possible mechanism by which low lung volume mechanical ventilation can cause injury is simply by regional hypoxia from atelectasis (52). Human studies have demonstrated that ‘‘higher’’ PEEP does not change mortality associated with ALI/ARDS and there is still controversy over recruitment maneuvers (53–55). Indeed, the answer to the exact mechanism(s) of injury during low lung volume mechanical ventilation would probably help us with regard to how aggressive we need to be with recruitment (i.e., if atelectasis is bad then it can be recruited, but care should be taken to not over-recruit—the ‘‘open lung concept’’) (56,57). In this regard, Chu et al. determined whether VILI and its associated cytokine release could be due to causes other than overdistension of the lung and examined the effect of cyclic opening and closing of small airways and alveoli at low lung volumes during mechanical ventilation (52). They randomized ex vivo rat lungs to three strategies of low volume mechanical ventilation, low VT with PEEP to minimize recruitment/derecruitment (VT 7 mL/kg with 5 cmH2O PEEP), cyclic opening and closing with a low VT without PEEP to maximize recruitment/derecruitment (VT 7 mL/kg with 0 cmH2O PEEP) and plain atelectasis (accomplished by sealing the mainstem bronchus) for three hours. They found the cyclic group with 0 cmH2O PEEP

The Role of Cytokines During the Pathogenesis of VALI/VILI

231

had more lung injury that was associated with a marked increase in BALF TNF-a protein levels. In addition, they demonstrated PEEP was protective by inhibiting recruitment/derecruitment and cytokine release (52). These data suggest that mechanical ventilation, even at low lung volumes, can cause injury and is due, in part, to recruitment/derecruitment of small airways and alveoli. Future investigations of lung injury during low lung volume mechanical ventilation in humans and translational studies using in vivo models will be required to substantiate these conclusions. E. The Role of TNF-a During the Pathogenesis of Animal VALI/VILI

Trembaly et al. performed translation studies using an ex vivo rat model of VILI (32). Their model consisted of excised lungs from rats that were pretreated with either intravenous (i.v.) lipopolysaccharide (LPS) or control. The lungs were then placed on a Harvard ventilator and randomized to four different ventilator strategies. This model system separated the in vivo cardiopulmonary instability of high VT mechanical ventilation, allowing them to mimic the effects of severe overdistension that occurs in speciﬁc regions of the ALI/ARDS lung (10,11,17). Furthermore, the use of LPS allowed them to mimic a SIRS/sepsis situation, which presumably can prime the lungs for exaggerated lung injury. Their four strategies of mechanical ventilation ranged from a noninjurious to severe injurious mechanical ventilation. The noninjurious strategy consisted of (low VT 7 cm3/kg with 3 cmH2O PEEP), intermediate protective strategy (VT 15 cm3/kg with 10 cmH2O PEEP), intermediate injurious strategy (VT 15 cm3/kg with 0 cmH2O PEEP), and an injurious strategy (VT 40 cm3/kg with 0 cmH2O PEEP). After two hours of mechanical ventilation the ex vivo lungs were evaluated for injury using lung compliance. Only the intermediate and injurious strategies from both the pretreated LPS and control groups developed signiﬁcant injury. Interestingly, the degree of injury (by compliance) was similar between the LPS and control pretreated groups. However, we speculate more subtle differences may have been demonstrated if lung histopathologic morphometrics, wet to dry ratio, and endothelial and epithelial cell permeability analysis were performed. The above study demonstrated a temporal increase in whole-lung TNF-a mRNA expression, by northern blot semiquantitative analysis that paralleled increasing injurious strategies of mechanical ventilation in the control pretreated groups (32). They conﬁrmed their results by BALF protein analysis of TNF-a. In addition, they demonstrated similar, but more exaggerated results in their LPS pretreated groups (32). Overall, this study conﬁrmed Ranieri and associates’ human study, corroborating that different strategies of mechanical ventilation can lead to changes in cytokine expression from the lung as well as lung injury. Although elegantly performed, Tremblay and

232

Belperio et al.

associates’ use of an ex vivo model system does have disadvantages. One major disadvantage is that the ex vivo lungs are not perfused and this ischemic lung will develop spontaneous cellular damage as well as the spontaneous release of cytokines depending on how long the ischemia period lasts (58). Ricard et al. expanded upon Tremblay’s results by using an in vivo rat model system of VILI (32,59). They placed sedated and paralyzed rats on two strategies of mechanical ventilation, a semi-injurious (low VT 7 mL/kg with 3 cmH2O PEEP) and injurious (high VT 42 mL/kg with 0 cmH2O PEEP) protocol, using a Harvard ventilator. Surprisingly, they did not ﬁnd detectable protein levels of TNF-a in the BALF or plasma from either group. They then turned to the Tremblay ex vivo model system and compared three strategies of mechanical ventilation, plain statically inﬂated lungs at VT 0 mL/kg with 7 cmH2O airway pressure, semi-injurious protocol with low VT 7 mL/kg with 3 cmH2O PEEP, and injurious protocol with VT 42 mL/kg with 0 cmH2O PEEP for two hours with or without DPS pretreatment. They found no signiﬁcant augmentation of TNF-a protein levels in BALF from all three groups without LPS pretreatment. However, they did ﬁnd marked elevations in TNF-a from both the semiinjurious and injurious groups as compared to the static inﬂated lungs when the lungs came from animals pretreated with LPS. This suggests that the effects of injurious mechanical ventilation are more profound when the lung has been primed by a source of systemic inﬂammation. The results from Tremblay and Ricard can together be interpreted in multiple ways, and can demonstrate some inconsistencies with regard to the release of TNF-a during VILI (32,59). One explanation for their inconsistencies may have to do with the ex vivo model. Being a nonperfused, ischemic model can in itself cause the generation of cytokines in a timedependent manner (58). Thus a lack of reproducibility between different laboratories is expected especially if the ‘‘total’’ lung ischemia time varies. Furthermore, both groups rely heavily on BALF protein analysis of TNF-a. This may underestimate total lung TNF-a levels as compared to measuring TNF-a expression and protein levels from whole-lung homogenates during VILI. The injury occurring during VILI is not only in the airspace, a signiﬁcant amount of injury and inﬂammation occurs in the lung interstitium. Moreover, performing a more rigorous kinetic analysis of TNF-a expression over a prolonged period of mechanical ventilation may rectify some of their inconsistencies. For example, many early response cytokines such as TNF-a can be upregulated within seconds to minutes after injury begins, and some studies have even demonstrated that TNF-a expression can be biphasic, having two peaks during lung injury (60). Overall, we feel that TNF-a is released during VILI both in vivo and ex vivo. However, the proper kinetics need to be performed on whole-lung homogenates and BALF to ﬁnd the time point(s) at which it peaks. Waiting for the lung injury to occur may cause us to miss the peak levels of this ‘‘early’’ response

The Role of Cytokines During the Pathogenesis of VALI/VILI

233

cytokine. Moreover, immunolocalization and in situ hybridization studies performed in a kinetic manner will also be important to determine which cells are expressing TNF-a (i.e., macrophages and/or other activated leukocytes, endothelial cells, and/or stromal cells). Continuing with a theme of increased cytokine expression and injurious mechanical ventilation, Takata et al. determined the effects of cmV, as compared to the presumably more protective high frequency oscillatory ventilation (HFOV) on TNF-a expression using surfactant-depleted rabbits (61). Rabbits were randomized to either cmV or HFOV at an FiO2 of 1.0 and a mean airway pressure of 13 cmH2O for one hour. The cmV group had increased TNF-a mRNA expression from BALF cell pellets as compared to the HFOV group. When these same experiments were carried out for a prolonged period of time, the cmV group developed signiﬁcantly more lung injury than the HFOV group. Taken together, these experiments suggest that activation of alveolar cells and their production of proinﬂammatory cytokines may play a pivotal role in the early stage of VALI and that the ventilator mode (i.e., cmV > HFOV) can substantially modulate alveolar cell activation and the degree of lung injury. Chu et al. determined the effect of cytokine release during cyclic overdistension versus static overdistension. The ARDS net, among other studies, clearly demonstrated that lung overdistension is an injurious form of mechanical ventilation (6,24,62–65). Similarly, in saline-lavaged rabbits, cmV had increased cytokine expression and lung injury as compared to HFOV (which limits the delivery of high VT) at a similar mean airway pressure (61,66). This supports the low–birth weight pediatric literature demonstrating that HFOV was a more protective ventilatory strategy than cmV (67). Interestingly, in vitro cell culture studies demonstrated that repetitive stretch causes a greater release of inﬂammatory mediators than continuous stretch alone (68). However, does static overstretch (i.e., constant overdistension) as compared to cyclic overdistension (i.e., swings in overdistension) cause the same amount of cytokine release and lung injury? To answer this question, Chu et al. randomized ex vivo rat lungs to three strategies of mechanical ventilation (52). Strategy I involves overdistension with swings in VT resulting in high peak inspiratory pressure (PIP) of 50 cmH2O with 8 cmH2O PEEP and a respiratory rate (RR) of 25 breaths/min. This strategy of mechanical ventilation leads to a mean airway pressure (PAW) of 31 cmH2O and no visible recruitment/derecruitment. Strategy II involves high constant overdistension with a continuous positive airway pressure (CPAP) of 50 cmH2O (i.e., the PIP of Strategy I). Strategy III involves low constant overdistension with a CPAP of 31 cmH2O (i.e., the PAW of strategy I). They found BALF TNF-a protein levels were signiﬁcantly elevated in the overdistended group with swings in VT as compared to the low constant overdistension (CPAP ¼ 31 cmH2O) group but not the high constant overdistension (CPAP ¼ 50 cmH2O) group. Similarly, the high constant overdistension (CPAP ¼ 50 cmH2O) group had higher BALF protein

234

Belperio et al.

levels of TNF-a, as compared to the low constant overdistension (CPAP ¼ 31 cmH2O) group. However, there was no difference in BALF TNF-a protein levels between the overdistension with swings in VT and the high constant overdistension (CPAP ¼ 50 cmH2O) group (i.e., overdistension with swings ¼ high constant overdistension at CPAP ¼ 50 cmH2O> low constant overdistension at CPAP ¼ 31 cmH2O). This suggests that lung overdistension (be it constant or with swings in VT without recruitment/derecruitment of alveoli) is more important for cytokine release than swings in VT during high-volume mechanical ventilation. F. The Role of TNF-a During the Pathogenesis of VALI/VILI in the Preterm Newborn

VALI/VILI can also occur in preterm newborns and can be much more aggressive than that seen in the adult (69). This presumably is due to the newborn lungs’ small gas volume, deﬁcient surfactant, and underdeveloped lung matrix and the fact that their airspaces contain residual lung ﬂuid (69). VALI/VILI in the preterm newborn has been associated with the chronic inﬂammatory/ﬁbroproliferative disorder, bronchopulmonary dysplasia (BPD)—the preterm equivalent to ALI/ARDS. Furthermore, BPD has been associated with increased granulocytes and proinﬂammatory cytokine levels (70). Naik and associates were interested in TNF-a expression during VALI in preterm newborn lungs. They used an animal model involving preterm lambs treated with recombinant surfactant protein C and ventilated with (VT 9–11 mL/kg with variable amounts of PEEP), as compared to nonventilated preterm lungs. Within hours of ventilation, there was histopathologic evidence of injury that was associated with whole-lung increased expression of TNF-a mRNA. This study demonstrates TNF-a is associated with VALI/VILI in preterm newborn lungs. G. The Role of TNF-a Release from VALI/VILI During the Pathogenesis of MODS

There is now clinical and scientiﬁc literature supporting the concept that VALI/VILI may initiate or at least perpetuate MODS (68,71–74). Murphy and associates were interested in exploring this concept and investigated the effects of different strategies of mechanical ventilation on the translocation of cytokines and endotoxin from the lung to the systemic circulation. They randomized lavaged rabbits to either protective ventilation (VT 5 mL/kg with 10–12.5 cmH2O PEEP) with and without instilled endotoxin as compared to injurious ventilation (VT 12 mL/kg with 0 cmH2O PEEP) with and without instilled endotoxin for three hours. Only in the injurious group with endotoxin did they ﬁnd marked elevations in plasma TNF-a that was associated with increased plasma endotoxin, decreased plasma bicarbonate, and mean arterial pressure. These ﬁndings expand upon the results of

The Role of Cytokines During the Pathogenesis of VALI/VILI

235

the study conducted by von Bethmann et al., which demonstrated in an isolated perfused murine model of VILI that there was an increased release of TNF-a from the lung into the perfusate (75). In addition, these studies conﬁrmed the ﬁndings from others demonstrating that bacteria and cytokines can translocate from the lung into the systemic circulation during lung injury (68,72–74). Furthermore, this study demonstrated that injurious ventilator strategies not only lead to increased translocation of cytokines and endotoxin, but also cause an increase in mortality (71). Lastly, several mechanisms have been proposed to account for improved outcome associated with protective ventilator strategies. This study indicates that protective ventilation strategies, in part, decrease cytokine production and cytokine/endotoxin/infectious translocation into the systemic circulation, theoretically causing a reduction in MODS and mortality (6,68,71–74). H. TNF-a Is Pivotal During the Pathogenesis of VALI/VILI

Imai et al. followed up on many of the above studies by evaluating the physiologic role of TNF-a during the pathogenesis of VALI/VILI (76). They used lavaged rabbits and placed them on cmV (VT 12–15 mL/kg with PEEP ¼ 5 cmH2O, FiO2 of 1.0, PIP ¼ 25 cmH2O, and PAW ¼ 15 cmH2O). They conﬁrmed that indeed cmV caused a marked increase in TNF-a protein levels in BALF and was associated with lung injury. The rabbits were then randomized to different doses of intratracheal anti-TNF-a antibody, as compared to appropriate controls. Pretreatment with intratracheal anti-TNF-a antibody reduced lung leukocyte sequestration, improved oxygenation, and attenuated lung injury in a dose-dependent response. These studies prove ‘‘proof of concept’’ that TNF-a plays an important role in leukocyte sequestration and lung injury during VALI/VILI.

VI. The Role of IL-1b During the Pathogenesis of VALI/VILI IL-1b is an early response cytokine, and like TNF-a is very proximal in the proinﬂammatory cascade (77). The release of mature active IL-1b requires cleavage of pro-IL-1b by the IL-1b–converting enzyme. IL-1b can then affect nearly every cell type and often acts in concert with other cytokines. IL-1b can bind to several receptors, all of which have different binding afﬁnity and biological activity. For instance, IL-1b preferentially binds to the extracellular domain of IL-1RII, yet this receptor does not transduce a signal (acting as a decoy receptor or a sink for IL-1b). IL-1b has low afﬁnity binding to the extracellular domain of IL-1RI; however there is a structural change in the third IgG-like domain of this receptor, allowing for the IL-1R accessory protein (IL-1AcP) to form a complex with the low afﬁnity

236

Belperio et al.

IL-1b/IL-lRI resulting in a high afﬁnity complex (IL-1AcP/ IL-1b/IL-1RI) that ultimately causes signal transduction. IL-1b receptor antagonist (IL-1Ra) binds primarily to IL-1RI, but does not result in a structural change in the receptor, so there is no signal transduction. Thus, IL-1Ra acts as a natural antagonist to IL-1b. IL-1b also can bind to soluble receptors found in the circulation and extracellular ﬂuid (i.e., sIL1RI and sIL-1RII). sIL-1RII can bind and enhance the inhibitory activity of IL-1Ra, and IL-1RI can be shed while maintaining its ability to bind IL-1Ra, hindering the overall IL-1b inhibitory activity of IL-1Ra (78–80). Despite near equal afﬁnity of IL-1Ra and IL-1b for IL-1RI, a 10-fold excess of IL-1Ra is usually required to inhibit IL-1b activity (81,82). The biology of IL-1b/receptor/IL-1Ra is very complex, with several studies demonstrating a decreased ratio of IL-1b:IL-1Ra during the pathogenesis of ﬁbroproliferative diseases such as pulmonary sarcoid, panbronchiolitis, bronchiolitis obliterans syndrome (BOS), idiopathic pulmonary, ﬁbrosis and pulmonary tuberculosis (83–88). Excess IL-1Ra in these diseases presumably enhances a local proﬁbrotic environment through the inhibition of the normal ‘‘ﬁbrolytic activity’’ mediated by IL-1b. The inhibition of IL-1b by IL-1Ra leads to a reduction in the production of PGE2, nitric oxide, and metalloproteinase, resulting in the promotion of excess deposition of extracellular matrix (ECM) (89,90). This has been substantiated in several animal models in which IL-1Ra via attenuation of IL-1b biology leads to augmented ﬁbrosis (91,92). However, there are other in vivo studies that demonstrate that vector overexpression of IL-1b in murine lungs causes an ALI, which eventual leads to pulmonary ﬁbrosis (93). Overall IL-1b/ receptor/IL-1Ra biology is complex, and depending on the type of lung injury may lead to increased or decreased ﬁbroproliferation. Elevated levels of IL-1b and its naturally occurring antagonist, IL-1Ra, have been identiﬁed in BALF from patients with ARDS (38,39,78,94–96). Speciﬁcally, the ratio of IL-1b:IL-1Ra was found to be 10:1 molar in BALF from patients with ARDS as compared to a 1:1 molar in BALF from healthy volunteers (78). These data suggest that IL-1b is contributing to the persistent inﬂammation occurring during the pathogenesis of ARDS (95). In addition, others have demonstrated that lower IL-1b levels in both serum and BALF from a pre-ARDS group exhibited a good negative predictive value for ARDS development (42). Furthermore, this group also demonstrated that elevated levels of IL-1b in both the serum and BALF correlated well with APACHE II scores and survival, and negatively with PaO2/FiO2 (42). Alternatively, other investigators have found IL-1b:IL-1Ra to be 1:10 molar in BALF for ARDS patients (39). Multicenter studies will be required to determine if IL-1b:IL-1Ra can be used to predict which patients are going to develop ALI/ARDS as well as overall patient mortality. The human study by Ranieri et al. involving the comparison of conventional versus protective ventilation in ARDS patients demonstrated an

The Role of Cytokines During the Pathogenesis of VALI/VILI

237

increase of IL-1b and IL-1Ra in BALF over a 36- to 40-hour period in the cmV group (2). Alternatively, patients in the lung-protective mechanical ventilation group had a reduction in BALF IL-1b levels and both plasma and BALF IL-1Ra levels over the same time period. Furthermore, at the end of the study period, IL-1b levels were signiﬁcantly lower in the protective group as compared to cmV and importantly, there was no difference in IL-1Ra levels between the two groups. This study demonstrates an association between cmV and increased IL-1b/IL-1Ra ratio, suggesting IL-1b may be biologically active and contributing to more inﬂammation during VALI in humans. Studies performed by Tremblay et al. using their ex vivo rat model system of VILI demonstrated a temporal increase in IL-1b protein levels in BALF, which paralleled the degree of VILI in both control and LPS pretreatment groups (32). Similarly, Ricard et al. demonstrated, in their ex vivo and in vivo rat models of VILI, that injurious mechanical ventilation leads to increased BALF protein levels of IL-1b (59). Their data like the above human data demonstrate that increased IL-1b levels are associated with VALI/VILI. Copland et al. expanded upon these studies by not only ﬁnding similar associations between IL-1b and VILI but also demonstrated using immunohistochemical techniques that IL-1b predominantly localized to the bronchiolar epithelium and this was conﬁrmed using microdissection gene expression techniques (97). Together, these studies suggest that injurious strategies of mechanical ventilation lead to deformation of bronchial epithelial cells, which release IL-1b causing increased VALI//VILI. Ribeiro et al. were also interested in cytokine biology during VILI and used the ex vivo rat model of VILI to explore the effects of heat shock protein (HSP) on IL-1b biology during the pathogenesis of VILI (98). The heat shock/stress response leads to the release of heat shock/stress proteins (HSP). These HSP, when triggered, either prior to or during inﬂammatory events (i.e., sepsis or lung injury) have been demonstrated to be protective (99–104). Presumably, these HSP prevent injury by binding cytokines or inhibiting their cellular release (99–105). This group determined if HSP could attenuate VILI (98). To test this concept, these authors randomized rats to receive either sham treatment or exposure to heat 18 hours prior to placing ex vivo lungs on the ventilator (VT 40 mL/kg in a warmed, humidiﬁed chamber) for two hours. Animals treated with heat stress demonstrated increased whole-lung protein levels of HSP77 as compared to controls. Mechanical ventilation of sham-operated control lungs produced a marked lung injury (decrease in compliance) associated with an increased IL-1b levels in BALF, as compared to a marked attenuation in both IL-1b levels and lung injury in those animals exposed to heat stress. Thus, the protective effects of heat, through the release of HSP, may be due, in part, to a reduction in IL-1b activity during the pathogenesis of VILI.

238

Belperio et al.

Lastly, Narimanbekov and Rozycki using the saline-lavaged rabbit model of VALI evaluated the physiologic role of IL-1b biology during the pathogenesis of hyperoxia and VALI (106). They randomized rabbits to two different groups, a control group consisting of lavaged rabbits placed on the ventilator with PIP (used to keep PaCO2 35–55 mmHg and an FiO2 of 0.21) and an injurious hyperoxia/VALI group with PIP (approximately double the control group and an FiO2 of 1.0) for eight hours. Rabbits placed on the injurious ventilator strategy had a marked elevation in lung lavage neutrophils, hypoxia, and lung injury scores. However, when the injurious group was treated with aerosolized IL-1Ra, there was a marked reduction in lung neutrophil counts, albumin, and elastase levels with a trend toward signiﬁcantly lower histopathological injury scores. This study demonstrates that IL-1Ra can inhibit some of the biological effects of IL-1b biology by inhibiting lung inﬂammation, which, in part, attenuates lung injury induced by hyperoxia and VALI. However, even though lung injury was decreased with IL-1Ra, there was still, albeit decreased, persistent lung inﬂammation/injury, thus reminding us that other cytokine networks operational in parallel and/or in series may be contributing to hyperoxia and VALI.

VII. The Role of IL-6 During the Pathogenesis of VALI/VILI IL-6 is a cytokine with both proinﬂammatory and anti-inﬂammatory properties and has been associated with a Type 2 inﬂammatory proﬁle and ﬁbrogenesis (107,108). IL-6 is produced by many cells including mononuclear phagocytes, endothelial cells, ﬁbroblasts, and smooth muscle cells in response to various stimuli including those caused by IL-1b, TNF-a, and endotoxin (109–113). On target cells, IL-6 ﬁrst binds to the IL-6 receptor (IL-6R) and this complex associates with the signal-transducing membrane protein gpl30, inducing its dimerization and initiation of signaling (109,114). All cells express gpl30, whereas mononuclear phagocytes, lymphocytes, and hepatocytes mainly express IL-6R. In addition, there is a naturally occurring soluble form of the IL-6R (sIL-6R), which has been found in various body ﬂuids (115–117). However, sIL-6R together with IL-6 not only stimulates cells, but can actually sensitize target cells to respond to IL-6 only in the presence of sIL-6R (117–122). Thus the soluble receptor of the IL-6 is a potential agonist. With regard to ARDS, circulating levels of IL-6 have been shown to be a predictor of the severity of ARDS (107,123,124). Ranieri et al. in their human study of VALI demonstrated that a patient on cmV with ARDS developed increased levels of IL-6 in both plasma and BALF over 36 to 40 hours (2). Yet, ARDS patients placed on a lung-protective ventilator strategy had a reduction in both plasma and BALF levels of IL-6 over

The Role of Cytokines During the Pathogenesis of VALI/VILI

239

the same time period. Lastly, they also found, at the end of the study, that the levels of IL-6 were signiﬁcantly lower in the lung-protective strategy. Similar results with regard to plasma IL-6 were demonstrated in the ARDSnet study (6). These studies imply that VALI is associated with increased levels of IL-6, which may be promoting ALI/ARDS and MODS. Steinberg et al. performed studies involving VALI in pigs by using an intratracheal instillation of Tween and then placing the pigs on a mechanical ventilator (125). The Tween caused a heterogeneous surfactant deactivation leading to regions of lung with unstable alveoli in an attempt to imitate ALI/ ARDS. They found that when they used an injurious strategy of mechanical ventilation (high VT and 0 cmH2O PEEP) the lungs developed signiﬁcant injury (congestion of the alveolar walls, edema, and what seemed to be interstitial leukocyte sequestration), which was associated with increases in both serum and BALF IL-6. Importantly, PEEP not only stabilized the alveoli during this lung injury, but was also associated with lower serum and BALF levels of IL-6. This in vivo animal model supports ﬁndings in humans demonstrating that lung-protective strategies of mechanical ventilation can lead to a reduction in both serum and BALF expression of IL-6 and lung injury (2,6). Results similar to those obtained in the above-mentioned study have been demonstrated by Rich et al. who performed an elegant study using an in vivo rat model of VILI (126). Rats were randomized to nonventilated controls, noninjurious (VT 7 mL/kg with an RR of either 20 or 40 breaths/ min) or injurious (VT 40 mL/kg with an RR of either 20 or 40 breaths/min) mechanical ventilation for one hour. As expected, there was no signiﬁcant lung injury or cytokine release in the noninjurious strategies of ventilation with or without a rapid RR. However, when the injurious mechanical ventilation was performed at a high RR, there was a marked increase in lung injury, inﬂammation, and the expression of IL-6. This study extends the overdistension/large VT data and demonstrates that at high VT, an increase in RR can further augment IL-6 release and lung injury presumably by increasing cyclic recruitment/derecruitment. Moreover, Tremblay and associates using their ex vivo rat model system of VILI conﬁrmed these results and demonstrated increased levels of IL-6 in BALF from injurious as compared to noninjurious strategies of mechanical ventilation (6). Overall, IL-6 may be an important cytokine released during VALI/VILI, which may propagate ALI/ARDS.

VIII. The Role of IFN-c During the Pathogenesis of VALI/VILI IFN-c is a Type 1 pleiotropic cytokine that can be induced by IL-12 and inhibited by IL-10 (127–129). It is expressed predominately by T cells,

240

Belperio et al.

NK cells, and B cells (130). IFN-c plays a complex role in concert with other cytokines in regulating inﬂammation and ﬁbrosis. With regard to ﬁbrosis, IFN-c can inhibit collagen production and proliferation of ﬁbroblasts (131–133). With regard to inﬂammation, it induces MHC class I and II antigens, the Fc-gamma receptor on mononuclear phagocytes, vascular cell adhesion molecule-1 (VCAM-1), and intracellular adhesion molecule-1 (ICAM-1) on ﬁbroblasts and inhibits T cell proliferation (131–140). IFN-c also has a regulatory role on other cytokines including TNF-a, IL1, IL-6, and platelet-derived growth factor (141,142). Interestingly, the administration of IFN-c to bleomycin-treated animals can ameliorate pulmonary ﬁbrosis (143). Furthermore, the beneﬁcial effects of IL-12 administration during bleomycin-induced pulmonary ﬁbrosis were neutralized by the administration of anti-IFN-c antibodies (144). Surprisingly, lung ﬁbrosis was not increased in IFN-c/ mice instilled with bleomycin. Moreover, in a clinical trial, IFN-c was beneﬁcial in a subgroup analysis of patients suffering from mild idiopathic pulmonary ﬁbrosis (145). The potential role of IFN in ALI is complex, and only future studies will determine if IFN therapy during ALI/ARDS will be protective from an inﬂammation/ﬁbroproliferation point of view or protective from an innate immunity point of view. Trembaly and associates, to our knowledge, is the only group that has studied IFN-c during VILI (32). Using their ex vivo model of VILI, they found increased IFN-c protein levels in BALF that paralleled the amount of lung injury occurring during VILI. Future studies of IFN-c both in vivo and in vitro will be required to determine the speciﬁc role of IFN during VALI/VILI and ALI/ARDS.

IX. The Role of IL-10 During the Pathogenesis of VALI/VILI IL-10 is a pleiotropic Type 2 cytokine with immunomodulatory bioactivity including the inhibition of cytotoxicity, MHC class II antigens, and proinﬂammatory cytokine production (88,129,146–159). Clinical studies have demonstrated low circulating and BALF levels of IL-10 in patients with ARDS (107), while other studies have found elevated levels of IL-10 in BALF to be associated with improved survival during ARDS (39,96). In addition, the administration of IL-10 in vivo has been demonstrated to be protective in multiple animal models of ALI (145,160–162). These studies implicate a potentially protective role for IL-10 during ALI/ARDS. Tremblay and associates were one of the ﬁrst groups to evaluate IL-10 during VILI (32). Using their ex vivo model system, they demonstrated increased IL-10 protein levels in BALF that temporally increased with increasing injurious mechanical ventilation in both their control and

The Role of Cytokines During the Pathogenesis of VALI/VILI

241

LPS-pretreated groups. This may indicate that the level of IL-10, while elevated during VILI, may not be high enough to produce enough anti-inﬂammatory stimulus to overcome the many proinﬂammatory cytokines (i.e., TNF-a, IL-1b, IL-6, and ELRþ CXC chemokines) produced during VILI. Alternatively, IL-10 may be promoting a Type 2 inﬂammatory response promoting ﬁbroproliferation during VILI. Future studies (i.e., overexpression and neutralization of IL-10) will need to be performed to indicate the speciﬁc role of IL-10 during the pathogenesis of VALI/VILI.

X. The Role of TGF-b During the Pathogenesis of VALI/VILI While TGF-b has a strong history of immunosuppressive activity, it is the most potent inducer of collagen synthesis, ﬁbroblast proliferation, and ﬁbroblast chemotaxis (163,164). In addition, TGF-b can regulate other biological activity including cell death and growth and is critically involved in wound repair (107,165). Human studies have demonstrated increased expression of TGF-b from transbronchial biopsies and in the epithelial lining ﬂuid from patients with BOS (166,167). Animal studies of BOS found TGF-b to be localized to inﬁltrating mononuclear cells and ﬁbrotic tissue while the inhibition of TGF attenuated the ﬁbroplasia involved in BOS (168). In addition, the inhibition of TGF protected animals from bleomycin- and hemorrhage-induced ALI, while vector overexpression of TGF in murine lungs caused an ALI with pulmonary ﬁbrosis (169–171). These studies implicate TGF as proﬁbrotic cytokines during the pathogenesis of certain types of lung injury. Imanka et al. evaluated the role of TGF during the pathogenesis of VILI (172). They used an in vivo rat model of VILI in which rats were randomized to either an injurious high pressure (PIP ¼ 45 cmH2O of peak with 0 cmH2O PEEP) or noninjurious low pressure (PIP ¼ 7 cmH2O with 0 PEEP) strategy of mechanical ventilation for 40 minutes. The injurious strategy of mechanical ventilation leads to an increase in lung injury (i.e., histologic ﬁndings of inﬁltrating neutrophils, destructive change of the alveolar wall, and deposition of matrix) and hypoxemia that was associated with increased whole-lung homogenate expression of TGF-b mRNA (173). Furthermore, others have shown that A549 cells, when cultured on a silicoelastic membrane and subjected to cyclic stretch, induced activation of protein kinase C and the release of TGF-b, supporting the notion that TGF is released from alveolar epithelial cells during deformation and may contribute to the chronic ﬁbroproliferative response associated with VALI/VILI and ALI/ARDS. Future studies will be required to determine which cells in vivo are expressing TGF and whether neutralization of TGF will attenuate VALI/VILI.

242

Belperio et al. XI. The Role of Chemokines and Chemokine Receptors During the Pathogenesis of VALI/VILI

The hallmarks of both ALI/ARDS and VALI/VILI are increased endothelial and epithelial cell permeability, leakage of proteinaceous exudates, and leukocyte and presumably nonleukocyte extravasation and recruitment, ultimately resulting in a robust ﬁbroproliferative process. The ability to maintain leukocyte and presumably nonleukocyte recruitment throughout this process is pivotal in the transition from the acute exudative phase of ALI/ARDS to the ﬁbroproliferative late stages of ALI/ARDS. This persistent elicitation of leukocytes and nonleukocytes requires intercellular communication between inﬁltrating leukocytes, endothelium, parenchymal cells, and components of the ECM. These events are mediated via the generation of adhesion molecules, cytokines, and chemokines. The chemokines, by virtue of their speciﬁc cell-surface receptor expression, can selectively mediate the local recruitment/activation of distinct leukocytes/cells allowing for migration across the endothelium and beyond the vascular compartment along established chemotactic gradients. The chemokine superfamily is divided into four subfamilies (C, CC, CXC, and CX3C) based on the presence of a conserved cysteine residue at the NH2-terminus (174–176). CXC chemokines depending on the presence or absence of the sequence glutamic acid–leucine–arginine (ELR) near the NH2-terminus can be neutrophil chemoattractants with angiogenic properties or chemoattractants of lymphocytes with angiostatic properties (177–181). CC chemokines predominantly recruit mononuclear cells (174,182). The C subfamily consists of lymphotactin-a/XCL1 and lymphotactin-b/XCL2, which attract lymphocytes, while Fractalkine/CX3CL1 is the only member of the CX3C subfamily, and its domain sits on a mucin stalk allowing for cellular adhesion (183–186). All chemokine action is mediated through seven-transmembranespanning G protein–coupled receptors (GPCRs) (187). These heterotrimeric G proteins are composed of a (deﬁnes the identity of the protein), b, and c subunits. The chemokine receptors generally undergo internalization and phosphorylation following ligand binding. Interaction of a ligand with its receptor leads to exchange of guanosine triphosphate (GTP) for guanosine diphosphate (GDP) and the dissociation of the a subunit from the bc subunit. The dissociated Ga and Gbc can activate downstream signal transduction events (188,189). A. The Role of ELR1 CXC Chemokines During the Pathogenesis of VALI/VILI

CXC chemokines consist of several members that have been shown to mediate leukocyte chemotaxis and regulate angiogenesis (176–181,185,186,190–196).

The Role of Cytokines During the Pathogenesis of VALI/VILI

243

The CXC chemokines can be further subclassiﬁed into ‘‘Glu–Leu–Arg’’ (ELR)þ and interferon-inducible ELR CXC chemokines, based on the presence or absence of tri–amino acid motif (ELR) of the NH2-terminus before the ﬁrst cysteine in the primary structure of these cytokines. This classiﬁcation correlates with functional differences, for instance interferon-inducible ELR CXC chemokines bind to CXCR3, are potent chemoattractants for lymphocytes and NK cells, and are angiostatic. ELRþ CXC chemokines bind to CXCR2 (note: only two ELRþ CXC chemokines bind to CXCR1, which is not present in either mice or rats), have potent chemotactic effects on neutrophils, and exhibit potent angiogenic activity (190–196). All ELRþ CXC chemokines bind to the seven-transmembrane GPCR; CXCR2 (found in both rodents and humans) while only CXC chemokine ligand 8 (CXCL8) and CXCL6 bind CXCR1 (not found in mice and rats) (188,197). Both receptors possess high ( > 80%) sequence homology at the amino acid level, except in their NH2-terminal portions where binding speciﬁcity is important. When ELRþ CXC chemokines bind to their receptor they activate pertussis toxin–sensitive and receptor-coupled G proteins, particularly Gai proteins (198). G proteins dissociate into Ga- and Gbcsubunits on conversion to the GTP-bound form. Gbc recruits and activates phosphatidylinositol 3-kinase-c, which in turn generates phosphatidylinositol 3,4,5-triphosphate (PIP3) (198). PIP3 activates protein kinase B (Akt) as well as GTPases, resulting in directed cell migration. These receptors have been found on both neutrophils and endothelial cells (188,199). However, studies in both human and rodent endothelial cells have found that CXCR2, not CXCR1, dictates the angiogenic activity of ELRþ CXC chemokines (199). Although the exact mechanism of ALI/ARDS and VALI/VILI is not fully understood, neutrophils are strongly implicated as having a causative role (200,201). In addition, neutrophil depletion has been shown to be protective in animal models of ALI (202). Additionally, BALF from patients with ARDS contained elevated concentrations of CXCL8, CXCL5, and CXCL1 and these ELRþ CXC chemokines were biologically active and correlated positively with increased BALF neutrophilia (95,203–208). Furthermore, elevated CXCL8 levels correlated with the development of ARDS in an at-risk patient group and CXCL8 levels declined with time in patients who survived ARDS (209,210). Moreover, animal models of ALI/ARDS have demonstrated elevated lung levels of multiple ELRþ CXC chemokines in association with lung neutrophilia and injury. When the interaction of CXCR2/CXCR2 ligands was abrogated, the recruitment of lung neutrophils and lung injury was attenuated (202,211–214). Angiogenesis is thought to support the persistent inﬂammation and ﬁbroplasia involved in ALI/ARDS (215). Lung biopsies from patients with ARDS have been shown to have signiﬁcant neovascularization and vascular remodeling. BALF from patients with ARDS has been show to have

244

Belperio et al.

signiﬁcant angiogenic activity (215). Importantly, elevated levels of CXCL1, CXCL5, and CXCL8 in BALF from patients with ARDS were demonstrated to be the predominant angiogenic molecules (215). Overall, these studies suggest that ELRþ CXC chemokines play a critical role in neutrophil recruitment and angiogenesis during the pathogenesis of ALI/ARDS. Kotani et al. were interested in mechanotransduction forces causing the production of CXCL8 during the pathogenesis of VALI/VILI (216). They used a rabbit model of mechanical ventilation and randomized rabbits to either noninjurious (small VT 8 mL/kg by maintaining PIP ¼ 8, 0 cmH2O PEEP with an RR of 34 breaths/min) or injurious (high VT 20 mL/kg by maintain PIP ¼ 20, 0 cmH2O PEEP with an RR of 13 breaths/min) strategies of mechanical ventilation for four hours. The rabbits in the injurious strategies had more lung injury as demonstrated by lower PaO2, higher PaCO2, and more BALF neutrophils. Associated with these lung changes were signiﬁcant elevations in BALF protein levels of CXCL8, which by immunohistochemistry were found to be localized to the alveolar epithelium and macrophages. Likewise, Wilson et al., using a murine model of hyperoxia and VALI, randomized mice to two types of ventilator strategies, an injurious (high VT, 0 cmH2O PEEP with an RR of 90 breaths/min, FiO2 ¼ 1.0, and 5% CO2) and a noninjurious (low VT, 2.5 cmH2O PEEP with an RR of 120 breaths/min, and FiO2 ¼ 1.0) strategy of mechanical ventilation for approximately 2.5 hours (217). The injurious group demonstrated significant lung injury including hyaline membranes lining alveoli. Associated with this lung injury were marked elevations of CXCL8 protein in BALF. Others have found similar results using the rat ex vivo model of VILI (32,59). Lastly, Ricard et al. using their in vivo rat model of VILI found an increased level of CXCL2/3 in BALF from both their protective and nonprotective strategies of mechanical ventilation as compared to controls (59). These studies suggest an association between increased levels of ELRþ CXC chemokines and VILI and suggest that CXCL2/3 may be a very sensitive marker of in vivo VILI. Altemeier et al. were interested in the in vivo effect of i.v. LPS on VILI. They found that rabbits, when pretreated with i.v. LPS and then placed on noninjurious forms of mechanical ventilation, would actually develop increased levels of CXCL1 and CXCL8 in BALF as well as VALI (218). This became very important in light of the data suggesting that sepsis is the most common risk factor for the development of ALI (219). However, many patients with sepsis do not develop clinically overt lung injury. Likewise, many animal models of bacterial sepsis do not develop ALI/ARDS without a second local lung insult (220,221). Taken together these studies suggest that the risk of developing ALI/ARDS during sepsis is increased by the presence of a cofactor such as mechanical ventilation. To further support the role of CXCR2 ligands in mediating VILI, experiments were carried out to prove ‘‘proof of concept’’ that CXCR2

The Role of Cytokines During the Pathogenesis of VALI/VILI

245

and CXCR2 ligand interaction contributes to the pathogenesis of VILI. Mice were placed on a high and low peak pressure/stretch protocol as compared to normal controls (214). The pattern of lung injury (by histopathology, vascular permeability, and neutrophil sequestration) was as follows: high peak pressure/stretch protocol low peak pressure/stretch protocol > normal lungs; this is consistent with other animal models of VALI/VILI (24,25,65,172,222–224). Both CXCL1 and CXCL2/3 (potent chemoattractants of neutrophils) expression were signiﬁcantly greater in the high peak pressure/stretch group, as compared to low peak pressure/stretch or normal nonventilated groups. In addition, the low peak pressure/stretch group had increased expression of CXCL1 and CXCL2/3, as compared to virtually no levels found in the lungs from normal nonventilated mice. This is in agreement with the results of the elegant in vitro studies that demonstrated increased CXCL8 expression from either alveolar macrophages, epithelial cells, or the combination of multiple pulmonary cell types cultured on an ‘‘artiﬁcial plastic lung’’ with continuous cyclic pressure-stretching (105,225,226). Together, these studies suggest that high peak pressure/stretch mechanical ventilation can induce the expression of ELRþ CXC chemokines by deforming/ injuring speciﬁc cell types (105,225,226). NF-jB is a transcription factor that can modulate the expression of cytokines and chemokines during cellular stress and has been implicated in multiple inﬂammatory injuries to the lung (227–232). There was more phosphorylation of IkBa and degradation of IjBa protein in the high peak pressure/stretch group than the low peak pressure/stretch group correlating with NF-jB activation and transcription of CXC chemokines (233,234). Similarly, others have found that NF-jB is upregulated in response to stretch in both in vitro and ex vivo lung preparations and most likely this event is pivotal in chemokine activation during VALI/VILI (235,236). Furthermore, in this murine study, the expression of CXCR2 in the lung and on leukocytes paralleled the production of both CXCL1 and CXCL2/3 ligands and neutrophil sequestration during VILI (214). Similarly, other studies of inﬂammatory diseases such as ALI from hyperoxia in newborn rats, immune arthritis, psoriasis, and pneumonia have demonstrated the importance of CXCR2 expression and its role in neutrophil recruitment during the pathogenesis of these diseases (237–240). Collectively, these studies demonstrate that augmented levels of ELRþ CXC chemokines are important in the recruitment of cells expressing CXCR2 during the pathogenesis of inﬂammatory diseases. Moreover, increased cell-surface expression of CXCR2 was found on nonleukocyte cell populations in the high peak pressure/stretch group (214). This implies that mechanical stimulation of nonleukocytes is involved in upregulating chemokine receptor expression. This is supported by other studies ﬁnding increased expression of CXCR2 on nonleukocytes such as ﬁbroblasts and epithelial and endothelial cells

246

Belperio et al.

(199,241,242), signifying a role for ligand/CXCR2 interactions that goes beyond neutrophil recruitment in mediating VALI/VILI. Mice placed on the high peak pressure/stretch protocol and treated with speciﬁc antibodies to CXCL1, CXCL2/3, or CXCL1 þ CXCL2/3 demonstrated signiﬁcant reductions in VILI scores. Interestingly, neutralization of ligands CXCL1, CXCL2/3, or the combination of CXCL1 and CXCL2/3 gave similar reductions of VILI. One explanation for this is the potential of homologous desensitization of the receptor, whereby neutralization of CXCL1 may overexpose the receptor to CXCL2/3 and vice versa, thereby resulting in desensitization of the receptor as is seen in chemotaxis assays at high concentration of chemokine ligands (243,244). Alternatively, this may also reﬂect the presence of other ELRþ CXC chemokines (i.e., CXCR5 and CXCL6) in the lung during VILI. Murine CXCR2 is the shared receptor not only for CXCL1 and CXCL2/3 but also CXCR5 and CXCL6. To evaluate the encompassing role of all ELRþ CXC chemokine ligand interactions with CXCR2, mice placed on the high peak pressure/stretch protocol were treated with speciﬁc antimCXCR2 antibody. The mice placed on the high peak pressure/stretch protocol treated with speciﬁc anti-mCXCR2 antibodies demonstrated signiﬁcant reductions in neutrophil sequestration, microvascular permeability, and histological injury. In addition, these results were conﬁrmed using CXCR2/ mice (214). Furthermore, the reductions in VILI scores from CXCR2/ mice were signiﬁcantly decreased as compared to the anti-CXCL1, antiCXCL2/3, or anti-CXCL1 þ anti-CXCL2/3 groups. Therefore inhibiting the receptor interactions with all ELRþ CXC chemokine ligands has a greater effect on attenuating the pathogenesis of VILI (214). These ﬁndings corroborate the results of Kawano et al. who found that when neutrophils were depleted by nitrogen mustard, there was a marked reduction in hyaline membrane formation, neutrophil inﬁltration, and only minimal presence of patchy lung necrosis during VALI (223). Together, these studies demonstrate the importance of neutrophils during the genesis of VILI. Importantly, these studies demonstrate a mechanism by which the neutrophils are recruited to the lungs via the important interaction of CXCR2 and CXCR2 ligands—a critical step for the pathogenesis of VILI. The ﬁndings of this study support the contention that CXCR2 and CXCR2 ligands, in part, mediate VILI, which may perpetuate the pathogenesis of ALI/ARDS.

XII. The Role of CC Chemokines During the Pathogenesis of VALI/VILI The CC chemokine family is predominantly responsible for the recruitment of mononuclear cells including lymphocytes, NK cells, mononuclear

The Role of Cytokines During the Pathogenesis of VALI/VILI

247

phagocytes, eosinophils, and basophils. Interestingly, in mice, the CC chemokine, CCL3, has also been shown to recruit neutrophils and mononuclear cells. With regard to ALI, deletion of CCR1, the receptor for CCL3 and CCL5 was found to be protective in a murine model of ALI secondary to acute pancreatitis, while anti-CCR1 antibodies were protective during bleomycin-induced ALI (245,246). Similarly, CCR2, the receptor for CCL2, has been implicated in multiple ﬁbroproliferative diseases of the lung including BOS, ALI from bleomycin and ALI from ﬂuorescein isothiocynate (FITC) instillation (247–249). Lastly CCR4/(CCL17 and CCL22) axis was also found to be critical during the pathogenesis of bleomycin-induced ALI (250). Presently, there are no studies to our knowledge evaluating the role of CC chemokines during the pathogenesis of VALI/ VILI. However, future studies of this ‘‘CC chemokine’’ family may lead to a better understanding of the inﬂammatory and ﬁbroproliferative phases occurring during both ALI/ARDS and VALI/VILI. XIII. Conclusion In summary, the above human studies have demonstrated the importance of multiple events including cytokine expression and chemokines/chemokine receptors that are ﬁnely orchestrated during the pathogenesis of VALI/VILI (Table 1). Table 1 Cytokine and Chemokine Receptors and Their Respective Ligands Implicated in Promoting VALI/VILI Cytokine/ chemokine TNF-a IL-1 and IL-1Ra IL-6 IFN-c

IL-10 TGF-beta CXCL1/2/3/5/ 6/8

Cytokine/chemokine receptor TNF-a-RI, TNF-a-RII and soluble TNF-a receptor IL-1RI, IL-1RII, and soluble IL-1 receptors IL-6R and soluble receptors IFN-cRa, a species-speciﬁc accessory factor, (AF-1 or IFNcRb) and sIFN-cRa IL-10R1 and IL-10R2 Type I, Type II, and Type III TGF-beta receptor CXCR1 and CXCR2

VALI/VILI references (2,32,52,59,61,69,75,76, 251–256) (2,32,59,97,98,106,249,255,256) (2,6,125,126) (32)

(32) (172,173) (32,59,214,216,218)

Abbreviations: VALI/VILI, ventilator-associated and ventilator-induced lung injury; TNF-a, tumor necrosis factor-alpha; TNF-a-R, TNF-a receptors; IL, interleukin; IFN-c, interferongamma; CXCL, CXC chemokine ligand.

248

Belperio et al.

Furthermore, translational studies in animal models of VALI/VILI have demonstrated ‘‘proof of principle’’ that precise cytokines and chemokine play a speciﬁc/important role in mediating the leukocyte inﬁltration and ECM deposition that perpetuate ALI/ARDS and propagate MODS (Fig. 2). The future studies of the mechanisms involved in each of these events will lead to the development of new paradigms to understand the pathogenesis of VALI/VILI. Furthermore, they should pave the way for the development of pharmaceutical agents that will target each of these biological events and provide new treatments that will ultimately enhance survival for patients with ALI/ARDS on mechanical ventilation.

Figure 2 Cytokine, endotoxin, and microbial agents released from the ALI/ ARDS lung during injurious mechanical ventilation. Mechanical ventilation– associated and –induced lung injury cause impairment of the integrity of the alveolar–capillary membrane and results in augmented cytokine release, leading to translocation of cytokines/endotoxin/microbial agents from the lung to the circulation contributing to systemic inﬂammation and increasing the risk for multiorgan dysfunction syndrome. Abbreviations: ALI, acute lung injury; ARDS, acute respiratory distress syndrome; IFN, interferon; IL, interleukin; TNF, tumor necrosis factor; TGF, transforming growth factor.

The Role of Cytokines During the Pathogenesis of VALI/VILI

249

Abbreviations BALF CCL2 CCL22 CCL17 CXCL1 CXCL5 CXCL6 CXCL8 IL-1b IL-6 IFF-c MIP-1a/CCL3 MIP-lb/CCL4 PDGF RANTES/CCL5 TGF-b TNF-a

Bronchoalveolar lavage ﬂuid CC chemokine ligand 2 CC chemokine ligand 22 CC chemokine ligand 17 CXC chemokine ligand 1 CXC chemokine ligand 5 CXC chemokine ligand 6 CXC chemokine ligand 8 Interleukin-1 beta Interleukin-6 Interferon-gamma CC chemokine ligand 3 CC chemokine ligand 4 Platelet derived growth factor CC chemokine ligand 5 Transforming growth factor-beta Tumor necrosis factor-alfa

Acknowledgments This work was supported, in part, by grants from the NIH (HL04493 and HL080206–01to JAB; P01HL67665; CA87879, P50CA90388, and HL66027 to RMS). References 1. Ware LB, Matthay MA. The acute respiratory distress syndrome. N Engl J Med 2000; 342(18):1334–1349. 2. Ranieri VM, Suter PM, Tortorella C, et al. Effect of mechanical ventilation on inﬂammatory mediators in patients with acute respiratory distress syndrome: a randomized controlled trial. JAMA 1999; 282(1):54–61. 3. Kollef MH, Schuster DP. The acute respiratory distress syndrome. N Engl J Med 1995; 332(1):27–37. 4. Dreyfuss D, Soler P, Basset G, Saumon G. High inﬂation pressure pulmonary edema. Respective effects of high airway pressure, high tidal volume, and positive end-expiratory pressure. Am Rev Respir Dis 1988; 137(5):1159–1164. 5. Corbridge TC, Wood LD, Crawford GP, Chudoba MJ, Yanos J, Sznajder JI. Adverse effects of large tidal volume and low PEEP in canine acid aspiration. Am Rev Respir Dis 1990; 142(2):311–315. 6. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342(18):1301–1308.

250

Belperio et al.

7. West JB, Mathieu-Costello O. Stress failure of pulmonary capillaries: role in lung and heart disease. Lancet 1992; 340(8822):762–767. 8. Hudson LD. Survival data in patients with acute and chronic lung disease requiring mechanical ventilation. Am Rev Respir Dis 1989; 140(2 Pt 2): S19–S24. 9. Gattinoni L, D’Andrea L, Pelosi P, Vitale G, Pesenti A, Fumagalli R. Regional effects and mechanism of positive end-expiratory pressure in early adult respiratory distress syndrome. JAMA 1993; 269(16):2122–2127. 10. Gattinoni L, Pesenti A, Baglioni S, Vitale G, Rivolta M, Pelosi P. Inﬂammatory pulmonary edema and positive end-expiratory pressure: correlations between imaging and physiologic studies. J Thorac Imaging 1988; 3(3):59–64. 11. Gattinoni L, Presenti A, Torresin A, et al. Adult respiratory distress syndrome proﬁles by computed tomography. J Thorac Imaging 1986; 1(3):25–30. 12. Bachofen H, Schurch S. Alveolar surface forces and lung architecture. Comp Biochem Physiol A Mol Integr Physiol 2001; 129(1):183–193. 13. West JB. Invited review: pulmonary capillary stress failure. J Appl Physiol 2000; 89(6):2483–2489; discussion 2497. 14. Martynowicz MA, Walters BJ, Hubmayr RD. Mechanisms of recruitment in oleic acid-injured lungs. J Appl Physiol 2001; 90(5):1744–1753. 15. Martynowicz MA, Minor TA, Walters BJ, Hubmayr RD. Regional expansion of oleic acid-injured lungs. Am J Respir Crit Care Med 1999; 160(1):250–258. 16. Wyszogrodski I, Kyei-Aboagye K, Taeusch HW Jr, Avery ME. Surfactant inactivation by hyperventilation: conservation by end-expiratory pressure. J Appl Physiol 1975; 38(3):461–466. 17. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28(5):596–608. 18. Liu M, Xu J, Liu J, Kraw ME, Tanswell AK, Post M. Mechanical strainenhanced fetal lung cell proliferation is mediated by phospholipase C and D and protein, kinase C. Am J Physiol 1995; 268(5 Pt 1):L729–L738. 19. Rannels DE. Role of physical forces in compensatory growth of the lung. Am J Physiol 1989; 257(4 Pt 1):L179–L189. 20. Wirtz HR, Dobbs LG. Calcium mobilization and exocytosis after one mechanical stretch of lung epithelial cells. Science 1990; 250(4985):1266–1269. 21. Watson PA. Function follows form: generation of intracellular signals by cell deformation. Faseb J 1991; 5(7):2013–2019. 22. Sachs F. Mechanical transduction by membrane ion channels: a mini review. Mol Cell Biochem 1991; 104(1–2):57–60. 23. Vandenburgh HH. Mechanical forces and their second messengers in stimulating cell growth in vitro. Am J Physiol 1992; 262(3 Pt 2):R350–R355. 24. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110(5):556–565. 25. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157(1):294–323. 26. Mantovani A, Allavena P, Sica A. Tumour-associated macrophages as a prototypic type II polarised phagocyte population: role in tumour progression. Eur J Cancer 2004; 40(11):1660–1667.

The Role of Cytokines During the Pathogenesis of VALI/VILI

251

27. Mantovani A, Schioppa T, Biswas SK, Marchesi F, Allavena P, Sica A. Tumor-associated macrophages and dendritic cells as prototypic type II polarized myeloid populations. Tumori 2003; 89(5):459–468. 28. Zhai Y, Kupiec-Weglinski JW. What is the role of regulatory T cells in transplantation tolerance? Curr Opin Immunol 1999; 11(5):497–503. 29. Mosmann TR, Cherwinski H, Bond MW, Giedlin MA, Cofﬁnan RL. Two types of murine helper T cell clone. I. Deﬁnition according to proﬁles of lymphokine activities and secreted proteins. J Immunol 1986; 136(7): 2348–2357. 30. Hall BM. Cells mediating allograft rejection published erratum appears in Transplantation 1991; 52(5):936; Transplantation 1991; 51(6):1141–1151. 31. Rosenberg AS, Singer A. Cellular basis of skin allograft rejection: an in vivo model of immune-mediated tissue destruction. Annu Rev Immunol 1992; 10:333–358. 32. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99(5):944–952. 33. Schottelius AJ, Moldawer LL, Dinarello CA, Asadullah K, Sterry W, Edwards CK III. Biology of tumor necrosis factor-alpha-implications for psoriasis. Exp Dermatol 2004; 13(4):193–222. 34. Frishman JI, Edwards CK III, Sonnenberg MG, Kohno T, Cohen AM, Dinarello CA. Tumor necrosis factor (TNF)-alpha-induced interleukin-8 in human blood cultures discriminates neutralization by the p55 and p75 TNF soluble receptors. J Infect Dis 2000; 182(6):1722–1730. 35. Moreland LW, McCabe DP, Caldwell JR, et al. Phase I/II trial of recombinant methionyl human tumor necrosis factor binding protein PEGylated dimer in patients with active refractory rheumatoid arthritis. J Rheumatol 2000; 27(3):601–609. 36. Beutler BA. The role of tumor necrosis factor in health and disease. J Rheumatol 1999; 26(suppl 57):16–21. 37. Ostensen ME, Thiele DL, Lipsky PE. Tumor necrosis factor-alpha enhances cytolytic activity of human natural killer cells. J Immunol 1987; 138(12): 4185–4191. 38. Suter PM, Suter S, Girardin E, Roux-Lombard P, Gran GE, Dayer JM. High bronchoalveolar levels of tumor necrosis factor and its inhibitors, interleukin-1, interferon, and elastase, in patients with adult respiratory distress syndrome after trauma, shock, or sepsis. Am Rev Respir Dis 1992; 145(5):1016–1022. 39. Park WY, Goodman RB, Steinberg KP, et al. Cytokine balance in the lungs of patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 164(10 Pt 1):1896–1903. 40. Hyers TM, Tricomi SM, Dettenmeier PA, Fowler AA. Tumor necrosis factor levels in serum and bronchoalveolar lavage ﬂuid of patients with the adult respiratory distress syndrome. Am Rev Respir Dis 1991; 144(2):268–271. 41. Millar AB, Foley NM, Singer M, Johnson NM, Meager A, Rook GA. Tumour necrosis factor in bronchopulmonary secretions of patients with adult respiratory distress syndrome. Lancet 1989; 2(8665):712–714.

252

Belperio et al.

42. Agouridakis P, Kyriakou D, Alexandrakis MG, et al. The predictive role of serum and bronchoalveolar lavage cytokines and adhesion molecules for acute respiratory distress syndrome development and outcome. Respir Res 2002; 3(1):25. 43. Piguet PF, Collart MA, Grau GE, Kapanci Y, Vassalli P. Tumor necrosis factor/cachectin plays a key role in bleomycin-induced pneumopathy and ﬁbrosis. J Exp Med 1989; 170(3):655–663. 44. He Y, Liu W. The preliminary research on the relationship between TNF- and egg-induced granuloma and hepatic ﬁbrosis of Schistosomiasis japonica. J Tongji Med Univ 1996; 16(4):205–208. 45. Broccard AF, Hotchkiss JR, Vannay C, et al. Protective effects of hypercapnic acidosis on ventilator-induced lung injury. Am J Respir Crit Care Med 2001; 164(5):802–806. 46. Broccard AF, Liaudet L, Aubert JD, Schnyder P, Schaller MD. Negative pressure post-tracheal extubation alveolar hemorrhage. Anesth Analg 2001; 92(1):273–275. 47. Montgomery AB, Stager MA, Carrico CJ, Hudson LD. Causes of mortality in patients with the adult respiratory distress syndrome. Am Rev Respir Dis 1985; 132(3):485–489. 48. von Luettichau I, Nelson PJ, Pattison JM, et al. RANTES chemokine expression in diseased and normal human tissues. Cytokine 1996; 8(1):89–98. 49. Laffey JG, Engelberts D, Kavanagh BP. Buffering hypercapnic acidosis worsens acute lung injury. Am J Respir Crit Care Med 2000; 161(1):141–146. 50. Laffey JG, Tanaka M, Engelberts D, et al. Therapeutic hypercapnia reduces pulmonary and systemic injury following in vivo lung reperfusion. Am J Respir Crit Care Med 2000; 162(6):2287–2294. 51. Muscedere JG, Mullen JB, Gan K, Slutsky AS. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med 1994; 149(5):1327–1334. 52. Chu EK, Whitehead T, Shitsky AS. Effects of cyclic opening and closing at low- and high-volume ventilation on bronchoalveolar lavage cytokines. Crit Care Med 2004; 32(1):168–174. 53. Brower RG, Lanken PN, Maclntyre N, et al. Higher versus lower positive endexpiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351(4):327–336. 54. Barbas CS, de Matos GF, Okamoto V, Borges IB, Amato MB, de Carvalho CR. Lung recruitment maneuvers in acute respiratory distress syndrome. Respir Care Clin N Am 2003; 9(4):401–418, vii. 55. Villagra A, Ochagavia A, Vatua S, et al. Recruitment maneuvers during lung protective ventilation in acute respiratory distress syndrome. Am J Respir Crit Care Med 2002; 165(2):165–170. 56. Meredith KS, deLemos RA, Coalson JJ, et al. Role of lung injury in the pathogenesis of hyaline membrane disease in premature baboons. J Appl Physiol 1989; 66(5):2150–2158. 57. Fukuse T, Hirata T, Nakamura T, Kawashima M, Hitomi S, Wada H. Inﬂuence of deﬂated and anaerobic conditions during cold storage on rat lungs. Am J Respir Crit Care Med 1999; 160(2):621–627.

The Role of Cytokines During the Pathogenesis of VALI/VILI

253

58. de Perrot M, Liu M, Waddell TK, Keshavjee S. Ischemia-reperfusion-induced lung injury. Am J Respir Crit Care Med 2003; 167(4):490–511. 59. Ricard JD, Dreyfuss D, Saumon G. Production of inﬂammatory cytokines in ventilator-induced lung injury: a reappraisal. Am J Respir Crit Care Med 2001; 163(5):1176–1180. 60. DeMeester SR, Rolfe MW, Kunkel SL, et al. The bimodal expression of tumor necrosis factor-alpha in association with rat lung reimplantation and allograft rejection. J Immunol 1993; 150(6):2494–2505. 61. Takata M, Abe J, Tanaka H, et al. Intraalveolar expression of tumor necrosis factor-alpha gene during conventional and high-frequency ventilation. Am J Respir Crit Care Med 1997; 156(1):272–279. 62. Hernandez LA, Coker PJ, May S, Thompson AL, Parker JC. Mechanical ventilation increases microvascular permeability in oleic acid-injured lungs. J Appl Physiol 1990; 69(6):2057–2061. 63. Bowton DL, Kong DL. High tidal volume ventilation produces increased lung water in oleic acid-injured rabbit lungs. Crit Care Med 1989; 17(9): 908–911. 64. Woo SW, Hedley-Whyte J. Macrophage accumulation and pulmonary edema due to thoracotomy and lung over inﬂation. J Appl Physiol 1972; 33(1):14–21. 65. Kolobow T, Moretti MP, Fumagalli R, et al. Severe impairment in lung function induced by high peak airway pressure during mechanical ventilation. An experimental study. Am Rev Respir Dis 1987; 135(2):312–315. 66. Imai Y, Kawano T, Miyasaka K, Takata M, Imai T, Okuyama K. Inﬂammatory chemical mediators during conventional ventilation and during high frequency oscillatory ventilation. Am J Respir Crit Care Med 1994; 150(6 Pt 1): 1550–1554. 67. Courtney SE, Durand DJ, Asselin JM, Hudak ML, Aschner JL, Shoemaker CT. High-frequency oscillatory ventilation versus conventional mechanical ventilation for very-low-birth-weight infants. N Engl J Med 2002; 347(9): 643–652. 68. Tschumperlin DJ, Oswaii J, Margulies AS. Deformation-induced injury of alveolar epithelial cells. Effect of frequency, duration, and amplitude. Am J Respir Crit Care Med 2000; 162(2 Pt 1):357–362. 69. Naik AS, Kallapur SG, Bachurski CJ, et al. Effects of ventilation with different positive end-expiratory pressures on cytokine expression in the preterm lamb lung. Am J Respir Crit Care Med 2001; 164(3):494–498. 70. Speer CP. Inﬂammation and bronchopulmonary dysplasia. Semin Neonatol 2003; 8(1):29–38. 71. Murphy DB, Cregg N, Tremblay L, et al. Adverse ventilatory strategy causes pulmonary-to-systemic translocation of endotoxin. Am J Respir Crit Care Med 2000; 162(1):27–33. 72. Johanson WG Jr, Higuchi JH, Woods DE, Gomez P, Coalson JJ. Dissemination of Pseudomonas aeruginosa during lung infection in hamsters. Role of oxygen-induced lung injury. Am Rev Respir Dis 1985; 132(2):358–361. 73. Slutsky AS, Tremblay LN. Multiple system organ failure. Is mechanical ventilation a contributing factor? Am J Respir Crit Care Med 1998; 157(6 Pt 1): 1721–1725.

254

Belperio et al.

74. Nahum A, Hoyt J, Schmitz L, Moody J, Shapiro R, Marini JJ. Effect of mechanical ventilation strategy on dissemination of intratracheally instilled Escherichia coli in dogs. Crit Care Med 1997; 25(10):1733–1743. 75. von Bethmann AN, Brasch F, Nusing R, et al. Hyperventilation induces release of cytokines from perfused mouse lung. Am J Respir Crit Care Med 1998; 157(1):263–272. 76. Imai Y, Kawano T, Iwamoto S, Nakagawa S, Takata M, Miyasaka K. Intratracheal anti-tumor necrosis factor-alpha antibody attenuates ventilatorinduced lung injury in rabbits. J Appl Physiol 1999; 87(2):510–515. 77. Dinarello CA. Interleukin-1. Cytokine Growth Factor Rev 1997; 8(4): 253–265. 78. Goodman RB, Pugin J, Lee JS, Matthay MA. Cytokine-mediated inﬂammation in acute lung injury. Cytokine Growth Factor Rev 2003; 14(6): 523–535. 79. Burger D, Chicheportiche R, Giri JG, Dayer JM. The inhibitory activity of human interleukin-1 receptor antagonist is enhanced by type II interleukin-1 soluble receptor and hindered by type I interleukin-1 soluble receptor. J Clin Invest 1995; 96(1):38–41. 80. Jensen LE, Muzio M, Mantovani A, Whitehead AS. IL-1 signaling cascade in liver cells and the involvement of a soluble form of the IL-1 receptor accessory protein. J Immunol 2000; 164(10):5277–5286. 81. Dinarello CA. Biologic basis for interleukin-1 in disease. Blood 1996; 87(6):2095–2147. 82. Arend WP, Welgus HG, Thompson RC, Eisenberg SP. Biological properties of recombinant human monocyte-derived interleukin 1 receptor antagonist. J Clin Invest 1990; 85(5):1694–1697. 83. Kadota J, Matsubara Y, Ishimatsu Y, et al. Signiﬁcance of IL-lbeta and IL-1 receptor antagonist (IL-1Ra) in bronchoalveolar lavage ﬂuid (BALF) in patients with diffuse panbronchiolitis (DPB). Clin Exp Immunol 1996; 103(3):461–466. 84. Smith DR, Kunkel SL, Standiford TJ, et al. Increased interleukin-1 receptor antagonist in idiopathic pulmonary ﬁbrosis: a compartmental analysis. Am J Respir Crit Care Med 1995; 151:1965–1973. 85. Rolfe MW, Standiford TJ, Kunkel SL, et al. Interleukin-1 receptor antagonist expression in sarcoidosis. Am Rev Respir Dis 1993; 148(5):1378–1384. 86. Firestein GS, Berger AE, Tracey DE, et al. IL-1 receptor antagonist protein production and gene expression in rheumatoid arthritis and osteoarthritis synovium. J Immunol 1992; 149(3):1054–1062. 87. Wilkinson RJ, Patel P, Llewelyn M, et al. Inﬂuence of polymorphism in the genes for the interleukin (IL)-1 receptor antagonist and IL-lbeta on tuberculosis. J Exp Med 1999; 189(12):1863–1874. 88. Belperio JA, DiGiovine B, Keane MP, et al. Interleukin-1 receptor antagonist as a biomarker for bronchiolitis obliterans syndrome in lung transplant recipients. Transplantation 2002; 73(4):591–599. 89. Dayer JM, de Rochemonteix B, Burrus B, Demczuk S, Dinarello CA. Human recombinant interleukin 1 stimulates collagenase and prostaglandin E2 production by human synovial cells. J Clin Invest 1986; 77(2):645–648.

The Role of Cytokines During the Pathogenesis of VALI/VILI

255

90. Wilborn J, Crofford LJ, Burdick MD, Kunkel SL, Strieter RM, Peters-Golden M. Cultured lung ﬁbroblasts isolated from patients with idiopathic pulmonary ﬁbrosis have a diminished capacity to synthesize prostaglandin E2 and to express cyclooxygenase-2. J Clin Invest 1995; 95(4):1861–1868. 91. Naruse K, Shimizu K, Muramatsu M, et al. Long-term inhibition of NO synthesis promotes atherosclerosis in the hypercholesterolemic rabbit thoracic aorta. PGH2 does not contribute to impaired endothelium-dependent relaxation [see comments]. Arterioscler Thromb 1994; 14(5):746–752. 92. Kallio EA, Koskinen PK, Aavik E, Vaali K, Lemstom KB. Role of nitric oxide in experimental obliterative bronchiolitis (chronic rejection) in the rat. J Clin Invest 1997; 100(12):2984–2994. 93. Kolb M, Margetts PJ, Anthony DC, Pitossi F, Gauldie J. Transient expression of IL-lbeta induces acute lung injury and chronic repair leading to pulmonary ﬁbrosis. J Clin Invest 2001; 107(12):1529–1536. 94. Siler TM, Swierkosz JE, Hyers TM, Fowler AA, Webster RO. Immunoreactive interleukin-1 in bronchoalveolar lavage ﬂuid of high-risk patients and patients with the adult respiratory distress syndrome. Exp Lung Res 1989; 15(6):881–894. 95. Goodman RB, Strieter RM, Martin DP, et al. Inﬂammatory cytokines in patients with persistence of the acute respiratory distress syndrome. Am J Respir Crit Care Med 1996; 154(3 Pt 1):602–611. 96. Donnelly SC, Strieter RM, Reid PT, et al. The association between mortality rates and decreased concentrations of interleukin-10 and interleukin-1 receptor antagonist in the lung ﬂuids of patients with the adult respiratory distress syndrome. Ann Intern Med 1996; 125(3):191–196. 97. Copland IB, Kavanagh BP, Engelberts D, McKerlie C, Belik J, Post M. Early changes in lung gene expression due to high tidal volume. Am J Respir Crit Care Med 2003; 168(9):1051–1059. 98. Ribeiro SP, Rhee K, Tremblay L, Veldhuizen R, Lewis JF, Slutsky AS. Heat stress attenuates ventilator-induced lung dysfunction in an ex vivo rat lung model. Am J Respir Crit Care Med 2001; 163(6):1451–1456. 99. Chiumello D, Pristine G, Slutsky AS. Mechanical ventilation affects local and systemic cytokines in an animal model of acute respiratory distress syndrome. Am J Respir Grit Care Med 1999; 160(1):109–116. 100. Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970; 227(259):680–685. 101. Fink L, Seeger W, Ennert L, et al. Real-time quantitative RT-PCR after laserassisted cell picking. Nat Med 1998; 4(11):1329–1333. 102. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using realtime quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 2001; 25(4):402–408. 103. Valafar F. Pattern recognition techniques in microarray data analysis: a survey. Ann N Y Acad Sci 2002; 980:41–64. 104. Tusher VG, Tibshirani R, Chu G. Signiﬁcance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 2001; 98(9):5116–5121. 105. Vlahakis NE, Schroeder MA, Limper AH, Hubmayr RD. Stretch induces cytokine release by alveolar epithelial cells in vitro. Am J Physiol 1999; 277(1 Pt 1):L167–L173.

256

Belperio et al.

106. Narimanbekov IO, Rozycki HJ. Effect of IL-1 blockade on inﬂammatory manifestations of acute ventilator-induced lung injury in a rabbit model. Exp Lung Res 1995; 21(2):239–254. 107. Bhatia M, Moochhala S. Role of inﬂammatory mediators in the pathophysiology of acute respiratory distress syndrome. J Pathol 2004; 202(2):145–156. 108. Elishimoto T. Interleukin-6 and its receptor in autoimmumty. J Autoimmun 1992; 5(suppl A):123–132. 109. Rose-John S. Interleukin-6 biology is coordinated by membrane bound and soluble receptors. Acta Biochim Pol 2003; 50(3):603–611. 110. Aasland D, Oppmann B, Grotzinger J, Rose-John S, Kallen KJ. The upper cytokine-binding module and the Ig-like domain of the leukaemia inhibitory factor (LEF) receptor are sufﬁcient for a functional LIP receptor complex. J Mol Biol 2002; 315(4):637–646. 111. Brunello AG, Weissenberger J, Kappeler A, et al. Astrocytic alterations in interleukin-6/soluble interleukin-6 receptor alpha double-transgenic mice. Am J Pathol 2000; 157(5):1485–1493. 112. Burger R, Neipel F, Fleckenstein B, et al. Human herpesvirus type 8 interleukin-6 homologue is functionally active on human myeloma cells. Blood 1998; 91(6):1858–1863. 113. Chow D, He X, Snow AL, Rose-John S, Garcia KC. Structure of an extracellular gpl30 cytokine receptor signaling complex. Science 2001; 291(5511): 2150–2155. 114. Taga T, Kishimoto T. Gpl30 and the interleukin-6 family of cytokines. Annu Rev Immunol 1997; 15:797–819. 115. Hundhausen C, Misztela D, Berkhout TA, et al. The disintegrin-like metalloproteinase ADAM10 is involved in constitutive cleavage of CX3CL1 (fractalkine) and regulates CX3CL1-mediated cell-cell adhesion. Blood 2003; 102(4):1186–1195. 116. Lust JA, Donovan KA, Kline MP, Greipp PR, Kyle RA, Maihle NJ. Isolation of an mRNA encoding a soluble form of the human inteleukin-6 receptor. Cytokine 1992; 4(2):96–100. 117. Rose-John S, Heinrich PC. Soluble receptors for cytokines and growth factors: generation and biological function. Biochem J 1994; 300(Pt 2): 281–290. 118. Taga T, Hibi M, Hirata Y, et al. Interleukin-6 triggers the association of its receptor with a possible signal transducer, gpl30. Cell 1989; 58(3):573–581. 119. Mullberg J, Althoff K, Jostock T, Rose-John S. The importance of shedding of membrane proteins for cytokine biology. Eur Cytokine Netw 2000; 11(1):27–38. 120. Peters M, Muller AM, Rose-John S. Interleukin-6 and soluble interleukin-6 receptor: direct stimulation of gpl30 and hematopoiesis. Blood 1998; 92(10):3495–3504. 121. Jones SA, Rose-John S. The role of soluble receptors in cytokine biology: the agonistic properties of the sIL-6R/IL-6 complex. Biochim Biophys Acta 2002; 1592(3):251–263. 122. Rose-John S. GP130 stimulation and the maintenance of stem cells. Trends Biotechnol 2002; 20(10):417–419.

The Role of Cytokines During the Pathogenesis of VALI/VILI

257

123. Leser HG, Gross V, Scheibenbogen C, et al. Elevation of serum interleukin-6 concentration precedes acute-phase response and reﬂects severity in acute pancreatitis. Gastroenterology 1991; 101(3):782–785. 124. Remick DG, Bolgos GR, Siddiqui J, Shin J, Nemzek JA. Six at six: interleukin-6 measured 6 h after the initiation of sepsis predicts mortality over 3 days. Shock 2002; 17(6):463–467. 125. Steinberg JM, Schiller HJ, Halter JM, et al. Alveolar instability causes early ventilator-induced lung injury independent of neutrophils. Am J Respir Crit Care Med 2004; 169(1):57–63. 126. Rich PB, Douillet CD, Hurd H, Boucher RC. Effect of ventilatory rate on airway cytokine levels and lung injury. J Surg Res 2003; 113(1):139–145. 127. Farrar MA, Schreiber RD. The molecular cell biology of interferon-gamma and its receptor. Annu Rev Immunol 1993; 11:571–611. 128. Thierfelder WE, van Deursen JM, Yamamoto K, et al. Requirement for Stat4 in interleukin-12-mediated responses of natural killer and T cells. Nature 1996; 382(6587):171–174. 129. Fiorentino DF, Zlotnik A, Mosmann TR, Howard M, O’Garra A. IL-10 inhibits cytokine production by activated macrophages. J Immunol 1991; 147(11):3815–3822. 130. Segel MJ, Izbicki G, Cohen PY, et al. Role of interferon-gamma in the evolution of murine bleomycin lung ﬁbrosis. Am J Physiol Lung Cell Mol Physiol 2003; 285(6): L1255–L1262. 131. Wallace WA, Ramage EA, Lamb D, Howie SE. A type 2 (Th2-like) pattern of immune response predominates in the pulmonary interstitium of patients with cryptogenic ﬁbrosing alveolitis (CFA). Clin Exp Immunol 1995; 101(3): 436–441. 132. Varga J, Olsen A, Herhal J, Constantine G, Rosenbloom J, Jimenez SA. Interferon-gamma reverses the stimulation of collagen but not ﬁbronectin gene expression by transforming growth factor-beta in normal human ﬁbroblasts. Eur J Clin Invest 1990; 20(5):487–493. 133. Stephenson ML, Krane SM, Amento EP, McCroskery PA, Byrne M. Immune interferon inhibits collagen synthesis by rheumatoid synovial cells associated with decreased levels of the procollagen mRNAs. FEBS Lett 1985; 180(1): 43–50. 134. Guyre PM, Morganelli PM, Miller R. Recombinant immune interferon increases immunoglobulin G Fc receptors on cultured human mononuclear phagocytes. J Clin Invest 1983; 72(1):393–397. 135. Spoelstra FM, Postma DS, Hovenga H, Noordhoek JA, Kauffman HF. Interferon-gamma and interleukin-4 differentially regulate ICAM-1 and VCAM-1 expression on human lung ﬁbroblasts. Eur Respir J 1999; 14(4):759–766. 136. Skoskiewicz MJ, Colvin RB, Schneeberger EE, Russell PS. Widespread and selective induction of major histocompatibility complex-determined antigens in vivo by gamma interferon. J Exp Med 1985; 162(5):1645–1664. 137. Hobart M, Ramassar V, Goes N, Urmson J, Halloran PF. The induction of class I and II major histocompatibility complex by allogeneic stimulation is dependent on the transcription factor interferon regulatory factor 1 (ERF-1): observations in IRF-1 knockout mice. Transplantation 1996; 62(12):1895–1901.

258

Belperio et al.

138. Hobart M, Ramassar V, Goes N, Urmson J, Halloran PF. IFN regulatory factor-1 plays a central role in the regulation of the expression of class I and II MHC genes in vivo. J Immunol 1997; 158(9):4260–4269. 139. Hassan AT, Dai Z, Konieczny BT, et al. Regulation of alloantigen-mediated T-cell proliferation by endogenous interferon-gamma: implications for longterm allograft acceptance. Transplantation 1999; 68(1):124–129. 140. Hidalgo LG, Halloran PF. Role of IFN-gamma in allograft rejection. Crit Rev Immunol 2002; 22(4):317–349. 141. Kamijo R, Le J, Shapiro D, et al. Mice that lack the interferon-gamma receptor have profoundly altered responses to infection with Bacillus CalmetteGuerin and subsequent challenge with lipopolysaccharide. J Exp Med 1993; 178(4):1435–1440. 142. Badgett A, Bonner JC, Brody AR. Interferon-gamma modulates lung macrophage production of PDGF-BB and ﬁbroblast growth. J Lipid Mediat Cell Signal 1996; 13(1):89–97. 143. Giri SN, Hyde DM, Maraﬁno BJ Jr. Ameliorating effect of murine interferon gamma on bleomycin-induced lung collagen ﬁbrosis in mice. Biochem Med Metab Biol 1986; 36(2):194–197. 144. Keane MP, Belperio JA, Burdick MD, Strieter RM. IL-12 attenuates bleomycin-induced pulmonary ﬁbrosis. Am J Physiol Lung Cell Mol Physiol 2001; 281(1):L92–L97. 145. Raghu G, Brown KK, Bradford WZ, et al. A placebo-controlled trial of interferon gamma-1b in patients with idiopathic pulmonary ﬁbrosis. N Engl J Med 2004; 350(2):125–133. 146. de Waal Malefyt R, Abrams J, Bennett B, Figdor CG, de Vries JE. Interleukin 10(IL-10) inhibits cytokine synthesis by human monocytes: an autoregulatory role of IL-10 produced by monocytes. J Exp Med 1991; 174(5):1209–1220. 147. de Waal Malefyt R, Haanen J, Spits H, et al. Interleukin 10 (IL-10) and viral IL10 strongly reduce antigen-speciﬁc human T cell proliferation by diminishing the antigen-presenting capacity of monocytes via downregulation of class II major histocompatibility complex expression. J Exp Med 1991; 174(4):915–924. 148. de Waal Malefyt R, Yssel H, Roncarolo MG, Spits H, de Vries JE. Interleukin-10. Curr Opin Immunol 1992; 4(3):314–320. 149. de Waal Malefyt R, Yssel H, de Vries JE. Direct effects of IL-10 on subsets of human CD4þ T cell clones and resting T cells. Speciﬁc inhibition of IL-2 production and proliferation. J Immunol 1993; 150(11):4754–4765. 150. te Velde AA, de Waal Malefyt R, Huijbens RJ, de Vries JE, Figdor CG. IL-10 stimulates monocyte Fc gamma R surface expression and cytotoxic activity. Distinct regulation of antibody-dependent cellular cytotoxicity by IFNgamma, IL-4, and IL-10. J Immunol 1992; 149(12):4048–4052. 151. Fiorentino DF, ZIotnik A, Vieira P, et al. IL-10 acts on the antigen-presenting cell to inhibit cytokine production by Th1 cells. J Immunol 1991; 146(10):3444–3451. 152. Bogdan C, Vodovotz Y, Nathan C. Macrophage deactivation by interleukin 10. J Exp Med 1991; 174(6):1549–1555. 153. Ralph P, Nakoinz I, Sampson-Johannes A, et al. IL-10, T lymphocyte inhibitor of human blood cell production of IL-1 and tumor necrosis factor. J Immunol 1992; 148(3):808–814.

The Role of Cytokines During the Pathogenesis of VALI/VILI

259

154. Hsu DH, de Waal Malefyt R, Fiorentino DF, et al. Expression of interleukin10 activity by Epstein-Barr virus protein BCRF1. Science 1990; 250(4982): 830–832. 155. Hsu DH, Moore KW, Spits H. Differential effects of IL-4 and IL-10 on IL-2induced IFN-gamma synthesis and lymphokine-activated killer activity. Int Immunol 1992; 4(5):563–569. 156. Chen WF, Zlotnik A. IL-10: a novel cytotoxic T cell differentiation factor. J Immunol 1991; 147(2):528–534. 157. MacNeil IA, Suda T, Moore KW, Mosmann TR, Zlotnik A. IL-10, a novel growth cofactor for mature and immature T cells. J Immunol 1990; 145(12): 4167–4173. 158. Hu S, Chao CC, Ehrlich LC, et al. Inhibition of microglial cell RANTES production by IL-10 and TGF-beta. J Leukoc Biol 1999; 65(6):815–821. 159. Bejarano MT, de Waal Malefyt R, Abrams JS, et al. Interleukin 10 inhibits allogeneic proliferative and cytotoxic T cell responses generated in primary mixed lymphocyte cultures. Int Immunol 1992; 4(12):1389–1397. 160. Inoue G. Effect of interleukin-10 (IL-10) on experimental LPS-induced acute lung injury. J Infect Chemother 2000; 6(1):51–60. 161. Kusske AM, Rongione AS, Ashley SW, McFadden DW, Reber HA. Interleukin-10 prevents death in lethal necrotizing pancreatitis in mice. Surgery 1996; 120(2):284–288; discussion 289. 162. Howard M, Muchamuel T, Andrade S, Menon S. Interleukin 10 protects mice from lethal endotoxemia. J Exp Med 1993; 177(4):1205–1208. 163. Penttinen RP, Kobayashi S, Bomstein P. Transforming growth factor beta increases mRNA for matrix proteins both in the presence and in the absence of changes in mRNA stability. Proc Natl Acad Sci USA 1988; 85(4): 1105–1108. 164. Roberts AB, Sporn MB, Assoian RK, et al. Transforming growth factor type beta: rapid induction of ﬁbrosis and angiogenesis in vivo and stimulation of collagen formation in vitro. Proc Natl Acad Sci USA 1986; 83(12):4167–4171. 165. Shull MM, Gnnsby I, Kier AB, et al. Targeted disruption of the mouse transforming growth factor-beta 1 gene results in multifocal inﬂammatory disease. Nature 1992; 359(6397):693–699. 166. Elssner A, Jaumann F, Dobmann S, et al. Elevated levels of interleukin-8 and transforming growth factor-beta in bronchoalveolar lavage ﬂuid from patients with bronchiolitis obliterans syndrome: proinﬂammatory role of bronchial epithelial cells. Munich Lung Transplant Group. Transplantation 2000; 70(2):362–367. 167. El-Gamel A, Sim E, Hasleton P, et al. Transforming growth factor beta (TGFbeta) and obliterative bronchiolitis following pulmonary transplantation. J Heart Lung Transplant 1999; 18(9):828–837. 168. Liu M, Suga M, Maclean AA, St George JA, Souza DW, Keshavjee S. Soluble transforming growth factor-beta type III receptor gene transfection inhibits ﬁbrous airway obliteration in a rat model of Bronchiolitis obliterans. Am J Respir Crit Care Med 2002; 165(3):419–423. 169. Harguindey S. Combined strategy of cancerous cells: fat and protein metabolism Conclusions (3rd of 3 parts). Ann Med Int 1994; 11(7):351–358.

260

Belperio et al.

170. Pittet JF, Grifﬁths MJ, Geiser T, et al. TGF-beta is a critical mediator of acute lung injury. J Clin Invest 2001; 107(12):1537–1544. 171. Sime PJ, Xing Z, Graham FL, Csaky KG, Gauldie J. Adenovector-mediated gene transfer of active transforming growth factor-beta1 induces prolonged severe ﬁbrosis in rat lung. J Clin Invest 1997; 100(4):768–776. 172. Imanaka H, Shimaoka M, Matsuura N, Nishimura M, Ohta N, Kiyono H. Ventilator-induced lung injury is associated with neutrophil inﬁltration, macrophage activation, and TGF-beta 1 mRNA upregulation in rat lungs. Anesth Analg 2001; 92(2):428–436. 173. Yamamoto H, Teramoto H, Uetani K, Igawa K, Shrmizu E. Cyclic stretch upregulates interleukin-8 and transforming growth factor-beta1 production through a protein kinase C-dependent pathway in alveolar epithelial cells. Respirology 2002; 7(2):103–109. 174. Rollins BJ. Chemokines. Blood 1997; 90(3):909–928. 175. Luster AD. Review articles: mechanisms of disease: chemokines—chemotactic cytokines that mediate inﬂammation. N Engl J Med 1998; 338(7):436–445. 176. Strieter RM, Kunkel SL. Chemokines and the lung. In: Crystal R, West J, Weibel E, et al., eds. Lung: Scientiﬁc Foundations. 2nd ed. NY: Raven Press, 1997:155–186. 177. Bleul CC, Fuhlbrigge RC, Casasnovas JM, Aiuti A, Springer TA. A highly efﬁcacious lymphocyte chemoattractant, stromal cell-derived factor 1 (SDF-1) see comments. J Exp Med 1996; 184(3):1101–1109. 178. Loetscher M, Gerber B, Loetscher P, et al. Chemokine receptor speciﬁc for IP 10 and mig: structure, function, and expression in activated T-lymphocytes [see comments]. J Exp Med 1996; 184(3):963–969. 179. Keane MP, Belperio JA, Arenberg DA, et al. IFN-gamma-inducible protein10 attenuates bleomycin-induced pulmonary ﬁbrosis via inhibition of angiogenesis [in process citation]. J Immunol 1999; 163(10):5686–5692. 180. Keane MP, Belperio JA, Moore TA, et al. Neutralization of the CXC chemokine, macrophage inﬂammatory protein-2, attenuates bleomycin-induced pulmonary ﬁbrosis. J Immunol 1999; 162(9):5511–5518. 181. Moore BB, Arenberg DA, Addison CL, Keane MP, Strieter RM. Tumor angiogenesis is regulated by CXC chemokines. J Lab Clin Med 1998; 132(2):97–103. 182. Baggiolini M, Dewald B, Moser B. Human chemokines: an update. Annu Rev Immunol 1997; 15:675–705. 183. Robinson LA, Nataraj C, Thomas DW, et al. A role for fractalkine and its receptor (CX3CR1) in cardiac allograft rejection. J Immunol 2000; 165(11):6067–6072. 184. Fong AM, Robinson LA, Steeber DA, et al. Fractalkine and CX3CR1 mediate a novel mechanism of leukocyte capture, ﬁrm adhesion, and activation under physiologic ﬂow. J Exp Med 1998; 188(8):1413–1419. 185. Pan Y, Lloyd C, Zhou H, et al. Neurotactin, a membrane-anchored chemokine upregulated in brain inﬂammation [published erratum appears in Nature 1997; 389(6646):100]. Nature 1997; 387(6633):611–617. 186. Bazan JF, Bacon KB, Hardiman G, et al. A new class of membrane-bound chemokine with a CX3C motif. Nature 1997; 385(6617):640–644.

The Role of Cytokines During the Pathogenesis of VALI/VILI

261

187. Zlotnik A, Yoshie O. Chemokines: a new classiﬁcation system and their role in immunity. Immunity 2000; 12(2):121–127. 188. Murphy PM, Baggiolini M, Charo IF, et al. International union of pharmacology. XXII. Nomenclature for chemokine receptors. Pharmacol Rev 2000; 52(1):145–176. 189. Segerer S, Nelson PJ, Schlondorff D. Chemokines, chemokine receptors, and renal disease: from basic science to pathophysiologic and therapeutic studies. J Am Soc Nephrol 2000; 11(1):152–176. 190. Ahuja SK, Murphy PM. The CXC chemokines growth-regulated oncogene (GRO) alpha, GRObeta, GROgamma, neutrophil-activating peptide-2, and epithelial cell-derived neutrophil-activating peptide-78 are potent agonists for the type B, but not the type A, human interleukin-8 receptor. J Biol Chem 1996; 271(34):20545–20550. 191. Bowman EP, Campbell JJ, Druey KM, Scheschonka A, Kehrl JH, Butcher EC. Regulation of chemotactic and proadhesive responses to chemoattractant receptors by RGS (regulator of G-protein signaling) family members. J Biol Chem 1998; 273(43):28040–28048. 192. Strieter RM, Polverini PJ, Kunkel SL, et al. The functional role of the ELR motif in CXC chemokine-mediated angiogenesis. J Biol Chem 1995; 270(45):27348–27357. 193. Strieter RM, Belperio JA, Phillips RJ, Keane MP. Chemokines: angiogenesis and metastases in lung cancer. Novartis Found Symp 2004; 256:173–84; discussion 184–188, 259–269. 194. Strieter RM, Belperio JA, Keane MP. CXC chemokines in vascular remodeling related to pulmonary ﬁbrosis. Am J Respir Cell Mol Biol 2003; 29(3 suppl):S67–S69. 195. Strieter RM, Belperio JA, Keane MP. CXC chemokines in angiogenesis related to pulmonary ﬁbrosis. Chest 2002; 122(6 suppl):298S–301S. 196. Strieter RM, Polverini PJ, Arenberg DA, et al. Role of C-X-C chemokines as regulators of angiogenesis in lung cancer. J Leukoc Biol 1995; 57(5): 752–762. 197. Murphy PM, Ahuja SK, Combadiere C, Gao J-L. Chemokine receptors. In: Koch AE, Strieter RM, eds. Chemokines in Disease. Austin, TX: R.G. Landes, Co., 1996:55–80. 198. Servant G, Werner OD, Herzmark P, Balla T, Sedat JW, Bourne HR. Polarization of chemoattractant receptor signaling during neutrophil chemotaxis. Science 2000; 287(5455):1037–1040. 199. Addison CL, Daniel TO, Burdick MD, et al. The CXC chemokine receptor 2, CXCR2, is the putative receptor for ELRþCXC chemokine-induced angiogenic activity. J Immunol 2000; 165(9):5269–5277. 200. Repine JE, Beehler CJ. Neutrophils and adult respiratory distress syndrome: two interlocking perspectives in 1991. Am Rev Respir Dis 1991; 144(2):251–252. 201. Lee WL, Downey GP. Neutrophil activation and acute lung injury. Curr Opin Crit Care 2001; 7(1):1–7. 202. Folkesson HG, Matthay MA, Hebert CA, Broaddus VC. Acid aspirationinduced lung injury in rabbits is mediated by interleukin-8-dependent mechanisms. J Clin Invest 1995; 96(1):107–116.

262

Belperio et al.

203. Aikawa A, McLaughlin PJ, McDicken IW, et al. TNF staining of graft biopsies in renal transplantation. Transplantation 1993; 56(1):231–233. 204. Goodman RB, Strieter RM, Frevert CW, et al. Quantitative comparison of C-X-C chemokines produced by endotoxin-stimulated human alveolar macrophages. Am J Physiol 1998; 275(1 Pt 1):L87–L95. 205. Proost P, De Wolf-Peeters C, Conings R, Opdenakker G, Billiau A, Van Damme J. Identiﬁcation of a novel granulocyte chemotactic protein (GCP2) from human tumor cells. In vitro and in vivo comparison with natural forms of GRO, IP-10, and IL-8. J Immunol 1993; 150(3):1000–1010. 206. Villard J, Dayer-Pastore F, Hamacher J, Aubert JD, Schlegel-Haueter S, Nicod LP. GRO alpha and interleukin-8 in Pneumocystis carinii or bacterial pneumonia and adult respiratory distress syndrome. Am J Respir Crit Care Med 1995; 152(5 Pt 1):1549–1554. 207. Miller EJ, Cohen AB, Nagao S, et al. Elevated levels of NAP-l/interleukin-8 are present in the airspaces of patients with the adult respiratory distress syndrome and are associated with increased mortality. Am Rev Respir Dis 1992; 146(2):427–432. 208. Jorens PG, Van Damme J, De Backer W, et al. Interleukin 8 (IL-8) in the bronchoalveolar lavage ﬂuid from patients with the adult respiratory distress syndrome (ARDS) and patients at risk for ARDS. Cytokine 1992; 4(6): 592–597. 209. Donnelly SC, Strieter RM, Kunkel SL, et al. Interleukin-8 and development of adult respiratory distress syndrome in at-risk patient groups. Lancet 1993; 341(8846):643–647. 210. Kurdowska A, Noble JM, Steinberg KP, Ruzinski JT, Hudson LD, Martin TR. Anti-interleukin 8 autoantibody: interleukin 8 complexes in the acute respiratory distress syndrome. Relationship between the complexes and clinical disease activity. Am J Respir Crit Care Med 2001; 163(2):463–468. 211. Welbourn CR, Goldman G, Paterson IS, Valeri CR, Shepro D, Hechtman HB. Pathophysiology of ischaemia reperfusion injury: central role of the neutrophil. Br J Surg 1991; 78(6):651–655. 212. Yokoi K, Mukaida N, Harada A, Watanabe Y, Matsushima K. Prevention of endotoxemia-induced acute respiratory distress syndrome-like lung injury in rabbits by a monoclonal antibody to IL-8. Lab Invest 1997; 76(3): 375–384. 213. Sue RD, Belperio JA, Burdick MD, et al. CXCR2 is critical to hyperoxiainduced lung injury. J Immunol 2004; 172(6):3860–3868. 214. Belperio JA, Keane MP, Burdick MD, et al. Critical role for CXCR2 and CXCR2 ligands during the pathogenesis of ventilator-induced lung injury. J Clin Invest 2002; 110(11):1703–1716. 215. Keane MP, Donnelly SC, Belperio JA, et al. Imbalance in the expression of CXC chemokines correlates with bronchoalveolar lavage ﬂuid angiogenic activity and procollagen levels in acute respiratory distress syndrome. J Immunol 2002; 169(11):6515–6521. 216. Kotani M, Kotani T, Ishizaka A, et al. Neutrophil depletion attenuates interleukin-8 production in mild-overstretch ventilated normal rabbit lung. Crit Care Med 2004; 32(2):514–519.

The Role of Cytokines During the Pathogenesis of VALI/VILI

263

217. Wilson MR, Choudhury S, Goddard ME, O’Dea KP, Nicholson AG, Takata M. High tidal volume upregulates intrapulmonary cytokines in an in vivo mouse model of ventilator-induced lung injury. J Appl Physiol 2003; 95(4):1385–1393. 218. Altemeier WA, Matute-Bello G, Frevert CW, et al. Mechanical ventilation with moderate tidal volumes synergistically increases lung cytokine response to systemic endotoxin. Am J Physiol Lung Cell Mol Physiol 2004; 287(3): L533–L542. 219. Hudson LD, Milberg JA, Anardi D, Maunder RJ. Clinical risks for development of the acute respiratory distress syndrome. Am J Respir Crit Care Med 1995; 151(2 Pt 1):293–301. 220. Czermak BJ, Breckwoldt M, Ravage ZB, et al. Mechanisms of enhanced lung injury during sepsis. Am J Pathol 1999; 154(4):1057–1065. 221. Matute-Bello G, Frevert CW, Kajikawa O, et al. Septic shock and acute lung injury in rabbits with peritonitis: failure of the neutrophil response to localized infection. Am J Respir Crit Care Med 2001; 163(1):234–243. 222. Dreyfuss D, Basset G, Soler P, Saumon G. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 1985; 132(4):880–884. 223. Kawano T, Mori S, Cybulsky M, et al. Effect of granulocyte depletion in a ventilated surfactant-depleted lung. J Appl Physiol 1987; 62(1):27–33. 224. Frank JA, Gutierrez JA, Jones KD, Allen L, Dobbs L, Matthay MA. Low tidal volume reduces epithelial and endothelial injury in acid-injured rat lungs. Am J Respir Crit Care Med 2002; 165(2):242–249. 225. Pugin J, Dunn I, Jolliet P, et al. Activation of human macrophages by mechanical ventilation in vitro. Am J Physiol 1998; 275(6 Pt 1):L1040–L1050. 226. Mourgeon E, Isowa N, Keshavjee S, Zhang X, Slutsky AS, Liu M. Mechanical stretch stimulates macrophage inﬂammatory protein-2 secretion from fetal rat lung cells. Am J Physiol Lung Cell Mol Physiol 2000; 279(4):L699–L706. 227. Blackwell TS, Christman JW. The role of nuclear factor-kappa B in cytokine gene regulation. Am J Respir Cell Mol Biol 1997; 17(1):3–9. 228. Beg AA, Ruben SM, Scheinman RI, Haskill S, Rosen CA, Baldwin AS Jr. I kappa B interacts with the nuclear localization sequences of the subunits of NF-kappa B: a mechanism for cytoplasmic retention. Genes Dev 1992; 6(10):1899–1913. 229. Beg AA, Finco TS, Nantermet PV, Baldwin AS Jr. Tumor necrosis factor and interleukin-1 lead to phosphorylation and loss of I kappa B alpha: a mechanism for NF-kappa B activation. Mol Cell Biol 1993; 13(6):3301–3310. 230. Baldwin AS Jr. The NF-kappa B and I kappa B proteins: new discoveries and insights. Annu Rev Immunol 1996; 14:649–683. 231. Maniatis T. Catalysis by a multiprotein IkappaB kinase complex. Science 1997; 278(5339):818–819. 232. Rodriguez MS, Michalopoulos I, Arenzana-Seisdedos F, Hay RT. Inducible degradation of I kappa B alpha in vitro and in vivo requires the acidic C-terminal domain of the protein. Mol Cell Biol 1995; 15(5):2413–2419. 233. Shattuck-Brandt RL, Richmond A. Enhanced degradation of I-kappaB alpha contributes to endogenous activation of NF-kappaB in Hs294T melanoma cells. Cancer Res 1997; 57(14):3032–3039.

264

Belperio et al.

234. Devalaraja MN, Wang DZ, Ballard DW, Richmond A. Elevated constitutive IkappaB kinase activity and IkappaB-alpha phosphorylation in Hs294T melanoma cells lead to increased basal MGSA/GRO-alpha transcription. Cancer Res 1999; 59(6):1372–1377. 235. Lentsch AB, Czermak BJ, Bless NM, Van Rooijen N, Ward PA. Essential role of alveolar macrophages in intrapulmonary activation of NF-kappaB. Am J Respir Cell Mol Biol 1999; 20(4):692–698. 236. Held HD, Boettcher S, Hamann L, Uhlig S. Ventilation-induced chemokine and cytokine release is associated with activation of nuclear factor-kappaB and is blocked by steroids. Am J Respir Crit Care Med 2001; 163(3 Pt 1): 711–716. 237. Auten RL, Richardson RM, White JR, Mason SN, Vozzelli MA, Whorton MH. Nonpeptide CXCR2 antagonist prevents neutrophil accumulation in hyperoxia-exposed newborn rats. J Pharmacol Exp Ther 2001; 299(1): 90–95. 238. Schimraer RC, Schrier DJ, Flory CM, et al. Streptococcal cell wall-induced arthritis. Requirements for neutrophils, P-selectin, intercellular adhesion molecule-1, and macrophage-inﬂammatory protein-2. J Immunol 1997; 159(8): 4103–4108. 239. Kulke R, Bornscheuer E, Schluter C, et al. The CXC receptor 2 is overexpressed in psoriatic epidermis. J Invest Dermatol 1998; 110(1):90–94. 240. Mehrad B, Strieter RM, Moore TA, Tsai WC, Lira SA, Standiford TJ. CXC chemokine receptor-2 ligands are necessary components of neutrophilmediated host defense in invasive pulmonary aspergillosis. J Immunol 1999; 163(11):6086–6094. 241. Nirodi CS, Devalaraja R, Nanney LB, et al. Chemokine and chemokine receptor expression in keloid and normal ﬁbroblasts. Wound Repair Regen 2000; 8(5):371–382. 242. Williams EJ, Haque S, Banks C, Johnson P, Sarsﬁeld P, Sheron N. Distribution of the interleukin-8 receptors, CXCR1 and CXCR2, in inﬂamed gut tissue. J Pathol 2000; 192(4):533–539. 243. Keane MP, Belperio JA, Burdick MD, Lynch JP, Fishbein MC, Strieter RM. ENA-78 is an important angiogenic factor in idiopathic pulmonary ﬁbrosis. Am J Respir Crit Care Med 2001; 164(12):2239–2242. 244. Ben-Baruch A, Michiel DF, Oppenheim JJ. Signals and receptors involved in recruitment of inﬂammatory cells. J Biol Chem 1995; 270(20):11,703–11,706. 245. Gerard C, Frossard JL, Bhatia M, et al. Targeted disruption of the betachemokine receptor CCR1 protects against pancreatitis-associated lung injury. J Clin Invest 1997; 100(8):2022–2027. 246. Tokuda A, Itakura M, Onai N, Kimura H, Kuriyama T, Matsushima K. Pivotal role of CCR1-positive leukocytes in bleomycin-induced lung ﬁbrosis in mice. J Immunol 2000; 164(5):2745–2751. 247. Gharaee-Kermani M, McCullumsmith RE, Charo IF, Kunkel SL, Phan SH. CC-chemokine receptor 2 required for bleomycin-induced pulmonary ﬁbrosis. Cytokine 2003; 24(6):266–276. 248. Moore BB, Paine R III, Christensen PJ, et al. Protection from pulmonary ﬁbrosis in the absence of CCR2 signaling. J Immunol 2001; 167(8):4368–4377.

The Role of Cytokines During the Pathogenesis of VALI/VILI

265

249. Belperio JA, Keane MP, Burdick MD, et al. Critical role for the chemokine MCP-1/CCR2 in the pathogenesis of bronchiolitis obliterans syndrome. J Clin Invest 2001; 108(4):547–556. 250. Belperio JA, Dy M, Murray L, et al. The role of the Th2 CC chemokine ligand CCL17 in pulmonary ﬁbrosis. J Immunol 2004; 173(7):4692–4698. 251. Belperio JA, Burdick MD, Keane MP, et al. The role of the CC chemokine, RANTES, in acute lung allograft rejection. J Immunol 2000; 165(1):461–472. 252. Suga M, Maclean AA, Keshavjee S, Fischer S, Moreira JM, Liu M. RANTES plays an important role in the evolution of allograft transplant-induced ﬁbrous airway obliteration. Am J Respir Crit Care Med 2000; 162(5):1940–1948. 253. Boehler A, Bai XH, Liu M, et al. Upregulation of T-helper 1 cytokines and chemokine expression in post-transplant airway obliteration. Am J Respir Crit Care Med 1999; 159(6):1910–1917. 254. Monti G, Magnan A, Fattal M, et al. Intrapulmonary production of RANTES during rejection and CMV pneumonitis after lung transplantation. Transplantation 1996; 61(12):1757–1762. 255. Panoskaltsis-Mortari A, Strieter RM, Hernmanson JR, et al. Induction of monocyte- and T-cell-attracting chemokines in the lung during the generation of idiopathic pneumonia syndrome following allogeneic murine bone marrow transplantation. Blood 2000; 96(3):834–839. 256. Sekine Y, Yasufuku K, Heidler KM, et al. Monocyte chemoattractant protein-1 and RANTES are chemotactic for graft inﬁltrating lymphocytes during acute lung allograft rejection. Am J Respir Cell Mol Biol 2000; 23(6):719–726.

This page intentionally left blank

10 Systemic Effects of Mechanical Ventilation

YUMIKO IMAI

ARTHUR S. SLUTSKY

Institute of Molecular Biotechnology of the Austrian Academy of Sciences (IMBA) Vienna, Austria

Interdepartmental Division of Critical Care Medicine and Division of Respirology, Department of Medicine, University of Toronto, and Department of Critical Care Medicine, St. Michael’s Hospital Toronto, Ontario, Canada

I. Introduction The acute respiratory distress syndrome (ARDS), the most severe form of acute lung injury (ALI), is a devastating clinical syndrome affecting approximately 1,000,000 people worldwide per year. Despite recent advances, the mortality rate is at least 30% (1). Predisposing factors for ARDS are diverse (2,3) and include sepsis, pneumonia, aspiration, trauma, and severe acute respiratory syndrome. No drug has been proven to improve the clinical outcome of ARDS, and therapy is largely supportive with mechanical ventilation (MV). However, MV can cause and/or worsen preexisting lung injury, the so-called ventilator-induced lung injury (VILI). Although the most obvious clinical and laboratory abnormalities of ARDS are related to the lung, death is usually due to the dysfunction of other organs, termed ‘‘multiple organ dysfunction syndrome’’ (MODS) (1). One hypothesis that has recently been advanced to explain this observation is that MV per se may not only be responsible for worsening of the preexisting lung injury but also, by a number of mechanisms including the development of systemic inﬂammatory response, contribute to the development of MODS (Fig. 1) (4). 267

268

Imai and Slutsky

Figure 1 Postulated mechanisms whereby ‘‘volutrauma,’’ ‘‘atelectrauma,’’ and ‘‘biotrauma’’ might contribute to MODS. The potential importance of biotrauma is not only that it can aggravate the ongoing lung injury, but also that it can contribute to the development of MODS, possibly through the release of proinﬂammatory mediators from the lung. Abbreviation: MODS, multiple organ dysfunction syndrome. Source: From Ref. 4.

Supportive evidence for this hypothesis comes from in vitro, ex vivo, and in vivo experimental studies as well as clinical studies. In vitro, by using cell stretch devices, mechanical strain has been shown to cause the release of a number of mediators from a variety of lung cells, including alveolar epithelial cells, endothelial cells, macrophages, ﬁbroblasts, and smooth muscle cells, as highlighted by several excellent reviews (5–13). An increase in the release of inﬂammatory mediators has been observed with the use of injurious ventilatory strategies in both isolated nonperfused rat lungs (14) and isolated perfused mouse lungs ex vivo (15), although the increased cytokine response is not universal in these studies (14,16). In vivo, injurious MV in ALI models has been shown to lead to increases in pulmonary and systemic inﬂammatory cytokines (17). Very recently, studies using an acid aspiration model of ARDS demonstrated that MV with a high tidal volume (VT) enhanced end-organ dysfunction (18), apoptosis (18), and inﬂammation (19). Importantly, clinical trials have demonstrated that protective ventilatory strategies are associated with decreased serum cytokine levels (20,21), decreased levels of organ dysfunction (20,22), and decreased mortality (20,23) in patients with ARDS. The concept of loss of the compartmentalization of local pulmonary inﬂammatory mediators by MV has been proposed to explain the translocation of inﬂammatory mediators from the lung into the systemic circulation, promoting a massive inﬂammatory

Systemic Effects of Mechanical Ventilation

269

response that may contribute to the development of MODS (4). MV with injurious ventilatory strategies can also induce bacterial and endotoxin translocation from the lung to the systemic circulation, which may also promote the development of MODS (24). In addition, it is well known that MV can affect the systemic and regional circulation as well as the oxygen delivery and consumption in critically ill patients (25), which may affect the development of MODS. This review is motivated by the clinical question of how MV might contribute to the development of MODS in patients with ARDS. We wish to highlight some of the most recent and pertinent ﬁndings that have contributed to our understanding of MODS. MODS is often irreversible, with mortality ranging from 60% to 98% (26). To date, there is neither an effective treatment for MODS nor an effective means for preventing its onset. By understanding the mechanisms by which MV might contribute to MODS, this new conceptualization of VILI could lead to a paradigm shift in which therapies aimed at various mediators may be used to mitigate ventilation-induced MODS.

II. Physiological Effects of MV It has been known for decades that MV can have important effects on systemic and regional hemodynamics, as well as global oxygen delivery and consumption, as highlighted in many excellent reviews (25,27,28). The application of positive end-expiratory pressure (PEEP) is one of the key components of the ventilatory management of ARDS. PEEP can cause a decrease in the cardiac output by decreasing venous return and increasing right ventricular afterload (29–31); as a result, PEEP may lead to a decrease in global oxygen delivery. This effect can usually be reversed by volume loading and inotropic agents. PEEP may also induce alterations in regional hemodynamics due to the combined effects of decreases in cardiac output and blood ﬂow redistribution (28). In addition, intra-abdominal pressure may increase in response to diaphragmatic swings, so that the splanchnic organs may be compressed (32). In experimental conditions, it has been reported that PEEP decreases splanchnic blood ﬂow (33,34), although splanchnic O2 consumption can usually be maintained by a compensatory increase in O2 extraction (35). A similar response to PEEP has been observed in patients with septic shock. Trager et al. (36) reported that cardiac output decreases in all patients having sepsis, and the hepatic vein O2 saturation decreased more at PEEP 15 cmH2O than at 10 cmH2O. In contrast, Kiefer et al. (37) demonstrated in patients with ALI that moderate levels of PEEP did not signiﬁcantly alter splanchnic blood or other indices of tissue hypoxia such as gastric mucosal PO2 and the blood lactate to pyruvate ratio, provided

270

Imai and Slutsky

ﬂuid resuscitation was able to maintain the cardiac output. SeemanLodding et al. (38) measured the arterio-venous concentration gradients of tissue-type plasminogen activator (t-PA) and the respective blood ﬂow across the pulmonary, coronary, hepatic, and preportal vascular beds in pigs, after the application of PEEP up to l0 cmH2O. They found that with increasing PEEP levels, the magnitude of the preportal net release of t-PA was markedly enhanced, with a concomitant decrease in liver blood ﬂow, thereby suggesting that clinically used levels of PEEP induce increases in the net release of t-PA within preportal organs. Reduction of Vt is a key component of the protective ventilatory strategies for ALI/ARDS. An increase in cardiac output resulting in increased oxygen delivery during Vt reduction has been observed in several studies of patients with ARDS (39,40). There are two mechanisms that could be responsible for the increased cardiac output with reduced Vt at a ﬁxed level of PEEP and inspiratory time:expiratory time (I:E) ratio. First, the decrease in airway pressure, via a reduction in pleural pressure, will lead to an increase in venous return. In addition, a decrease in the transpulmonary pressure during Vt reduction could decrease the overdistension of the lung, leading to a decrease in the resistance of alveolar microvessels, hence imposing a relatively reduced transient impedance to the right ventricular output. Another mechanism leading to an increase in the cardiac output during VT reduction is the resulting hypercapnia. Hypercapnia can induce an increase in sympathetic activity, which may enhance the cardiac output (41,42). In addition to the central hemodynamic effects, VT reduction may also induce regional hemodynamic alterations such as effects on the gut circulation, although the effect of VT reduction on gut regional blood ﬂow is controversial. Cardenas et al. (43) reported a parallel increase in the cardiac output and mesenteric blood ﬂow after VT reduction in animals. In contrast, despite an increase in cardiac output, Sitbon et al. (44) observed no increase in the gastric mucosal blood ﬂow after the VT reduction, resulting in hypercapnia in ARDS patients. They speculated that the heterogeneity in the individual response of gastric mucosal blood ﬂow during VT reduction resulting in hypercapnia could be due to the opposing direct (i.e., local vasodilatory effect) and indirect (i.e., global sympathetic stimulation) effects of hypercapnia on gut vessels. III. Mechanical Strain–Induced Release of Inﬂammatory Mediators In Vitro Not surprisingly, with the complexity of the lung structure, the variety of cell types, and the variety of mechanical forces to which these cells are exposed, there are a number of mechanisms by which mechanical stimulation may alter cellular responses. There are several excellent reviews of this topic (5–10,45–47). Similar to other organs, the lung has the ability to ‘‘sense’’

Systemic Effects of Mechanical Ventilation

271

stretch. Common mechanosensors include stretch-activated ion channels, integrin receptors, the focal adhesion complex, and growth factor receptors. These mechanosensors subsequently activate the mitogen-activated protein kinases (MAPKs), extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase, and p38, which activate transcription factors such as activated protein 1 and early growth response 1. Using cultured pulmonary epithelial cells (A549 cells), Vlahakis et al. (13) found that interleukin (IL)-8 mRNA increased fourfold after four hours of cyclic strain sufﬁcient to change cell surface area by 30%. Continued strain for up to 48 hours resulted in a nearly 50% increase in the IL-8 secretion when compared with nonstrained controls. Quinn et al. (12) conﬁrmed these ﬁndings and also found that the increase in IL-8 secretion was associated with MAPKs. Recently, it was suggested that IL-8 production is also stimulated by the stretch-induced release of hyaluronan from ﬁbroblasts, a pathway possibly mediated by Janus-activated kinase 2 (48). Pulmonary endothelial cells form a continuous monolayer on the luminal surface of the lung vasculature. During MV, pulmonary endothelial cells are exposed to shear stress, as well as to changes in transluminal pressure during alveolar inﬂation. Cyclic stretch of endothelial cells has been shown to induce the activation of p38 and ERK, and the phosphorylation of myosin light chain and actomyosin (49). Gan et al. (50,51) found that various combinations of shear and intraluminal pressures in human umbilical vascular endothelial cells increased the mRNA and protein expression of vascular endothelial growth factor (VEGF) as well as of c-jun and c-fos. Pugin et al. (11) exposed human alveolar macrophages to cyclic stretch for up to 32 hours. They found that cyclic strain increased the secretion of IL-8 and matrix metalloproteinase-9 (gelatinase B—a type IV collagenase), and interestingly, led to the nuclear translocation of nuclear factor jB (NFjB). Moriyama et al. (52) demonstrated that alveolar macrophages isolated from the lungs after high VT ventilation displayed higher expression of CD14, a lipopolysaccharide (LPS) recognition molecule, compared with the macrophages isolated from animals after small VT ventilation. These studies suggest that the mechanical stretch of alveolar macrophages may be linked to CD14 expression and nuclear translocation NFjB that shift host defense balance toward a proinﬂammatory state, although further studies are needed to elucidate it.

IV. Pulmonary and Systemic Release of Inﬂammatory Mediators in Ex Vivo and In Vivo Models of VILI Studies using in vivo animal models of ALI (preinjured lungs) have shown that injurious ventilatory strategies (high VT and/or low PEEP) can induce the release of proinﬂammatory cytokines into the airspaces and bloodstream,

272

Imai and Slutsky

an increased neutrophil inﬁltration into the lung, and the activation of alveolar macrophages. Chiumello et al. (17) demonstrated that MV for four hours with a VT of 16 mL/kg, zero PEEP versus VT of 9 mL/kg, PEEP of 5 cmH2O resulted in an increased release of proinﬂammatory cytokines [tumor necrosis factor (TNF) -a and macrophage inﬂammatory protein-2 (MIP-2)] in the lung as well as into the systemic circulation in an in vivo acid aspiration–induced ALI model in rats. Haitsma et al. (53) found that MV for 20 minutes with peak inspiratory pressure/PEEP of 45/0 versus 45/10 increased the release of TNF-a into the systemic circulation in rats pretreated with LPS intratracheally, while an increased bronchoalveolar lavage (BAL) TNF-a in rats pretreated with intraperitoneal LPS suggested a loss of alveolar and systemic compartmentalization of TNF-a. These studies indicate that VILI can produce proinﬂammatory mediators in preinjured lungs, which is clinically relevant to VILI in patients with ALI/ARDS. The studies quoted above utilized ventilatory strategies in lungs with preexisting injury. The literature on ventilation-induced release of mediators from lungs without preexisting injury is less clear. Tremblay et al. (14) found that ventilation of the isolated, nonperfused rat lungs without preinjury, receiving a VT of 40 mL/kg and zero PEEP for two hours ex vivo, resulted in large increases in the lavage concentrations of TNF-a, IL-1b, IL-6, and MIP-2. The increase in these cytokines was greater if the rats were pretreated with LPS. Northern blot analysis of whole lung homogenates revealed an increased expression of c-fos mRNA, with both high and moderate VT and/or zero PEEP ventilation. An isolated, nonperfused ex vivo model has advantages, including the fact that it ampliﬁes the impact of zero PEEP because the lungs collapse completely owing to the fact that there is no chest wall, and that there is clearly no impact on hemodynamics. There are, however, disadvantages in this model including the fact that it is nonphysiologic, and the lungs are ischemic, and hence are not normal, albeit with no overt ALI. Ricard et al. (16) followed up Tremblay’s study using the same isolated, nonperfused rat lungs. They could not replicate the ﬁndings of increases in the lavage concentration of TNF-a, but found an approximately fourfold increase in IL-1b in the lungs receiving VT of 40 mL/kg and zero PEEP for two hours. The reasons for this discrepancy between studies are not clear but may relate to the age or weight of the rats (VT was set based on the animal’s body weight) or sensitivity of the antibody using TNF-a ELISA. Interestingly, in an in vivo murine model of VILI without preexisting injury, Wilson et al. (54) demonstrated that high VT (34.5 2.9 mL/kg) ventilation was associated with an increased TNF-a in lung lavage ﬂuid at the early stage (120 minutes) but not at the later stage (156 17 minutes until mean blood pressure fell below 45 mmHg), whereas lavage ﬂuid MIP-2 was increased in all high VT ventilation, suggesting the transient nature of

Systemic Effects of Mechanical Ventilation

273

TNF-a upregulation by VILI. Recently, Belperio et al. (55) found that the CXC chemokines KC/CXCL1 and MTP-2/CXCL2/3 were increased in in vivo murine lungs without preinjury receiving a high peak pressure (40 cmH2O) ventilation for six hours, and that the pharmacological or genetic inhibition of CXCR2, a receptor of KC/CXCL1 and MIP-2/ CXCL2/3, reduced the lung injury caused by high peak pressure ventilation. These studies indicate that injurious ventilation per se, without preexisting lung injury, also can initiate proinﬂammatory mediator–evoked lung injury. Recently Held et al. (56) demonstrated in isolated perfused mouse lungs that both LPS and overventilation (OV) caused nuclear translocation of NFjB, leading to the release of MIP-2. Interestingly, they found that LPS-induced, but not OV-induced nuclear translocation of NFjB and release of MIP-2 were reduced in C3H/HeJ mice, which have a mutation in toll-like receptor-4 (TLR-4), suggesting that the initial signaling steps due to LPS differ from those due to OV, and that the NFjB translocation elicited by OV is independent of TLR-4.

V. Passage of Mediators from Lung to Bloodstream In healthy lungs, the alveolar barrier restricts the transport of macromolecules of a size similar to that of cytokines (15–20 kDa). Cytokines such as TNF-a remain in the alveolar space, and leak into the circulation only if there is injury of the alveolar–capillary barrier. The loss of compartmentalization of local pulmonary inﬂammatory mediators due to MV likely explains the increased serum levels of cytokines in the experimental models discussed above (4). Tutor et al. (57) employed an isolated perfused rat lung model in which they injected TNF-a into the lung and measured its appearance in the perfusate. They found that the perfusate TNF-a concentrations were increased only when the alveolar–capillary permeability was increased, and not in the normal lung, suggesting that the loss of compartmentalization of alveolar TNF-a could occur, but only in the context of damage to the alveolar–capillary membrane. von Bethmann et al. (15) reported that in an isolated perfused murine lung model, ventilation with a higher transpulmonary pressure (25 cmH2O) versus normal pressure (10 cmH2O) led to a signiﬁcant increase in the concentration of both TNF-a and IL-6 in the perfusate. Because the compartmentalization of the local pulmonary response is lost, the systemic release of inﬂammatory mediators may play a role in massive inﬂammatory response that underlies MODS (see below). The same concept seems to hold in humans as well. In patients with ARDS, concentrations of TNF-a, IL-1b, and IL-6 were higher in the pulmonary arterialized capillary blood (obtained via a Swan-Ganz catheter), as compared with mixed venous blood, suggesting the translocation of local

274

Imai and Slutsky

Figure 2 Apoptotic index in lung, kidney, and small intestine. Values are mean standard error. p < 0.01 versus NON-INJ. Quantiﬁcation of apoptotic indices in various organs of rabbits with acid aspiration–induced ALI, ventilated with either a relatively NON-INJ or a more INJ with a larger VT, and lower PEEP. (A) Apoptotic index in lung. The apoptotic index was signiﬁcantly higher in the NON-INJ group than the INJ group. (B) Apoptotic index in the kidney. The major apoptotic cell type was the tubular epithelial cell. The apoptotic index was signiﬁcantly higher in the INJ group than the NON-INJ group. (C) Apoptotic index in the villi of the small intestine. The apoptotic index was signiﬁcantly higher in the INJ group than the NON-INJ group. (D) Apoptotic index in the crypts of the small intestine. The apoptotic index was not signiﬁcantly different between the groups. Abbreviations: ALI, acute lung injury; NON-INJ, noninjurious ventilatory strategy; INJ, injurious strategy. Source: From Ref. 18.

cytokines to the systemic circulation in these patients (58). This did not occur in patients with disease processes other than ARDS. ARDS is characterized by a loss of integrity of the alveolar–capillary barrier due to severe diffuse alveolar damage, leading to a bidirectional protein ﬂux. Therefore, not only the proinﬂammatory cytokines but also the locally secreted proteins, particularly the surfactant-associated protein (SP), may pass into the systemic circulation. SP-A, SP-B, and SP-D have been detected in the serum of ARDS patients, and have been associated with outcome in patients with ARDS (59–61). The balance between the pro- and anti-inﬂammatory cytokines passing from the lung to the bloodstream may be more important in determining subsequent effects than the absolute values of any single mediator.

Systemic Effects of Mechanical Ventilation

275

VI. Injurious Ventilatory Strategies Can Enhance End-Organ Dysfunction, Apoptosis, and Inﬂammation The data summarized above indicate that certain ventilatory strategies can cause the release of cytokines from the lung, which may translocate into the systemic circulation. But can these cytokines impact on organs distal to the lung? Using a rabbit model of acid aspiration–induced ARDS, we have recently demonstrated that MV with an injurious ventilatory strategy could lead to the apoptosis of certain cells in some organs, which may be an important down-stream mechanism leading to the development of MODS consequent to VILI/biotrauma (Fig. 1) (18). We found that animals with an acidinduced ALI and ventilated with an injurious strategy had a marked increase in epithelial cell apoptosis in the kidney (Fig. 2B) and small intestine (Fig. 2C and D), accompanied by evidence of renal dysfunction. Furthermore, plasma obtained from animals subjected to injurious ventilation induced greater apoptosis in renal tubular epithelial cells in vitro. This apoptosis was attenuated by Fas:Ig, a fusion protein that binds to and blocks FasL in vitro. We also found a signiﬁcant correlation between changes in sFasL and changes in serum creatinine in ARDS patients. These data suggest that the Fas–FasL pathway may play a pivotal role in the end-organ apoptosis and end-organ dysfunction caused by VILI or biotrauma. Recently, using a mouse model of acid aspiration–induced ARDS, Gurkan et al. (19) demonstrated that mice ventilated with high VT (17 mL/kg) manifested lung injury as well as increased IL-6 and VEGF receptor-2 (VEGFR-2) in the lung, liver, and kidney, and that MV with low VT (6 mL/kg) attenuated lung injury as well as IL-6 and VEGFR-2 expression in lung and systemic organs. These data suggest that ventilation with large VTs can lead to inﬂammatory changes in organs other than the lung. VII. Bacterial Translocation in MV Another mechanism whereby MV may contribute to the development of a systemic inﬂammatory response is by promoting bacterial translocation from the air spaces into the circulation. Experimental studies have evaluated the inﬂuence of ventilator strategy on the translocation of bacteria from the lung into the bloodstream. After intratracheal instillation of bacteria (Klebsiella pneumoniae), rats were ventilated with a high transpulmonary pressure (30 cmH2O) and minimal (0–3 cmH2O) or 10 cmH2O PEEP. Bacteremia was greater in animals ventilated with a high VT and a low PEEP compared to control animals that were ventilated with a low peak airway pressure. Ventilation with the large VT, but with 10 cmH2O PEEP, resulted in rates of bacteremia as low as in controls (62). In a saline-lavaged rabbit lung injury model, MV with a VT of 12 mL/kg with

276

Imai and Slutsky

zero PEEP resulted in the translocation of intratracheally instilled endotoxin into the systemic circulation, but ventilation with a VT of 5 mL/kg and a PEEP of 10 cmH2O did not lead to the same result. The appearance of endotoxin in the blood stream was associated with an increase in plasma TNF-a (24).

VIII. Does the Release of Mediators by VILI Have Any Pathophysiologic Relevance? Many of the studies cited above demonstrate that injurious ventilatory strategies can lead to the release of various mediators. A key question is whether these mediators have any pathophysiologic relevance causing further damage to the lungs or other organs. The literature related to this question is much more sparse than the evidence demonstrating the release of mediators. Evidence for the importance of these inﬂammatory mediators in the development of VILI comes from experimental studies of the effects of an anti-TNF antibody and an IL-1 receptor antagonist on lung injury following a saline lavage. Imai et al. (63) employed anti-TNF-a antibodies and observed improvements in oxygenation and respiratory compliance, decreased lavage neutrophil counts, as well as reduced histological evidence of lung injury. Narimanbekov and Rozycki (64) used an IL-1b receptor antagonist and found reduced lung lavage concentrations of a number of markers of lung injury (i.e., albumin, elastase, and neutrophils) in a salinelavaged rabbit model. Guery et al. (65) demonstrated that gut permeability assessed by the leakage of 125I-labeled albumin was increased by high VT (30 mL/kg) ventilation when compared to the controls (VT 10 mL/kg), and that anti-TNF-a antibodies abrogated the increase in gut permeability as well as lung permeability caused by high VT ventilation, suggesting the functional signiﬁcance of TNF-a in the end-organ dysfunction caused by high VT ventilation. Further studies are necessary to clarify the pathophysiological relevance of the release of mediators by VILI in end-organ dysfunction that underlie MODS.

IX. Pulmonary and Systemic Inﬂammatory Mediators in VILI in Clinical Studies Most of the data cited above were obtained in animal studies. Elevated levels of proinﬂammatory mediators have also been measured in the lavage ﬂuid and the plasma of patients with ARDS. Ranieri et al. (21) measured BAL and plasma levels of several proinﬂammatory cytokines in 44 patients

Systemic Effects of Mechanical Ventilation

277

with ARDS. At study entry, patients were randomized to receive MV with a conventional strategy (mean VT 11.1 mL/kg, mean plateau airway pressure 31 cmH2O, and mean PEEP 6.5 cmH2O), or a protective ventilatory strategy (mean VT 7.6 mL/kg, mean plateau airway pressure 24.6 cmH2O, and mean PEEP 14.8 cmH2O). PEEP in the latter group was set above the lower inﬂection point of the respiratory system pressure–volume curve. Baseline measurements of cytokines were made at the time of admission (study entry), and were then measured serially for three days. By 36 hours, BAL ﬂuid from patients in the protective ventilation group had signiﬁcantly lower concentrations of TNF-a, IL-1b, IL-6, and IL-8. Plasma levels of IL-6 were also signiﬁcantly lower in the patients receiving protective ventilation. The National Institutes of Health (NIH) ARDS Network Study (20) found lower levels of plasma IL-6 at three days in patients ventilated with low VT compared with conventional VT. These patients had a 22% relative decrease in mortality compared to the higher VT group. The NIH ARDS Network group (66) recently reported a post hoc analysis with the data of 861 patients enrolled in this study, and found that baseline plasma levels of IL-6, IL-8, and IL-10 were each associated with an increased risk of death in both logistic regression analyses controlling for ventilator group and multivariate analyses controlling for ventilation strategy, acute physiology and chronic health evaluation III score, PaO2/FiO2 ratio, creatinine, platelet count, and vasopressor use. IL-6 and IL-8 levels were also associated with a signiﬁcant decrease in ventilator free and organ failure–free days. Importantly, by day 3, the low VT (6 mL/kg) strategy was associated with a greater decrease in IL-6 and IL-8 levels compared with the high VT (12 mL/kg) group. These data add further evidence about the clinical signiﬁcance of plasma levels of proinﬂammatory cytokines associated with ventilatory strategy. The relatively quick time course of change in the mediator release was demonstrated by Stuber et al. (67). They studied patients with ALI and found that switching to conventional MV (VT of 12 mL/kg, PEEP of 5 cmH2O) from a lung-protective strategy (VT of 5 mL/kg, PEEP of 15 cmH2O) was associated with a marked increase in plasma cytokine levels within one hour, whereas plasma cytokine levels returned to baseline when lung-protective settings were re-established. In contrast, in 39 patients without ARDS, Wrigge et al. (68) found that ventilation with a high VT (15 mL/kg) and zero PEEP did not affect plasma levels of IL-6, TNF-a, IL-l receptor antagonist, or IL-10. These studies, along with the animal studies cited above, suggest that circulating cytokines can come from the lung due to the loss of compartmentalization, and that this largely occurs in patients with increases in alveolar–capillary permeability (e.g., patients with ALI/ARDS), and not in patients with relatively normal lungs. However, injurious ventilatory strategies can certainly lead to increased alveolar–capillary

278

Imai and Slutsky

permeability, and hence these strategies may lead to the translocation of cytokines, even in lungs that start off normal.

X. Multiple Organ Dysfunction and VILI in Clinical Studies Ranieri et al. reported that the use of a lung-protective strategy in patients with ARDS attenuated an increase in the pulmonary and systemic cytokine levels, including TNF-a and IL-6 (69). In a subsequent post hoc analysis, they reported a higher incidence of renal failure in ARDS patients ventilated with conventional ventilation compared with those ventilated with a lung-protective strategy. Furthermore, they found a signiﬁcant correlation between overall MODS score and changes in plasma concentration of a number of inﬂammatory mediators (IL-6, TNFa, ILl-b, and IL-8), which have been previously shown to be potentially important in the development of MODS (22). Similarly, the NIH ARDS Network (20) reported the results of a randomized, clinical trial comparing a VT of 12 mL/kg with a VT of 6 mL/kg (predicted body weight). They found lower levels of plasma IL-6 in the 6 mL/kg VT group. This was associated with a greater number of organ failure–free days (circulatory failure, coagulation failure, and renal failure) and a 22% reduction in mortality rate in the 6 mL/kg group. It is important to emphasize that MODS is a complex syndrome, often precipitated and intensiﬁed by a series of events rather than a single event. A likely scenario is that there is an ongoing inﬂammatory response as a result of the persistence of the factors that either initiated or exacerbated the response, and/or failure of intrinsic regulatory mechanisms. In this context, the speciﬁc contribution (if any) of ventilator-induced biotrauma to the development of MODS in patients is at present unclear.

XI. Conclusions Based on the paradigm developed in this review, it is suggested that VILI may play an important role in initiating and/or propagating a systemic inﬂammatory response leading to MODS. Patients at the greatest risk are those at greatest risk of VILI—patients with ALI and ARDS. As such, protective ventilatory strategies in concert with other novel therapies could reduce the development of MODS and decrease mortality in mechanically ventilated patients. The biotrauma paradigm also suggests that pharmacological modulation of the cellular and molecular sequelae of VILI maybe useful in abrogating the development of MODS.

Systemic Effects of Mechanical Ventilation

279

References 1. Esteban A, Anzueto A, Frutos F, et al. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study. JAMA 2002; 287:345–355. 2. Hudson LD, Milberg JA, Anardi D, Maunder RJ. Clinical risks for development of the acute respiratory distress syndrome. Am J Respir Crit Care 1995; Med 151:293–301. 3. Vincent JL, Sakr Y, Ranieri VM. Epidemiology and outcome of acute respiratory failure in intensive care unit patients. Crit Care Med 2003; 31:S296–S299. 4. Slutsky AS, Tremblay LN. Multiple system organ failure. Is mechanical ventilation a contributing factor? Am J Respir Crit Care Med 1998; 157: 1721–1725. 5. Waters CM, Sporn PH, Liu M, Fredberg JJ. Cellular biomechanics in the lung. Am J Physiol Lung Cell Mol Physiol 2002; 283:L503–L509. 6. Edwards YS. Stretch stimulation: its effects on alveolar type II cell function in the lung. Comp Biochem Physiol A Mol Integr Physiol 2001; 129:245–260. 7. Liu M, Post M. Invited review: mechanochemical signal transduction in the fetal lung. J Appl Physiol 2000; 89:2078–2084. 8. Ishida T, Takahashi M, Corson MA, Berk BC. Fluid shear stress-mediated signal transduction: how do endothelial cells transduce mechanical force into biological responses? Ann NY Acad Sci 1997; 811:12–23. 9. Uhlig S. Ventilation-induced lung injury and mechanotransduction: stretching it too far? Am J Physiol Lung Cell Mol Physiol 2002; 282:L892–L896. 10. Dos Santos CC, Slutsky AS. Invited review: mechanisms of ventilator-induced lung injury: a perspective. J Appl Physiol 2000; 89:1645–1655. 11. Pugin J, Dunn I, Jolliet P, et al. Activation of human macrophages by mechanical ventilation in vitro. Am J Physiol 1998; 275:L1040–L1050. 12. Quinn D, Tager A, Joseph PM, Bonventre JV, Force T, Hales CA. Stretchinduced mitogen-activated protein kinase activation and interleukin-8 production in type II alveolar cells. Chest 1999; 116:89S–90S. 13. Vlahakis NE, Schroeder MA, Limper AH, Hubmayr RD. Stretch induces cytokine release by alveolar epithelial cells in vitro. Am J Physiol 1999; 277:L167–L173. 14. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99:944–952. 15. von Bethmann AN, Brasch F, Nusing R, et al. Hyperventilation induces release of cytokines from perfused mouse lung. Am J Respir Crit Care Med 1998; 157:263–272. 16. Ricard JD, Dreyfuss D, Saumon G. Production of inﬂammatory cytokines in ventilator-induced lung injury: a reappraisal. Am J Respir Crit Care Med 2001; 163:1176–1180. 17. Chiumello D, Pristine G, Slutsky AS. Mechanical ventilation affects local and systemic cytokines in an animal model of acute respiratory distress syndrome. Am J Respir Crit Care Med 1999; 160:109–116.

280

Imai and Slutsky

18. Imai Y, Parodo J, Kajikawa O, et al. Injurious mechanical ventilation and end-organ epithelial cell apoptosis and organ dysfunction in an experimental model of acute respiratory distress syndrome. JAMA 2003; 289:2104–2112. 19. Gurkan OU, O’Donnell C, Brower R, Ruckdeschel E, Becker PM. Differential effects of mechanical ventilatory strategy on lung injury and systemic organ inﬂammation in mice. Am J Physiol Lung Cell Mol Physiol 2003; 285: L710–L718. 20. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342: 1301–1308. 21. Ranieri VM, Suter PM, Tortorella C, et al. Effect of mechanical ventilation on inﬂammatory mediators in patients with acute respiratory distress syndrome: a randomized controlled trial [see comments]. JAMA 1999; 282:54–61. 22. Ranieri VM, Giunta F, Suter PM, Slutsky AS. Mechanical ventilation as a mediator of multisystem organ failure in acute respiratory distress syndrome. JAMA 2000; 284:43–44. 23. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome [see comments]. N Engl J Med 1998; 338:347–354. 24. Murphy DB, Cregg N, Tremblay L, et al. Adverse ventilatory strategy causes pulmonary-to-systemic translocation of endotoxin. Am J Respir Crit Care Med 2000; 162:27–33. 25. Pinsky MR. Recent advances in the clinical application of heart-lung interactions. Curr Opin Crit Care 2002; 8:26–31. 26. Le Gall JR, Klar J, Lemeshow S, et al. The logistic organ dysfunction system. A new way to assess organ dysfunction in the intensive care unit. ICU Scoring Group. JAMA 1996; 276:802–810. 27. Russell JA, Phang PT. The oxygen delivery/consumption controversy. Approaches to management of the critically ill. Am J Respir Crit Care Med 1994; 149:533–537. 28. De Backer D. The effects of positive end-expiratory pressure on the splanchnic circulation. Intensive Care Med 2000; 26:361–363. 29. Fessler HE, Brower RG, Wise RA, Permutt S. Effects of positive end-expiratory pressure on the gradient for venous return. Am Rev Respir Dis 1991; 143:19–24. 30. Jardin F, Farcot JC, Boisante L, Curien N, Margairaz A, Bourdarias JP. Inﬂuence of positive end-expiratory pressure on left ventricular performance. N Engl J Med 1981; 304:387–392. 31. Robotham JL, Lixfeld W, Holland L, et al. The effects of positive endexpiratory pressure on right and left ventricular performance. Am Rev Respir Dis 1980; 121:677–683. 32. Brienza N, Revelly JP, Ayuse T, Robotham JL. Effects of PEEP on liver arterial and venous blood ﬂows. Am J Respir Crit Care Med 1995; 152:504–510. 33. Dorinsky PM, Hamlin RL, Gadek JE. Alterations in regional blood ﬂow during positive end-expiratory pressure ventilation. Crit Care Med 1987; 15: 106–113.

Systemic Effects of Mechanical Ventilation

281

34. Matuschak GM, Pinsky MR, Rogers RM. Effects of positive end-expiratory pressure on hepatic blood ﬂow and performance. J Appl Physiol 1987; 62:1377–1383. 35. Sha M, Saito Y, Yokoyama K, Sawa T, Amana K. Effects of continuous positive-pressure ventilation on hepatic blood ﬂow and intrahepatic oxygen delivery in dogs. Crit Care Med 1987; 15:1040–1043. 36. Trager K, Radermacher P, Georgieff M. PEEP and hepatic metabolic performance in septic shock. Intensive Care Med 1996; 22:1274–1275. 37. Kiefer P, Nunes S, Kosonen P, Takala J. Effect of positive end-expiratory pressure on splanchnic perfusion in acute lung injury. Intensive Care Med 2000; 26:376–383. 38. Seeman-Lodding H, Haggmark S, Jern C, et al. Systemic levels and preportal organ release of tissue-type plasminogen activator are enhanced by PEEP in the pig. Acta Anaesthesiol Scand 1999; 43:623–633. 39. Kiiski R, Takala J, Kari A, Milic-Emili J. Effect of tidal volume on gas exchange and oxygen transport in the adult respiratory distress syndrome. Am Rev Respir Dis 1992; 146:1131–1135. 40. Thorens JB, Jolliet P, Ritz M, Chevrolet JC. Effects of rapid permissive hypercapnia on hemodynamics, gas exchange, and oxygen transport and consumption during mechanical ventilation for the acute respiratory distress syndrome. Intensive Care Med 1996; 22:182–191. 41. Brofman JD, Leff AR, Munoz NM, Kirchhoff C, White SR. Sympathetic secretory response to hypercapnic acidosis in swine. J Appl Physiol 1990; 69:710–717. 42. Rose CE Jr, Althaus JA, Kaiser DL, Miller ED, Carey RM. Acute hypoxemia and hypercapnia: increase in plasma catecholamines in conscious dogs. Am J Physiol 1983; 245:H924–H929. 43. Cardenas VJ Jr, Zwischenberger JB, Tao W, et al. Correction of blood pH attenuates changes in hemodynamics and organ blood ﬂow during permissive hypercapnia. Crit Care Med 1996; 24:827–834. 44. Sitbon P, Teboul JL, Duranteau J, Anguel N, Richard C, Samii K. Effects of tidal volume reduction in acute respiratory distress syndrome on gastric mucosal perfusion. Intensive Care Med 2001; 27:911–915. 45. Vlahakis NE, Hubmayr RD. Cellular stress failure in ventilator injured lungs. Am J Respir Crit Care Med 2005; 171(12):1328–1342. 46. Vlahakis NE, Hubmayr RD. Response of alveolar cells to mechanical stress. Curr Opin Crit Care 2003; 9:2–8. 47. Vlahakis NE, Hubmayr RD. Invited review: plasma membrane stress failure in alveolar epithelial cells. J Appl Physiol 2000; 89:2490–2496. 48. Mascarenhas MM, Day RM, Ochoa CD, et al. Low molecular weight hyaluronan from stretched lung enhances interleukin-8 expression. Am J Respir Cell Mol Biol 2004; 30:51–60. 49. Birukov KG, Jacobson JR, Flores AA, et al. Magnitude-dependent regulation of pulmonary endothelial cell barrier function by cyclic stretch. Am J Physiol Lung Cell Mol Physiol 2003; 285:L785–L797. 50. Gan L, Doroudi R, Hagg U, Johansson A, Selin-Sjogren L, Jern S. Differential immediate-early gene responses to shear stress and intraluminal pressure in intact human conduit vessels. FEBS Lett 2000; 477:89–94.

282

Imai and Slutsky

51. Gan L, Miocic M, Doroudi R, Selin-Sjogren L, Jern S. Distinct regulation of vascular endothelial growth factor in intact human conduit vessels exposed to laminar ﬂuid shear stress and pressure. Biochem Biophys Res Commun 2000; 272:490–496. 52. Moriyama K, Ishizaka A, Nakamura M, et al. Enhancement of the endotoxin recognition pathway by ventilation with a large tidal volume in rabbits. Am J Physiol Lung Cell Mol Physiol 2004; 286:L1114–L1121. 53. Haitsma JJ, Uhlig S, Goggel R, Verbrugge SJ, Lachmann U, Lachmann B. Ventilator-induced lung injury leads to loss of alveolar and systemic compartmentalization of tumor necrosis factor-alpha. Intensive Care Med 2000; 26:1515–1522. 54. Wilson MR, Choudhury S, Goddard ME, ODea KP, Nicholson AG, Takata M. High tidal volume upregulates intrapulmonary cytokines in an in vivo mouse model of ventilator-induced lung injury. J Appl Physiol 2003; 95: 1385–1393. 55. Belperio JA, Keane MP, Burdick MD, et al. Critical role for CXCR2 and CXCR2 ligands during the pathogenesis of ventilator-induced lung injury. J Clin Invest 2002; 110:1703–1716. 56. Held HD, Boettcher S, Hamann L, Uhlig S. Ventilation-induced chemokine and cytokine release is associated with activation of nuclear factor-kappaB and is blocked by steroids. Am J Respir Crit Care Med 2001; 163:711–716. 57. Tutor JD, Mason CM, Dobard E, Beckerman RC, Summer WR, Nelson S. Loss of compartmentalization of alveolar tumor necrosis factor after lung injury. Am J Respir Crit Care Med 1994; 149:1107–1111. 58. Douzinas EE, Tsidemiadou PD, Pitaridis MT, et al. The regional production of cytokines and lactate in sepsis-related multiple organ failure. Am J Respir Crit Care Med 1997; 155:53–59. 59. Greene KE, Wright JR, Steinberg KP, et al. Serial changes in surfactant-associated proteins in lung and serum before and after onset of ARDS. Am J Respir Crit Care Med 1999; 160:1843–1850. 60. Greene KE, Ye S, Mason RJ, Parsons PE. Serum surfactant protein-A levels predict development of ARDS in at-risk patients. Chest 1999; 116:90S–91S. 61. Doyle IR, Bersten AD, Nicholas TE. Surfactant proteins-A and -B are elevated in plasma of patients with acute respiratory failure. Am J Respir Crit Care Med 1997; 156:1217–229. 62. Verbrugge SJ, Sorm V, van V, Mouton JW, Gommers D, Lachmann B. Lung overinﬂation without positive end-expiratory pressure promotes bacteremia after experimental Klebsiella pneumoniae inoculation. Intensive Care Med 1998; 24:172–177. 63. Imai Y, Kawano T, Iwamoto S, Nakagawa S, Takata M, Miyasaka K. Intratracheal anti-tumor necrosis factor-alpha antibody attenuates ventilator-induced lung injury in rabbits. J Appl Physiol 1999; 87:510–515. 64. Narimanbekov IO, Rozycki HJ. Effect of IL-1 blockade on inﬂammatory manifestations of acute ventilator-induced lung injury in a rabbit model. Exp Lung Res 1995; 21:239–254. 65. Guery BP, Welsh DA, Viget NB, et al. Ventilation-induced lung injury is associated with an increase in gut permeability. Shock 2003; 19:559–563.

Systemic Effects of Mechanical Ventilation

283

66. Parsons PE, Eisner MD, Thompson BT, et al. Lower tidal volume ventilation and plasma cytokine markers of inﬂammation in patients with acute lung injury. Crit Care Med 2005; 33:1–6. 67. Stuber F, Wrigge H, Schroeder S, et al. Kinetic and reversibility of mechanical ventilation-associated pulmonary and systemic inﬂammatory response in patients with acute lung injury. Intensive Care Med 2002; 28:834–841. 68. Wrigge H, Zinserling J, Stuber F, et al. Effects of mechanical ventilation on release of cytokines into systemic circulation in patients with normal pulmonary function. Anesthesiology 2000; 93:1413–1417. 69. Ranieri VM, Slutsky AS. Respiratory physiology and acute lung injury: the miracle of Lazarus [editorial; comment’. Intensive Care Med 1999; 25:1040–1043.

This page intentionally left blank

11 Alveolar Fluid Reabsorption During VILI

¨ KHAN M. MUTLU, EMILIA LECUONA, and JACOB I. SZNAJDER GO Pulmonary and Critical Care Medicine, Feinberg School of Medicine, Northwestern University Chicago, Illinois, U.S.A.

I. Introduction It is well recognized that the mechanisms regulating alveolar ﬂuid clearance differ from those contributing to edema formation, where changes in the pulmonary ﬁltration coefﬁcient and the hydrostatic and oncotic pressure gradients determine the extent of edema formation. Lung edema clearance is effected by active Naþ transport where Naþ moves vectorially across the alveolar epithelial barrier via the apical Naþ channels and the basolaterally located Na,K-ATPases, with water following isosmotically. Ventilator-induced lung injury (VILI) results from high alveolar distending volume leading to the injury of the alveolo-capillary barrier, resulting in increased permeability and hyaline membrane formation (1–3). Along with increased edema formation, there is impairment in alveolar ﬂuid reabsorption (4). While the mainstay of clinical management of patients with pulmonary edema includes the use of diuretics along with salt and water restriction, better understanding of the mechanisms regulating the clearance of pulmonary edema should lead to more speciﬁc therapeutic approaches to accelerate alveolar ﬂuid reabsorption. 285

286

Mutlu et al. II. Alveolar Epithelial Sodium Transport

Under normal conditions, the alveolo-capillary barrier has a very low permeability to solutes (5,6). Tight junctions are critical for the alveolar epithelial barrier function because they connect adjacent epithelial cells and modulate dynamic permeability with distinction selective pores (7,8). Both alveolar type I (AT1) and alveolar type II (AT2) cells appear to be important in active Naþ transport. Because of technical limitations, AT2 cell has been better studied and was thought to be responsible for the majority of the alveolar epithelial transport of Naþ (9–15). However, recently, an important role for AT1 cell in vectorial Naþ transport has been demonstrated and it appears that AT1 cells are responsible for approximately 60% of Naþ transport (15–18). In addition to alveolar epithelial cells, distal airway epithelium may also play a role in ﬂuid clearance because it actively transports Naþ (19–24). A. Sodium Channels þ

Na enters the apical surface of alveolar epithelial cells predominantly via amiloride-sensitive epithelial Naþ channels (ENaC) and also via other less well-characterized, cationic channels (25–29). The ENaC are heterodimers of up to three subunits, a, b, and c ENaC, which confer Naþ ﬂux speciﬁcity (30,31). These subunits are expressed along the respiratory tract epithelium and the apical surface of the alveolar epithelial cells (32–35). The importance of ENaC in alveolar epithelial Naþ transport is supported by studies in transgenic mice where animals lacking functional ENaC do not survive (36). The contribution of ENaC to alveolar active Naþ transport seems to be species dependent. While amiloride can inhibit up to 90% of Naþ transport in mice, the relative contribution of ENaC to basal alveolar ﬂuid clearance is approximately 40% to 60% in the lungs of rats, sheep, rabbits, and humans. B. Na,K-ATPase

After Naþ enters the alveolar epithelial cells, the sodium pumps (Na, K-ATPase) expressed on the basolateral surface actively transport Naþ against a gradient, consuming adenosine triphosphate. The Na,K-ATPase is expressed both in AT1 and AT2 cells (15,17,18,37,38). It is a heterodimeric protein composed of an a and a b-subunit, and both are required for a functional Na,K-ATPase (39,40). The short-term regulation of Na,K-ATPase activity is by the recruitment of Naþ pump proteins from intracellular compartments and their insertion into the plasma membrane (40–45). Long-term regulation occurs via G-protein coupled receptor (GPCR) activation and mitogen-activated protein kinase/extracellular

Alveolar Fluid Reabsorption During VILI

287

signal-regulated kinase (MAPK/ERK), and mammalian target of Rapamycin (mTOR) pathways, as reviewed previously (46). C. Chloride Channels

Apical membrane chloride channels have functional and pharmacological properties similar to those of cystic ﬁbrosis transmembrane conductance regulator (CFTR). The CFTR is a cAMP-regulated Cl channel found on the apical surface of many epithelial cells including airway and alveolar epithelial cells (47). Although lack of CFTR gene (targeted deletion of deltaF508 -/-) does not affect normal lung ﬂuid homeostasis at baseline, it appears to be important in the regulation of ﬂuid clearance during hydrostatic pulmonary edema (48). D. Aquaporins

Transcellular water channels or aquaporins (AQPs) are localized to the lung (49,50). In mice and rats, AQP1 is expressed in the endothelial cells and ﬁbroblasts (51), and AQP3, AQP4, and AQP5 are found on both apical and basolateral membranes at different locations of respiratory tract epithelium (52). AQP5 is expressed at the apical surface of AT1 cells and in the nasopharyngeal epithelium, and AQP3 is expressed at the apical membrane of columnar epithelial, basal, and AT2 cells (53). III. Alveolar Fluid Reabsorption During VILI It has been reported that the alveolar ﬂuid reabsorption is inhibited in the models of lung injury even when there is no gross disruption of the alveolocapillary barrier (54–56). Most patients with noncardiogenic pulmonary edema have impaired alveolar ﬂuid clearance (57). However, patients with a normal rate of edema ﬂuid clearance have better outcomes (46,57–59). Pulmonary edema, one of the hallmarks of VILI, is thought to result from high-volume overdistension of the alveolar epithelium, and increased alveolo-capillary permeability (1,3). Because the epithelium becomes more permeable, accumulation of pulmonary edema is enhanced by the fact that overinﬂation decreases alveolar ﬂuid reabsorption in association with Na,K-ATPase inhibition (4). Pharmacological agents such as b-adrenergic agonists (terbutaline and isoproterenol) (60) or dopamine (56) have been shown to improve alveolar ﬂuid reabsorption after ventilation of rat lungs with high tidal volumes by restoring Naþ transport across the alveolar epithelium and upregulation of the Na,K-ATPase activity. The importance of an adequate Naþ transport in preventing the accumulation of ventilator-induced pulmonary edema is suggested by a report in which adenoviral-mediated gene

288

Mutlu et al.

transfer of the Na,K-ATPase b1-subunit normalized the alveolar ﬂuid reabsorption in rats that were subjected to high tidal volume ventilation (61). Experiments carried out in alveolar epithelial cells subjected to mechanical stretch suggest an opposite effect on the Naþ transport machinery, with a reported increase in Na,K-ATPase activity (62,63). These apparently discordant data could be due to the fact that mild stretch has protective effects in that stretch increased intracellular Ca2þ and secreted surfactant (64). However, excessive alveolar distension associated with injurious mechanical ventilation also impaired alveolo-capillary permeability and increased the production of inﬂammatory mediators (2,65–67), which could have an adverse effect on active Naþ transport by inhibiting the Naþ channels and Na,K-ATPase function. IV. Summary During mechanical ventilation with high tidal volumes, alveoli appear to be exposed to stretch and shear forces causing variable degrees of alveolocapillary barrier disruption. Paralleling these changes, Naþ transport mechanisms responsible for the clearance of edema are inhibited, contributing to the overall increase in edema and lung injury. How stretch or shear stress impairs active Naþ transport and alveolar ﬂuid reabsorption is not fully understood. Do mechanical strain signals trigger endocytosis of ENaC and Na,K-ATPase? Do these signals involve reactive oxygen species and phosphorylation and nitration of proteins? These are complex biologic events that warrant further elucidation. Importantly, the inhibition of Na,K-ATPase and clearance during VILI can be reversible as suggested by reports showing that isoproterenol and dopamine restore alveolar ﬂuid clearance to normal levels during injurious high tidal volume ventilation. References 1. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 2. Dreyfuss D, Basset G, Soler P, Saumon G. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 1985; 132:880–884. 3. Sznajder JI, Ridge K, Saumon G, Dreyfuss D. Lung injury induced by mechanical ventilation. In: Matthay MA, Ingbar DH, eds. Pulmonary Edema. Marcel Dekker, 1998:413–430. 4. Lecuona E, Saldias F, Comellas A, Ridge K, Guerrero C, Sznajder JI. Ventilator-associated lung injury decreases lung ability to clear edema in rats. Am J Respir Crit Care Med 1999; 159:603–609. 5. Taylor AE, Guyton AC, Bishop VS. Permeability of the alveolar epithelium to solutes. Circ Res 1965; 16:353–362.

Alveolar Fluid Reabsorption During VILI

289

6. Gorin AB, Stewart PA. Differential permeability of endothelial and epithelial barriers to albumin ﬂux. J Appl Physiol 1979; 47:1315–1324. 7. Gumbiner BM. Breaking through the tight junction barrier. J Cell Biol 1993; 123:1631–1633. 8. Schneeberger EE, Lynch RD. Structure, function, and regulation of cellular tight junctions. Am J Physiol 1992; 262:L647–L661. 9. Goodman BE, Fleischer RS, Crandall ED. Evidence for active Naþ transport by cultured monolayers of pulmonary alveolar epithelial cells. Am J Physiol 1983; 245:C78–C83. 10. Mason RJ, Williams MC, Widdicombe JH, Sanders MJ, Misfeldt DS, Berry LC Jr. Transepithelial transport by pulmonary alveolar type II cells in primary culture. Proc Natl Acad Sci USA 1982; 79:6033–6037. 11. Matalon S, Benos DJ, Jackson RM. Biophysical and molecular properties of amiloride-inhibitable Naþ channels in alveolar epithelial cells. Am J Physiol 1996; 271:L1–L22. 12. Matthay MA, Folkesson HG, Verkman AS. Salt and water transport across alveolar and distal airway epithelial in the adult lung. Am J Physiol 1996; 270:L487–L503. 13. Michaut P, Planes C, Escoubet B, Clement A, Amiel C, Clerici C. Rat lung alveolar type II cell line maintains sodium transport characteristics of primary culture. J Cell Physiol 1996; 169:78–86. 14. Filippatos GS, Hughes WF, Qiao R, Sznajder JI, Uhal BD. Mechanisms of liquid ﬂux across pulmonary alveolar epithelial cell monolayers. In Vitro Cell Dev Biol Anim 1997; 33:195–200. 15. Ridge KM, Olivera WG, Saldias F, et al. Alveolar type 1 cells express the alpha2 Na,K-ATPase, which contributes to lung liquid clearance. Circ Res 2003; 92:453–460. 16. Ridge KM, Rutschman DH, Factor P, Katz AI, Bertorello AM, Sznajder JL. Differential expression of Na-K-ATPase isoforms in rat alveolar epithelial cells. Am J Physiol 1997; 273:L246–L255. 17. Borok Z, Liebler JM, Lubman RL, et al. Na transport proteins are expressed by rat alveolar epithelial type I cells. Am J Physiol Lung Cell Mol Physiol 2002; 282:L599–L608. 18. Johnson MD, Widdicombe JH, Allen L, Barbry P, Dobbs LG. Alveolar epithelial type I cells contain transport proteins and transport sodium, supporting an active role for type I cells in regulation of lung liquid homeostasis. Proc Natl Acad Sci USA 2002; 99:1966–1971. 19. Al-Bazzaz FJ. Regulation of Na and Cl transport in sheep distal airways. Am J Physiol 1994; 267:L193–L198. 20. Ballard ST, Taylor AE. Bioelectric properties of proximal bronchiolar epithelium. Am J Physiol 1994; 267:L79–L84. 21. Boucher RC. Human airway ion transport. Part one. Am J Respir Crit Care Med 1994; 150:271–281. 22. Boucher RC. Human airway ion transport. Part two. Am J Respir Crit Care Med 1994; 150:581–593. 23. Inglis SK, Corboz MR, Taylor AE, Ballard ST. Regulation of ion transport across porcine distal bronchi. Am J Physiol 1996; 270:L289–L297.

290

Mutlu et al.

24. Willumsen NJ, Boucher RC. Sodium transport and intracellular sodium activity in cultured human nasal epithelium. Am J Physiol 1991; 261:C319–C331. 25. Berthiaume Y, Lesur O, Dagenais A. Treatment of adult respiratory distress syndrome: plea for rescue therapy of the alveolar epithelium. Thorax 1999; 54:150–160. 26. Jayr C, Garat C, Meignan M, Pittet JF, Zelter M, Matthay MA. Alveolar liquid and protein clearance in anesthetized ventilated rats. J Appl Physiol 1994; 76:2636–2642. 27. Norlin A, Folkesson HG. Alveolar ﬂuid clearance in late-gestational guinea pigs after labor induction: mechanisms and regulation. Am J Physiol Lung Cell Mol Physiol 2001; 280:L606–L616. 28. Norlin A, Finley N, Abedinpour P, Folkesson HG. Alveolar liquid clearance in the anesthetized ventilated guinea pig. Am J Physiol 1998; 274:L235–L243. 29. Smedira N, Gates L, Hastings R, et al. Alveolar and lung liquid clearance in anesthetized rabbits. J Appl Physiol 1991; 70:1827–1835. 30. Canessa CM, Merillat AM, Rossier BC. Membrane topology of the epithelial sodium channel in intact cells. Am J Physiol 1994; 267:C1682–C1690. 31. McDonald JV Jr, Gonzales LW, Ballard PL, Pitha J, Roberts JM. Lung betaadrenoreceptor blockade affects perinatal surfactant release but not lung water. J Appl Physiol 1986; 60:1727–1733. 32. Renard S, Voilley N, Bassilana F, Lazdunski M, Barbry P. Localization and regulation by steroids of the alpha, beta and gamma subunits of the amiloridesensitive Naþ channel in colon, lung and kidney. Pﬂugers Arch 1995; 430: 299–307. 33. Matsushita K, McCray PB Jr, Sigmund RD, Welsh MJ, Stokes JB. Localization of epithelial sodium channel subunit mRNAs in adult rat lung by in situ hybridization. Am J Physiol 1996; 271:L332–L339. 34. Farman N, Talbot CR, Boucher R, et al. Noncoordinated expression of alpha-, beta-, and gamma-subunit mRNAs of epithelial Naþ channel along rat respiratory tract. Am J Physiol 1997; 272:C131–C141. 35. Burch LH, Talbot CR, Knowles MR, Canessa CM, Rossier BC, Boucher RC. Relative expression of the human epithelial Naþ channel subunits in normal and cystic ﬁbrosis airways. Am J Physiol 1995; 269:C511–C518. 36. Hummler E, Barker P, Gatzy J, et al. Early death due to defective neonatal lung liquid clearance in alpha-ENaC-deﬁcient mice. Nat Genet 1996; 12: 325–328. 37. Schneeberger EE, McCarthy KM. Cytochemical localization of Naþ-KþATPase in rat type II pneumocytes. J Appl Physiol 1986; 60:1584–1589. 38. O’Brodovich H, Staub O, Rossier BC, Geering K, Kraehenbuhl JP. Ontogeny of alpha 1- and beta 1-isoforms of Na(þ)-K(þ)-ATPase in fetal distal rat lung epithelium. Am J Physiol 1993; 264:C1137–C1143. 39. McDonough AA, Geering K, Farley RA. The sodium pump needs its beta subunit. Faseb J 1990; 4:1598–1605. 40. Dada LA, Chandel NS, Ridge KM, Pedemonte C, Bertorello AM, Sznajder JI. Hypoxia-induced endocytosis of Na,K-ATPase in alveolar epithelial cells is mediated by mitochondrial reactive oxygen species and PKC-zeta. J Clin Invest 2003; 111:1057–1064.

Alveolar Fluid Reabsorption During VILI

291

41. Bertorello AM, Katz AI. Short-term regulation of renal Na-K-ATPase activity: physiological relevance and cellular mechanisms. Am J Physiol 1993; 265:F743–F755. 42. Ridge KM, Dada L, Lecuona E, et al. Dopamine-induced exocytosis of Na, K-ATPase is dependent on activation of protein kinase C-epsilon and -delta. Mol Biol Cell 2002; 13:1381–1389. 43. Bertorello AM, Ridge KM, Chibalin AV, Katz AI, Sznajder JI. Isoproterenol increases Naþ-Kþ-ATPase activity by membrane insertion of alpha-subunits in lung alveolar cells. Am J Physiol 1999; 276:L20–L27. 44. Lecuona E, Garcia A, Sznajder JI. A novel role for protein phosphatase 2A in the dopaminergic regulation of Na,K-ATPase. FEBS Lett 2000; 481: 217–220. 45. Bertorello AM, Komarova Y, Smith K, et al. Analysis of Na(þ),K(þ)-ATPase motion and incorporation into the plasma membrane in response to G proteincoupled receptor signals in living cells. Mol Biol Cell 2003; 14:1149–1157. 46. Sznajder JI. Alveolar edema must be cleared for the acute respiratory distress syndrome patient to survive. Am J Respir Crit Care Med 2001; 163:1293–1294. 47. Naren AP, Cobb B, Li C, et al. A macromolecular complex of beta 2 adrenergic receptor, CFTR, and ezrin/radixin/moesin-binding phosphoprotein 50 is regulated by PKA. Proc Natl Acad Sci USA 2003; 100:342–346. 48. Fang X, Barbry P, Fukuda N, Matthay BA. Upregulation of isosmolar alveolar ﬂuid clearance in mice depends on CFTR. Pediatr Pulmonol 2000; 20:69A. 49. Verkman AS, Matthay MA, Song Y. Aquaporin water channels and lung physiology. Am J Physiol Lung Cell Mol Physiol 2000; 278:L867–L879. 50. Agre P, Kozono D. Aquaporin water channels: molecular mechanisms for human diseases. FEBS Lett 2003; 555:72–78. 51. King LS, Nielsen S, Agre P. Aquaporin-1 water channel protein in lung: ontogeny, steroid-induced expression, and distribution in rat. J Clin Invest 1996; 97:2183–2191. 52. Kim YH, Earm JH, Ma T, et al. Aquaporin-4 expression in adult and developing mouse and rat kidney. J Am Soc Nephrol 2001; 12:1795–1804. 53. Kreda SM, Gynn MC, Fenstermacher DA, Boucher RC, Gabriel SE. Expression and localization of epithelial aquaporins in the adult human lung. Am J Respir Cell Mol Biol 2001; 24:224–234. 54. Olivera WG, Ridge KM, Sznajder JI. Lung liquid clearance and Na,K-ATPase during acute hyperoxia and recovery in rats. Am J Respir Crit Care Med 1995; 152:1229–1234. 55. Comellas AP, Pesce LM, Azzam Z, Saldias FJ, Sznajder JI. Scorpion venom decreases lung liquid clearance in rats. Am J Respir Crit Care Med 2003; 167:1064–1067. 56. Saldias FJ, Comellas AP, Pesce L, Lecuona E, Sznajder JI. Dopamine increases lung liquid clearance during mechanical ventilation. Am J Physiol Lung Cell Mol Physiol 2002; 283:L136–L143. 57. Ware LB, Matthay MA. Alveolar ﬂuid clearance is impaired in the majority of patients with acute lung injury and the acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163:1376–1383.

292

Mutlu et al.

58. Matthay MA, Wiener-Kronish JP. Intact epithelial barrier function is critical for the resolution of alveolar edema in humans. Am Rev Respir Dis 1990; 142:1250–1257. 59. Ware LB, Golden JA, Finkbeiner WE, Matthay MA. Alveolar epithelial ﬂuid transport capacity in reperfusion lung injury after lung transplantation. Am J Respir Crit Care Med 1999; 159:980–988. 60. Saldias FJ, Lecuona E, Comellas AP, Ridge KM, Rutschman DH, Sznajder JI. Beta-adrenergic stimulation restores rat lung ability to clear edema in ventilator-associated lung injury. Am J Respir Crit Care Med 2000; 162:282–287. 61. Adir Y, Factor P, Dumasius V, Ridge KM, Sznajder JI. Na,K-ATPase gene transfer increases liquid clearance during ventilation-induced lung injury. Am J Respir Crit Care Med 2003; 168:1445–1448. 62. Waters CM, Ridge KM, Sunio G, Venetsanou K, Sznajder JI. Mechanical stretching of alveolar epithelial cells increases Naþ-Kþ-ATPase activity. J Appl Physiol 1999; 87:715–721. 63. Fisher JL, Margulies SS. Naþ-Kþ-ATPase activity in alveolar epithelial cells increases with cyclic stretch. Am J Physiol Lung Cell Mol Physiol 2002; 283:L737–L746. 64. Wirtz HR, Dobbs LG. Calcium mobilization and exocytosis after one mechanical stretch of lung epithelial cells. Science 1990; 250:1266–1269. 65. Corbridge TC, Wood LD, Crawford GP, Chudoba MJ, Yanos J, Sznajder JI. Adverse effects of large tidal volume and low PEEP in canine acid aspiration. Am Rev Respir Dis 1990; 142:311–315. 66. Al-Jamal R, Ludwig MS. Changes in proteoglycans and lung tissue mechanics during excessive mechanical ventilation in rats. Am J Physiol Lung Cell Mol Physiol 2001; 281:L1078–L1087. 67. Uhlig S. Mechanotransduction in the lung: ventilation-induced lung injury and mechanotransduction: stretching it too far? Am J Physiol Lung Cell Mol Physiol 2002; 282:L892–L896.

12 Interaction of VILI with Previous Lung Alterations

JEAN-DAMIEN RICARD

DIDIER DREYFUSS

Service de Re´animation Me´dicale, Assistance Publique-Hoˆpitaux de Paris, Hoˆpital Louis Mourier Colombes, France Inserm U 722, Paris 7-Denis Diderot Medical School Paris, France

Paris 7-Denis Diderot Medical School Paris, France Service de Re´animation Me´dicale, Assistance Publique-Hoˆpitaux de Paris, Hoˆpital Louis Mourier Colombes, France

GEORGES SAUMON EA 3512, IFR 02 Claude Bernard, Paris 7-Denis Diderot Medical School Paris, France

I. Introduction Initial description of ventilator-induced lung injury (VILI) was made by ventilating normal lungs with high peak inspiratory pressures (PIP) (1). However, mechanical ventilation is most often used in patients with diseased lungs. These lungs are inhomogenous, normal zones coexisting with edematous or atelectatic ones (2). Thus it is conceivable that inﬂation of lungs with heterogeneously distributed lesions may lead to greater regional stress and local overinﬂation than that of uniform uninjured ones. Mead et al. were the ﬁrst to conceptualize the increased risk of tissue injury during inﬂation when lungs had zones of atelectasis (3). They calculated that the pressure tending to expand an atelectatic region surrounded by a fully expanded lung would be approximately 140 cmH2O at a transpulmonary pressure of 30 cmH2O (3). They further speculated that ‘‘mechanical ventilators, by applying high transpulmonary pressure to the nonuniformly expanded lungs of some patients who would otherwise die of respiratory insufﬁciency, may cause the hemorrhage and the formation of hyaline membranes found in such patients’ lungs at death’’ (3). Other abnormalities that 293

294

Ricard et al.

may increase lung tissue stress are the presence of signiﬁcant areas excluded from the ventilation. In that case, the bulk of ventilation is delivered to smaller (and likely less damaged) lung volume that thus would be at a greater risk of overinﬂation. Using different models, many experimental studies have investigated this area of VILI. This chapter recalls the main characteristics of the models used to assess the inﬂuence of previous lung injury on VILI and reviews some of these studies.

II. Surfactant Depletion and Deactivation In a pivotal experimental study, Lachmann et al. described a model of neonate respiratory distress syndrome (RDS) (4) induced in the guinea pig by 10 bilateral lung lavage with 35 mL/kg warm (37 C) saline in an aim to remove surfactant. This procedure led to a dramatic decrease in oxygenation and impaired respiratory mechanics with increased resistance and decreased compliance. After several hours of ventilation, light microscopy ﬁndings included atelectasis and desquamation of bronchial and bronchiolar epithelium accompanied by formation of hyaline membranes. Electron microscopy ﬁndings included necrosis and desquamation of pneumocytes, and alveolar basement membrane denudation. This well-established and reproducible model helped examine the effects of high frequency oscillatory ventilation and of positive end-expiratory pressure (PEEP) during RDS. A. Effects of Conventional Mechanical Ventilation and High Frequency Oscillatory Ventilation on Premature and Surfactant-Deﬁcient Lungs

Studies on prematurely delivered lambs (5), baboons (6), and adult rabbits made surfactant deﬁcient by repeated saline lavage (7–9) indicate that the efﬁciency of high-frequency oscillatory ventilation (HFO) on lung lesions depends on the performance of a preliminary sustained static inﬂation (also called ‘‘lung conditioning’’) to recruit the greatest possible number of lung units before starting HFO (10). Hamilton et al. (7) compared oxygenation and lung pathology in rabbits with saline-lavaged lungs ventilated by conventional mechanical ventilation with a 6 cmH2O PEEP and HFO at similar mean airway pressure (15 cmH2O). Both groups underwent static inﬂation at 25 to 30 cmH2O for 15 seconds. HFO-treated animals had considerably higher PaO2. More importantly, whereas conventionally ventilated rabbits had extensive hyaline membrane formation, the lungs of HFO-treated animals had few, if any, hyaline membranes.

Interaction of VILI with Previous Lung Alterations

295

Meredith et al., working on premature baboons, showed that hyaline membrane disease was prevented when HFO was preceded by a recruitment maneuver (6). The importance of successful recruitment for preventing lung injury during HFO was illustrated by the severity of the changes in microvascular and alveolar permeability and histological damage, which were similar to those caused by conventionally ventilating premature newborn lambs, when recruitment was not successful (5). Failure to achieve recruitment was ascribed to the inability of premature lungs to secrete enough surfactant (5). Another study (8) also indicated the pivotal role of lung recruitment. Rabbits made surfactant-deﬁcient (by repeated lung lavage) were subjected to conventional mechanical ventilation with a PEEP (8 cmH2O) below the inﬂection point on the pressure–volume curve and a mean airway pressure of 18 to 19 cmH2O, or to HFO at two levels of mean airway pressure [9–10 and 15–16 cmH2O] resulting in low or high lung volumes. All animals underwent recruitment by static lung inﬂation at an airway pressure of 30 cmH2O for 15 seconds and were then connected to the conventional or HFO ventilator. Lung mechanical properties were better preserved in the HFO–high lung volume animals. Indeed, at the end of the experimental period (seven hours) lung compliance was signiﬁcantly greater in HFO– high lung volume animals than in those ventilated with HFO–low volume or conventional mechanical ventilation. Consequently, HFO–high lung volume animals had a lung volume above functional residual capacity (FRC), three times that of animals ventilated with HFO at low lung volume and ﬁve times that of animals conventionally ventilated. These preserved mechanical properties resulted in markedly better oxygenation. HFO–high lung volume animals had also considerably less hyaline membrane and bronchiolar epithelium necrosis. This study suggests that reopening an atelectasis-prone lung is not sufﬁcient to prevent injury due to shear stress when ventilation causes the repeated collapse and opening of terminal airways. It is thus important to keep the lung open (11) by applying sufﬁcient mean airway pressure during HFO. Avoiding large pressure–volume variations with HFO does not totally prevent lung injury if sufﬁcient FRC cannot be maintained. Prevention of VILI by HFO was essentially demonstrated in RDS models (surfactant deﬁciency). Its efﬁciency during other types of lung injury is largely unknown (10,12). B. Importance of Maintaining Lung Volume During Conventional Mechanical Ventilation: Effect of PEEP

The hypothesis that maintenance of an ‘‘open lung’’ during the whole ventilatory cycle (11) by setting an appropriate level of PEEP that prevents distal lung injury was also tested during conventional mechanical ventilation of diseased lungs (surfactant depletion, HCl instillation, and oleic-acid edema).

296

Ricard et al.

Sykes and coworkers (13,14) studied this issue by ventilating rabbits whose lungs were depleted of surfactant by lavage. PIP was 15 mmHg at the beginning of the experiment and 25 mmHg ﬁve hours later, because lung compliance decreased [tidal volume (VT) was set but not stated]. PEEP was adjusted so that FRC was either above or below the lower inﬂection point on the inspiratory limb of the pressure–volume curve. This gave PEEP levels of about 1 to 2 mmHg (below inﬂection) and 8 to 12 mmHg (above inﬂection). The mortality rates in the two groups were identical, but the arterial PaO2 was better preserved and there was less hyaline membrane formation in the high PEEP group (13,14). This lessening of pathological alterations occurred even when the mean airway pressures in the low- and high-PEEP groups were kept at the same level by adjusting the inspiratory/expiratory time ratio (14). Muscedere et al. (15) reported similar results for isolated, unperfused, lavaged rabbit lungs ventilated with a low [5–6 mL/kg body weight (BW)] tidal volume and with a PEEP set below or above the inﬂection point. However, Sykes and colleagues could not replicate these ﬁndings in rabbits with hydrochloric acid–injured lungs using the same ventilation settings (16). Whether the protective effect of PEEP during lung injury is restricted to the peculiar situation of surfactant depletion remains unsettled. A study in isolated rat lungs reported much greater increases in proinﬂammatory cytokines [tumor necrosis factor-alpha (TNF-a) and interleukin (IL)-1b] and in macrophage inﬂammatory protein (MIP)-2 in bronchoalveolar lavage (BAL) ﬂuid when ventilation was conducted at low end-expiratory lung volume (without PEEP) than with PEEP (17). The increase in MIP-2, however, was not conﬁrmed by the same team in a subsequent study using the same ventilator settings (18). Thus, it is conceivable that the protective effect of PEEP set above the lower inﬂection point of the pressure–volume curve is observed only in the very special context of surfactant deﬁciency but not during severe alveolar edema because lung instability and airspace collapse is observed only during the former. Using in vivo videomicroscopy, Nieman and colleagues (19–23) directly observed and quantiﬁed the dynamic changes in alveolar size throughout the ventilatory cycle during tidal ventilation, in normal lungs and in Tween surfactant–deactivated lungs. In normal lungs, they found that alveolar volume did not change appreciably during ventilation (21), in accordance with former ﬁndings from Wilson and Bachofen (24). Tidal change in alveolar shape and increase in alveolar size at endinspiration were observed in surfactant-deactivated lungs (21). This suggests alveolar ducts rather than alveoli shape change during ventilation in normal lungs, whereas it is the opposite in surfactant-deactivated lungs. The alveolar overdistension seen in surfactant-deactivated lungs may increase the risk for VILI. In a subsequent study (22), the same team documented the effect of increasing end-expiratory pressure. They found that application of PEEP to a surfactant-deactivated lung reversed the observed alveolar size increase,

Interaction of VILI with Previous Lung Alterations

297

Figure 1 Number of alveoli per microscopic ﬁeld (solid columns) and alveolar stability as measured by change in alveolar area during tidal ventilation (I–ED; hatched columns). Compared with before RM, the number of alveoli was signiﬁcantly greater during RM and after RM at both PEEP levels. However, those supported with higher PEEP demonstrated a signiﬁcant improvement in stability (low I–ED). Data represent mean SEM. Abbreviations: PEEP, positive end-expiratory pressures; RM, recruitment maneuver. Source: From Ref. 22.

leading to a return to control levels (Fig. 1). This stabilization of lung tissue may help explain the reduction of lung lesions observed with PEEP. III. Toxic Lung Injuries A. Oleic Acid

Injection of oleic acid in the central venous circulation produces a permeability type edema because of the release of inﬂammatory mediators and reactive oxygen species, and activation of coagulation in microvessels leading to acute endothelial and alveolar epithelial cell necrosis (25). Edema is not conﬁned to necrotic areas and may overﬂow to uninjured regions. Oleic acid increases vascular permeability of the pulmonary endothelium and extravascular lung water content. Grossman et al. found that the loss of ventilatable units, secondary to alveolar ﬂooding, was responsible for the decreased static lung compliance

298

Ricard et al.

observed in oleic acid–induced pulmonary edema (26). Obstruction of airways by edema foam rather than the collapse of terminal units is likely responsible for the loss of ventilated lung during pulmonary edema (27) (see chapter on ‘‘Lung Mechanics and Pathological Features During VILI’’). Hemodynamic consequences of oleic acid include mild to moderate pulmonary hypertension, due in part to vasoconstriction and vascular derecruitment in edematous lung regions, mediated by prostaglandins (25). Bowton and Kong (28), using the oleic acid model, showed that isolated rabbit lungs ventilated with 18 mL/kg VT had a signiﬁcantly greater weight gain than those ventilated with 6 mL/kg VT. Moreover, the latter had signiﬁcantly greater weight gains than lungs ventilated with the same small VT but not administered oleic acid. Hernandez et al. (29) showed that whereas low doses of oleic acid or 25 cmH2O PIP mechanical ventilation did not affect ﬁltration coefﬁcient and wet–dry ratio, the combination of both did (Fig. 2). The same group also reported that the ﬁltration coefﬁcient increase observed after high PIP [30–45 cmH2O] ventilation of isolated perfused rabbit lungs was more marked when surfactant was inactivated by dioctyl succinate instillation (30). Moreover, whereas light microscopic examination showed only mild abnormalities (minimal hemorrhage and vascular congestion) in the animals subjected to high PIP ventilation only or surfactant inactivation only, the combination of surfactant inactivation and high PIP ventilation caused severe damage (extensive hemorrhage, pulmonary edema, and formation of hyaline membranes). Thus, ventilation-induced lung edema seems to develop at lower airway pressures in already edematous lungs. These studies were performed in isolated lungs in which chemical and ventilator settings may not have the same consequences as in intact animals. B. ANTU

Alpha-naphthyl-thiourea (ANTU) induces a permeability-type pulmonary edema (31,32). Five hours after intraperitoneal injection, extravascular lung water increased by 50% in rats given 5 mg/kg ANTU in comparison with controls (33). The main ﬁnding of light microscopy was a widespread interstitial edema predominantly located in peribronchovascular cuffs while normal lung architecture was preserved. Electron microscopy ﬁndings mainly showed endothelial lesions. Endothelial cells were swollen and vacuolized. Small blebs, resulting from the detachment of alveolar capillary endothelium from the basement membrane, were observed (33). Two hours after intravenous injection, ANTU infusion caused moderate interstitial pulmonary edema of the permeability type. The respiratory system compliance was signiﬁcantly lower in rats given ANTU than in control animals (34). The effect of high PIP ventilation on ANTU injured lungs was investigated in rats by comparing different degrees of lung distention (34).

Interaction of VILI with Previous Lung Alterations

299

Figure 2 Effect of single (A) and combined (B) insults on the ﬁltration coefﬁcient of isolated rabbit lungs. A low dose of oleic acid or a moderately high peak pressure (24 cmH2O) alone failed to induce Kfc changes, whereas the combination of both insults was responsible for a signiﬁcant increase in Kfc. , p < 0.05. Abbreviation: Kfc, ﬁltration coefﬁcient. Source: From Ref. 29.

Mechanical ventilation resulted in a permeability edema whose severity depended on the tidal volume amplitude. It was possible to calculate how much mechanical ventilation would theoretically injure lungs diseased by ANTU by summing up the separate effect of mechanical ventilation alone or ANTU alone on edema severity. The results showed that in animals with lungs injured by ANTU and ventilated at high PIP and high VT (45 mL/kg BW), permeability edema was more severe than predicted, indicating

300

Ricard et al.

synergy between the two insults. Even minor alterations such as those produced by spontaneous ventilation during prolonged anesthesia (which degrades surfactant activity and promotes focal atelectasis) were sufﬁcient to synergistically increase the harmful effects of high-volume ventilation. The extent of lung mechanical properties deterioration prior to ventilation is a key factor in this synergy. The amount of edema produced by high-volume mechanical ventilation in the lungs of animals given ANTU or that had undergone prolonged anesthesia was inversely proportional to the respiratory system compliance measured at the very beginning of mechanical ventilation (34,35). The same observation was made with the volume of the upper inﬂection point of the pressure–volume curve of the respiratory system (35). The reason for this synergy requires clariﬁcation. The presence of local alveolar ﬂooding in animals given the most harmful ventilation protocol was the most evident difference from those ventilated with lower, less harmful tidal volumes (34). It is conceivable that edema foam in airways reduced the number of alveoli that received the tidal volume, exposing them to overinﬂation and rendering them more susceptible to injury, further reducing the aerated lung volume and resulting in positive feedback. To explore this possibility, alveolar ﬂooding was produced by instilling saline into the trachea of rats that were immediately ventilated with tidal volumes of up to 33 mL/kg (36). Flooding with saline did not signiﬁcantly affect microvascular permeability when tidal volume was low. As tidal volume was increased, capillary permeability alterations were larger in ﬂooded animals than in intact animals, reﬂecting further impairment of their endothelial barrier (Fig. 3). There was also a correlation between end-inspiratory airway pressure, the pressure at which was found the lower inﬂection point on the pressure–volume curve, and capillary permeability alterations in ﬂooded animals ventilated with a high tidal volume. Thus, the less compliant and recruitable the lung was after saline ﬂooding, the more severe were the changes in permeability caused by lung distention. These studies support the conclusion that the risk of overinﬂation is more important in edematous than in healthy lungs. Strategies to prevent VILI should oppose the synergy between ventilation and previous lung injury. C. Hydrochloric Acid

Since Mendelson’s classic description of acute respiratory failure following gastric content aspiration in women during labor, acid aspiration–induced injury has been recognized as important cause of acute respiratory distress syndrome (ARDS). Not surprisingly, many investigators have used hydrochloric acid as a model of acute lung injury. Hydrochloric acid damages the alveolar-capillary membrane and promotes polymorphonuclear neutrophil adhesion, activation and sequestration

Interaction of VILI with Previous Lung Alterations

301

Figure 3 Effect of increasing VT during mechanical ventilation for 10 minutes on lung permeability indices (extravascular albumin distribution space in lungs) of rats with intact lungs (open bars) or with alveolar ﬂooding (closed bars) produced by saline instillation. There was a moderate increase in the albumin space in intact rats at the larger VT. Lung ﬂooding did not produce signiﬁcant increases of albumin space when the VT was normal or moderately increased. A VT of 24 mL/kg signiﬁcantly increased albumin space. The increase in albumin space greatly exceeded additivity, indicating a positive interaction between the two insults. Key: þ, p < 0.05 as compared with other groups of intact rats; , p < 0.001, as compared with intact animals. Abbreviation: VT, tidal volume. Source: From Ref. 36.

through the release (among other mediators) of TNF-a and IL-8, and ﬁnally a permeability-type pulmonary edema with gas exchange deterioration. Usually, 2 to 4 mL/kg HCl (pH 1.5) are used to induce lung injury in rats. Severe hypoxemia and impaired respiratory mechanics rapidly follow HCl instillation. To study the effects of different VT and PEEP levels on VILI development in acid-injured lungs, Corbridge et al. compared a large VT and low PEEP strategy (30 mL/kg and 3 cmH2O) with a lower VT and high PEEP one (15 mL/kg and 12.5 cmH2O) in a canine model (37). End-inspiratory pressure and lung volume were kept similar in the two groups. They found that the large VT–low PEEP strategy resulted in more edema and venous mixture than the small VT–high PEEP one (37). Authors hypothesized that in the presence of hydrochloric acid–induced lung injury, the large VT–low PEEP strategy depleted surfactant and that PEEP not only redistributed existing edema and reopened previously collapsed alveoli but perhaps also protected against irreversible surfactant compression and subsequent depletion (37). Results obtained ex vivo by Sykes and colleagues (13,14) are in agreement with these, obtained in vivo.

302

Ricard et al. D. Hyperoxia

The deleterious effects of hyperoxia on lung function have been recognized in a few humans reports (38,39) and in numerous experimental studies (40,41). Hyperoxia-induced lung injury results from direct oxygen toxicity that promotes the release of inﬂammatory mediators in the lungs. These processes cause alveolar hyaline membrane formation, edema, hyperplasia and proliferation of type II alveolar epithelial cells, destruction of type I alveolar epithelial cells, interstitial ﬁbrosis, and pulmonary vascular remodeling. In vivo oxygen-induced lung injury is well characterized in rodents and has therefore been used as a valuable model of ARDS (42). Quinn et al. used the hyperoxia model to study the interaction between previous lung injury and mechanical ventilation (43). In their study, rats were ventilated either in 100% oxygen or in room air with 7 or 20 mL/kg VT for two hours. Lung wet–dry weight ratio was signiﬁcantly higher in animals exposed to hyperoxia and ventilated with 20 mL/kg VT than those ventilated with the same VT but in room air. As could have been expected, lung wet–dry weight ratio did not increase in animals ventilated with 7 mL/kg VT with room air (in comparison with non ventilated animals) but did in those in hyperoxia. In this model, and contrary to ﬁndings with ANTU (34), the effect of oxygen was additive and not synergistic (43). Using moderate hyperoxia (50% FiO2), Sinclair et al. conﬁrmed these results using 25 mL/kg VT in rabbits (44). They found no effect, however, when VT was only 10 mL/kg (44). It may be that the severity of previous lung injury and, in particular, the presence of zones of alveolar ﬂooding, which reduces the amount of ventilatable lung, modulates the harmful effect of mechanical ventilation with moderate to high VTs. IV. Inﬂammation and Infection: The Importance of Lung Priming and the Two-Hit Theory Along with a physical lung preinjury (one that overtly reduces ventilatable lung volume), studies have investigated the effect of inﬂammation and/or infection as a means to sensitize lungs to the deleterious effects of mechanical ventilation. Indeed, mechanical ventilation with overinﬂation may act as a ﬁrst hit (see below) and induce the release of large amounts of proinﬂammatory mediators into the lung and systemic circulation. Whether this release is a consequence or the cause of the lung injury observed during high tidal volume is still a matter of debate (45) and will not be addressed here. Nonetheless, the question arises as to whether even larger amounts of proinﬂammatory mediators can be released by injurious mechanical ventilation when the lung is preinjured. Bouadma et al. used the mesenteric ischemia-reperfusion model to address this issue (46). Mesenteric ischemiareperfusion is a well-established model of systemic inﬂammation and lung

Interaction of VILI with Previous Lung Alterations

303

injury (47). Ischemia-reperfusion is supposed to increase inﬂammatory response to a second stimulus, building up the two-hit hypothesis (48). In the Bouadma et al. study (46), lung concentrations of proinﬂammatory cytokines such as TNF-a, IL-1b or IL-6, and chemoattractants such MIP-2, after four hours of mechanical ventilation with high tidal volume were signiﬁcantly greater in animals with mesenteric ischemia-reperfusion than in sham animals. Interestingly, this difference in concentration of mediators was also present with noninjurious ventilation strategies (i.e., 6 and 10 mL/kg VT), without, however, always reaching statistical signiﬁcance (Fig. 4). Plasma concentrations of these mediators were also signiﬁcantly

Figure 4 Effect of mesenteric ischemia-reperfusion on lung cytokine concentrations. Comparison between groups of rats with mesenteric ischemia-reperfusion (ﬁlled circles) and sham operated rats (open circles) ventilated with the same modality. Lung cytokine concentrations were higher in mesenteric ischemia-reperfusion than in shamoperated rats for the same ventilatory modality. Key: , p < 0.05 vs. mesenteric ischemia-reperfusion, for the same ventilatory modality. Abbreviations: TNF, tumor necrosis factor; IL, interleukin; MIP, macrophage inﬂammatory protein; PEEP, positive end-expiratory pressure; VT, tidal volume. Source: From Ref. 46.

304

Ricard et al.

greater in animals with mesenteric ischemia-reperfusion than in sham animals, whatever the ventilation strategy. This study suggests that, in the presence of systemic inﬂammation (here induced by mesenteric ischemia-reperfusion), otherwise harmless mechanical ventilation strategies induce local and systemic release of inﬂammatory mediators. Savel et al. investigated the effect of two ventilatory strategies (6 and 15 mL/kg) on lung injury parameters of rabbits instilled beforehand with Pseudomonas aeruginosa (49). In the absence of P. aeruginosa instillation, they found that alveolar permeability to proteins was increased in animals ventilated with 15 mL/kg VT as compared with animals ventilated with 6 mL/kg VT at 240 minutes and remained signiﬁcantly elevated four hours later. The addition of P. aeruginosa to the right lung dramatically increased the epithelial permeability, in both the 15 and 6 mL/kg VT groups (49). Schortgen et al. studied the effect of various mechanical ventilation strategies [low (6 mL/kg) VT with or without PEEP (8 cmH2O), low VT in the left lateral position with PEEP; high VT (27 mL/kg) with Zero end-expiratory pressure (ZEEP) and ﬁnally partial liquid ventilation] on bacterial dissemination in rats with documented P. aeruginosa unilateral pneumonia (50). All mechanical ventilation strategies, with the exception of low VT–PEEP, promoted contralateral dissemination of P. aeruginosa. Overall bacterial dissemination was less in nonventilated controls and low VT–PEEP than in the

Figure 5 Effect of ventilation strategy on the overall bacterial dissemination. Overall dissemination was deﬁned as the percentage of positive left lung, spleen, or liver cultures. Key: , p < 0.05 vs. NV and LV/8. Abbreviations: (NV non ventilated controls, LV/0 ¼ 6 mL/kg VT ZEEP, LV/8 ¼ 6 mL/kg VT 8 cmH2O PEEP, HV/0 ¼ VT set such as end-inspiratory pressure was 30 cmH2O no PEEP, LLP/ 8 ¼ 6 mL/kg VT 8 cmH2O PEEP, rat in the left lateral position.) PLV, partial liquid ventilation; VT, tidal volume; LLP, left lateral position; NV, nonventilated; HV, high tidal volume; LV, low tidal volume. Source: From Ref. 50.

Interaction of VILI with Previous Lung Alterations

305

Figure 6 Changes in plasma TNF-a concentration after two hours of mechanical ventilation in rats ventilated with the different strategies. Abbreviations: (NV nonventilated controls, LV/0 ¼ 6 mL/kg VT ZEEP, LV/8 ¼ 6 mL/kg VT 8 cmH2O PEEP, HV/0 ¼ VT set such as end-inspiratory pressure was 30 cmH2O no PEEP, LLP/8 ¼ 6 mL/kg VT 8 cmH2O PEEP, rat in the left lateral position.) ND, not detectable; PLV, partial liquid ventilation; TNF-a; tumor necrosis factor-alpha; PEEP, positive end-expiratory pressure; VT, tidal volume. Source: From Ref. 50.

other groups (Fig. 5). Plasma TNF-a concentration increased signiﬁcantly after mechanical ventilation with no PEEP at both VT (Fig. 6) (50). These results suggest that in the setting of acute lung infection, ventilation even with a normally harmless VT may act as a ‘‘second hit’’ leading to the worsening of lung injury (49), increased bacterial dissemination, and TNF-a release (50). PEEP seems to reduce these deleterious effects (50) as previously suggested in other models (51,52). In the above-mentioned studies, mechanical ventilation always acted as the ‘‘second hit.’’ In one study, mechanical ventilation was used as the ﬁrst hit (53). In this study, Lin et al. examined the hypothesis that mechanical ventilation with a potentially injurious strategy would predispose animals to the detrimental effects of subsequent instillation of bacteria. Animals received mechanical ventilation with either 7 mL/kg VT and 5 cmH2O PEEP or 21 mL/kg VT and ZEEP for one hour (53). They were then instilled with P. aeruginosa and mortality rate was assessed up to 48 hours. Mortality rate tended to be higher in the high VT group than in the low VT group (40% vs. 28%) although this difference was not signiﬁcant. Blood cultures were more often positive in the high VT group than in the low one (33% vs. 11%, p < 0.05). Cytokines were also measured

306

Ricard et al.

and, surprisingly, blood TNF-a and lung and blood MIP-2 concentrations were lower in the rats ventilated with the high VT than in those with the low VT. These results suggest that mechanical ventilation can cause otherwise healthy lungs to become vulnerable to bacterial challenge, contributing to bacteremia (53). Whether this vulnerability occurs through an impaired host defense remains to be determined, although some clinical data seem to indicate that high VT mechanical ventilation may induce immunomodulation toward anti-inﬂammation (54,55), that may, conceptually, impede appropriate inﬂammatory response. V. Consequences of Previous Lung Injury on Lung Mechanics These issues are addressed in the chapter ‘‘Lung mechanics and pathological features during VILI.’’ VI. Counteracting Previous Lung Injury Reducing VT is obviously of paramount importance to avoid VILI. In the presence of preexisting lung injury, this may, however, be insufﬁcient, as shown in several studies where VT between 6 and 10 mL/kg are associated with increased VILI. Thus, investigators have tried to ﬁnd means to eliminate or at least reduce the amount of preexisting injury to minimize the deleterious effects of mechanical ventilation. Some of these approaches are reviewed. A. Counteracting Inﬂammation

As stated above, mechanical ventilation with overinﬂation may induce the release of large amounts of proinﬂammatory mediators into the lung and systemic circulation. Several attempts to counteract this release during injurious ventilation have been done, and found to reduce VILI. For example, the beneﬁt of intratracheal anti-TNF antibody administration on lung injury was investigated by Imai et al. in a saline-lavaged rabbit lung model (56). With the initiation of mechanical ventilation, saline lavage induced a dramatic drop in oxygenation in controls. In animals treated with anti-TNF antibodies, however, oxygen decrease was signiﬁcantly attenuated, and PaO2 remained around 400 mmHg at the end of the experiment while it dropped down to 100 mmHg in controls (56). Lung mechanics and histopathological changes were also milder in animals that received anti-TNF antibodies. As acknowledged by the authors, lung injury was not, however, completely abrogated by the treatment, suggesting involvement of other factors in the process of VILI. For example, CXC-chemokines (potent

Interaction of VILI with Previous Lung Alterations

307

neutrophils chemoattractants) such as MIP-2 have repeatedly been found elevated in the BAL ﬂuid of lungs submitted to high-volume ventilation (45). Belperio et al. investigated CXC-mediated lung injury during highstretch mechanical ventilation in mice (57). Following high-stretch and high-peak pressure ventilation, lung expression of KC and MIP-2 paralleled lung injury and neutrophil sequestration in comparison with low-stretch/ low-peak pressure ventilation. Inhibition of CXCR2/CXC chemokine– ligand interactions led to a marked reduction in neutrophil sequestration and lung injury. These ﬁndings were conﬁrmed using CXCR2–/– mice (57). These experiments thus support the notion that increased expression of KC and MIP-2 and their interaction with CXCR2 are important in the pathogeneses of VILI. Held et al. compared the effects of ventilation or lipopolysaccharide (LPS) on nuclear factor (NF)-kappa B activation, chemokine release, and cytokine release in isolated perfused lungs. They found that both LPS and ventilation with a high distending pressure caused translocation of NF-kappa B, which was abolished by pretreatment with the steroid dexamethasone. Both injuries resulted in similar increases in perfusate levels of a variety of inﬂammatory mediators, which were largely prevented by dexamethasone pretreatment. It is conceivable that such interferences may also be beneﬁcial in the presence of a previous inﬂammatory state. However, no such study has been yet done. The only works dealing with VILI prevention were aimed at ameliorating lung mechanical properties of injured lungs. B. Restoring Lung Mechanics

As detailed above, previous injury renders the lung more susceptible to mechanical ventilation because have been its mechanical properties altered. One way to avoid the aggravation of lung injury with mechanical ventilation in the presence of previous injury would be to restore lung mechanics. This could be achieved by administrating compounds with tensioactive properties, such as exogenous surfactant or perﬂuorocarbons. Exogenous Surfactant

Vazquez de Anda et al. studied the effect of exogenous surfactant administration during injurious mechanical ventilation. VILI was induced in rats by increasing PIP to 45 cmH2O without PEEP for 20 minutes. Animals were thereafter randomly divided into three groups: VILI-control (animals were killed immediately after the high PIP ventilation period), animals that underwent mechanical ventilation with a PIP of 30 and 10 cmH2O PEEP and no other treatment, and animals that were subjected to the same settings plus a bolus of surfactant. A BAL was performed to determine protein content, minimal surface tension, and surfactant composition in

308

Ricard et al.

the BAL ﬂuid. Oxygenation, lung mechanics, and surfactant function and composition were signiﬁcantly improved in the surfactant-treated group compared to the ventilated and nonventilated control groups. In this setting, administration of exogenous surfactant was able to restore impaired respiratory mechanics to a level comparable to that of healthy animals (58). Perﬂuorocarbons

As indicated above, one means by which the lung can be sensitized to the deleterious effects of mechanical ventilation is by reducing its ventilatable lung volume. Experimentally, this can be achieved by instilling saline into the lungs. By doing so, an otherwise undisruptive VT induces a signiﬁcant increase in lung microvascular permeability as assessed by the distribution space of albumin (36). Figure 3 shows that in the absence of alveolar ﬂooding, ventilating rats with 24 mL/kg VT does not increase their distribution space of albumin in comparison with rats ventilated with 7 mL/kg VT. After instillation of saline, however, this VT becomes harmful, as assessed by a signiﬁcant increase in albumin space. Any further increase in VT considerably worsens lung injury (Fig. 3) (36). Administration of perﬂubron partly obviates the detrimental effect of saline instillation (Fig. 7) as assessed by a signiﬁcant reduction in distribution space of albumin in animals with alveolar ﬂooding given perﬂubron in comparison with those not given

Figure 7 Effect of PFC instillation on indices of permeability pulmonary edema in rats ventilated with a VT of 33 mL/kg. Flooding increased albumin space (p < 0.001). PFC given as a bolus dose before ﬂooding (A), by slow infusion before ﬂooding (B), or as a bolus dose after ﬂooding (C) resulted in signiﬁcant decreases in albumin space (p < 0.001), whose values remained, however, higher than in controls (p < 0.05). Closed circles with error bars indicate mean SEM. Abbreviations: VT, tidal volume; PFC, perﬂuorocarbons. Source: From Ref. 36.

Interaction of VILI with Previous Lung Alterations

309

Figure 8 Correlation between the lower inﬂection point pressure and albumin space in controls (open circles), ﬂooded animals (closed circles), and animals given a bolus dose of perﬂubron before ﬂooding (open squares), a slow infusion before ﬂooding (open triangles), or a bolus dose after ﬂooding (open diamonds). All animals were ventilated with a VT of 33 mL/kg. Segmented regression analysis revealed that the best ﬁt was obtained with two joined linear segments. The slope of the ﬁrst segment is essentially zero. Animals in which perﬂubron reduced the lower inﬂection point pressure had normal or near-normal values for albumin space. There was a threshold for a pressure value of around 15 cmH2O. Abbreviation: VT, tidal volume. Source: From Ref. 36.

perﬂubron. This beneﬁcial effect is obtained through an improvement in lung mechanics properties (reduction in the pressure of the lower inﬂection point and normalization of maximal airway pressure). Restoration of lung mechanics to explain the beneﬁcial effects of perﬂubron was further conﬁrmed by the correlation observed between lung injury and the lower inﬂection point (Fig. 8). Similar results were obtained with ANTU as a model of preinjury (59). This study showed that perﬂubron was able to eliminate the synergy between ANTU and high-volume mechanical ventilation by restoring lung mechanics. Indeed, perﬂubron administration signiﬁcantly improved respiratory mechanics (decrease in end-inspiratory pressure and increase in respiratory system compliance) and reduced the ventilation-induced permeability alterations in animals exposed to both insults to the level observed in control rats that were not ventilated (59).

VII. Clinical Considerations All patients with ARDS are, by deﬁnition, in the setting of previously injured lungs submitted to mechanical ventilation. Therefore, given the bulk

310

Ricard et al.

of evidence recalled above, use of an otherwise harmless VT should be, in the presence of ARDS, potentially deleterious. This hypothesis has been convincingly veriﬁed in the ARDS net study (60) where patients ventilated with 12 mL/kg VT had a signiﬁcantly greater mortality than those ventilated with 6 mL/kg. Moreover, patients ventilated with 6 mL/kg VT had greater reduction in plasma levels of IL-6 and IL-8 than did those ventilated with 12 mL/kg (60,61). In the absence of previous lung injury, an even larger VT (15 mL/kg) did not lead to a signiﬁcant systemic release of proinﬂammatory cytokines in comparison with 6 mL/kg VT in patients undergoing general anesthesia for elective surgery (62). This is in agreement with the recommended use of VT ranging between 10 and 15 mL/kg in patients with near-normal lungs (63). If there is no doubt that VT should be reduced in the presence of ARDS, use of high levels of PEEP remains controversial. In the same way as some experimental studies have shown that response to PEEP and recruitment is model dependent (64), human studies indicate that this is also the case in the clinical setting, where response to PEEP may differ depending on the pulmonary or extrapulmonary origin of ARDS (65). Ventilatory strategies aiming at further reducing VILI should therefore perhaps take into account the etiology of ARDS. References 1. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110:556–565. 2. Gattinoni L, Caironi P, Pelosi P, Goodman LR. What has computed tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001; 164:1701–1711. 3. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28:596–608. 4. Lachmann B, Robertson B, Vogel J. In vivo lung lavage as an experimental model of the respiratory distress syndrome. Acta Anaesth Scand 1980; 24:231–236. 5. Solimano A, Bryan AC, Jobe A, Ikegami M, Jacobs H. Effects of highfrequency and conventional ventilation on the premature lamb lung. J Appl Physiol 1985; 59:1571–1577. 6. Meredith KS, DeLemos RA, Coalson JJ, et al. Role of lung injury in the pathogenesis of hyaline membrane disease in premature baboons. J Appl Physiol 1989; 66:2150–2158. 7. Hamilton PP, Onayemi A, Smyth JA, et al. Comparison of conventional and high-frequency ventilation: oxygenation and lung pathology. J Appl Physiol 1983; 55:131–138. 8. McCulloch PR, Forkert PG, Froese AB. Lung volume maintenance prevents lung injury during high frequency oscillatory ventilation in surfactant-deﬁcient rabbits. Am Rev Respir Dis 1988; 137:1185–1192.

Interaction of VILI with Previous Lung Alterations

311

9. Sugiura M, McCulloch PR, Wren S, Dawson RH, Froese AB. Ventilator pattern inﬂuences neutrophil inﬂux and activation in atelectasis-prone rabbit lung. J Appl Physiol 1994; 77:1355–1365. 10. Froese AB, Bryan AC. High frequency ventilation. Am Rev Respir Dis 1987; 135:1363–1374. 11. Lachmann B. Open up the lung and keep the lung open. Intensive Care Med 1992; 18:319–321. 12. Slutsky AS. High frequency ventilation. Intensive Care Med 1991; 17:375–376. 13. Argiras EP, Blakeley CR, Dunnill MS, Otremski S, Sykes MK. High peep decreases hyaline membrane formation in surfactant deﬁcient lungs. Br J Anaesth 1987; 59:1278–1285. 14. Sandhar BK, Niblett DJ, Argiras EP, Dunnill MS, Sykes MK. Effects of positive end-expiratory pressure on hyaline membrane formation in a rabbit model of the neonatal respiratory distress syndrome. Intensive Care Med 1988; 14:538–546. 15. Muscedere JG, Mullen JB, Gan K, Slutsky AS. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med 1994; 149:1327–1334. 16. Sohma A, Brampton WJ, Dunnill MS, Sykes MK. Effect of ventilation with positive end-expiratory pressure on the development of lung damage in experimental acid aspiration pneumonia in the rabbit. Intensive Care Med 1992; 18:112–117. 17. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99:944–952. 18. Whitehead TC, Zhang H, Mullen B, Slutsky AS. Effect of mechanical ventilation on cytokine response to intratracheal lipopolysaccharide. Anesthesiology 2004; 101:52–58. 19. McCann UG 2nd, Schiller HJ, Carney DE, Gatto LA, Steinberg JM, Nieman GF. Visual validation of the mechanical stabilizing effects of positive end-expiratory pressure at the alveolar level. J Surg Res 2001; 99:335–342. 20. Schiller HJ, McCann UG 2nd, Carney DE, Gatto LA, Steinberg JM, Nieman GF. Altered alveolar mechanics in the acutely injured lung. Crit Care Med 2001; 29:1049–1055. 21. Steinberg J, Schiller HJ, Halter JM, et al. Tidal volume increases do not affect alveolar mechanics in normal lung but cause alveolar overdistension and exacerbate alveolar instability after surfactant deactivation. Crit Care Med 2002; 30:2675–2683. 22. Halter JM, Steinberg JM, Schiller HJ, et al. Positive end-expiratory pressure after a recruitment maneuver prevents both alveolar collapse and recruitment/ derecruitment. Am J Respir Crit Care Med 2003; 167:1620–1626. 23. Steinberg JM, Schiller HJ, Halter JM, et al. Alveolar instability causes early ventilator-induced lung injury independent of neutrophils. Am J Respir Crit Care Med 2004; 169:57–63. 24. Wilson TA, Bachofen H. A model for mechanical structure of the alveolar duct. J Appl Physiol 1982; 52:1064–1070. 25. Schuster D. Clinical lessons form the oleic acid model of acute lung injury. Am J Respir Crit Care Med 1994; 149:245–260.

312

Ricard et al.

26. Grossman RF, Jones JG, Murray JF. Effects of oleic acid-induced pulmonary edema on lung mechanics. J Appl Physiol 1980; 48:1045–1051. 27. Martynowicz MA, Minor TA, Walters BJ, Hubmayr RD. Regional expansion of oleic acid-injured lungs. Am J Respir Crit Care Med 1999; 160:250–258. 28. Bowton DL, Kong DL. High tidal volume ventilation produces increased lung water in oleic acid-injured rabbit lungs. Crit Care Med 1989; 17:908–911. 29. Hernandez LA, Coker PJ, May S, Thompson AL, Parker JC. Mechanical ventilation increases microvascular permeability in oleic acid-injured lungs. J Appl Physiol 1990; 69:2057–2061. 30. Coker PJ, Hernandez LA, Peevy KJ, Adkins K, Parker JC. Increased sensitivity to mechanical ventilation after surfactant inactivation in young rabbit lungs. Crit Care Med 1992; 20:635–640. 31. Richter CP. The physiology and cytology of pulmonary edema and pleural effusion produced in rats by alpha-naphthyl thiourea (ANTU). J Thorac Surg 1952; 23:66–91. 32. Cunningham AL, Hurley JV. Alpha-naphthyl-thiourea-induced pulmonary oedema in the rat: a topographical and electron-microscope study. J Pathol 1972; 106:25–35. 33. Vivet P, Brun-Pascaud M, Mansour H, Pocidalo JJ. Non-hypoxaemic pulmonary oedema induced by alpha-naphthyl thiourea in the rat. Br J Exp Pathol 1983; 64:361–366. 34. Dreyfuss D, Soler P, Saumon G. Mechanical ventilation-induced pulmonary edema. Interaction with previous lung alterations. Am J Respir Crit Care Med 1995; 151:1568–1575. 35. Martin-Lefevre L, Ricard JD, Roupie E, Dreyfuss D, Saumon G. Signiﬁcance of the changes in the respiratory system pressure-volume curve during acute lung injury in rats. Am J Respir Crit Care Med 2001; 164:627–632. 36. Dreyfuss D, Martin-Lefevre L, Saumon G. Hyperinﬂation-induced lung injury during alveolar ﬂooding in rats: effect of perﬂuorocarbon instillation. Am J Respir Crit Care Med 1999; 159:1752–1757. 37. Corbridge TC, Wood LDH, Crawford GP, Chudoba MJ, Yanos J, Sznadjer JI. Adverse effects of large tidal volume and low PEEP in canine acid aspiration. Am Rev Respir Dis 1990; 142:311–315. 38. Nash G, Blennerhassett JB, Pontoppidan H. Pulmonary lesions associated with oxygen therapy and artiﬁcial ventilation. N Engl J Med 1967; 276: 368–374. 39. Davis WB, Rennard SI, Bitterman PB, Crystal RG. Pulmonary oxygen toxicity. Early reversible changes in human alveolar structures induced by hyperoxia. N Engl J Med 1983; 309:878–883. 40. Jenkinson SG. Oxygen toxicity. New Horiz 1993; 1:504–511. 41. Carvalho CR, de Paula Pinto Schettino G, Maranhao B, Bethlem EP. Hyperoxia and lung disease. Curr Opin Pulm Med 1998; 4:300–304. 42. Pagano A, Barazzone-Argiroffo C. Alveolar cell death in hyperoxia-induced lung injury. Ann N Y Acad Sci 2003; 1010:405–416. 43. Quinn D, Tager A, Joseph PM, Bonventre JV, Force T, Hales CA. Stretchinduced mitogen-activated protein kinase activation and interleukin-8 production in type II alveolar cells. Chest 1999; 116:89S–90S.

Interaction of VILI with Previous Lung Alterations

313

44. Sinclair SE, Altemeier WA, Matute-Bello G, Chi EY. Augmented lung injury due to interaction between hyperoxia and mechanical ventilation. Crit Care Med 2004; 32:2496–2501. 45. Dreyfuss D, Ricard J-D, Saumon G. On the physiologic and clinical relevance of lung-borne cytokines during ventilator-induced lung injury. Am J Respir Crit Care Med 2003; 167:1467–1471. 46. Bouadma L, Schortgen F, Ricard JD, Martet G, Dreyfuss D, Saumon G. Ventilation strategy affects cytokine release after mesenteric ischemia-reperfusion in rats. Crit Care Med 2004; 32:1563–1569. 47. Koike K, Moore FA, Moore EE, Poggetti RS, Tuder RM, Banerjee A. Endotoxin after gut ischemia/reperfusion causes irreversible lung injury. J Surg Res 1992; 52:656–662. 48. Moore FA, Moore EE. Evolving concepts in the pathogenesis of postinjury multiple organ failure. Surg Clin North Am 1995; 75:257–277. 49. Savel RH, Yao EC, Gropper MA. Protective effects of low tidal volume ventilation in a rabbit model of Pseudomonas aeruginosa-induced acute lung injury. Crit Care Med 2001; 29:392–398. 50. Schortgen F, Bouadma L, Joly-Guillou ML, Ricard JD, Dreyfuss D, Saumon G. Infectious and inﬂammatory dissemination are affected by ventilation strategy in rats with unilateral pneumonia. Intensive Care Med 2004; 30:693–701. 51. Nahum A, Hoyt J, Schmitz L, Moody J, Shapiro R, Marini JJ. Effect of mechanical ventilation strategy on dissemination of intratracheally instilled Escherichia coli in dogs. Crit Care Med 1997; 25:1733–1743. 52. Verbrugge SJ, Sorm V, vant Veen A, Mouton JW, Gommers D, Lachmann B. Lung overinﬂation without positive end-expiratory pressure promotes bacteremia after experimental Klebsiella pneumoniae inoculation. Intensive Care Med 1998; 24:172–177. 53. Lin CY, Zhang H, Cheng KC, Slutsky AS. Mechanical ventilation may increase susceptibility to the development of bacteremia. Crit Care Med 2003; 31: 1429–1434. 54. Stuber F, Wrigge H, Schroeder S, et al. Kinetic and reversibility of mechanical ventilation-associated pulmonary and systemic inﬂammatory response in patients with acute lung injury. Intensive Care Med 2002; 28:834–841. 55. Pugin J. Is the ventilator responsible for lung and systemic inﬂammation? Intensive Care Med 2002; 28:817–819. 56. Imai Y, Kawano T, Iwanoto S, Nakagawa S, Takata M, Miyasaka K. Intratracheal anti-tumor necrosis factor-alpha antibody attenuates ventilator-induced injury in rabbits. J Appl Physiol 1999; 87:510–515. 57. Belperio JA, Keane MP, Burdick MD, et al. Critical role for CXCR2 and CXCR2 ligands during the pathogenesis of ventilator-induced lung injury. J Clin Invest 2002; 110:1703–1716. 58. Vazquez de Anda GF, Lachmann RA, Gommers D, Verbrugge SJ, Haitsma J, Lachmann B. Treatment of ventilation-induced lung injury with exogenous surfactant. Intensive Care Med 2001; 27:559–565. 59. Iserin F, Ricard J-D, Dreyfuss D, Saumon G. Partial liquid ventilation (PLV) reduces high volume induced pulmonary edema in rats [abstract]. Am J Respir Crit Care Med 2001; 163:A764.

314

Ricard et al.

60. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342:1301–1308. 61. Parsons PE, Eisner MD, Thompson BT, et al. Lower tidal volume ventilation and plasma cytokine markers of inﬂammation in patients with acute lung injury. Crit Care Med 2005; 33:1–6. 62. Wrigge H, Zinserling J, Stuber F, et al. Effects of mechanical ventilation on release of cytokines into systemic circulation in patients with normal pulmonary function. Anesthesiology 2000; 93:1413–1417. 63. Mador M. Assist-control ventilation. In: Tobin MJ, ed. Principles and Practice of Mechanical Ventilation. New York: McGraw-Hill, 1994:207–219. 64. Kloot TEVd, Blanch L, Melynne Youngblood A, et al. Recruitment maneuvers in three experimental models of acute lung injury. Effect on lung volume and gas exchange. Am J Respir Crit Care Med 2000; 161:1485–1494. 65. Gattinoni L, Pelosi P, Suter PM, Pedoto A, Vercesi P, Lissoni A. Acute respiratory distress syndrome caused by pulmonary and extrapulmonary disease. Different syndromes? Am J Respir Crit Care Med 1998; 158:3–11.

13 Biological Markers of Ventilator-Induced Lung Injury

THOMAS R. MARTIN

MICHAEL A. MATTHAY

Pulmonary Research Laboratories, VA Puget Sound Health Care System, and Division of Pulmonary and Critical Care Medicine, Department of Medicine, University of Washington School of Medicine Seattle, Washington, U.S.A.

Cardiovascular Research Institute, University of California at San Francisco San Francisco, California, U.S.A.

I. Introduction Mechanical ventilation is a lifesaving treatment in patients with acute respiratory failure, regardless of whether respiratory failure is associated with acute hypoxemia, acute hypercarbia, or both. Ventilator-induced lung injury (VILI) occurs when mechanical breaths overdistend the alveolar units (1,2). In patients with normal lungs, such as patients with neuromuscular disease or injuries, or patients with acute poisonings who have not aspirated, mechanical ventilation with relatively large tidal volumes (e.g., 10–12 mL/kg measured body weight) does not seem to harm the lungs. In fact, some recommendations for ventilating patients with neuromuscular disease call for the use of larger tidal volumes (> 10 mL/kg) to prevent atelectasis. However, when major areas of alveolar collapse occur, as in acute lung injury (ALI), the effective alveolar volume is substantially reduced, and a set tidal volume of 10 mL/kg body weight may be equivalent to a set tidal volume of 20 mL/kg body weight or greater in a reduced alveolar space. Such relatively high tidal volumes have the potential to overdistend and damage the alveolar walls, but as yet there is no direct way to 315

316

Martin and Matthay

estimate the effective alveolar volume in order to appropriately set the tidal volume. Investigators have worked to identify biomarkers of VILI so that protective measures can be initiated as soon as possible to reduce the chances of further lung injury. Biological markers also could be used to make predictions about prognosis and to stratify patients into different risk groups in order to apply appropriate new treatments. Lastly, biological markers are useful in studying the pathophysiology of lung injury. Efforts to identify markers of VILI have overlapped with broader efforts to identify markers of ALI, as virtually all patients with ALI are treated with mechanical ventilation. Although the acute respiratory distress syndrome (ARDS) in adults was ﬁrst identiﬁed in 1967, the speciﬁc biological markers of injury to the lungs still need to be identiﬁed (3,4). In contrast, sensitive and speciﬁc markers of injury exist for other key organs, including the heart, liver, kidneys, pancreas, skeletal muscle and others which are affected in patients in whom ALI and VILI occur. Markers of lung injury should be present in patients with ALI and absent in those without ALI. To be conﬁdent that a marker of ALI is also a marker of VILI, the marker should correlate with the degree of distension applied to the alveolar walls. For example, markers of VILI should increase in patients ventilated with higher tidal volumes, and higher alveolar pressures. Importantly, ﬁndings about biomarkers in observational studies using preexisting cohorts of patients must be validated prospectively in additional groups of patients, preferably patients randomized to different types of treatments. Samples from recent clinical trials of different ventilator strategies have provided an opportunity to study the relationship between biological markers and different levels of alveolar distension, and to perform both observational and prospective studies. The purpose of this chapter is to review the progress that has been made in evaluating biological markers of ALI and VILI since the last comprehensive review of markers of ALI (5). In addition, we will review new technological developments that promise to improve signiﬁcantly on the current biological markers, such as the use of proteomics to identify groups of marker proteins.

II. Rationale for Biological Markers of VILI Clinical predictors of ALI have been used for more than 25 years to separate patients into broad risk categories for the onset and prognosis of ALI. Sepsis syndrome carries the highest single risk for the onset of lung injury and mortality (6,7). The presence of two or more clinical risks increases the likelihood of ALI and signiﬁcantly worsens the likelihood of death once ALI occurs (8). While clinical predictors are useful, they are imprecise, and attempts have been made to improve clinical prediction by developing and incorporating biological markers of lung injury (Table 1).

Biological Markers of VILI

317

Table 1 Examples of Biological Markers of Lung Injury in Humans, Grouped by Anatomic Location Location Alveolar epithelium Airway epithelium Alveolar space

Biological markers KL-6 glycoprotein Surfactant proteins SP-A, SP-B, SP-D None Neutrophils Neutrophil myeloperoxidase Collagenase Metalloproteinases Procollagen peptide III

Interstitial compartment Vascular VWF antigen endothelium Vascular space Plasma cytokines: IL-8, IL-6, IL-10

Validated in humans Ya Y (SP-D) a

References (33) (11,29–31)

N N N N Y

(43) (94) (50,94) (51,95) (14,15)

Y

(42)

Ya

(77)

Y means that the marker is associated with onset or outcome of ALI in humans. a The marker is reduced by a protective ventilation strategy in humans. For a more comprehensive listing of earlier studies, see Ref. 5. Abbreviations: SP, surfactant proteins; VWF, von Willebrand factor; IL, interleukin; ALI, acute lung injury.

Physiological markers have also been proposed, and one of the earliest was the ratio of the PaO2 to FiO2 (P/F ratio). However, while the P/F ratio is a deﬁning characteristic for ALI, it is not a sensitive marker of mortality. In the ﬁrst study conducted by the National Institutes of Health (NIH) Acute Respiratory Distress Syndrome Network (ARDSNet), oxygenation was actually worse in the subgroup of patients treated with low tidal volume ventilation, who had the best prognosis (9). Recently, the dead space fraction (Vd/Vt) has been found to predict mortality and may be a more important physiological measure than oxygenation alone (10). Marked increases in shunt fraction can increase the calculated Vd/Vt, so that in patients with severe lung injury, the Vd/Vt is likely to reﬂect a combination of increased dead space and increased shunt fraction. Static compliance and peak airway pressure have also been measured, but are insensitive markers of outcome. Although these physiological markers reﬂect altered function in the lungs, they provide little insight into the actual pathogenesis of the injury. Biological markers of injury include biochemical constituents (typically proteins, lipids, or cells) that are measurable in biological ﬂuids before or after the onset of injury, in concentrations that parallel the course of injury, and are plausibly related to the pathophysiological process. In the lungs and other tissues, the structural constituents of the tissue that are

318

Martin and Matthay

released into local or systemic ﬂuids during injury or repair can be used as the markers of injury. For example, troponin is a structural protein in myocardial cells that leaks into the circulation when cardiac muscle is injured. Alternatively, normal constituents of alveolar ﬂuids may change in the setting of injury, or escape from the alveolar spaces into the systemic circulation before or after the onset of lung injury. Surfactant proteins (SP)-D and -A are examples, because the concentration of SP-D in the airspaces falls with lung injury, and both SP-A and SP-D escape from injured airspaces and circulate in plasma (11). Because tissue injury is associated with acute inﬂammation, the inﬂammatory markers in the alveolar spaces and the systemic circulation have been tested as markers of lung injury (5,12,13). Biological markers of the repair process have also been evaluated, as collagen production begins soon after the onset of injury and is detectable for more than a week after the onset of injury in patients who remain mechanically ventilated (14,15). Biological markers can be measured in lung ﬂuids at the site of injury, or in the plasma compartment, reﬂecting the markers that escape from the local lung environment. The plasma compartment is easier to sample, but reﬂects events that are happening not only in the lungs, but also in the entire body. Sampling the lung compartment yields markers that are less affected by systemic events and that are measurable in much higher concentrations than in the plasma. Sampling edema ﬂuid directly from the airways is relatively easy, but edema ﬂuid is present only at the onset of lung injury, and only in patients with more severe lung injury. Bronchoscopy with bronchoalveolar lavage (BAL) can be performed at any time in an intubated patient, but the samples are diluted and the bronchoscopy itself is an invasive and expensive procedure outside of a research setting. Mini-BAL with a balloon-tipped or mushroom-tipped catheter inserted blindly into the lower airway and wedged into a lower lobe bronchus has also been used, but the correlations between the cell counts and the total proteins in mini-BAL and traditional bronchoscopic lavage are poor. The mini-BAL procedure may be useful to evaluate infection in the lower airways, but it is not recommended for sampling biological markers in the lungs.

III. Recent Progress in Identifying Biological Markers of VILI A. Structural Markers

For some tissues, like the heart and liver, speciﬁc intracellular proteins that reﬂect tissue injury have been identiﬁed (e.g., troponin and hepatocellular enzymes), but similar markers have not been validated for human lungs. Structural markers in the lungs include the membrane and intracellular proteins from epithelial cells in the alveolar walls and/or in the small

Biological Markers of VILI

319

airways; the junctional proteins that connect epithelial cells and are released when epithelial junctions are disrupted; the matrix glycoproteins released when basement membranes or interstitial matrix is fractured or degraded; and even the endothelial proteins released when the lung microvascular endothelium is damaged [reviewed in (16)]. Because VILI is likely to involve overdistension of some alveolar units with the rupture of cellular membranes, the search for structural markers of lung injury has a priori merit, but signiﬁcant progress must still be made. An antigen on rat Type I pneumocytes (RTI40) was identiﬁed by Dobbs et al., and shown to correlate with experimental lung injury in rats (17–21). The RTI40 antigen is interesting because it is not expressed on freshly isolated alveolar Type II cells, but is detectable as cultured Type II cells ﬂatten and assume more of a Type I morphology (17). A related antigen has been identiﬁed in human lungs, but its use has not been thoroughly validated, in part because of the lack of a puriﬁed standard for use in immunoassays (22). Several other integral membrane proteins have been identiﬁed on rodent Type II cells and Clara cells, including a protein called p172 (23), aminopeptidase N (24), alkaline phosphatase (25), and a glycoprotein called pneumocin (26), but these have not been validated as markers of lung injury in humans. A recently described protein antigen on rat Type II cells, designated MMC4, is expressed on the apical surface of alveolar Type II cells and not on Type I cells, but it is also expressed in rat kidney and gut epithelium (27). If a human counterpart is identiﬁed, it could be useful in measuring Type II epithelial injury in lung specimens, but it may not be speciﬁc for lung injury if measured in the plasma. Additional markers of structural injury include intracellular proteins secreted during normal homeostasis in the lungs, but which are released in large quantities when lung epithelial cells are disrupted. Examples include surfactant proteins (SP) and mucus glycoproteins released from Type II cells and airway glandular epithelium, respectively, as well as the von Willebrand factor (VWF) antigen released from Weibel–Palade bodies in endothelial cells. Doyle et al. were the ﬁrst to report the appearance of surfactant-associated proteins in the plasma of patients with lung injury, and that SP-A, SP-B and the SP-B/SP-A ratio, were all inversely related to oxygenation and lung compliance (28,29). In a small single-center study, the plasma SP-B concentration on admission was a better predictor of the onset of ARDS than was the lung injury score (30). In two different singlecenter studies, the alveolar concentration of SP-D was a marker of the severity of alveolar epithelial injury and outcome (11,31). The BAL and edema ﬂuid SP-D concentrations were lower in patients who died, and also in those who were on the ventilator for a longer time and had more episodes of organ failure. In these relatively small studies, the plasma SP-A and SP-D concentrations did not predict outcome, but the SP-A concentrations were higher in patients with more severe alveolar epithelial injury, consistent with the ﬁndings of Doyle et al. (28,31).

320

Martin and Matthay

The KL-6 protein is a mucus glycoprotein (MUC-1) originally isolated from a lung cancer cell line that is expressed in human Type II cells and airway glandular epithelium (32,33). Ishizaka et al. found that the KL-6 protein is released into the edema ﬂuids of patients with ALI, presumably reﬂecting type II injury and/or proliferation as part of epithelial repair (33). The alveolar and plasma concentrations of KL-6 were signiﬁcantly higher in patients who died, suggesting a strong relationship between KL-6 and alveolar epithelial injury. The VWF antigen is a marker of endothelial cell activation and disruption, which is released from the intracellular Weibel–Palade bodies, as well as from the platelets (34,35). The concentration of VWF antigen was found to be higher in the plasma of patients with systemic sepsis, reﬂecting endothelial cell abnormalities, and was a predictor of the onset of ARDS (36). Several subsequent studies in other centers did not conﬁrm this association (37–40), but a more recent single-center study of 51 patients with sepsis found that plasma VWF antigen was independently associated with mortality (41). In a recent large multicenter trial of ALI, the concentration of VWF antigen was signiﬁcantly higher at the onset of illness in patients who later died, and high VWF antigen concentrations were associated with more days on mechanical ventilation, regardless of whether or not the patients had clinical sepsis (42). Thus, the weight of the evidence supports the conclusion that the VWF concentration is a good marker of endothelial activation and/or injury in patients with lung injury, and that higher concentrations are associated with a worse outcome. B. Markers of Inﬂammation

Acute inﬂammation is a prominent feature of ALI. Initially, investigators hypothesized that uncontrolled inﬂammation caused ALI, and efforts were made to identify single markers of the inﬂammatory response that predicted the onset of ALI in patients at risk and the outcome of ALI once it occurred. Many of the initial studies were reviewed by Pittet et al. (5), and at that time, none of the single markers of the inﬂammatory response in BAL ﬂuid or plasma were consistent predictors of either the onset or the outcome of ALI. These included leukocytes in BAL ﬂuid, proinﬂammatory cytokines such as tumor necrosis factor (TNF)a and interleukins, markers of the coagulation pathways, and others. Virtually all of these earlier studies were limited by relatively small sample sizes, and by the heterogeneous causes of ALI. Studies of cellular proﬁles using sequential BAL in patients with ARDS showed that the inﬂammatory response tended to resolve with time in patients with ARDS following trauma, whereas it persisted in patients with sepsis as the cause of ARDS (43). In patients with persistent ARDS, the cell and cytokine proﬁles were very similar on days 1 and 3 after the

Biological Markers of VILI

321

onset of ARDS and single cytokine measurements at these early times did not predict survival. However, on day 7 after the onset of ARDS, cytokines such as interleukin (IL)-1b were signiﬁcantly higher in patients who later died, but there was signiﬁcant overlap in the individual values, making the IL-1b values poor predictors of late outcome (12). Meduri et al. found that several plasma cytokines were consistently higher in patients with ARDS following sepsis, and that sustained elevations of IL-1b and IL-6 were more common in patients who died (44). The cluster of differentiation (CD)-14 pattern recognition receptor, which mediates responses to lipopolysaccharide (LPS), is shed from the surface of macrophages and other cells and accumulates in the lung ﬂuids of patients with ALI (45–47). Interestingly, soluble CD14 (sCD14) was a strong predictor of two independent measures of lung inﬂammation and permeability changes, the BAL total polymorphonuclear leukocyte (PMN) and protein concentrations (46). This provides a clue that pathways that mediate LPS responsiveness in the lungs are important determinants of lung injury. The complexity inherent in making individual cytokine measurements in patients with ALI became apparent when investigators studied agonist/ antagonist pairs in BAL ﬂuid of patients with lung injury, and the concept of ‘‘cytokine balance’’ in the lungs emerged (13,48). Soluble receptors for individual cytokines are shed from the surface of macrophages and other cells and typically antagonize the bioactivity of their respective ligands, although this is not true in all cases. For example, soluble monomeric forms of the two TNFa receptors, TNFRI(p55) and TNFRII(p75), are cleaved from the macrophage surface by the action of speciﬁc metalloproteinases and antagonize the effects of TNFa in solution. Similarly, the soluble IL-1R is an antagonist of IL-1b, as is the lower molecular weight IL-1 receptor antagonist (IL-IRA), which blocks signaling by the IL-1 receptor on the cell surface. Surprisingly, the soluble concentrations of TNFRI and TNFRII are substantially higher than the concentration of TNFa in BAL ﬂuid, and little TNFa activity is detectable in BAL ﬂuid (13,48). As with the TNF system, the concentration of IL-1RA exceeds the concentration of free IL-1b in ARDS BAL ﬂuid by an order of magnitude, but IL-1b bioactivity is detectable because very few molecules of free IL-1b are needed to cause a biological effect (13,49). This concept is also true for proteinases in ARDS BAL ﬂuid. Activated neutrophils and macrophages that accumulate in the airspaces of ARDS patients release metalloproteinases such as matrix metalloproteinase (MMP)-2 and MMP-9 (also called gelatinase A and B, respectively), and collagenolytic activity is detectable in ARDS BAL ﬂuids (50–52,95). However, the concentrations of the naturally occurring inhibitors of MMPs, tissue inhibitors of metalloproteinases, exceed the concentrations of the MMPs, signiﬁcantly reducing the biological activity of these proteinases in solution (51). The BAL procedure samples only the water-soluble phase of the alveolar inﬂammatory reaction, and may

322

Martin and Matthay

underestimate the activity of cytokines, proteinases, and other proinﬂammatory products in the alveolar microenvironment. Coagulation cascades are also activated in the alveolar spaces of patients with ALI (39,53), but coagulation markers in the lungs or the plasma compartment have not proven to be speciﬁc markers of lung injury (5). C. Markers of Apoptosis

Alveolar epithelial injury is characteristic in ALI, and efforts have been made to understand the mechanisms involved. An initial hypothesis was that the inﬂammatory response in the airspaces caused direct epithelial injury through neutrophil migration and activation, and the release of oxidants, proteinases, and other toxic products in the alveolar spaces. Subsequently, it was shown in humans that large numbers of neutrophils could migrate into the alveolar spaces without causing signiﬁcant epithelial injury, and that large numbers of neutrophils could enter the airspaces of sheep lungs in response to bacterial LPS without increasing epithelial permeability, as measured by the movement of radio-labeled albumin out of the alveolar spaces into the circulation (54,55). These observations led to studies of additional mechanisms of alveolar epithelial death, including apoptosis and necrosis, and both have been implicated in VILI. Apoptosis is a form of regulated cell death in which cells involute and shrink, and then are ingested by neighboring leukocytes and other cells. Necrosis involves cell swelling, disruption of cellular membranes, and dysregulated release of intracellular products directly into the local environment. Apoptosis is regulated by receptor-mediated pathways, and also by a mitochondrial pathway common to most cells. The membrane Fas protein is the prototypical death receptor, but the ‘‘death receptor’’ pathway includes the TNF receptors, the IL-1 receptor, and others (56). Clustering of membrane Fas by Fas ligand (FasL) on the surface of lymphocytes, or by soluble FasL (sFasL) in the extracellular environment leads to the assembly of intracellular docking proteins that share the Fas-associated death domain and the sequential activation of a series of intracellular proteinases (caspases) that eventually cause DNA cleavage and cell death. The Fas protein is widely distributed on the airway and the alveolar epithelium in the lungs, as well as on the alveolar macrophages (57,58), and FasL is expressed on airway epithelium as well as on myeloid cells (59). The sFasL is released from the cell surface by the action of metalloproteinases such as MMP-3 and MMP-7 (60–62), and membrane-bound MMP-7 can cleave and degrade sFasL via speciﬁc cleavage sequences in sFasL (63). Thus, metalloproteinases are involved in the release of sFasL from the cell surface, and also in the inactivation of sFasL. The activity of sFasL is also modulated by a decoy receptor (DcR3), which is shed from the cell surface and can bind and inactivate sFasL (56,64). Instilling recombinant sFasL

Biological Markers of VILI

323

into the lungs of rabbits causes apoptosis in the alveolar walls, alveolar hemorrhage, and the production of proinﬂammatory cytokines by alveolar macrophages (65). In the lungs of mice, Fas activation using a speciﬁc monoclonal antibody causes alveolar injury and inﬂammation (66,67). Markers of the Fas-dependent apoptosis pathway are detectable in the lungs of patients before and after the onset of ALI and VILI. The sFasL peptide is detectable in BAL ﬂuids of patients at risk for ARDS, at the onset of ARDS, and for more than seven days in the lungs of patients with sustained ARDS (68). The sFasL concentration was higher in patients who died, but there was considerable overlap in the groups, so that sFasL by itself is not a good predictor of onset or severity of ALI. Interestingly, biologically active sFasL was detectable in the BAL ﬂuids only at the onset of ARDS, despite the fact that sFasL was detectable by immunoassay before and for more than seven days after the onset of lung injury. Both Fas and sFasL are detectable in the lung edema ﬂuids from patients at the onset of ALI, and Fas expression is increased in lung tissue sections of patients who die of severe lung injury (69). The pathways by which sFasL increases in the lungs, and the mechanisms that control its biological activity are not completely clear. sFasL is released from activated monocytes (70), but it is not a major product of alveolar macrophages. It may accumulate in lung ﬂuids when it is cleaved from epithelial surfaces by activated MMPs, or it may move from the plasma compartment into the lungs when endothelial and epithelial permeability are altered (60,63). Although the DcR3 inhibitor of sFasL is also detectable in BAL ﬂuids at the onset of ARDS, its presence does not completely explain the biological activity of sFasL at the onset of ARDS. The aggregation state of sFasL may be an important determinant of biological activity because monomeric sFasL has little biological activity. Clustered forms of sFasL have greater biological activity in vitro, but the importance of clustering in controlling biological activity in vivo is not certain (71). D. Markers of Repair

Procollagen peptides and the transforming growth factor (TGF) family of proteins have been used as markers of inﬂammation and repair. The procollagen peptide precursor of Type III collagen (PCPIII) is elevated at the onset of ARDS, as soon as lung injury is clinically detectable, and is signiﬁcantly higher in patients who die (15). In patients with sustained ARDS, the BAL PCPIII concentration on day 7 identiﬁed patients who were more likely to die, but the overlap between survivors and nonsurvivors was substantial (14). TGFa has been identiﬁed in BAL ﬂuid of patients after the onset of ARDS and in the edema ﬂuid of patients at the onset of ARDS (72,73). TGFa is an alveolar macrophage product that is a mitogen for epithelial cells and mesenchymal cells (74). As such, TGFa is a marker of

324

Martin and Matthay

the ﬁbroproliferative phase that follows ALI. Interestingly, patients with high concentrations of both TGFa and PCPIII were signiﬁcantly more likely to die, presumably reﬂecting worsening of an underlying or ongoing injury in the lungs (73). E. Speciﬁc Studies of Markers of Injury in VILI

The paradigm for markers of lung injury shifted dramatically with the publication by Ranieri et al., of a small prospective randomized study designed to test the hypothesis that a lung-protective ventilatory strategy would preserve or improve lung function, and that this would be reﬂected by a reduction in the inﬂammatory responses in the lungs (75). Because acute inﬂammation is characteristic of ALI, many earlier studies had focused on trying to ﬁnd single inﬂammatory mediators that were involved in pathophysiology, and which could predict the onset and/or outcome of ALI. The study by Ranieri et al. assumed that the injury might occur ﬁrst, mediated by either overdistension or rapid recruitment and derecruitment of alveolar units, and then would be reﬂected by acute inﬂammation. In the patients randomized to the lung-protective mechanical ventilation strategy, the BAL cell and cytokine concentrations (e.g., TNFa, IL-8, and IL-6) fell signiﬁcantly over 48 hours (Figs. 1 and 2). The effect was general, reﬂecting a lessening of the inﬂammatory response, rather than being speciﬁc for any single cytokine or cell type. Similar ﬁndings were observed in the plasma samples. This study was important for several reasons: ﬁrst, it suggested that injury might be driving inﬂammation, rather than the reverse; second, it showed that reduction of injury would be associated with falls in a range of cytokine concentrations, suggesting that they were all manifestations of the same injury; third, the study correctly predicted the outcome of the largest controlled randomized trial in ARDS, the ARDSNet study of different mechanical ventilation strategies in ARDS, in which patients ventilated with the lower tidal volume strategy had better outcomes (9). The study by Ranieri and associates was the ﬁrst experiment speciﬁcally designed to evaluate the relationship between the two different ventilatory strategies and markers of inﬂammation in the lungs and plasma of patients with lung injury. The NIH ARDSNet was established in 1997 as a national network of 10 centers in the United States with the purpose of conducting controlled randomized clinical trials of therapies for ARDS, emphasizing treatments that are not being developed by industry. When the initial studies were planned, the ARDSNet investigators recognized that the strategy used for mechanical ventilation would be an important confounding variable in all of the trials of biological modiﬁers that were under consideration. Therefore, the ARDSNet investigators designed a trial to test the concept that mechanical ventilation with a low tidal volume would protect the lungs from ventilator-associated injury and improve outcome, and the results

Biological Markers of VILI

325

Figure 1 Bronchoalveolar lavage PMN% and IL-1b in two groups of patients ventilated with standard methods (Vt ¼ 11.1 mL/Kg) and a lung protective strategy (Vt ¼ 7.6 mL/Kg). Source: From Ref. 75.

conﬁrmed this hypothesis (9). Because plasma samples were collected and stored from all of the patients at entry into the trial and three days later, this trial provided the ﬁrst large-scale opportunity to test the relationship between biological markers, ventilatory strategy, and outcome in a clinical trial in which differences in mechanical ventilation produced signiﬁcant differences in outcome (76). Lung ﬂuid specimens were not collected in this trial, so it was not possible to test markers in the BAL or edema ﬂuid, as had been done by Ranieri et al. (75).

326

Martin and Matthay

Figure 2 Bronchoalveolar lavage and plasma cytokines in two groups of patients ventilated with standard methods (Vt ¼ 11.1 mL/Kg) and a lung protective strategy (Vt ¼ 7.6 mL/Kg). Source: From Ref. 75.

Using the ARDSNet plasma samples, Parsons et al. addressed one of the persistent questions about biological markers in patients with lung injury—whether plasma cytokine markers would predict the onset or outcome of ARDS (77). When many different cytokines were measured in plasma at the onset of clinically deﬁned lung injury, IL-8, IL-6, and IL-10 were signiﬁcantly associated with an increased risk of death (Table 2). Patients with sepsis had the highest cytokine concentrations in plasma and the greatest risk of death associated with the elevations of each of the cytokines. Interestingly, the patients ventilated with the lower tidal volume strategy had a fall in cytokine concentrations of approximately 25% for IL-6, and 10% for IL-8 and IL-10 (Table 3 and Fig. 3). This suggests that the stretch applied to the lungs accounts for a portion of the elevation in these circulating cytokines, and that ventilation with a lower tidal volume is associated with a reduction in the inﬂammatory response in the lungs. In the ARDSNet trial, the absolute difference in plasma cytokine concentrations at the onset of lung injury between survivors and nonsurvivors was approximately two- to threefold, and the reduction in plasma cytokine

Biological Markers of VILI

327

Table 2 Plasma Cytokine Concentrations and Mortality Alive

Time IL-6 Baseline Day 3 IL-8 Baseline Day 3 IL-10 Baseline Day 3

n

Cytokine level median (25th–75th IQR)

505 478

Dead

n

Cytokine level median (25th–75th IQR)

p valuea

227 (94–630) 80 (39–179)

276 240

411 (133–1471) 208 (80–635)

< 0.0001 < 0.0001

505 478

33 (0–78) 24 (0–51)

275 240

67 (24–180) 66 (25–144)

< 0.0001 < 0.000l

330 360

14 (0–54) 0 (0–23)

213 184

34 (0–85) 22 (0-59)

< 0.000l < 0.0001

a

Wilcoxon test comparing cytokine levels among alive versus dead at each time point. All cytokine levels are in pg/mL. Abbreviations: IQR, interquartile range; IL, interleukin. Source: From Ref. 77.

concentrations in the low tidal volume group was small, indicating that large clinical trials are needed to test the value of biological markers in plasma in lung injury. In contrast, the study by Ranieri et al., suggests that much smaller trials can be used to test biological markers measured directly in BAL (or edema) ﬂuid obtained directly from the lungs. The ARDSNet trial included plasma samples from 781 of the 861 patients in the trial. Approximately 63% of the patients had either sepsis or pneumonia as the clinical cause of their ARDS, 14% had gastric aspiration, and only 10% had trauma as the primary cause, so the conclusions are less certain for trauma-associated ARDS than for ARDS associated with a primary inﬂammatory process in the lungs or the systemic circulation. Importantly, the predictive value of the cytokine measurements at the onset of ARDS needs to be validated prospectively in another large cohort, and this is underway. Parsons et al., measured the concentrations of TNFa and the two soluble TNFa receptors, TNFRI (p55) and TNFRII (p75), in the plasma of the patients in the ARDSNet study and compared the results with the measurements obtained from a single-center study of patients before and after the onset of ARDS (78). The circulating concentrations of TNFa were very low and the TNFa values were not helpful in any way. The concentrations of TNFRI and TNFRII were similar, so that only TNFRI was measured in all patients. In the single-center study of 35 patients at risk for ARDS, and 50 patients with established ARDS, the TNF receptor concentrations in plasma were higher in patients who died, but this was not statistically signiﬁcant because of the variability in the data and the relatively

328 Table 3 Plasma Cytokine Concentrations in Patients Ventilated with Lower and Higher Tidal Volumes in the NIH ARDSNet study 6 mL/kg Measurement IL-6 Day 0 Day 3 IL-8 Day 0 Day 3 IL-10 Day 0 Day 3

12 mL/kg

Total

n

Median (IQR)

n

Median (IQR)

n

Median (IQR)

393 364

264 (109–766) 96 (46–213)

388 354

284 (109–1069) 126 (49–388)

781 718

273 (109–8991) 104 (47–291)

393 364

43 (0–93) 32 (0–73)

387 354

41 (0–114) 35 (0–83)

780 718

42 (0–101) 34 (0–77)

302 278

18 (0–57) 1 (0–30)

291 266

24 (0–71) 1 (0–36)

593 544

19 (0–63) 1 (0–32)

p Valuesa Change over time

Ventilator group

<0.0001

0.0006

<0.0001

0.04

<0.0001

0.15

Martin and Matthay

All cytokine levels are in pg/mL. a p for ventilator group based on analysis of covariance for model: log-10 cytokine level at day 3 ¼ ventilator group þ log-10 cytokine level at day 0. Abbreviation: IQR, interquartile range. Source: From Ref. 77, Table 5.

Biological Markers of VILI

329

Figure 3 Percent reduction from study entry to day 3 in three different cytokines in patients ventilated with two different tidal volumes (6 mL/Kg vs. 12 mL/Kg). The 95% conﬁdence intervals are 12% to 37% for IL-6, 1% to 23% for IL-8, and 4% to 25% for IL-10. Source: From Ref. 77.

small sample size. However, in the larger ARDSNet study, the plasma concentration of soluble TNFRI was strongly related to mortality, longer time on mechanical ventilation, and more organ failures. The data for TNFRII were consistent with this, but the TNFRII measurements were obtained only in a subset of the patients. Interestingly, the low tidal volume ventilatory strategy was associated with a fall in plasma TNFRI whereas the plasma TNFRI rose in patients ventilated with higher tidal volumes. Thus, the plasma TNFa values are not helpful, but high concentrations of TNFRI in plasma are associated with increased morbidity and mortality. The source of the plasma TNFRI is not clear, because it could derive from circulating leukocytes or cells of the reticuloendothelial system, or it could leak out of the lungs after shedding from the surface of lung epithelial cells (78). Eisner et al., evaluated the predictive value of plasma surfactant protein (SP) in 565 patients in the ARDSNet cohort (79). Consistent with prior reports, SP-A and SP-D were detectable in the plasmas of all patients at the onset of ARDS. The plasma SP-D concentration at study entry was higher in patients who later died, and plasma SP-D levels were associated with a longer time on mechanical ventilation and more organ failure. In contrast, the plasma SP-A concentrations were not associated with mortality or any other clinical outcome measure. Interestingly, the lower tidal volume strategy attenuated the rise in plasma SP-D that occurred during the ﬁrst three days of the study, but had no effect on SP-A. This large trial suggests that plasma SP-D may have value as a marker of the severity of lung injury, and that the lung-protective strategy reduces the leak of SP-D out of the lungs into the systemic circulation. As noted previously, the VWF antigen is a marker of endothelial and platelet activation. In the ARDSNet patients, the initial plasma concentration of VWF antigen was signiﬁcantly higher in patients who died, and an

330

Martin and Matthay

elevated VWF antigen was strongly associated with a longer time on mechanical ventilation and the number of organ failures (42). This was true whether or not the patients were septic. However, in contrast to the markers of inﬂammation and the surfactant-associated proteins, the plasma concentration of VWF was not reduced by the lower tidal volume strategy. This suggests that the lung-protective ventilation protects the lungs, but not the systemic vascular compartment, and/or that the lungs are not a major source of VWF antigen in patients with systemic inﬂammation. IV. Future Approaches to Identifying Markers of VILI Progress has been made in identifying biological markers of lung injury in humans since the last major review of the subject (5). For the most part, this progress occurred because of the completion of the large multicenter ARDSNet trial, which had the statistical power to evaluate individual candidate markers, and the fact that different mechanical ventilation strategies produced different outcomes. Yet we still do not have the equivalent of a ‘‘troponin’’ for identifying ALI. Although markers like SP-D, TNFRI, VWF, and the cytokines IL-8, IL-6, and IL-10 are associated with higher mortality and more severe clinical illness, they do not separate groups as clearly as the troponin separates patients with and without myocardial infarction (76). A. Proteomics

The ﬁeld of proteomics, in which many different proteins are analyzed simultaneously in biological ﬂuids, is expanding rapidly. Proteomics technology has been used to identify markers of malignancy, and it is also being applied to lung diseases (80,81). In the simplest approach, proteins in biological ﬂuids are resolved in one dimension by isoelectric focusing, and then in a second dimension according to molecular size. This two-dimensional electrophoresis produces a protein ‘‘map’’ with protein spots spread over the gel like marbles spilled on a ﬂoor. The identity of each spot must be established, typically by mass spectroscopy. This method can be improved by a technique called ‘‘differential in gel electrophoresis’’ (DIGE). The DIGE method involves labeling proteins from two different experimental samples with different ﬂuorescent markers, and then labeling a standard protein mixture with a third marker. Aliquots of these three protein samples are mixed together and then run in the same gel. By scanning the gels at three different wavelengths speciﬁc for the ﬂuorescent markers, the abundance of each protein in the two different experimental samples can be measured relative to the same internal standard in each spot on the gel. Using this approach, the protein maps of specimens from individual patients can be compared in at least three different ways: proteins

Biological Markers of VILI

331

that increase or decrease relative to the standard can be identiﬁed; proteins that change due to posttranslational modiﬁcations can be identiﬁed; and new proteins in a sample can be identiﬁed, for example, proteins that appear in BAL ﬂuid at the onset of lung injury. Analysis of proteins by two-dimensional protein electrophoresis depends on the ability to produce reproducible protein maps and on the ability to identify as many of the proteins in the standard map as possible, typically by analyzing excised spots from the gels using mass spectroscopic methods, which is both time consuming and expensive. Protein maps produced in different laboratories may be very different if the experimental conditions are not identical. So far, protein maps have been made and published for normal human BAL ﬂuid, for patients with interstitial lung diseases, and for patients with ALI (82,83). A major limitation of the protein maps published to date is the lack of internal standards in the gels, so that the maps from different laboratories are not comparable. Using the gel electrophoresis method, Bowler and associates compared the protein proﬁles in the lung edema ﬂuid and plasma samples from 16 patients with ALI, and BAL and plasma samples from 12 normal subjects (83). Over 300 protein spots could be identiﬁed in the edema ﬂuid of patients with ALI and the BAL ﬂuid of the normal subjects, and 158 of these were lifted from the gels and identiﬁed by mass spectroscopy (using matrixassisted laser desorption/ionization time-of-ﬂight mass spectrometry, MALDI-TOF). Most of the proteins in the ALI edema ﬂuid were also found in plasma, consistent with the concept that ALI is a high-permeability edema. Interestingly, some of the proteins in the ALI edema ﬂuid were modiﬁed into smaller forms, suggesting proteolysis, and some, like SP-A, were much less abundant in the ALI edema ﬂuids than in normal BAL ﬂuid. For example, SP-A was detectable in the edema ﬂuid of only 1 of 16 patients with ALI, but in the BAL ﬂuid from all of the normal subjects. This result differs significantly from studies of SP-A in BAL of patients with ARDS, in which SP-A was detectable in all patients using a sandwich immunoassay (11). An alternative approach to comprehensive proteomics analysis involves labeling the proteins with isotopes of different molecular weights, and then digesting them with trypsin to yield a mixture of a large number of peptides [isotope coded afﬁnity tag (ICAT) method]. These peptides can be identiﬁed by mass and charge using different types of mass spectroscopy, and complex computer algorithms can be used to predict the original mixture or proteins that yielded that mixture of speciﬁc peptides (84,85). This approach to proteomics analysis is very powerful, and more than 1000 proteins in biological specimens can be identiﬁed. Importantly, this method is not dependent on gel electrophoresis in the ﬁrst step. A major drawback is that it is time consuming and expensive, and it is not widely available. Current improvements in technology and computer processing should shorten the analytical time, and reduce the costs.

332

Martin and Matthay

All the published studies using proteomics in lung diseases have some limitations that must be addressed to make proteomics analysis a more valuable tool. First, 50% to 75% of the BAL proteins are albumin and immunoglobulins, and these highly abundant proteins mask the presence of less abundant, but possibly more informative proteins. Methods are available to deplete these high abundance proteins, and when this is done, many more proteins are visible in the gels. Second, as noted, there is a need for internal standards in the proteomics methods, so that protein changes in the injured lungs can be measured relative to a reference protein in the same sample. For example, in the DIGE method, a mixture of normal proteins derived from human BAL ﬂuid can be run in all of the gels from a series of patients, facilitating comparisons between the patients. The ICAT method makes an internal standard possible, as proteins from two different biological samples (one the unknown and one the standard) can be labeled with different molecular weight isotopes, mixed, and analyzed at the same time. Third, and very importantly, methods are just being developed to use much more of the information available from the proteomics analyses. For example, computational methods will be needed to analyze clusters of protein spots in gels in order to move away from single proteins and determine how groups of proteins change at the same time. Aside from cytokines and other products of the inﬂammatory response, groups of intracellular proteins that are not normally present in lung ﬂuids should be detectable when lung injury causes signiﬁcant damage to the alveolar epithelium. Methods are also needed to ‘‘subtract’’ the protein information that is constant in two different clinical situations in order to focus on proteins that are appearing, disappearing, or changing signiﬁcantly with lung injury. If these challenges can be met, then proteomics technology could revolutionize the way in which we evaluate and think about ALI. Cluster analysis of lung proteins could provide a group of lung or plasma proteins that perform similarly to the troponin for the heart. B. Markers of Bacterial Products in the Lungs

Although investigators have focused on deﬁning endogenous biological markers of ALI and VILI, another frontier is the need to better deﬁne the bacterial products that are in the alveolar spaces, or in the plasma. Bacterial LPS is known to sensitize the lungs of experimental animals to the effects of mechanical stretch. Human alveolar macrophages respond to mechanical stretching by producing proinﬂammatory cytokines, and this is enhanced if LPS is added to the incubation medium (86). LPS sensitizes rat lungs to the effects of mechanical stretch ex vivo (87), and LPS sensitizes rabbit lungs ventilated with moderate or conventional tidal volumes in vivo (88,89). LPS and other bacterial products are recognized by a series of pattern recognition receptors on the cell surface, termed toll-like

Biological Markers of VILI

333

receptors (TLRs), and many share common intracellular signaling pathways (90). The bacterial epitopes recognized by TLRs include LPS, peptidoglycans, lipoproteins, ﬂagellin, bacterial DNA, and double-stranded RNA, among others. It is likely that a number of different bacterial products that enter the lungs via the oropharyngeal route or via the bloodstream have the potential to sensitize the lungs to the effects of mechanical stretch, and worsen the lung injury. Therefore, it is possible that markers of microbial entry into the lungs will have value in identifying patients who are at risk for the most injurious effects of mechanical ventilation, and the worst VILI. C. Host Determinants of Lung Injury

Lastly, emerging data show that humans vary in the intensity of their endogenous innate immune responses and it is possible that host inﬂammatory response determinants affect the likelihood of developing VILI. The severity of meningococcal disease is modulated by genetic determinants (91), and more than 50% of the innate immune response is inﬂuenced by genetic factors (92). Using a simple whole blood LPS challenge assay, it is possible to identify individuals in the normal population who have ‘‘high’’ and ‘‘low’’ response phenotypes to LPS and other stimuli of TLR-mediated innate immune pathways (93). Although simple biological assays are not yet available to identify high or low responders at the onset of illness, the development of stable genetic markers such as clusters of single nucleotide polymorphisms, and the development of rapid genotyping platforms, could make it possible to include host inﬂammatory markers in the list of biological markers of VILI. V. Summary and Conclusions The completion of several large randomized clinical trials of ventilator management in patients with ALI has made it possible to test biological markers of lung injury in more precise ways than could be done in earlier small studies that had produced conﬂicting results. Several plasma markers have value as initial markers of the severity of illness and/or outcome, including the cytokines IL-8, IL-6, and IL-10, SP-D, and the VWF antigen. Less is known about the predictive value of lung lavage markers or pulmonary edema ﬂuid markers, because the lung ﬂuids were not sampled consistently in the large trials of mechanical ventilation strategies, and what we do know is based on smaller studies. Simple methods to sample lung ﬂuids are needed so that large-scale studies of lung markers of injury can be conducted in multicenter trials. New technologies on the horizon may make it possible to evaluate clusters of biological markers in the lungs at the same time, recognizing the complexity of lung injury in humans. By embracing the complexity of lung injury related to mechanical ventilation, we may

334

Martin and Matthay

identify better methods to predict who is at the greatest risk for VILI, and who will have the worst outcome once it develops, so that those patients can be targeted for new and better treatments.

Acknowledgments This work was supported in part by the Medical Research Service of the Department of Veterans Affairs, and by grants HL69852 and HL073996 from the National Institutes of Health. References 1. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 2. Dos Santos CC, Slutsky AS. Mechanisms of ventilator-induced lung injury: a perspective. J Appl Physiol 2000; 89:1645–1655. 3. Ashbaugh DG, Bigelow DB, Petty TL, Levine BE. Acute respiratory distress in adults. Lancet 1967; 2:319–323. 4. Petty TL, Ashbaugh DG. The adult respiratory distress syndrome. Clinical features and factors inﬂuencing prognosis and principles of management. Chest 1971; 60:233–239. 5. Pittet JF, Mackersie R, Martin TR, Matthay MA. Biological markers of acute lung injury: prognostic and pathogenetic signiﬁcance. Am J Respir Crit Care Med 1997; 155:1187–1205. 6. Fowler AA III, Hamman RF, Good JT, et al. Adult respiratory distress syndrome: risk with common predispositions. Ann Intern Med 1983; 98:593–597. 7. Montgomery AB, Stager MA, Carrico CJ, Hudson LD. Causes of mortality in patients with the adult respiratory distress syndrome. Am Rev Respir Dis 1985; 132:485–489. 8. Hudson LD, Milberg JA, Anardi D, Maunder RJ. Clinical risks for development of the acute respiratory distress syndrome. Am J Respir Crit Care Med 1995; 151:293–301. 9. NIH ARDSNet Group. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342:1301–1308. 10. Nuckton TJ, Alonso JA, Kallet RH, et al. Pulmonary dead-space fraction as a risk factor for death in the acute respiratory distress syndrome. N Engl J Med 2002; 346:1281–1286. 11. Greene KB, Wright JR, Steinberg KP, et al. Serial changes in surfactantassociated proteins in lung and serum before and after onset of ARDS. Am J Respir Crit Care Med 1999; 160:1843–1850. 12. Goodman RB, Strieter RM, Steinberg KP, et al. Inﬂammatory cytokines in patients with persistence of the acute respiratory distress syndrome. Am J Respir Crit Care Med 1996; 154:602–611.

Biological Markers of VILI

335

13. Park WY, Goodman RB, Steinberg KP, et al. Cytokine balance in the lungs of patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 164:1896–1903. 14. Clark JG, Milberg JA, Steinberg KP, Hudson LD. Type III procollagen peptide in the adult respiratory distress syndrome: association of increased peptide levels in bronchoalveolar lavage ﬂuid with increased risk for death. Ann Intern Med 1995; 122:17–23. 15. Chesnutt AN, Matthay MA, Tibayan FA, Clark JG. Early detection of type III procollagen peptide in acute lung injury: pathogenetic and prognostic signiﬁcance. Am J Respir Crit Care Med 1997; 156:840–845. 16. Hermans C, Bernard A. Lung epithelium-speciﬁc proteins: characteristics and potential applications as markers. Am J Respir Crit Care Med 1999; 159:646–678. 17. Dobbs LG, Williams MC, Gonzalez R. Monoclonal antibodies speciﬁc to apical surfaces of rat alveolar type I cells bind to surfaces of cultured, but not freshly isolated, type II cells. Biochim Biophys Acta 1988; 970:146–156. 18. McElroy MC, Pittet JF, Hashimoto S, Allen L, Wiener Kronish XP, Dobbs LG. A type I cell-speciﬁc protein is a biochemical marker of epithelial injury in a rat model of pneumonia. Am J Physiol 1995; 268:L181. 19. McElroy MC, Pittet XF, Allen L, Wiener-Kronish XP, Dobbs LG. Biochemical detection of type I cell damage after nitrogen dioxide-induced lung injury in rats. Am J Physiol 1997; 273:L1228–L1234. 20. Frank JA, Gutierrez JA, Jones KD, Allen L, Dobbs L, Matthay MA. Low tidal volume reduces epithelial and endothelial injury in acid-injured rat lungs. Am J Respir Crit Care Med 2002; 165:242–249. 21. Frank JA, McAuley DF, Gutierrez JA, Daniel BM, Dobbs L, Matthay MA. Differential effects of sustained inﬂation recruitment maneuvers on alveolar epithelial and lung endothelial injury. Crit Care Med 2005; 33:181–188. 22. Newman V, Gonzalez RF, Matthay MA, Dobbs LG. A novel alveolar type I cell-speciﬁc biochemical marker of human acute lung injury. Am J Respir Crit Care Med 2000; 161:990–995. 23. Girod CE, Shin DH, Hershenson MB, Solway J, Dahl R, Miller YE. pl72: an alveolar type II and Clara cell speciﬁc protein with late developmental expression upregulation by hyperoxic lung injury. Am J Respir Cell Mol Biol 1996; 14:538–547. 24. Funkhouser JD, Tangada SD, Jones M, SJ O, Peterson RD. pl46 type II alveolar epithelial cell antigen is identical to aminopeptidase N. Am J Physiol 1991; 260:L274–L279. 25. Edelson JD, Shannon JM, Mason RJ. Alkaline phosphatase: a marker of alveolar type II cell differentiation. Am Rev Respir Dis 1988; 138:1268–1275. 26. Lwebuga-Mukasa JS. Identiﬁcation of pneumocin, a developmentally regulated apical membrane glycoprotein in rat lung type II and Clara cells. Am J Respir Cell Mol Biol 1991; 4:489–496. 27. Boylan GM, Pryde JG, Dobbs LG, McElroy MC. Identiﬁcation of a novel antigen on the apical surface of rat alveolar epithelial type II and Clara cells. Am J Physiol Lung Cell Mol Physiol 2001; 280:L1318–L1326. 28. Doyle IR, Nicholas TE, Bersten AD. Serum surfactant protein-A levels in patients with acute cardiogenic pulmonary edema and adult respiratory distress syndrome. Am J Respir Crit Care Med 1995; 152:307–317.

336

Martin and Matthay

29. Doyle IR, Bersten AD, Nicholas TE. Surfactant proteins-A and -B are elevated in plasma of patients with acute respiratory failure. Am J Respir Crit Care Med 1997; 156:1217–1229. 30. Bersten AD, Hunt T, Nicholas TE, Doyle IR. Elevated plasma surfactant protein-B predicts development of acute respiratory distress syndrome in patients with acute respiratory failure. Am J Respir Crit Care Med 2001; 164:648–652. 31. Cheng IW, Ware LB, Greene KE, Nuckton TJ, Eisner MD, Matthay MA. Prognostic value of surfactant proteins A and D in patients with acute lung injury. Crit Care Med 2003; 31:20–27. 32. Finkbeiner WE, Carrier SD, Teresi CE. Reverse transcription-polymerase chain reaction (RT-PCR) phenotypic analysis of cell cultures of human tracheal epithelium, tracheobronchial glands, and lung carcinomas. Am J Respir Cell Mol Biol 1993; 9:547–556. 33. Ishizaka A, Matsuda T, Albertine KH, et al. Elevation of KL-6, a lung epithelial cell marker, in plasma and epithelial lining ﬂuid in acute respiratory distress syndrome. Am J Physiol Lung Cell Mol Physiol 2004; 286:L1088–L1094. 34. Ribes JA, Francis CW, Wagner DD. Fibrin induces release of von Willebrand factor from endothelial cells. J Clin Invest 1987; 79:117–123. 35. Hamilton KK, Sims PJ. Changes in cytosolic Ca2þ associated with von Willebrand factor release in human endothelial cells exposed to histamine. Study of microcarrier cell monolayers using the ﬂuorescent probe indo-1. J Clin Invest 1987; 79:600–608. 36. Rubin DB, Wiener-Kronish JP, Murray JF, et al. Elevated von Willebrand factor antigen is an early plasma predictor of acute lung injury in nonpulmonary sepsis syndrome. J Clin invest 1990; 86:474–480. 37. Moss M, Ackerson L, Gillespie MK, Moore FA, Moore EE, Parsons PE. von Willebrand factor antigen levels are not predictive for the adult respiratory distresssyndrome. Am J Respir Crit Care Med 1995; 151:15–20. 38. Moss M, Gillespie MK, Ackerson L, Moore FA, Moore EE, Parsons PE. Endothelial cell activity varies in patients at risk for the adult respiratory distress syndrome. Crit Care Med 1996; 24(11):1782–1786. 39. Sabharwal AK, Bajaj SP, Ameri A, et al. Tissue factor pathway inhibitor and von Willebrand factor antigen levels in adult respiratory distress syndrome and in a primate model of sepsis. Am J Respir Crit Care Med l995; 151:758–767. 40. Bajaj MS, Tricomi SM. Plasma levels of the three endothelial-speciﬁc proteins von Willebrand factor, tissue factor pathway inhibitor, and thrombomodulin do not predict the development of acute respiratory distress syndrome. Intensive Care Med 1999; 25:1259–1266. 41. Ware LB, Conner ER, Matthay MA. von Willebrand factor antigen is an independent marker of poor outcome in patients with early acute lung injury. Crit Care Med 2001; 9:2325–2331. 42. Ware LB, Eisner MD, Thompson BT, Parsons P, Matthay MA. The Acute Respiratory Distress Syndrome Network. Signiﬁcance of von Willebrand factor in septic and non-septic patients with acute lung injury. Am J Respir Crit Care Med 2004. 43. Steinberg KP, Milberg JA, Martin TR, Maunder RJ, Cockrill BA, Hudson LD. Evolution of bronchoalveolar cell populations in the adult respiratory distress syndrome. Am J Respir Crit Care Med 1994; 150:113–122.

Biological Markers of VILI

337

44. Meduri GU, Headley S, Kohler G, et al. Persistent elevation of inﬂammatory cytokines predicts a poor outcome in ARDS. Plasma IL-1 beta and IL-6 levels are consistent and efﬁcient predictors of outcome over time. Chest 1995; 107:1062–1073. 45. Martin TR, Mongovin SM, Tobias PS, et al. The CD14 differentiation antigen mediates the development of endotoxin responsiveness during differentiation of mononuclear phagocytes. J Leukoc Biol 1994; 56:1–9. 46. Martin TR, Ruzinski JT, Radella II FR, et al. Endotoxin ampliﬁcation pathways in lungs during acute respiratory distress syndrome (ARDS). Am J Respir Crit Care Med 1996; 153:A593. 47. Hasday JD, Dubin W, Mongovin S, et al. Bronchoalvelar macrophage CD14 expression. Shift between the membrane-associated and soluble pools. Am J Physiol (Lung Cell Mol Physiol) 1997; 272:L925–L933. 48. Suter PM, Suter S, Girardin E, Roux-Lombard P, Grau GE, Dayer JM. High bronchoalveolar levels of tumor necrosis factor and its inhibitors, interleukin-1, interferon, and elastase, in patients with adult respiratory distress syndrome after trauma, shock, or sepsis. Am Rev Respir Dis 1992; 145:1016–1022. 49. Pugin J, Ricou B, Steinberg KP, Suter PM, Martin TR. Proinﬂammatory activity in bronchoalveolar lavage ﬂuids from patients with ARDS, a prominent role for interleukin-1b. Am J Respir Crit Care Med 1996; 153:1850–1856. 50. Christner P, Fein AM, Goldberg S, Lipmann NM, Abrams W, Weinbaum G. Collagenase in the lower respiratory tract of patients with adult respiratory distress syndrome. Am Rev Respir Dis 1985; 131:690–695. 51. Ricou B, Nicod L, Lacraz S, Welgus HG, Suter PM, Payer JM. Matrix metalloproteinases and TIMP in acute respiratory distress syndrome. Am J Respir Crit Care Med 1996; 154:346–352. 52. Pugin J, Verghese G, Widmer MC, Matthay MA. The alveolar space is the site of intense inﬂammatory and proﬁbrotic reactions in the early phase of acute respiratory distress syndrome. Crit Care Med 1999; 27:304–312. 53. Idell S, Gonzalez K, Bradford H, et al. Procoagulant activity in bronchoalveolar lavage in the adult respiratory distress syndrome. Contribution of tissue factor associated with factor VII. Am Rev Respir Dis 1987; 136:1466–1474. 54. Martin TR, Pistorese BP, Chi EY, Goodman KB, Matthay MA. Effects of leukotriene B4 in the human lung. Recruitment of neutrophils into the alveolar spaces without a change in protein permeability. J Clin Invest 1989; 84:1609–1619. 55. Wiener-Kronish JP, Albertine KH, Matthay MA. Differential responses of the endothelial and epithelial barriers of the lung in sheep to Escherichia coli endotoxin. J Clin Invest 1991; 88:864–875. 56. Ashkenazi A, Dixit VM. Apoptosis control by death and decoy receptors. Curr Opin Cell Biol 1999; 11:255–260. 57. Hamann KJ, Dorscheid DR, Ko FD, et al. Expression of Fas (CD95) and FasL (CD95L) in human airway epithelium. Am J Respir Cell Mol Biol 1998; 19:537–542. 58. Fine A, Anderson NL, Rothstein TL, Williams MC, Gochuico BR. Fas expression in pulmonary alveolar type II cells. Am J Physiol (Lung Cell Mol Physiol) 1997; 273:L64–L71. 59. Gochuico BR, Miranda KM, Hessel EM, et al. Airway epithelial Fas ligand expression: potential role in modulating bronchial inﬂammation. Am J Physiol 1998; 274:L444–L449.

338

Martin and Matthay

60. Powell WC, Fingleton B, Wilson CL, Boothby M, Matrisian LM. The metalloproteinase matrilysin proteolytically generates active soluble Fas ligand and potentiates epithelial cell apoptosis. Curr Biol 1999; 9:1441–1447. 61. Matsuno H, Yudoh K, Watanabe Y, Nakazawa F, Aono H, Kimura T. Stromelysin-1 (MMP-3) in synovial ﬂuid of patients with rheumatoid arthritis has potential to cleave membrane bound Fas ligand. J Rheumatol 2001; 28:22–28. 62. Ethell DW, Kinloch R, Green DR. Metalloproteinase shedding of Fas ligand regulates beta-amyloid neurotoxicity. Curr Biol 2002; 12:1595–1600. 63. Vargo-Gogola T, Crawford HC, Fingleton B, Matrisian LM. Identiﬁcation of novel matrix metalloproteinase-7 (matrilysin) cleavage sites in murine and human Fas ligand. Arch Biochem Biophys 2002; 408:155–161. 64. Pitti RM, Marsters SA, Lawrence DA, et al. Genomic ampliﬁcation of a decoy receptor for Fas ligand in lung and colon cancer. Nature (Lond) 1998; 396:699–703. 65. Matute-Bello G, Liles WC, Frevert CW, et al. Recombinant human Fas ligand induces alveolar epithelial cell apoptosis and lung injury in rabbits. Am J Physiol Lung Cell Mol Physiol 2001; 281:L328–L335. 66. Matute-Bello G, Winn RK, Jonas M, Chi EY, Martin TR, Liles WC. Activation of Fas (CD95) induces lung injury and apoptosis of type I and II pneumocytes in mice. Am J Respir Crit Care Med 1999; 159:A697. 67. Neff TA, Guo RF, Neff SB, et al. Relationship of acute lung inﬂammatory injury to Fas/FasL system. Am J Pathol 2005; 166:685–694. 68. Matute-Bello G, Liles WC, Steinberg KP, et al. Soluble Fas-Ligand induces epithelial cell apoptosis in humans with acute lung injury (ARDS). J Immunol 1999; 163:2217–2225. 69. Albertine KH, Soulier MF, Wang Z, et al. Fas and fas ligand are up-regulated in pulmonary edema ﬂuid and lung tissue of patients with acute lung injury and the acute respiratory distress syndrome. Am J Pathol 2002; 161:1783–1796. 70. Kiener PA, Davis PM, Rankin BM, et al. Human monocytic cells contain high levels of intracellular Fas ligand: rapid release following cellular activation. J Immunol 1997; 159:1594–1598. 71. Holler N, Tardivel A, Kovacsovics-Bankowski M, et al. Two adjacent trimeric Fas ligands are required for Fas signaling and formation of a death-inducing signaling complex. Mol Cell Biol 2003; 23:1428–1440. 72. Chesnutt AN, Kheradmand F, Folkesson HG, Alberts M, Matthay MA. Soluble transforming growth factor-alpha is present in the pulmonary edema ﬂuid of patients with acute lung injury. Chest 1997; 111:652–656. 73. Madtes DK, Klima LD, Rubenfeld G, et al. Elevated transforming growth factor-a levels in bronchoalveolar lavage ﬂuid in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 1998; 158:424–430. 74. Madtes DK, Raines EW, Sakaraissen KS, et al. Induction of transforming growth factor-alpha in activated human alveolar macrophages. Cell 1988; 53:285–293. 75. Ranieri VM, Suter PM, Tortorella C, et al. Effect of mechanical ventilation on inﬂammatory mediators in patients with acute respiratory distress syndrome: a randomized controlled trial. JAMA 1999; 282:54–61. 76. Martin TR. Cytokines and lung injury: searching for useful biomarkers. Crit Care Med 2005; 33:230–232.

Biological Markers of VILI

339

77. Parsons PE, Eisner MD, Thompson BT, et al. Lower tidal volume ventilation and plasma cytokine markers of inﬂammation in patients with acute lung injury. Crit Care Med 2005; 33:1–6. 78. Parsons PE, Matthay MA, Ware LB, Eisner MD. Elevated plasma levels of soluble TNF receptors are associated with morbidity and mortality in patients with acute lung injury. Am J Physiol Lung Cell Mol Physiol 2005; 288: L426–L431. 79. Eisner MD, Parsons P, Matthay MA, Ware L, Greene K. Plasma surfactant protein levels and clinical outcomes in patients with acute lung injury. Thorax 2003; 58:983–988. 80. Noel-Georis I, Bernard A, Falmagne P, Wattiez R. Proteomics as the tool to search for lung disease markers in bronchoalveolar lavage. Dis Markers 2001; 17:271–284. 81. Hiisch J, Hansen KC, Burlingame AL, Matthay MA. Proteomics: current techniques and potential applications to lung disease. Am J Physiol Lung Cell Mol Physiol 2004; 287:L1–L23. 82. Wattiez R, Hermans C, Cruyt C, Bernard A, Falmagne P. Human bronchoalveolar lavage ﬂuid protein two-dimensional database: study of interstitial lung diseases. Electrophoresis 2000; 21:2703–2712. 83. Bowler RP, Duda B, Chan ED, et al. Proteomic analysis of pulmonary edema ﬂuid and plasma in patients with acute lung injury. Am J Physiol Lung Cell Mol Physiol 2004; 286:L1095–L1104. 84. Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH, Aebersold R. Quantitative analysis of complex protein mixtures using isotope-coded afﬁnity tags. Nat Biotechnol 1999; 17:994–999. 85. Aebersold R, Goodlett DR. Mass spectrometry in proteomics. Chem Rev 2001; 101:269–295. 86. Pugin J, Dunn I, Jolbet P, et al. Activation of human macrophages by mechanical ventilation in vitro. Am J Physiol (Lung Cell Mol Physiol) 1999; 275:L1040–L1050. 87. Tremblay L, Valenza F, Ribeiro S, Li J, Slutslcy AS. Injurious ventilatory strategies increase cytokines and cfos mRNA in an isolated rat lung model. J Clin Invest 1997; 5:944–952. 88. Altemeier WA, Matute-Bello G, Frevert CW, et al. Mechanical ventilation with moderate tidal volumes synergistically increases lung cytokine response to systemic endotoxin. Am J Physiol Lung Cell Mol Physiol 2004; 287:L533–L542. 89. Bregeon F, Delpierre S, Chetaille B, et al. Mechanical ventilation affects lung function and cytokine production in an experimental model of endotoxemia. Anesthesiology 2005; 102:331–339. 90. Beutler B, Hoebe K, Du X, Ulevitch PJ. How we detect microbes and respond to them: the Toll-like receptors and their transducers. J Leukoc Biol 2003; 74:479–485. 91. Westendorp RG, Langermans JA, Huizinga TW, et al. Genetic inﬂuence on cytokine production and fatal meningococcal disease. Lancet 1997; 349:170–173. 92. de Craen AJ, Posthuma D, Remarque EJ, van den Biggelaar AH, Westendorp RG, Boomsma DI. Heritability estimates of innate immunity: an extended twin study. Genes Immun 2005; 6:167–170.

340

Martin and Matthay

93. Wurfel MM, Park WY, Radella FR II, et al. Identiﬁcation of high and low responders to lipopolysaccharide in normal subjects: an unbiased approach to identify modulators of innate immunity. J Immunol 2005; 175:2570–2578. 94. Weiland JE, Davis WB, Holter JF, Mohammed JR, Dorinsky PM, Gadek JE. Lung neutrophils in the adult respiratory distress syndrome. Clinical and pathophysiologic signiﬁcance. Am Rev Respir Dis 1986; 133:218–225. 95. Torii K, Lida K, Miyazaki Y, et al. Higher concentrations of matrix metalloproteinases in bronchoalveolar lavage ﬂuid of patients with adult respiratory distress syndrome. Am J Respir Crit Care Med 1997; 155:43–46.

14 Modulation of Lung Injury by Hypercapnia

JOHN G. LAFFEY

BRIAN P. KAVANAGH

Department of Anesthesia and Intensive Care Medicine, Clinical Sciences Institute and National Centre for Biomedical Engineering Sciences, National University of Ireland and University College Hospital Galway, Ireland

Departments of Critical Care Medicine and Anesthesia, Hospital for Sick Children, and University of Toronto Toronto, Ontario, Canada

I. Introduction—Historical Context The ﬁnding that hypocapnia is injurious to lungs was noted over 30 years ago with the observations of Edmunds and Holm (1,2). They demonstrated that hemorrhagic consolidation occurred when alveolar hypocapnia was produced by unilateral pulmonary artery ligation (1,2). They demonstrated in addition, that correction of alveolar hypocapnia, by addition of inhaled CO2 to the inspired gas, attenuated such adverse effects. In the clinical context, Trimble et al. (3) documented that hypocapnia was associated with adverse effects on gas exchange and administered CO2 to patients with what was then termed ‘‘post-traumatic pulmonary insufﬁciency,’’ which would now be called acute respiratory distress syndrome (ARDS). Trimble et al. recorded that addition of CO2 in the inspired gas improved multiple markers of systemic oxygenation, including arterial oxygen tension and shunt fraction (3). Because the investigators produced modest hypercapnia by addition of inspired CO2, and not by reducing tidal volume, this was in effect the ﬁrst recorded application of ‘‘therapeutic hypercapnia’’ in the clinical setting, although it was not recorded as such. 341

342

Laffey and Kavanagh

After the description of ventilator-induced lung injury in the laboratory by Webb and Tierney (4), little attention was paid at the bedside in the following decade, and patients with acute lung injury were managed primarily with attention directed toward normalization of arterial blood gases. In the mid-1980s two landmark studies were published, one in adults with status asthmaticus (5) and the second in neonates with hypoxemic respiratory failure (6). Both of these studies emphasized gentle ventilation and avoidance of high tidal volumes and inspired pressures, fearing that there was a far greater danger from mechanical lung injury than from hypercapnia associated with reduced minute ventilation. The levels of hypercapnic acidosis (HCA) are not reported in these studies (5,6), but the authors make it clear that signiﬁcant HCA developed. The outcomes of these case series represented a major improvement compared with conventional results, with zero mortality in both the series of asthmatic patients (5) and in the neonatal series (6). These reports paved the way for the critically important work of Hickling et al. (7,8). The ﬁrst paper from Hickling’s group was on a retrospective series of patients with ARDS in whom the pressure- and volumelimited strategy resulting in hypercapnia was adopted (7). This ‘‘permissive hypercapnia’’ was a breakthrough in the management of ARDS, and the subsequent prospective series indicated far greater survival than that predicted by the admission APACHE II scores (8). Thus the term ‘‘permissive hypercapnia’’ was coined. Although many authors have emphasized the dangers of HCA (for example, in cases of raised intracranial pressure or pulmonary hypertension), this pressure and volume limitation (with the attendant hypercapnia) was associated with a reduction in conventionally used tidal volumes from the late 1980s onward. Indeed, this approach to the management of ARDS prompted multiple reviews of protective ventilatory management and permissive hypercapnia, and the reduction of tidal volume formed the rationale for the ARDS Network study in 2000, which documented lower mortality associated with lower versus higher tidal volumes (9). The idea that hypercapnia, independent of the effects of reduced lung tidal stretch, could be protective was proposed by Laffey and Kavanagh (10). In that article, it was hypothesized that therapeutic hypercapnia— the deliberate elevation of CO2 independent of reduced minute ventilation—might be protective through the multiple anti-inﬂammatory effects of HCA, as well as through the previously described beneﬁcial effects on tissue O2 supply–demand balance (Fig. 1).

II. Hypercapnia—Deﬁnitions and Terminology Carbon dioxide (CO2) is the universal ‘‘waste product’’ of aerobic cellular respiration, produced by all aerobic life forms, and a key function of

Modulation of Lung Injury by Hypercapnia

343

Figure 1 This ﬁgure illustrates multiple mechanisms whereby hypercapnic acidosis might attenuate pathogenic processes in critically ill patients, including antiinﬂammatory effects and alteration of global and regional O2 supply–demand balance. Source: From Ref. 10.

ventilation is to eliminate CO2. Traditional approaches to CO2 management, including patients with acute respiratory failure requiring mechanical ventilatory support, have focused on the potential for hypercapnia to exert deleterious effects. Accordingly, ventilatory goals have emphasized the need to avoid hypercapnia, and the beneﬁts of aiming for normocapnia, or even hypocapnia (Table 1) (11). However, this approach has been increasingly questioned in recent years, and it is increasingly recognized that aggressive ventilatory support with the aim of maintaining normal CO2 levels may be deleterious (6–8,12). Mechanical ventilation has been demonstrated to potentiate and even cause lung injury, and to worsen the outcome in ARDS patients (6–9,12,13). The use of aggressive ventilation strategies, which use high tidal volumes and transpulmonary pressures, may result in ‘‘ventilatorassociated lung injury’’ (VALI) via several mechanisms. Direct mechanical trauma results from repetitive overstretching and damage of lung tissue and cyclic alveolar recruitment and derecruitment (14–19). Because regional lung overdistention, rather than high airway pressure per se, appears to

344

Laffey and Kavanagh

Table 1 Evolution of Concepts Regarding CO2 Traditional Hypercapnia

Avoidance

Hypocapnia

Advantageous/ benign

Contemporary

Therapeutic hypercapnia

Tolerate (bystander) Advantageous/ benign

Deliberately elevate Potentially pathogenic

Abbreviation: CO2, carbon dioxide.

be the chief cause of mechanical injury, the term volutrauma (as opposed to barotrauma) has been used to describe this injury (17). In addition, increased mechanical stress appears to directly activate the cellular and humoral immune response in the lung (20–22), although the exact role of this mechanism in the pathogenesis of lung and systemic organ injury remains unclear (23,24). The potential for intrapulmonary mediators and pathogens to access the systemic circulation is clear from experiments demonstrating the translocation of intrapulmonary prostaglandins (25), cytokines (21), endotoxins (26), and bacteria (27) across an impaired alveolar–capillary barrier following high stretch mechanical ventilation. VALI may be limited by permitting hypoventilation in order to reduce mechanotrauma and the resulting inﬂammatory effects. This invariably involves a reduction in the tidal volume, and generally leads to an elevation in PaCO2, an approach that has been termed ‘‘permissive hypercapnia’’ (Table 1). These protective lung ventilation strategies have been demonstrated to improve survival in ARDS patients (6–8,12). Current paradigms attribute the protective effect of ventilatory strategies incorporating permissive hypercapnia solely to reductions in lung stretch, with hypercapnia permitted in order to achieve this goal. However, protective ventilatory strategies that involve hypoventilation result in both the limitation of lung stretch and the elevation of systemic PCO2. Lung stretch is distinct from elevated PCO2, and can be separately controlled, at least to some extent, by manipulation of the respiratory parameters (frequency, tidal volume, dead-space, inspired CO2, etc.). The potential exists for hypercapnia to contribute to the beneﬁcial effects of low-stretch ventilation strategies. If hypercapnia was proven to have independent beneﬁt, then deliberately elevating PaCO2 might provide an additional advantage over reducing lung stretch alone. Conversely, in patients managed with conventional permissive hypercapnia, adverse effects of elevated PaCO2 might be concealed by the beneﬁts of lessened lung stretch. These issues are further underlined by the fact that hypercapnia has potentially severe adverse effects in some clinical settings such as critically elevated intracranial pressure or pulmonary vascular resistance. Because patient outcome following a

Modulation of Lung Injury by Hypercapnia

345

critical illness might be related to a systemic injury rather than just a lung injury, consideration of the effects of hypercapnia on the heart and brain, as well as on the lung, is required. This has led several researchers to investigate the direct effects of induced hypercapnia per se in models of lung and systemic organ injury. As will be explored further in the next section, these studies raise the possibility that deliberate induction of hypercapnia, termed ‘‘therapeutic hypercapnia,’’ may constitute a testable clinical approach to intensive care unit (ICU) patients (Table 1) (10). III. Hypercapnia—Physiologic Effects This section reviews clinical studies of permissive hypercapnia in lung injury states over the last 15 years. Extensive documentation of the detailed effects of hypercapnia on clinical and laboratory physiological responses are covered in detail in an outstanding review by Feihl and Perret (28). A critical point in assessing the effects of permissive hypercapnia on clinical physiology is the deﬁnition of permissive hypercapnia. Reduction of tidal volume does indeed result in permissive hypercapnia, but the altered pulmonary physiology occurs in association with the reduced tidal volume and the ensuing HCA, as well as the potential contribution of altered respiratory frequency. No clinical studies to date have been able to dissect out these changes; indeed it is difﬁcult to conceive of how that can be done. Nonetheless, conventional application of permissive hypercapnia involves reduction of tidal volume, some increase in respiratory frequency, and tolerance of the HCA. Reports from several case series suggest that profound levels of hypercapnia might be well tolerated (29,30). Indeed, one case report describes a 46-year-old man who, on undergoing general anesthesia, developed HCA with a PaCO2 of 375 and a pH of 6.6 (29). In this case, the HCA resolved with the institution of mechanical ventilation and was associated with no neurologic deﬁcits (29). The cardiopulmonary effects of permissive hypercapnia have been examined in various degrees of detail by several experimental groups (31–41). Reduction of tidal volume from approximately 10 mL/kg to approximately 7.7 mL/kg in adult trauma patients with ARDS, was associated with reduced levels of mean plateau pressure (from 45.4 to 36.7 cmH2O) as well as an increase in PaCO2 from 37.9 to 56.7 mmHg (31). In this study pH was signiﬁcantly decreased from 7.41 to 7.31, without changes in either the hemodynamics or the derived measures of oxygen delivery or consumption (31), and PaO2 was unaltered. However, the reduction in tidal volume was relatively modest and Feihl et al. (33) provided a more detailed assessment of pulmonary physiology associated with a larger reduction in tidal volume, from a mean of 10 to 6 mL/kg. They were able to document that the reduction of tidal volume markedly increased intrapulmonary shunt from 32% to 48%, and through the use

346

Laffey and Kavanagh

of multiple inert gas elimination technique (MIGET) scanning, documented that there was no effect on dispersion of ventilation to perfusion (33). In addition to the increase in shunt, the permissive hypercapnic regimen increased cardiac output, and decreased PaO2 from 109 to 92 mmHg. When the parameters were reverted to the prestudy conditions, and the cardiac output was increased using dobutamine, the shunt fraction remained at an intermediate level, and the PaO2 remained low (33). This suggested that the reduced PaO2 was a combined effect of reduced tidal volume, increased cardiac output and shunt fraction, as well as decreased alveolar ventilation (33). Similar ﬁndings (increased venous mixture, cardiac index as well as mean pulmonary artery pressure) were reported in the study of 11 patients with ARDS (35). Tidal volumes were reduced such that PaCO2 rose from mean values of 40.3 to 59.3 and pH decreased from 7.4 to 7.26 (35). In this case, the authors cautioned about the use of permissive hypercapnia where oxygenation or cardiac function was severely impaired or in the context of signiﬁcant pulmonary hypertension. Finally, Pfeiffer et al. (41) studied 22 patients with ARDS, categorized into hyperdynamic sepsis or nonseptic groups. Multiple inert gas uptake studies were performed on each of these patients, and the overall result from this study was that the permissive hypercapnia induced by tidal volume reduction was associated with an increase in intrapulmonary shunt (either statistically signiﬁcant or with a demonstrable trend), and maintenance or increase in arterial oxygen tension (41). The authors suggested that atelectasis, resulting from reduced tidal volume, resulted in increased intrapulmonary shunt, the effects of which were countered by increased mixed venous O2 associated with hypercapnia-induced increases in cardiac output (41). Such ﬁndings have been the subject of speculation using complex mathematical models, which although interesting, are limited by the number of validated variables that can be included (42). Complications of hypercapnia include worsening of pulmonary arterial hypertension. An elegant study by Puybasset et al. (32) documented that in patients with ARDS, inhaled nitric oxide has important effects in normocapnia and permissive hypercapnic conditions (32). During normocapnia, inhaled nitric oxide decreased the true pulmonary shunt but was ineffective in reducing pulmonary shunt during hypercapnia (32). However, inhaled nitric oxide increased arterial oxygenation by comparable amounts during normocapnic and hypercapnic conditions (32). The known pulmonary effects of permissive hypercapnia have been studied in detail in the presence and absence of tromethamine buffering (38). Weber et al. studied 12 patients with ARDS and following baseline parameters evaluation, induced permissive hypercapnia with a target PaCO2 of 80 mmHg for two hours (38). Patients in this study were then randomized to no correction of arterial pH versus pH correction using intravenous tromethamine. Permissive hypercapnia resulted in signiﬁcantly decreased systemic vascular resistance and increases in cardiac output.

Modulation of Lung Injury by Hypercapnia

347

In addition, myocardial contractility was decreased and mean pulmonary artery pressure increased (38). These ﬁndings were consistent with previously documented data (see above). However, buffering with tromethamine attenuated but did not fully reverse the altered myocardial contractility and pulmonary artery pressure changes (38). Thus, it could be argued that permissive hypercapnia is generally well tolerated in patients with ARDS and buffering with tris-hydroxymethyl aminomethane (THAM) (38) or judicious use of inhaled nitric oxide (32) might be important cotherapies. Beyond the pulmonary and cardiovascular effects of hypercapnia, Kiefer et al. (39) studied the effects of CO2 on splanchnic blood ﬂow and oxygen kinetics in patients who were suffering from acute lung injury. They reported that the resultant hypercapnia had no important effects on the splanchnic circulation; however in this study, CO2 was increased by the addition of ventilatory dead space and not by the reduction of tidal volume (39). Two studies of patients with congenital heart disease deserve mention (43,44). In the context of single ventricle physiology, pulmonary vascular resistance can be controlled by inducing either alveolar hypoxia or alveolar hypercapnia. One study documented that the addition of inspired CO2 increased cerebral oxygenation and mean arterial pressure whereas reducing inspired O2 was far less effective in elevating the pulmonary vascular resistance following cavo-pulmonary connection (43). Hypoventilation will result in arterial hypercapnia and may, through reduced intrathoracic pressure, augment pulmonary vascular ﬂow. In the context of buffered hypercapnia, reduced minute ventilation has been demonstrated to increase cerebral blood ﬂow as well as systemic oxygenation (44). Furthermore, reduction of tidal volume (level of 6 mL/kg) while attempting to maintain a minute ventilation with increased respiratory frequency is associated with signiﬁcant respiratory acidosis. It has also been reported that such a strategy results in gas trapping and intrinsic positive end-expiratory pressure (PEEP) necessitating the reduction of external PEEP (45). Indeed, another group corroborated these ﬁndings (46) and suggested that the reduced mortality observed with the ARDSNet strategy may have been partly due to the protective effect of increased PEEP. If this is true, then beneﬁcial effects associated with permissive hypercapnia may have three potential components: reduced tidal stretch, direct beneﬁt from HCA, and regional lung recruitment associated with auto-PEEP.

IV. Acute Organ Injury: Evidence That CO2 Is Protective There is a growing body of evidence suggesting that hypercapnia and acidosis exert biologically important beneﬁcial effects in experimental acute lung injury (ALI) (Table 2). However, as patients with ARDS tend not to die of hypoxia per se, but rather because of the development of multiorgan

348

Table 2 Published Studies of Induced Hypercapnic Acidosis in ALI Model Shibata et al., 1998 (47) Laffey et al., 2000 (48) Laffey et al., 2000 (49)

Sinclair et al., 2002 (51)

Ex vivo isolated perfused (rabbit) lung In vivo whole animal (rabbit) model

Strand et al., 2003 (52)

In vivo (premature lamb) model

Key ﬁndings

Lung free-radical–induced ALI Lung ischemia-reperfusion ALI

HCA attenuated indices of ALI (Kfc, PAW, PCAP, PISO) HCA attenuated indices of ALI (Kfc, PAW, PCAP, PISO)

Lung ischemia-reperfusion

Acidosis, hypercapnic more than metabolic, attenuated indices of ALI (Kfc, PAW, PCAP, PISO). Buffering HCA abolished protective effect HCA attenuated indices of ALI (lung permeability, A–a O2 gradient, compliance, PAW) and inﬂammation (BALF TNFa, free-radical injury). Mechanisms included attenuation of nitrotyrosine and apoptosis HCA attenuated indices of ALI (lung permeability, BALF protein, Kfc). Potential mechanism attenuation of lung NO formation HCA attenuated indices of ALI (lung permeability, A–a O2 gradient, compliance, histologic injury) and inﬂammation (BALF neutrophils) HCA well tolerated. HCA attenuated (compliance, histologic injury) indices of ALI. Trend toward reduced indices of inﬂammation with HCA

Lung ischemia-reperfusion

Ventilator-induced high lung stretch Ventilator-induced high lung stretch (VT 25 mL/kg for 4.0 hr) Moderate lung stretch (VT 12– 15 mL/kg, 0.5 hr) followed by (6–9 mL/kg, 5.5 hr)

Laffey and Kavanagh

Broccard et al., 2001 (50)

Ex vivo isolated perfused (rabbit) lung Ex vivo isolated perfused (rabbit) lung In vivo whole animal (rabbit) model

Injury process

In vivo whole animal (rabbit) model

Laffey et al., 2003 (54)

In vivo whole animal (rat) model

Rai et al., 2004 (55)

In vivo whole animal (rabbit) model

Laffey et al., 2004 (56)

In vivo whole animal (rat) model

Ventilator-induced moderate lung stretch (VT 12 mL/kg for 4.0 hr) Mesenteric ischemia-reperfusion

Lung surfactant depletion followed by ventilator-induced moderate lung stretch (VT 12 mL/kg for 2.5 hr) Endotoxin-induced ALI

HCA modestly improved (A–a O2 gradient) while hypocapnia worsened (PAW) indices of ALI. Effects of CO2 independent of surfactant HCA attenuated indices of ALI (lung permeability, A–a O2 gradient, compliance, PAW,). Degree of protection with HCA was dose dependent HCA effective when commenced following initiation of the injury process HCA did not attenuate indices of ALI (A–a O2 gradient, compliance, PAW). HCA attenuated BAL MCP-1 levels but not other indices of inﬂammation HCA attenuated ALI (A–a O2 gradient, compliance, PAW, Histologic injury), and inﬂammation (BALF neutrophils). Effective following initiation of injury, and decreased production of higher oxides of NO. Protective effects not mediated by inhibition of peroxynitrite-induced nitration

Modulation of Lung Injury by Hypercapnia

Laffey et al., 2003 (53)

Abbreviations: ALI, acute lung injury; IR, ischemia-reperfusion; Kfc, capillary ﬁltration coefﬁcient; PAW, peak airway pressure; PCAP, pulmonary capillary pressure; PISO, pulmonary capillary isographic pressure; VT, tidal volume; A–a O2 gradient, alveolar–arterial oxygen gradient; BALF, bronchoalveolar lavage ﬂuid; TNF-a, tumor necrosis factor-a; NO, nitric oxide; MCP-1, monocyte chemoattractant protein-1; HCA, hypercapnic acidosis.

349

350

Laffey and Kavanagh

failure (57), any consideration of HCA in critical illness must also include its effects in extrapulmonary organs. In this regard, HCA has been demonstrated to exert protective effects in experimental brain, myocardial, hepatic, and renal injury (Table 3). A. Acute Lung Injury

The ﬁrst experiments to examine the direct effects of HCA in the lung utilized the buffer-perfused isolated rabbit lung. Shibata et al. demonstrated that HCA attenuated the increases in lung permeability seen following free radical as well as ischemia-reperfusion–induced ALI (Fig. 2) (47). These investigators demonstrated that HCA attenuated the activity of xanthine oxidase (XO), a key enzyme complex in the pathogenesis of ischemia-reperfusion injury (47). In subsequent experiments, they investigated the contribution of acidosis and hypercapnia per se to the protective effects of HCA in ischemia-reperfusion–induced ALI in the ex vivo lung. Laffey et al. reported that, while both hypercapnic and normocapnic (i.e., metabolic) acidosis were protective, HCA was the most protective (48). Buffering of the HCA, i.e., buffered hypercapnia, attenuated its protective effect (48). In fact the degree of injury with ‘‘buffered hypercapnia’’ was not different from that seen in control lungs. Both hypercapnic and metabolic acidosis similarly inhibit XO activity in vitro, so the different levels of

Figure 2 Protection against reperfusion injury in the isolated perfused rabbit lung with 25% (HCA) verses 5% (normocapnia). Injury is assessed as elevation in the pulmonary microvascular ﬁltration coefﬁcient (Kf,c) before versus after reperfusion. Abbreviation: HCA, hypercapnic acidosis. Source: From Ref. 47.

Modulation of Lung Injury by Hypercapnia

351

protection seen could not be explained solely on the basis of extracellular activity (48). These ﬁndings, which raised the potential for protective effects of HCA in ischemia-reperfusion–induced ALI, were conﬁrmed and further developed in subsequent in vivo studies. In a whole animal rabbit model of primary pulmonary ischemia-reperfusion, produced by ligation of the left pulmonary hilum followed by reperfusion, HCA preserved lung mechanics, attenuated protein leakage, reduced pulmonary edema, and improved oxygenation compared to control conditions (49). HCA also attenuated indices of systemic injury, as indicated by an improved acid–base and lactate proﬁle, and attenuated free-radical–induced injury as evidenced by a reduction in isoprostane levels (49). In a subsequent in vivo study of secondary ALI, Laffey et al. reported that HCA directly attenuated indices of ALI such as oxygenation, lung mechanics, and lung permeability following mesenteric ischemia-reperfusion compared to control conditions (54). The degree of primary, that is, bowel injury was not inﬂuenced by HCA. The protective effects of HCA were not mediated via a decrease in pulmonary artery resistance. In fact, lung protection with hypercapnia occurred despite pulmonary artery pressures that were greater than those observed with normocapnia (54). The direct effects of HCA in the setting of VALI have been examined in both ex vivo and in vivo models. In a key study, Broccard et al. reported for the ﬁrst time the potential for HCA to attenuate ventilator-induced ALI in the isolated rabbit lung (Fig. 3) (50). HCA attenuated the increase in indices of lung permeability, including the ultraﬁltration coefﬁcient, lung weight gain, and the protein and hemoglobin concentrations in bronchoalveolar lavage (BAL) ﬂuid. Sinclair et al. subsequently demonstrated the potential for HCA to potently attenuate the physiologic and histologic indices of lung injury induced by high lung stretch (tidal volume of 25 mL/kg) (51). HCA markedly attenuated the increase in the physiologic indices of injury, such as lung plateau pressures, arterial oxygenation, and lung wet–dry weight. BAL protein concentration and cell count, and overall histologic lung injury score, were reduced by HCA in comparison to lungs exposed to control conditions (51). Alterations in systemic CO2 tension appear to exhibit more modest protective effects in the context of more clinically relevant tidal stretch. Strand et al. (52) examined the potential for HCA to attenuate mild ventilator-induced lung injury in surfactant-treated preterm lamb lungs. Animals were randomized to receive either hypercapnia (mean PaCO2 levels 95 mmHg) or control conditions, and to ventilation with moderate tidal lung stretch (tidal volume, 12–15 mL/kg) for 30 minutes followed by a conventional tidal volume (6–9 mL/kg) for 5.5 hours. They reported that HCA (mean PaCO2 levels 95 mmHg) was well tolerated with minimal hemodynamic effects, and that the postnatal hemodynamic adaptation appeared unaffected. HCA attenuated the physiologic indices of ALI and reduced

352

Laffey and Kavanagh

Figure 3 Panels (A) and (B) demonstrate the effects of HPNC ventilation, compared with control and HPHC ventilation, on lung weight gain and microvascular ﬂuid leakage at three time points throughout ventilation in the isolated perfused rabbit lung. Panels (C) and (D) demonstrate the amount of protein and hemoglobin leakage, respectively, at completion of the experiment. , p 0.05 within-group comparison (T3 vs. T1) y, p < 0.05 versus C group; z, p < 0.05 versus C group and HPHC group. Abbreviations: HPHC, high pressure hypercapnic; HPNC, high pressure normocapnic. Source: From Ref. 50.

inﬁltration of inﬂammatory cells into the airspaces. However, inﬁltration of inﬂammatory cells into the lung tissue was not altered. In addition, there was a consistent trend to reduced indicators of inﬂammation in the lung tissue and BAL ﬂuid with HCA, although these differences did not achieve statistical signiﬁcance (52). Laffey et al. examined the effects of hypocapnia, normocapnia, and hypercapnia in the context of a more clinically relevant high tidal volume strategy (tidal volume, 12 mL/kg; PEEP, 0 cmH2O; rate, 42 breaths/min) in the in vivo rabbit (53). They reported that CO2 modulated key physiologic indices of lung injury, including alveolar–arterial oxygen gradient and airway pressure, with hypocapnia being potentially deleterious and HCA being potentially protective (53). However, alterations in systemic CO2 tensions did not result in differences in static lung compliance or surfactant chemistry (total surfactant, aggregates, or composition) in this study (53).

Modulation of Lung Injury by Hypercapnia

353

HCA did not attenuate lung injury induced by surfactant depletion, an atelectasis-prone model of ALI, which mimics neonatal respiratory distress syndrome (55). Following surfactant depletion, animals were randomized to either an injurious (tidal volume, 12 mL/kg; PEEP, 0 cmH2O) or a protective ventilatory strategy (tidal volume, 5 mL/kg; PEEP, 12.5 cmH2O), and to receive either control conditions or HCA. Injurious ventilation resulted in a clear lung injury, as evidenced by a large alveolar–arterial O2 gradient, elevated peak airway pressure, increased protein leak, and impaired lung compliance. HCA did not alter these physiologic indices of ALI (55). However, hypercapnia did abolish the stretch-induced increase in BAL monocyte chemoattractant protein-1, although it did not affect any of the other mediators studied, raising questions regarding the pathogenic role of this cytokine in lung injury (55). Taken together, these ﬁndings suggest that while HCA substantially attenuates VALI due to excessive stretch, its effects in the context of more clinically relevant lung stretch may be more modest, and it does not attenuate the extent of lung damage due to collapse and re-expansion of the atelectatic lung. In the clinical setting, ARDS develops most commonly in the context of severe pulmonary or extrapulmonary sepsis, in both adults (58,59) and children (60–62). Sepsis is also a very common complication of ARDS due to other causes (63). Of all the causes of ARDS, sepsis is associated with the poorest outcome (60,61,64–66). The mechanisms that initiate lung injury in sepsis-induced ARDS are quite distinct from those that do so in the models of lung injury previously examined. Lipopolysaccharide, a key endotoxin of gram-negative bacteria, initiates lung injury by activating a speciﬁc receptor (toll-like receptor-4) of the innate immune system, a pathway that shows evolutionary conservation across a wide range of eukaryotic species (67,68). HCA has been demonstrated to directly protect against ALI induced by intratracheal endotoxin instillation, a model of sterile sepsis-induced ARDS (56). HCA attenuated the physiologic indices of lung injury, including lung compliance and oxygenation. In addition, HCA attenuated the increase in alveolar lung tissue density and the decrease in alveolar airspace density compared to control conditions (56). In most clinical scenarios, therapeutic intervention is only possible following initiation of the injury process. The process of ALI is generally well established before the presentation of the patient for speciﬁc therapy in an ICU. The therapeutic potential of HCA is underlined by the ﬁnding that it appears to be effective even when instituted after initiation of the lung injury process, in the setting of both mesenteric ischemia-reperfusion and endotoxin-induced ALI models (54,56). This contrasts with many other initially promising experimental strategies that demonstrate potential when used prior to the injury process, but lose their effectiveness when utilized after the development of organ injury, thus minimizing their potential clinical utility.

354

Laffey and Kavanagh B. Acute Myocardial Injury

HCA appears to protect the heart from the deleterious effects of ischemicreperfusion injury (Table 3). Nomura et al., in a model of cardioplegic ischemia in isolated blood-perfused neonatal lamb hearts, reported that acidiﬁcation of the perfusate during the early phase of reperfusion potentiated myocardial recovery following prolonged cold cardioplegic ischemia (70). Preparations exposed to the greatest degree of HCA demonstrated the best functional recovery. Conversely, the preparations exposed to comparable metabolic acidemia (pH 6.8), followed by buffering to normal pH with bicarbonate and THAM, were not protected (70). In contrast, Kitakaze et al. found that metabolic acidosis (pH 6.6) prevented myocardial stunning following global ischemia in the isolated Langendorff-perfused ferret heart, when compared to control conditions (69). A potential explanation for these divergent ﬁndings regarding acidosis in these studies is the fact that Nomura et al. buffered their acidotic perfusate back to normal pH using NaHCO3 and tromethamine (70), whereas Kitakaze et al. replaced the acidotic perfusate with a perfusate of normal pH (69). Kitakaze et al. (71) have also demonstrated the potential for acidosis to be protective in in vivo studies. They reported that reperfusion with both hypercapnic and metabolic acidotic reperfusates were equally effective in reducing infarct size in an in vivo canine model of left anterior descending coronary artery ischemia and reperfusion, when compared to control conditions. Preckel et al. (75) found that regional reperfusion with acidotic blood (produced by the addition of hydrochloric acid) following left anterior descending coronary artery occlusion and reperfusion reduced infarct size and improved myocardial function in in vivo anesthetized open-chest dogs. Possible mechanisms for the protective effects of acidosis include the reduction of calcium loading to the myocardium through Hþ inhibition of calcium uptake, and, in the case of HCA, the induction of coronary vasodilation. Nomura et al. found that the greatest coronary artery blood ﬂow, consistent with hypercapnia-induced vasodilation, was present with maximal hypercapnia (70). In contrast, increases in regional coronary artery blood ﬂow do not contribute to the protective effects of normocapnic acidosis (75). C. Acute Brain Injury

Several studies have demonstrated protective effects of hypercapnia in brain injury (Table 3). In the brain, HCA has been demonstrated to attenuate hypoxic–ischemic brain injury in the immature rat (72,73). Vannucci et al. developed a model of hypoxic–ischemic injury in the immature rat, comprised of unilateral common carotid artery ligation followed by exposure to hypoxia (FiO2 8%) combined with varying inspired CO2 concentrations of 0%, 3%, 6%, or 9%, to produce a range of PaCO2 levels from hypo- to hypercapnia, for a duration of two hours. Thereafter, the animals

Model Myocardial injury Kitakaze et al., 1988 (69) Nomura et al., 1994 (70) Kitakaze et al., 1997 (71) Neurologic injury Vannucci et al., 1995 (72)

Injury process

Ex vivo isolated perfused (ferret) heart Ex vivo isolated perfused (neonatal lamb) heart

Myocardial ischemia-reperfusion Myocardial ischemia-reperfusion

In vivo whole animal (rabbit) model

Myocardial ischemia-reperfusion

In vivo whole animal (rat) model

Unilateral common carotid artery occlusion followed by hypoxia Unilateral common carotid artery occlusion followed by hypoxia

Vannucci et al., 1997 (73)

In vivo whole animal (rat) model

Vannucci et al., 2001 (74)

In vivo whole animal (rat) model

Unilateral common carotid artery occlusion followed by hypoxia

Key ﬁndings Acidosis [metabolic] during reperfusion decreased myocardial infarct size HCA improved postischemic myocardial function. Metabolic acidosis to an equivalent pH did not improve postischemic function Acidosis (hypercapnic and metabolic) during reperfusion decreased myocardial infarct size HCA decreased histologic brain damage. Dose response seen with 6% CO2 more neuroprotective than 9% CO2 HCA decreased histologic brain damage. Mechanisms may include improved cerebral blood ﬂow and attenuation of NMDA receptor activation Severe HCA (15% CO2) worsened histologic brain damage

Modulation of Lung Injury by Hypercapnia

Table 3 Published Studies of Induced Hypercapnic Acidosis in Extrapulmonary Injury

Abbreviations: NMDA, N-methyl-D-aspartate; HCA, hypercapnic acidosis.

355

356

Laffey and Kavanagh

were returned to their dams until 30 days postnatal age, at which time a neuropathologic examination was performed. The results demonstrated that hypocapnia was deleterious, and that elevated CO2 was protective. However, as will be discussed later, the relationship between FiCO2, PaCO2 and neuroprotection was not linear. In a subsequent study, Vannucci et al. demonstrated that cerebral blood ﬂow was better preserved during hypercapnia, and that the greater oxygen delivery promoted cerebral glucose utilization and oxidative metabolism for optimal maintenance of tissue high-energy phosphate reserves (73). Furthermore, cerebrospinal ﬂuid glutamate levels were lowest with hypercapnia, giving rise to speculation that the inhibition of excitatory amino acid neurotransmitter secretion might be involved in central nervous system protection (73). Additional mechanistic insights include the ﬁndings that HCA reduces indices of reactive oxygen species–induced injury following hypoxia/reoxygenation-induced injury in the neonatal porcine brain (76), and that acidosis attenuates neuronal apoptosis (77). D. Other Organs

In isolated hepatocytes exposed to anoxia (78) and chemical hypoxia (79,80), acidosis markedly delays the onset of cell death. Correction of the pH actually accelerated cell death. This phenomenon may represent a protective adaptation against hypoxic and ischemic stress. Isolated renal cortical tubules exposed to anoxia have improved ATP levels on reoxygenation at a pH of 6.9, when compared with tubules incubated at pH 7.5 (78). V. Mechanisms of CO2-Induced Protection A clear understanding of the cellular and biochemical mechanisms underlying the protective effects of hypercapnia is essential for several reasons. It constitutes a prerequisite if we are to successfully translate laboratory ﬁndings regarding hypercapnia to the bedside, as it allows us to more clearly deﬁne the potential therapeutic utility of hypercapnia in ALI. Of particular importance, a greater understanding of the mechanisms of action of hypercapnia allows us to predict its potential side effects in the clinical context. This should facilitate the identiﬁcation of patient groups for which hypercapnia may have deleterious effects, and in whom hypercapnia should be avoided. A second advantage is that an increased mechanistic understanding facilitates extrapolation of these insights to a variety of other disease states. In this regard, the ﬁnding that the protective effects of HCA in stretch-induced lung injury appear independent of effects on the surfactant (53) may have implications for surfactant-deﬁcient disease states such as respiratory distress syndrome in the premature neonate. Finally, a greater understanding of the protective actions of hypercapnia

Modulation of Lung Injury by Hypercapnia

357

and acidosis in ALI may lead to the discovery of other promising therapeutic modalities for this devastating disease process. A. Hypercapnia—Protection via Acidosis vs. Hypercapnia

The protective effects of HCA in experimental lung and systemic organ injury appear to be primarily a function of the acidosis generated (48,81). In the isolated lung, the protective effect of HCA in ischemiareperfusion–induced ALI was greatly attenuated if the pH was buffered toward normal (48). There appeared to be no signiﬁcant protective effects detectable with buffered hypercapnia (Table 2). Conversely, normocapnic (i.e., metabolic) acidosis attenuates primary ischemia-reperfusion–induced ALI in an ex vivo model, although it is less effective than HCA in this model (48). The protective effects of HCA in models of systemic organ injury also appear to be a function of the acidosis. The myocardial protective effects of HCA are also seen with metabolic acidosis both in ex vivo (69) and in vivo (71,75) models. Metabolic acidosis appears to exert protective effects in other models of organ injury. In the liver, metabolic acidosis delays the onset of cell death in isolated hepatocytes exposed to anoxia (78) and to chemical hypoxia (79,80). Correcting the pH to 7.4 abolished the protective effect and in fact accelerated hepatocyte cell death (79). Furthermore, isolated renal cortical tubules exposed to anoxia have improved ATP levels on reoxygenation at acidotic- compared to alkalotic-environmental pH levels (78). The type of acidosis, i.e., hypercapnic versus metabolic, does appear to be of importance. While metabolic acidosis attenuates primary ischemiareperfusion–induced ALI in an ex vivo model, it is less effective than HCA (48). In addition, Pedoto et al. have reported the potential for lung (82) and intestinal (83) injury following induction of metabolic acidosis by hydrochloric acid infusion in whole animal models. However, it is important to recognize that infusion of hyperosmolar solutions of strong acids into whole animal preparations may produce toxic effects unrelated to any change in pH (84). This is due to the localized damaging effects of strong acids within the bloodstream, at or near the point of infusion, which activate the inﬂammatory response, and may at least partially explain the ﬁndings of Pedoto et al. B. Anti-inﬂammatory Effects

Several key components of the inﬂammatory response, which contribute substantially to tissue injury and damage in ARDS patients, appear to be attenuated by HCA. HCA appears to interfere with the coordination of the immune response by reducing cytokine signaling (85–87). HCA inhibits the release of tumor necrosis factor-a (TNF-a) and interleukin-1 (IL-1)

358

Laffey and Kavanagh

from stimulated macrophages in vitro (85). The potential for HCA to attenuate pulmonary and systemic levels of key cytokines in vivo is clear from the ﬁnding that it decreased BAL levels of TNF-a following pulmonary ischemia-reperfusion (49). The cellular and molecular mechanisms underlying the inhibitory effects of HCA in the neutrophil are increasingly well understood. HCA modulates neutrophil expression of selectins and intercellular adhesion molecules, which are necessary for neutrophil binding to the vascular surface during inﬂammation (88). Hypercapnia and acidosis may impair neutrophil intracellular pH regulation. Intracellular pH decreases when neutrophils are activated by immune stimuli (87,89–91). If milieu pH is normal, there tends to be a recovery in neutrophil intracellular pH back toward normal levels. Hypercapnia decreases extracellular and intracellular pH in the local milieu, resulting in a rapid fall in neutrophil cytosolic pH (92– 94), potentially overwhelming the capacity of neutrophils, and in particular activated neutrophils (95), to regulate cytosolic pH. Failure to restore neutrophil cytosolic pH has been demonstrated to impair functions such as chemotaxis (96,97). The potential for HCA to attenuate neutrophil activity in vivo is clear from the ﬁnding that it attenuates lung neutrophil recruitment following ventilator-induced (51) as well as endotoxin-induced (56) ALI. C. Effects on Free-Radical Generation and Activity

HCA appears to attenuate free-radical production and modulate freeradical–induced tissue damage. In common with most biological enzymes, the enzymes that produce these oxidizing agents function optimally at neutral physiologic pH levels. Oxidant generation by both basal and stimulated neutrophils appears to be regulated by ambient CO2 levels, with oxidant generation reduced by hypercapnia and increased by hypocapnia (94). The production of superoxide by stimulated neutrophils in vitro is decreased at an acidic pH (93,98,99). In the brain, HCA attenuates glutathione depletion and lipid peroxidation, indices of free-radical activity and tissue damage, respectively (76). In the lung, HCA has been demonstrated to reduce freeradical tissue injury following pulmonary ischemia-reperfusion (49). HCA appears to attenuate the production of the higher oxides of NO, such as NO2 and NO3, following both ventilator-induced (50) and endotoxin-induced (56) ALI. HCA inhibits ALI mediated by xanthine oxidase (XO), a complex enzyme system produced in increased amounts during periods of tissue injury, which is a potent source of free radicals, in the isolated lung (47,48). D. Regulation of Gene Expression

HCA has been demonstrated to regulate the expression of genes central to the inﬂammatory response. Nuclear factor kappa B (NF-jB) is a key regulator of the expression of multiple genes involved in the inﬂammatory

Modulation of Lung Injury by Hypercapnia

359

Figure 4 Panel (A) demonstrates a gel shift assay with a time-dependent increase of NF-jB to the nucleus, following stimulation with LPS. The attenuation of this effect under conditions of HA contrasts with that observed under control (NC) conditions. Panel (B) demonstrates differential effects of HA, IA, and BH on LPSinduced DNA binding activity of NF-jB. HA, IA, and BH attenuated LPS-induced NF-jB activation, but HA showed the greatest effect. Panel (C) demonstrates nuclear extracts from HPAEC without LPS (lane 1) and with 1-h LPS stimulation (lanes 2–5) under NC. Lane 3: presence of 100-fold excess of unlabeled NF-jB. Lane 4: presence of anti-p50 monoclonal antibody. Lane 5: pressence of anti-p65 monoclonal antibody. Abbreviations: HA, hypercapnic acidosis; LPS, lipopolysaccharides; NC, normocapnic; NF-jB, nuclear factor kappa-B; HA, hypercapnic acidosis; IA, isocopnic acidosis; BH, buffered hypercapnia; HPAEC, human pulmonary artery endothelial cells. Source: From Ref. 101.

response and its activation represents a pivotal early step in the activation of the inﬂammatory response (100). NF-jB is found in the cytoplasm in an inactive form bound to inhibitory proteins called inhibitory protein jB (IjB), of which the important isoforms are IjB-a and IjB-b. IjB proteins are phosphorylated by the IjB kinase complex and subsequently degraded, thus allowing NF-jB to translocate into the nucleus, bind to speciﬁc promoter sites, and activate target genes (100). HCA has been demonstrated to signiﬁcantly inhibit endotoxin-induced NF-jB activation and DNA binding activity in human pulmonary endothelial cells via a mechanism mediated through a decrease in IjB-a degradation (Fig. 4) (101). HCA was demonstrated to suppress endothelial cell production of intercellular adhesion molecule-1 (ICAM) and IL-8 mRNA and protein, which are thought to be mainly regulated by the NF-jB–related pathway, and suppressed indices of cell injury (101).

360

Laffey and Kavanagh VI. Molecular Mechanisms of Hypercapnia-Induced Tissue Injury A. Tissue Nitration

Concerns exist regarding the potential for hypercapnia to potentiate tissue nitration by peroxynitrite, a potent free radical. Peroxynitrite is produced in vivo largely by the reaction of nitric oxide with superoxide radical, and causes tissue damage by oxidizing a variety of biomolecules and by nitrating phenolic amino acid residues in proteins (102–106). The potential for buffered hypercapnia to promote the formation of nitration products from peroxynitrite has been clearly demonstrated in recent in vitro experiments (81,107). However, the potential for HCA to promote nitration of lung tissues in vivo appears to depend on the injury process. HCA decreased tissue nitration following pulmonary ischemia-reperfusion–induced ALI (49), but increased nitration following endotoxin-induced lung injury (Fig. 5) (56). B. Sepsis

The potential for hypercapnia to exert deleterious effects in the context of live bacterial sepsis is increasingly a matter of concern (56,108). The importance of these concerns is clear given the prevalence of sepsis as a cause of admission to ICU (109), the frequency of nosocomial infection in the critically ill (110), and the fact that severe sepsis associated with multiorgan failure remains a leading cause of death in these patients. Laboratory studies of HCA to date have been in sterile, nonsepsis models of acute lung and systemic organ injury (108). While HCA has been shown to be protective against endotoxin-induced lung injury (56), this pathway is only one of several mechanisms by which live proliferating bacteria cause lung injury. Hypercapnia and/or acidosis may modulate the interaction between host and bacterial pathogen via several mechanisms, as discussed earlier. The potent anti-inﬂammatory properties of HCA may impair the host response to live bacterial sepsis. The potential for hypercapnia to alter intracellular pH regulation may inhibit neutrophil, microbicidal (96,97), and chemotactic activity (111). The production of free radicals such as the superoxide radical (O2), hydrogen peroxide (H2O2), and hypochlorous acid (HOCl) is central to the bactericidal activity of neutrophils and macrophages. The potential for HCA to attenuate free-radical production is clear. This is of importance given that the phagocytic activity and bactericidal capacity of neutrophils and macrophages is central to an effective host response to invading bacteria. Acidosis may render some antibiotics less effective (112). In addition, acidosis may alter the mechanism of neutrophil cell death from apoptosis to necrosis, which may result in increased tissue destruction (87,113). Conversely, hypercapnia may retard pathogen growth, and thereby decrease the overall septic insult (114,115). At the

Modulation of Lung Injury by Hypercapnia

361

Figure 5 Following intratracheal administration of lipolysaccharide tissue nitrotyrosine formation was increased in the setting of added CO2 (B) and less in the absence of added CO2 (A). This was despite the ﬁnding that addition of CO2 lessened all other indices of lung injury (see text). Source: From Ref. 56.

362

Laffey and Kavanagh

cellular level, mitochondrial dysfunction and cellular dysoxia are central to the pathogenesis of sepsis (116,117). Hypercapnia might favorably modulate cellular supply–demand balance in favor of cellular survival, given its effects in other contexts (10). However, the potential interactions between hypercapnia and sepsis at a cellular level remain to be elucidated. The overall effect of the degree of hypercapnia seen with protective lung ventilation on the host response to sepsis remains unclear. Many in vitro studies examining the effects of CO2 on indices of immune function utilize levels well beyond those seen in the clinical context. Nevertheless, the potential for hypercapnia to exert deleterious effects in the context of sepsis, and to result in signiﬁcant adverse consequences, is clear.

VII. Administration and Dose Response Hypercapnia can be achieved by reduction of minute ventilation, addition of circuit dead space, or directly adding CO2 to the inspired gas. In permissive hypercapnia, the obvious primary aim is to reduce the tidal volume, and thus any increase in PaCO2 that results will be because of reduced minute ventilation. When deliberately inducing hypercapnia in the experimental setting (or, as in previous eras, for hastening emergence from general anesthesia), reliance on reduced minute ventilation or on the addition of circuit dead space has the disadvantage of uneven distribution of CO2 to poorly perfused lung regions. This is not an issue with direct addition of CO2 to the inspired gas because as part of the inspired gas, it is distributed in accordance with the alveolar ventilation; thus, underperfused lung regions are exposed to the same inspired CO2 concentration as other regions (108). This point might be important if the investigator is seeking a lung-protective effect from the deliberate hypercapnia. Indeed, recent work from Swenson’s group in Seattle exploited the principles of phased delivery of inspired gas (118). These investigators demonstrated that addition of CO2 to the late inspiratory phase during mechanical ventilation resulted in the effects of CO2 being restricted to the airways and airspaces, and having little pulmonary vascular effects (118). This was conﬁrmed by the MIGET scan data; indeed, the systemic effects were striking, because addition of CO2 late in inspiration resulted in signiﬁcant augmentation of oxygenation, which was comparable to that achieved with addition of CO2 throughout the ventilatory cycle, but with signiﬁcantly less systemic PaCO2 and pH changes (118). This technique might allow one to obtain beneﬁcial effects from ‘‘therapeutic hypercapnia’’ while minimizing systemic acid–base effects. There is experimental evidence that the beneﬁcial effects of moderate hypercapnia may be counterbalanced by a potential for adverse effects at higher levels. This is supported by the experimental evidence demonstrating

Modulation of Lung Injury by Hypercapnia

363

Figure 6 The addition of CO2 resulted in a dose-dependent decrease in lung injury (within the ranged illustrated), following in vivo mesenteric ischemia-reperfusion. Source: From Ref. 54.

that protection from the adverse effects of brain ischemia was better when the inspired CO2 was set at 6% (corresponding to PaCO2 of 54 mmHg), rather than at 9% (72). Of concern, in a more recent study by this group, severe hypercapnia produced by 15% CO2 has been demonstrated to worsen neurologic injury in this context (74). In the lung, the degree of protection seen with HCA following mesenteric reperfusion-induced lung injury is dose dependent (Fig. 6) (54). Laffey et al. demonstrated an incremental increase in the degree of protection seen at doses of CO2 from 2.5% to 10%. However, there was little additional lung protection at inspired CO2 concentrations above 5%, and increasing hemodynamic instability was seen with 10% CO2, while doses of 20% CO2 were not tolerated in this study. The potential for hypercapnia to increase H2O was clearly demonstrated, with a stepwise increase in preinjury H2O demonstrated with increasing FiCO2 (54). Finally, in isolated hepatocytes, the degree of protection from anoxic injury conferred by a metabolic acidosis was greater at pH 6.9 than at pH 6.6 (78).

VIII. Role of Buffering Buffering of the acidosis induced by hypercapnia in ARDS patients remains a common, albeit controversial clinical practice (119,120). In fact, buffering

364

Laffey and Kavanagh

with sodium bicarbonate was permitted in the ARDSnet study, and cited as a potential reason for the success of the study (9), although this was subsequently disputed (121). While it is widely accepted that reduction in pH has profound effects on normal tissue function, it is also clear that hypercapnia per se, in the absence of alterations in pH, may exert biologically important physiologic effects distinct from those produced by acidosis. Of potential importance in the context of ALI, hypercapnia per se exerts effects on systemic (122) and pulmonary vascular tone (122,123) and pulmonary vascular remodeling (124) that are increasingly well characterized. Thus, the protective effects of HCA may be a function of the acidosis or the hypercapnia per se. This issue is of particular relevance when considering the appropriateness of buffering in the clinical context. Acidosis is common in a critical illness and is often a poor prognostic sign. However, as previously discussed, this effect is associative rather than causative, and patient prognosis depends on the underlying condition rather than the acidosis per se. If any protective effects of HCA were found to result from the acidosis, then efforts to buffer HCA would lessen such protection and should be discouraged. Conversely, if hypercapnia per se (and not the acidemia) were found to be protective, then further research efforts should be directed toward ﬁnding better buffering strategies in order to maximize the beneﬁts of hypercapnia. In this regard, as previously discussed, there is increasing evidence that the protective effects of HCA in experimental models of both acute lung and systemic organ injury appear to be a function of the acidosis, rather than elevated CO2 per se. Speciﬁc concerns exist regarding the use of bicarbonate to correct an acidosis. These concerns have resulted in the removal of bicarbonate therapy from routine use in cardiac arrest algorithms (125,126). The effectiveness of bicarbonate infusion as a buffer is dependent on its ability to excrete CO2, rendering it less effective in buffering HCA. In fact, bicarbonate may further raise systemic CO2 levels under conditions of reduced alveolar ventilation, such as ARDS (127). Furthermore, while bicarbonate may correct arterial pH, it may worsen an intracellular acidosis, because the CO2 produced when the bicarbonate reacts with metabolic acids diffuses readily across cell membranes, whereas the bicarbonate cannot (128). Bicarbonate may also exert detrimental effects when used to buffer a lactic acidosis. The potential for bicarbonate infusion to augment the production of lactic acid has been demonstrated in the experimental and clinical setting (129–135). Bicarbonate infusion exerted deleterious cardiovascular effects in a model of hypoxia-induced lactic acidosis (132,133). The safety of bicarbonate in diabetic patients has also been questioned. Bicarbonate administration slowed the rate of decrease of ketoacids in patients with diabetic ketoacidosis (136). Of even more concern, bicarbonate administration is associated with a fourfold increase in the risk of cerebral edema in children with diabetic ketoacidosis (137).

Modulation of Lung Injury by Hypercapnia

365

The administration of sodium bicarbonate constitutes a signiﬁcant osmolar load, which may exert beneﬁcial effects independent of any associated changes in pH. Osmolar loads, such as hypertonic saline, may improve the hemodynamic proﬁle in hemorrhagic shock (138), attenuate key aspects of the immune response (138–140), and prevent organ injury in experimental models (139–141). In fact, when compared to an equimolar dose of sodium chloride, bicarbonate administration does not improve the hemodynamic status of critically ill patients who have lactic acidosis (142). A follow-up study in an in vivo model of lactic acidemia found that bicarbonate exerted hemodynamic effects (mean arterial pressure, cardiac output, left ventricular contractility, etc.), which were indistinguishable from those seen in response to an equimolar dose of sodium chloride (143). These data give cause for concern about the practice of buffering metabolic acidosis, and comparable questions may exist in the setting of HCA. These concerns do not exclude a role for the use of other buffers, such as the amino alcohol tromethamine (THAM), in speciﬁc situations where the physiologic effects of HCA are of concern. THAM penetrates cells easily and can buffer pH changes and simultaneously reduce PCO2 (144), making it effective in situations where CO2 excretion is limited, such as ARDS (38). THAM has been demonstrated to ameliorate the hemodynamic consequences of a rapidly induced HCA in ARDS patients (38). In summary, although it is a widely accepted clinical practice, there are no long-term clinical outcome data (e.g., survival, duration of hospital stay, etc.) to support the practice of buffering HCA. Taken together, the above literature suggests that, in the absence of correcting the primary problem, buffering HCA with bicarbonate is not likely to be of beneﬁt. If the clinician elects to buffer HCA, the rationale for this practice should be clear (e.g., to ameliorate potentially deleterious hemodynamic consequences of acidosis). THAM may have a role in these clinical situations.

IX. Hypercapnia—Clinical Studies Clinical studies of hypercapnia in ARDS include the two studies by Hickling’s group, one retrospective (7) and the other, a prospective (casecontrolled) study (8). Both of these studies, as well as the earlier case series of permissive hypercapnia by Wung et al. in neonates (6) and Darioli and Perret in adults with status asthmaticus (5), strongly suggested survival beneﬁts with the permissive hypercapnia approach (i.e., pressure/ volume limitation). The ﬁrst––and to date—only randomized clinical trial of the technique of permissive hypercapnia was conducted by Mariani et al., in 49 children with neonatal respiratory failure (145). Although not powered to detect differences in survival, an important intermediate outcome (i.e., duration

366

Laffey and Kavanagh

Figure 7 The PaCO2 and duration of mechanical ventilation in a randomized, prospective, clinical study of permissive hypercapnia in neonatal respiratory failure. (A) illustrates the PaCO2 levels, and (B) illustrates the duration of mechanical ventilation over time. Source: From Ref. 145.

of mechanical ventilation) was signiﬁcantly shorter in the permissive hypercapnia group, in the absence of obvious adverse effects (Fig. 7). However, although encouraging, it must be recognized that such a small study would not detect a subtle incidence of adverse effect. In a potentially important preliminary communication (146), Kregenow et al. have exploited the ARDS Network database (9), and examined mortality as a function of permissive hypercapnia on day 1 postrandomization in patients enrolled in the ARDSNet tidal volume study. Using multivariate logistic regression analysis, and controlling for other comorbidities and severity of lung injury, they found that in the high tidal volume arm of the study, permissive hypercapnia was an independent predictor of survival (146). However, there was no additional protective effect of permissive hypercapnia in patients randomized to receive the lower tidal volume (6 mL/kg) (146). X. Future Directions A clearer understanding of the effects and mechanisms of action of hypercapnia and acidosis is essential in order to facilitate identiﬁcation of the optimum response to, and tolerance of, hypercapnia in the setting of protective ventilator strategies, and to more clearly deﬁne the safety and potential therapeutic utility of hypercapnia in ARDS. An important limitation when considering the clinical implications of current experimental insights regarding hypercapnia is the relatively short duration of the ALI models in which HCA has been studied to date. The common clinical scenario in ARDS patients is that of a more prolonged hypercapnia, during which time the acidosis may be partially, or even completely, compensated. As we have

Modulation of Lung Injury by Hypercapnia

367

seen, there is reason to believe that the acidosis generated by acute hypercapnia may be the protective factor in acute models of ALI. Future studies should address the effects of hypercapnia in ALI models of considerably longer duration, which more closely replicate the clinical context. The effects of hypercapnia in sepsis must be further explored.

XI. Summary Hypercapnia is a central component of current protective lung ventilatory strategies in the clinical context. The data linking ventilatory strategies involving hypercapnia to improved outcome are persuasive, although not conclusive. Furthermore, induced HCA appears to demonstrate considerable protective effects in several laboratory models of acute lung and systemic organ injury. However, there are concerns regarding the potential for hypercapnia and/or acidosis to exert deleterious effects, which suggest the need for caution and further investigation into the effects of hypercapnia in the clinical context. At present, ventilatory strategies that involve hypercapnia are clinically acceptable only provided the clinician is primarily targeting reduced tidal stretch. There are insufﬁcient clinical data to suggest that hypercapnia per se should be independently induced, outside the context of a protective ventilatory strategy. References 1. Edmunds LH Jr, Holm JC. Effect of inhaled CO2 on hemorrhagic consolidation due to unilateral pulmonary arterial ligation. J Appl Physiol 1969; 26: 710–715. 2. Edmunds LH Jr, Holm JC. Effect of atelectasis on lung changes after pulmonary arterial ligation. J Appl Physiol 1968; 25:115–123. 3. Trimble C, Smith DE, Rosenthal MH, Fosburg RG. Pathophysiologic role of hypocarbia in post-traumatic pulmonary insufﬁciency. Am J Surg 1971; 122: 633–638. 4. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110:556–565. 5. Darioli R, Perret C. Mechanical controlled hypoventilation in status asthmaticus. Am Rev Respir Dis 1984; 129:385–387. 6. Wung JT, James LS, Kilchevsky E, James E. Management of infants with severe respiratory failure and persistence of the fetal circulation, without hyperventilation. Pediatrics 1985; 76:488–494. 7. Hickling KG, Henderson SJ, Jackson R. Low mortality associated with low volume pressure limited ventilation with permissive hypercapnia in severe adult respiratory distress syndrome. Intensive Care Med 1990; 16:372–377.

368

Laffey and Kavanagh

8. Hickling KG, Walsh J, Henderson S, Jackson R. Low mortality rate in adult respiratory distress syndrome using low-volume, pressure-limited ventilation with permissive hypercapnia: a prospective study. Crit Care Med 1994; 22:1568–1578. 9. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342:1301–1308. 10. Laffey JG, Kavanagh BP. Carbon dioxide and the critically ill – too little of a good thing? (Hypothesis Paper). Lancet 1999; 354:1283–1286. 11. Laffey JG, Kavanagh BP. Hypocapnia. N Engl J Med 2002; 347:43–53. 12. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338:347–354. 13. Pinhu L, Whitehead T, Evans T, Grifﬁths M. Ventilator-associated lung injury. Lancet 2003; 361:332–340. 14. Boussarsar M, Thierry G, Jaber S, Roudot-Thoraval F, Lemaire F, Brochard L. Relationship between ventilatory settings and barotrauma in the acute respiratory distress syndrome. Intensive Care Med 2002; 28:406–413. 15. Maggiore SM, Jonson B, Richard JC, Jaber S, Lemaire F, Brochard L. Alveolar derecruitment at decremental positive end-expiratory pressure levels in acute lung injury: comparison with the lower inﬂection point, oxygenation, and compliance. Am J Respir Crit Care Med 2001; 164:795–801. 16. Ricard JD, Dreyfuss D, Saumon G. Ventilator-induced lung injury. Curr Opin Crit Care 2002; 8:12–20. 17. Dreyfuss D, Saumon G. Barotrauma is volutrauma, but which volume is the one responsible? Intensive Care Med 1992; 18:139–141. 18. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 19. Dreyfuss D, Saumon G. From ventilator-induced lung injury to multiple organ dysfunction? Intensive Care Med 1998; 24:102–104. 20. Slutsky AS, Tremblay LN. Multiple system organ failure. Is mechanical ventilation a contributing factor? Am J Respir Crit Care Med 1998; 157:1721–1725. 21. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos mRNA expression in an isolated rat lung model. J Clin Invest 1997; 99:944–952. 22. Tremblay LN, Slutsky AS. Ventilator-induced injury: from barotrauma to biotrauma. Proc Assoc Am Physicians 1998; 110:482–488. 23. Ricard JD, Dreyfuss D. Cytokines during ventilator-induced lung injury: a word of caution. Anesth Analg 2001; 93:251–252. 24. Dreyfuss D, Ricard JD, Saumon G. On the physiologic and clinical relevance of lung-borne cytokines during ventilator-induced lung injury. Am J Respir Crit Care Med 2003; 167:1467–1471. 25. Edmonds JF, Berry E, Wyllie JH. Release of prostaglandins caused by distension of the lungs. Br J Surg 1969; 56:622–623. 26. Murphy D, Cregg N, Tremblay L, et al. Adverse ventilator strategy causes pulmonary to systemic translocation of endotoxin. Am J Resp Crit Care Med 2000; 162:27–33.

Modulation of Lung Injury by Hypercapnia

369

27. Nahum A, Hoyt J, Schmitz L, Moody J, Shapiro R, Marini JJ. Effect of mechanical ventilation strategy on dissemination of intratracheally instilled Escherichia coli in dogs. Crit Care Med 1997; 25:1733–1743. 28. Feihl F, Perret C. Permissive hypercapnia. How permissive should we be? Am J Respir Crit Care Med 1994; 150:1722–1737. 29. Potkin RT, Swenson ER. Resuscitation from severe acute hypercapnia. Determinants of tolerance and survival. Chest 1992; 102:1742–1745. 30. Goldstein B, Shannon DC, Todres ID. Supercarbia in children: clinical course and outcome. Crit Care Med 1990; 18:166–168. 31. McIntyre RC Jr, Haenel JB, Moore FA, Read RR, Burch JM, Moore EE. Cardiopulmonary effects of permissive hypercapnia in the management of adult respiratory distress syndrome. J Trauma 1994; 37:433–438. 32. Puybasset L, Stewart T, Rouby JJ, et al. Inhaled nitric oxide reverses the increase in pulmonary vascular resistance induced by permissive hypercapnia in patients with acute respiratory distress syndrome. Anesthesiology 1994; 80: 1254–1267. 33. Feihl F, Eckert P, Brimioulle S, et al. Permissive hypercapnia impairs pulmonary gas exchange in the acute respiratory distress syndrome. Am J Respir Crit Care Med 2000; 162:209–215. 34. Roupie E, Dambrosio M, Servillo G, et al. Titration of tidal volume and induced hypercapnia in acute respiratory distress syndrome. Am J Respir Crit Care Med 1995; 152:121–128. 35. Thorens JB, Jolliet P, Ritz M, Chevrolet JC. Effects of rapid permissive hypercapnia on hemodynamics, gas exchange, and oxygen transport and consumption during mechanical ventilation for the acute respiratory distress syndrome. Intensive Care Med 1996; 22:182–191. 36. Mador MJ, Wendel T, Kufel TJ. Effect of acute hypercapnia on diaphragmatic and limb muscle contractility. Am J Respir Crit Care Med 1997; 155: 1590–1595. 37. Carvalho CR, Barbas CS, Medeiros DM, et al. Temporal hemodynamic effects of permissive hypercapnia associated with ideal PEEP in ARDS. Am J Respir Crit Care Med 1997; 156:1458–1466. 38. Weber T, Tschernich H, Sitzwohl C, et al. Tromethamine buffer modiﬁes the depressant effect of permissive hypercapnia on myocardial contractility in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2000; 162:1361–1365. 39. Kiefer P, Nunes S, Kosonen P, Takala J. Effect of an acute increase in PCO2 on splanchnic perfusion and metabolism. Intensive Care Med 2001; 27:775–778. 40. Natalini G, Seramondi V, Fassini P, et al. Acute respiratory acidosis does not increase plasma potassium in normokalaemic anaesthetized patients. A controlled randomized trial. Eur J Anaesthesiol 2001; 18:394–400. 41. Pfeiffer B, Hachenberg T, Wendt M, Marshall B. Mechanical ventilation with permissive hypercapnia increases intrapulmonary shunt in septic and nonseptic patients with acute respiratory distress syndrome. Crit Care Med 2002; 30:285–289. 42. Joyce CJ, Hickling KG. Permissive hypercapnia and gas exchange in lungs with high Qs/Qt: a mathematical model. Br J Anaesth 1996; 77:678–683.

370

Laffey and Kavanagh

43. Ramamoorthy C, Tabbutt S, Kurth CD, et al. Effects of inspired hypoxic and hypercapnic gas mixtures on cerebral oxygen saturation in neonates with univentricular heart defects. Anesthesiology 2002; 96:283–288. 44. Bradley SM, Simsic JM, Mulvihill DM. Hypoventilation improves oxygenation after bidirectional superior cavopulmonary connection. J Thorac Cardiovasc Surg 2003; 126:1033–1039. 45. Richard JC, Brochard L, Breton L, et al. Inﬂuence of respiratory rate on gas trapping during low volume ventilation of patients with acute lung injury. Intensive Care Med 2002; 28:1078–1083. 46. de Durante G, del Turco M, Rustichini L, et al. ARDSNet lower tidal volume ventilatory strategy may generate intrinsic positive end-expiratory pressure in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2002; 165:1271–1274. 47. Shibata K, Cregg N, Engelberts D, Takeuchi A, Fedorko L, Kavanagh BP. Hypercapnic acidosis may attenuate acute lung injury by inhibition of endogenous xanthine oxidase. Am J Resp Crit Care Med 1998; 158:1578–1584. 48. Laffey JG, Engelberts D, Kavanagh BP. Buffering hypercapnic acidosis worsens acute lung injury. Am J Resp Crit Care Med 2000; 161:141–146. 49. Laffey JG, Tanaka M, Engelberts D, et al. Therapeutic hypercapnia reduces pulmonary and systemic injury following in vivo lung reperfusion. Am J Respir Crit Care Med 2000; 162:2287–2294. 50. Broccard AF, Hotchkiss JR, Vannay C, et al. Protective effects of hypercapnic acidosis on ventilator-induced lung injury. Am J Respir Crit Care Med 2001; 164:802–806. 51. Sinclair SE, Kregenow DA, Lamm WJ, Starr IR, Chi EY, Hlastala MP. Hypercapnic acidosis is protective in an in vivo model of ventilator-induced lung injury. Am J Respir Crit Care Med 2002; 166:403–408. 52. Strand M, Ikegami M, Jobe AH. Effects of high PCO2 on ventilated preterm lamb lungs. Pediatr Res 2003; 53:468–472. 53. Laffey JG, Engelberts D, Duggan M, Veldheuizen R, Lewis J, Kavanagh BP. Carbon dioxide attenuates pulmonary impairment resulting from hyperventilation. Crit Care Med 2003; 31:2634–2640. 54. Laffey JG, Jankov RP, Engelberts D, et al. Effects of therapeutic hypercapnia on mesenteric ischemia-reperfusion injury. Am J Respir Crit Care Med 2003; 168:1383–1390. 55. Rai S, Engelberts D, Laffey JG, et al. Therapeutic hypercapnia is not protective in the in vivo surfactant-depleted rabbit lung. Pediatr Res 2004; 55: 42–49. 56. Laffey JG, Honan D, Hopkins N, Hyvelin JM, Boylan JF, McLoughlin P. Hypercapnic acidosis attenuates endotoxin-induced acute lung injury. Am J Respir Crit Care Med 2004; 169:46–56. 57. Montgomery AB, Stager MA, Carrico CJ, Hudson LD. Causes of mortality in patients with the adult respiratory distress syndrome. Am Rev Resp Dis 1985; 132:485–489. 58. Valta P, Uusaro A, Nunes S, Ruokonen E, Takala J. Acute respiratory distress syndrome: frequency, clinical course, and costs of care. Crit Care Med 1999; 27:2367–2374.

Modulation of Lung Injury by Hypercapnia

371

59. TenHoor T, Mannino DM, Moss M. Risk factors for ARDS in the United States: analysis of the 1993 National Mortality Followback Study. Chest 2001; 119:1178–1179. 60. Goh AY, Chan PW, Lum LC, Roziah M. Incidence of acute respiratory distress syndrome: a comparison of two deﬁnitions. Arch Dis Child 1998; 79:256–259. 61. Redding GJ. Current concepts in adult respiratory distress syndrome in children. Curr Opin Pediatr 2001; 13:261–266. 62. Paret G, Ziv T, Augarten A, et al. Acute respiratory distress syndrome in children: a 10 year experience. Isr Med Assoc J 1999; 1:149–153. 63. Markowicz P, Wolff M, Djedaini K, et al. Multicenter prospective study of ventilator-associated pneumonia during acute respiratory distress syndrome. Incidence, prognosis, and risk factors. ARDS Study Group. Am J Respir Crit Care Med 2000; 161:1942–1948. 64. Ferring M, Vincent JL. Is outcome from ARDS related to the severity of respiratory failure? Eur Respir J 1997; 10:1297–1300. 65. Eisner MD, Thompson T, Hudson LD, et al. Efﬁcacy of low tidal volume ventilation in patients with different clinical risk factors for acute lung injury and the acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 164:231–236. 66. Gupta D, Ramanathan RP, Aggarwal AN, Jindal SK. Assessment of factors predicting outcome of acute respiratory distress syndrome in North India. Respirology 2001; 6:125–130. 67. Beutler B. Endotoxin, toll-like receptor 4, and the afferent limb of innate immunity. Curr Opin Microbiol 2000; 3:23–28. 68. Lien E, Ingalls RR. Toll-like receptors. Crit Care Med 2002; 30:S1–S11. 69. Kitakaze M, Weisfeldt ML, Marban E. Acidosis during early reperfusion prevents myocardial stunning in perfused ferret hearts. J Clin Invest 1988; 82: 920–927. 70. Nomura F, Aoki M, Forbess JM, Mayer JE. Effects of hypercarbic acidotic reperfusion on recovery of myocardial function after cardioplegic ischemia in neonatal lambs. Circulation 1994; 90:321–327. 71. Kitakaze M, Takashima S, Funaya H, et al. Temporary acidosis during reperfusion limits myocardial infarct size in dogs. Am J Physiol 1997; 272: H2071–H2078. 72. Vannucci RC, Towﬁghi J, Heitjan DF, Brucklacher RM. Carbon dioxide protects the perinatal brain from hypoxic-ischemic damage: an experimental study in the immature rat. Pediatrics 1995; 95:868–874. 73. Vannucci RC, Brucklacher RM, Vannucci SJ. Effect of carbon dioxide on cerebral metabolism during hypoxia-ischemia in the immature rat. Pediatr Res 1997; 42:24–29. 74. Vannucci RC, Towﬁghi J, Brucklacher RM, Vannucci SJ. Effect of extreme hypercapnia on hypoxic-ischemic brain damage in the immature rat. Pediatr Res 2001; 49:799–803. 75. Preckel B, Schlack W, Obal D, et al. Effect of acidotic blood reperfusion on reperfusion injury after coronary artery occlusion in the dog heart. J Cardiovasc Pharmacol 1998; 31:179–186.

372

Laffey and Kavanagh

76. Barth A, Bauer R, Gedrange T, Walter B, Klinger W, Zwiener U. Inﬂuence of hypoxia and hypoxia/hypercapnia upon brain and blood peroxidative and glutathione status in normal weight and growth-restricted newborn piglets. Exp Toxicol Pathol 1998; 50:402–410. 77. Xu L, Glassford AJ, Giaccia AJ, Giffard RG. Acidosis reduces neuronal apoptosis. Neuroreport 1998; 9:875–879. 78. Bonventre JV, Cheung JY. Effects of metabolic acidosis on viability of cells exposed to anoxia. Am J Physiol 1985; 249:C149–C159. 79. Gores GJ, Nieminen AL, Fleishman KE, Dawson TL, Herman B, Lemasters JJ. Extracellular acidosis delays onset of cell death in ATP-depleted hepatocytes. Am J Physiol 1988; 255:C315–C322. 80. Gores GJ, Nieminen AL, Wray BE, Herman B, Lemasters JJ. Intracellular pH during ‘‘chemical hypoxia’’ in cultured rat hepatocytes. Protection by intracellular acidosis against the onset of cell death. J Clin Invest 1989; 83:386–396. 81. Lang JD Jr, Chumley P, Eiserich JP, et al. Hypercapnia induces injury to alveolar epithelial cells via a nitric oxide-dependent pathway. Am J Physiol Lung Cell Mol Physiol 2000; 279:L994–L1002. 82. Pedoto A, Caruso JE, Nandi J, et al. Acidosis stimulates nitric oxide production and lung damage in rats. Am J Respir Crit Care Med 1999; 159:397–402. 83. Pedoto A, Nandi J, Oler A, Camporesi EM, Hakim TS, Levine RA. Role of nitric oxide in acidosis-induced intestinal injury in anesthetized rats. J Lab Clin Med 2001; 138:270–276. 84. Shams H, Peskar BA, Scheid P. Acid infusion elicits thromboxane A2mediated effects on respiration and pulmonary hemodynamics in the cat. Respir Physiol 1988; 71:169–183. 85. West MA, Baker J, Bellingham J. Kinetics of decreased LPS-stimulated cytokine release by macrophages exposed to CO2. J Surg Res 1996; 63:269–274. 86. West MA, Hackam DJ, Baker J, Rodriguez JL, Bellingham J, Rotstein OD. Mechanism of decreased in vitro murine macrophage cytokine release after exposure to carbon dioxide: relevance to laparoscopic surgery. Ann Surg 1997; 226:179–190. 87. Coakley RJ, Taggart C, McElvaney NG, O’Neill SJ. Cytosolic pH and the inﬂammatory microenvironment modulate cell death in human neutrophils after phagocytosis. Blood 2002; 100:3383–3391. 88. Serrano CV, Fraticelli A, Paniccia R, et al. pH dependence of neutrophilendothelial cell adhesion and adhesion molecule expression. Am J Physiol 1996; 271:C962–C970. 89. Grinstein S, Furuya W. Cytoplasmic pH regulation in activated human neutrophils: effects of adenosine and pertussis toxin on Naþ/Hþ exchange and metabolic acidiﬁcation. Biochim Biophys Acta 1986; 889:301–309. 90. Grinstein S, Furuya W. Cytoplasmic pH regulation in phorbol ester-activated human neutrophils. Am J Physiol 1986; 251:C55–C65. 91. Grinstein S, Furuya W, Biggar WD. Cytoplasmic pH regulation in normal and abnormal neutrophils. Role of superoxide generation and Naþ/Hþ exchange. J Biol Chem 1986; 261:512–514.

Modulation of Lung Injury by Hypercapnia

373

92. Trevani AS, Andonegui G, Giordano M, et al. Extracellular acidiﬁcation induces human neutrophil activation. J Immunol 1999; 162:4849–4857. 93. Simchowitz L, Roos A. Regulation of intracellular pH in human neutrophils. J Gen Physiol 1985; 85:443–470. 94. Coakley RJ, Taggart C, Greene C, McElvaney NG, O’Neill SJ. Ambient pCO2 modulates intracellular pH, intracellular oxidant generation, and interleukin-8 secretion in human neutrophils. J Leukoc Biol 2002; 71:603–610. 95. Hackam DJ, Grinstein S, Nathens A, Watson WG, Marshall JC, Rotstein OD. Exudative neutrophils show impaired pH regulation compared with circulating neutrophils. Arch Surg 1996; 131:1296–1301. 96. Simchowitz L, Cragoe EJ Jr. Regulation of human neutrophil chemotaxis by intracellular pH. J Biol Chem 1986; 261:6492–6500. 97. Demaurex N, Downey GP, Waddell TK, Grinstein S. Intracellular pH regulation during spreading of human neutrophils. J Cell Biol 1996; 133:1391–1402. 98. Leblebicioglu B, Lim JS, Cario AC, Beck FM, Walters JD. pH changes observed in the inﬂamed gingival crevice modulate human polymorphonuclear leukocyte activation in vitro. J Periodontol 1996; 67:472–477. 99. Gabig TG, Bearman SI, Babior BM. Effects of oxygen tension and pH on the respiratory burst of human neutrophils. Blood 1979; 53:1133–1139. 100. Tak PP, Firestein GS. NF-kappaB: a key role in inﬂammatory diseases. J Clin Invest 2001; 107:7–11. 101. Takeshita K, Suzuki Y, Nishio K, et al. Hypercapnic acidosis attenuates endotoxin-induced nuclear factor-kappa B activation. Am J Respir Cell Mol Biol 2003; 29:124–132. 102. Beckman JS, Koppenol WH. Nitric oxide, superoxide, and peroxynitrite: the good, the bad, and ugly. Am J Physiol 1996; 271:C1424–C1437. 103. Pryor WA, Squadrito GL. The chemistry of peroxynitrite: a product from the reaction of nitric oxide with superoxide. Am J Physiol 1995; 268:L699–L722. 104. Stamler JS. Redox signaling: nitrosylation and related target interactions of nitric oxide. Cell 1994; 78:931–936. 105. Squadrito GL, Pryor WA. Oxidative chemistry of nitric oxide: the roles of superoxide, peroxynitrite, and carbon dioxide. Free Radic Biol Med 1998; 25:392–403. 106. van der Vliet A, Eiserich JP, Shigenaga MK, Cross CE. Reactive nitrogen species and tyrosine nitration in the respiratory tract: epiphenomena or a pathobiologic mechanism of disease? Am J Respir Crit Care Med 1999; 160:1–9. 107. Zhu S, Basiouny KF, Crow JP, Matalon S. Carbon dioxide enhances nitration of surfactant protein A by activated alveolar macrophages. Am J Physiol Lung Cell Mol Physiol 2000; 278(5):L1025–L1031. 108. Swenson ER. Therapeutic hypercapnic acidosis: pushing the envelope. Am J Respir Crit Care Med 2004; 169:8–9. 109. Rubenfeld GD. Epidemiology of acute lung injury. Crit Care Med 2003; 31:S276–S284. 110. Vincent JL, Bihari DJ, Suter PM, et al. The prevalence of nosocomial infection in intensive care units in Europe. Results of the European Prevalence of Infection in Intensive Care (EPIC) Study. EPIC International Advisory Committee. JAMA 1995; 274:639–644.

374

Laffey and Kavanagh

111. Rotstein OD, Fiegel VD, Simmons RL, Knighton DR. The deleterious effect of reduced pH and hypoxia on neutrophil migration in vitro. J Surg Res 1988; 45:298–303. 112. Simmen HP, Blaser J. Analysis of pH and pO2 in abscesses, peritoneal ﬂuid, and drainage ﬂuid in the presence or absence of bacterial infection during and after abdominal surgery. Am J Surg 1993; 166:24–27. 113. Zhu WH, Loh TT. Effects of Naþ/Hþ antiport and intracellular pH in the regulation of HL-60 cell apoptosis. Biochim Biophys Acta 1995; 1269:122–128. 114. Hays GL, Burroughs JD, Warner RC. Microbiological aspects of pressure packaged foods. Food Technol 1959; 13:567–570. 115. Dixon NM, Kell DB. The inhibition by CO2 of the growth and metabolism of micro-organisms. J Appl Bacteriol 1989; 67:109–136. 116. Brealey D, Brand M, Hargreaves I, et al. Association between mitochondrial dysfunction and severity and outcome of septic shock. Lancet 2002; 360:219–223. 117. Gutierrez G, Wulf ME. Lactic acidosis in sepsis: a commentary. Intensive Care Med 1996; 22:6–16. 118. Brogan TV, Robertson HT, Lamm WJ, Souders JE, Swenson ER. Carbon dioxide added late in inspiration reduces ventilation-perfusion heterogeneity without causing respiratory acidosis. J Appl Physiol 2004; 96:1894–1898. 119. Tobin MJ. Mechanical ventilation (review). N Engl J Med 1994; 330:1056–1061. 120. Kollef MH, Schuster DP. The acute respiratory distress syndrome (review). N Engl J Med 1995; 332:27–37. 121. Laffey JG, Kavanagh BP. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury [letter]. N Engl J Med 2000; 343:812. 122. Sweeney M, Beddy D, Honner V, Sinnott B, O’Regan RG, McLoughlin P. Effects of changes in pH and CO2 on pulmonary arterial wall tension are not endothelium dependent. J Appl Physiol 1998; 85:2040–2046. 123. Sweeney M, O’Regan RG, McLoughlin P. Effects of changes in pH and PCO2 on wall tension in isolated rat intrapulmonary arteries. Exp Physiol 1999; 84:529–539. 124. Ooi H, Cadogan E, Sweeney M, Howell K, O’Regan RG, McLoughlin P. Chronic hypercapnia inhibits hypoxic pulmonary vascular remodeling. Am J Physiol Heart Circ Physiol 2000; 278:H331–H338. 125. Levy MM. An evidence-based evaluation of the use of sodium bicarbonate during cardiopulmonary resuscitation. Crit Care Clin 1998; 14:457–483. 126. Grillo JA, Gonzalez ER. Changes in the pharmacotherapy of CPR. Heart Lung 1993; 22:548–553. 127. Sun JH, Filley GF, Hord K, Kindig NB, Bartle EJ. Carbicarb: an effective substitute for NaHCO3 for the treatment of acidosis. Surgery 1987; 102:835–839. 128. Goldsmith DJ, Forni LG, Hilton PJ. Bicarbonate therapy and intracellular acidosis. Clin Sci 1997; 93:593–598. 129. Abu Romeh S, Tannen RL. Amelioration of hypoxia-induced lactic acidosis by superimposed hypercapnea or hydrochloride acid infusion. Am J Physiol 1986; 250:F702–F709. 130. Arieff AI, Leach W, Park R, Lazarowitz VC. Systemic effects of NaHCO3 in experimental lactic acidosis in dogs. Am J Physiol 1982; 242:F586–F591.

Modulation of Lung Injury by Hypercapnia

375

131. Benjamin E, Oropello JM, Abalos AM, et al. Effects of acid-base correction on hemodynamics, oxygen dynamics, and resuscitability in severe canine hemorrhagic shock. Crit Care Med 1994; 22:1616–1623. 132. Graf H, Leach W, Arieff AI. Evidence for a detrimental effect of bicarbonate therapy in hypoxic lactic acidosis. Science 1985; 227:754–756. 133. Graf H, Leach W, Arieff AI. Metabolic effects of sodium bicarbonate in hypoxic lactic acidosis in dogs. Am J Physiol 1985; 249:F630–F635. 134. Fraley DS, Adler S, Bruns FJ, Zett B. Stimulation of lactate production by administration of bicarbonate in a patient with a solid neoplasm and lactic acidosis. N Engl J Med 1980; 303:1100–1102. 135. Rhee KH, Toro LO, McDonald GG, Nunnally RL, Levin DL. Carbicarb, sodium bicarbonate, and sodium chloride in hypoxic lactic acidosis. Effect on arterial blood gases, lactate concentrations, hemodynamic variables, and myocardial intracellular pH. Chest 1993; 104:913–918. 136. Okuda Y, Adrogue HJ, Field JB, Nohara H, Yamashita K. Counterproductive effects of sodium bicarbonate in diabetic ketoacidosis. J Clin Endocrinol Metab 1996; 81:314–320. 137. Glaser N, Barnett P, McCaslin I, et al. Risk factors for cerebral edema in children with diabetic ketoacidosis. The Pediatric Emergency Medicine Collaborative Research Committee of the American Academy of Pediatrics. N Engl J Med 2001; 344:264–269. 138. Rotstein OD. Novel strategies for immunomodulation after trauma: revisiting hypertonic saline as a resuscitation strategy for hemorrhagic shock. J Trauma 2000; 49:580–583. 139. Shields CJ, Sookhai S, Winter DC, et al. Attenuation of pancreatitis-induced pulmonary injury by aerosolized hypertonic saline. Surg Infect (Larchmt) 2001; 2:215–224. 140. Pascual JL, Khwaja KA, Ferri LE, et al. Hypertonic saline resuscitation attenuates neutrophil lung sequestration and transmigration by diminishing leukocyte-endothelial interactions in a two-hit model of hemorrhagic shock and infection. J Trauma 2003; 54:121–130; discussion 130–122. 141. Rizoli SB, Kapus A, Parodo J, Fan J, Rotstein OD. Hypertonic immunomodulation is reversible and accompanied by changes in CD11b expression. J Surg Res 1999; 83:130–135. 142. Cooper DJ, Walley KR, Wiggs BR, Russell JA. Bicarbonate does not improve hemodynamics in critically ill patients who have lactic acidosis. A prospective, controlled clinical study. Ann Intern Med 1990; 112:492–498. 143. Cooper DJ, Herbertson MJ, Werner HA, Walley KR. Bicarbonate does not increase left ventricular contractility during L-lactic acidemia in pigs. Am Rev Respir Dis 1993; 148:317–322. 144. Nahas GG, Sutin KM, Fermon C, et al. Guidelines for the treatment of acidaemia with THAM. Drugs 1998; 55:191–224. 145. Mariani G, Cifuentes J, Carlo WA. Randomized trial of permissive hypercapnia in preterm infants. Pediatrics 1999; 104:1082–1088. 146. Kregenow DA, Rubenfeld G, Hudson L, Swenson ER. Permissive hypercapnia reduces mortality with 12 mL/kg tidal volumes in acute lung injury. Am J Resp Crit Care Med 2003; 167:A616.

This page intentionally left blank

15 Alveolar Epithelial Function in Ventilator-Injured Lungs

JAMES A. FRANK and MICHAEL A. MATTHAY Cardiovascular Research Institute, University of California at San Francisco San Francisco, California, U.S.A.

I. Introduction A. Alveolar Epithelium in Acute Lung Injury

A healthy alveolar epithelium functions to maintain a surface for gas exchange, protect against invasion from airborne pathogens, regulate airspace ﬂuid content (including removing airspace edema ﬂuid), synthesize and secrete surfactants, and repair itself following injury. All of these functions are potentially impaired as a consequence of injurious mechanical ventilation (Table 1). Clinical studies have found that preservation of the alveolar epithelial barrier correlates with better outcomes including mortality in patients with acute lung injury (ALI) and the acute respiratory distress syndrome (ARDS) (1,2). For example, Ware and Matthay reported that ALI/ARDS patients with the most preserved epithelial ﬂuid–transport rates had a signiﬁcantly lower death rate and a shorter duration of mechanical ventilation compared with patients who had submaximal or absent alveolar ﬂuid transport capacity (2). Because mechanical ventilation induces abnormal mechanical stresses on the alveolar epithelium, ventilator-induced alveolar epithelial injury could signiﬁcantly affect outcomes in ARDS patients. 377

378

Frank and Matthay

Table 1 Functions of the Alveolar Epithelium Adversely Affected by Injurious Ventilation Function Gas exchange Barrier function, compartmentalization Ion and ﬂuid transport, regulation of airspace ﬂuid content Surfactant homeostasis

Potential mechanism for impairment Alveolar and airway ﬂooding, interstitial edema, loss of epithelial cells Epithelial cell injury, necrosis or apoptosis, increased paracellular permeability, increased transepithelial movement of bacteria and peptides Decreased sodium–potassium ATPase activity, decreased epithelial sodium channel expression and function, loss of cAMP-dependent transport, increased epithelial permeability Decreased surfactant protein expression, altered regulation of secretion, decreased surface active properties

Abbreviations: ATP, adenosine triphosphate; cAMP, cyclic adenosine monophosphate.

This hypothesis was supported by the recent ARDS Clinical Trials Network study comparing ventilation with 12 mL/kg tidal volumes and ventilation with 6 mL/kg tidal volumes (3). In this study, lower tidal volume ventilation resulted in an absolute reduction in mortality of 9%, translating into one life saved for every 11 ALI/ARDS patients managed with the low tidal volume strategy (3). This chapter primarily considers the effects of mechanical ventilation on alveolar epithelial barrier function including lung epithelial protein permeability and ion and ﬂuid transport. The effects of mechanical ventilation on epithelial-derived inﬂammatory mediators and surfactant function are brieﬂy reviewed; however, these topics are reviewed in more detail in Chapters 7, 8, 9, and 19 of this text. B. Stresses Encountered by the Alveolar Epithelium During Mechanical Ventilation

Nearly all patients with ARDS require mechanical ventilation and are therefore at risk for ventilator-induced alveolar epithelial injury. This risk appears to be due in part to the uneven distribution of lung injury and edema in ALI/ARDS. Studies using computed tomography (CT) scanning have demonstrated that the distribution of air and ﬂuid in the lungs of ALI/ ARDS patients is not uniform (4). Heterogeneity in the lung results in the functional reduction of lung volume and predisposes the lung to mechanical forces not encountered in normal physiology. These potentially pathogenic forces include excessive tensile strain (stretch) from overdistension and interdependence, and shear stress to the epithelial cells of the airspaces due to the movement of air and ﬂuid during tidal ventilation. The latter

Alveolar Epithelial Function in Ventilator-Injured Lungs

379

might be especially important when collapsed lung units are reexpanded. Under this paradigm, regions of the injured lung exist in one of three conditions: (i) ﬂuid-ﬁlled or collapsed and never inﬂated; (ii) collapsed or ﬂuid containing at end exhalation, but reexpanded at end inhalation; or (iii) aerated throughout the respiratory cycle, but prone to overdistension due to the uneven distribution of an inﬂated breath and interdependence. ‘‘Interdependence’’ refers to the forces exerted on an alveolus by the surrounding alveoli. In a normal lung, the alveolar distending force is equal to the transpulmonary pressure. In the injured lung, local distending forces will differ so as to counterbalance heterogeneity, and restore lung expansion (5). For example, Mead et al. (6) proposed that at a transpulmonary pressure of 30 cmH2O, the pressure across an atelectatic region surrounded by a fully expanded lung would be approximately 140 cmH2O. Therefore, in the heterogeneously injured lung, strain may be greater at areas where the inﬂated region is adjacent to atelectatic or ﬂuid-ﬁlled region, due to interdependence. Mechanical stresses on the alveolar epithelium may directly injure epithelial cells, resulting in breaks in the epithelial barrier. Mechanical stress could also trigger more subtle changes in alveolar epithelial cells, resulting from mechanical sensing and signaling by the cytoskeleton and plasma membrane, or indirectly via effects of mechanical forces on nonepithelial cells that interact with the epithelium. Mechanotransduction in ventilatorinduced lung injury (VILI) is reviewed in Chapters 1 and 4. II. Effects of Mechanical Ventilation on Alveolar Epithelial Barrier Function Type I and type II pneumocytes populate the alveolar barrier and function largely to restrict the movement of water and proteins from the interstitial space into the airspace. The alveolar epithelium also regulates airspace ﬂuid content. If pulmonary edema ﬂuid is present in the alveolus, alveolar epithelial cells actively transport ﬂuid from the airspace into the interstitial space (Fig. 1) (7). Disruption of the alveolar epithelial barrier and ﬂooding of the airspaces with edema ﬂuid are hallmarks of ALI/ARDS and VILI (Fig. 2). The potential mechanisms of the loss of epithelial barrier function during VILI is discussed in the following section. A. Ventilator-Induced Alveolar Epithelial Permeability—Overview

Although our understanding of the regulation of vascular endothelial permeability has grown substantially over the past several years, relatively little is known about how the permeability of the alveolar epithelium is regulated. However, several experimental studies have demonstrated that mechanical ventilation can undermine the barrier properties of the alveolar epithelium. In a seminal study, Webb and Tierney (8) reported that high tidal volume

380

Frank and Matthay

Figure 1 Mechanisms of alveolar epithelial ion and ﬂuid transport and their regulation. Sodium–potassium (Na/K) ATPase activity in the basolateral membrane is the driving force, creating an osmotic gradient down which water follows passively via paracellular routes and through aquaporin channels (e.g., AQP 5 in type I cells). Sodium enters the apical surface of type I and type II pneumocytes through the ENaC and other channels. b-Adrenergic agonist binding to its receptor (b-AR) activates adenylate cyclase (AC) via a G-protein–coupled signaling pathway and results in an increase in intracellular cAMP. cAMP stimulates sodium and therefore ﬂuid transport by increasing Na/K ATPase activity and localization to the basolateral membrane, increasing ENaC localization to the apical membrane and the open probability of the channel, as well as increasing the chloride inﬂux necessary for maximal ﬂuid transport rates. Abbreviations: ATP, adenosine triphosphate; ENaC, epithelial sodium channel; cAMP, cyclic adenosine monophosphate. Source: From Ref. 7.

ventilation induced pulmonary edema and diffuse alveolar damage histologically indistinguishable from ARDS in a rat model (Fig. 3). They also found that high volume (high inspiratory pressure) ventilation was much less injurious when positive end-expiratory pressure (PEEP) was used. Rats ventilated with a peak airway pressure of 45 cmH2O and no PEEP developed signiﬁcantly more edema than rats ventilated with the same peak inspiratory pressure and a PEEP of 10 cmH2O. Of course the tidal volume used to achieve a comparable peak pressure was considerably lower when PEEP was added (43 mL/kg compared with 15 mL/kg with PEEP). Dreyfuss et al. (9) subsequently found that high tidal volume ventilation induced increased permeability edema in the airspaces, and that transpulmonary pressure rather than peak airway pressure was the most important determinant of edema formation. Transpulmonary pressure, or alveolar distending pressure, is analogous to lung volume. These investigators ventilated rats with a peak airway pressure of 45 cmH2O using either positive

Alveolar Epithelial Function in Ventilator-Injured Lungs

381

Figure 2 Histology of rat lung with ventilator-induced lung injury. High (A) and low (B) magniﬁcation views demonstrating airspaces ﬂooded with protein-rich edema ﬂuid. Numerous inﬂammatory cells are present in the interstitial and airspaces. Alveolar septae are thickened, indicating interstitial edema. A low-power view of a peribronchovascular region (C) demonstrates signiﬁcant interstitial edema (arrows) surrounding an artery (a), vein (v), and airway (aw), and airway epithelial cell injury (arrowheads). These histologic lesions are indistinguishable from the acute respiratory distress syndrome.

or negative pressure ventilation and found similar increases in lung edema and epithelial protein permeability measured by the appearance of a vascular tracer in the airspaces. They also ventilated rats that had rubber bands applied to the chest and abdomen such that peak airway pressure was the same, but tidal volume was reduced by roughly half, and found that no edema developed. These ﬁndings correlated with scanning electron micrograph studies of lungs exposed to high distending pressures that reported endothelial and epithelial plasma membrane breaks (9,10). Therefore, tensile strain resulting in physical breaks in the alveolar epithelium is one mechanism by which barrier function is lost during VILI. Parker and Ivey (11) expanded on these ﬁndings, showing that changes in intracellular signaling also contributed to the increased permeability edema associated with high tidal volume ventilation. In isolated, perfused

382

Frank and Matthay

Figure 3 High tidal volume ventilation induces severe pulmonary edema and airspace ﬂooding. The addition of PEEP and reduction of tidal volume reduce the injury. y, p < 0.05 compared with all other groups. , p < 0.05 compared with 12 mL/kg, 0 PEEP group. Abbreviation: PEEP, positive end-expiratory pressure. Source: From Ref. 8.

lungs, the administration of a b-adrenergic agonist or a phosphodiesterase inhibitor to increase intracellular cyclic adenosine monophosphate (cAMP) resulted in signiﬁcantly less lung edema and lower protein permeability during high tidal volume ventilation. Furthermore, blocking strain-activated calcium channels with gadolinium also reduced the severity of ventilatorinduced pulmonary edema and protein permeability (12). The same group has also reported that inhibition of tyrosine kinase, inhibition of calcium/ calmodulin, or inhibition of phosphorylation of myosin light chain kinase also reduces edema and protein permeability in rats ventilated with large tidal volumes (13,14). It should be noted that the latter studies did not distinguish between endothelial and epithelial permeability. Therefore, the potential intracellular signaling pathways speciﬁc to epithelial permeability changes remain uncertain. As with lung endothelial permeability, alveolar epithelial permeability can increase with a progressive increase in lung volume. For example, increasing lung volume by the application of PEEP during mechanical ventilation results in increased clearance of inhaled 99mTc-diethylene triaminepentaacetic acid (molecular weight 393 Da), in excess of what would be predicted from a change in surface area alone (15,16). The alveolar epithelial permeability to albumin also increases with increasing lung volume (17,18). In one study, the epithelium of isolated lung lobes distended with ﬂuid to a pressure of 40 cmH2O became more permeable to albumin (18).

Alveolar Epithelial Function in Ventilator-Injured Lungs

383

This correlated with an increase in the equivalent pore radius from approximately 1 nm to 5 nm. When entire lungs rather than isolated lobes were tested, the effect was less pronounced because regional differences in transpulmonary pressure were prevented (17). Interestingly, epithelial breaks resulting from a high lung inﬂation volume are rapidly reversible. In one study, many of the epithelial breaks present immediately after a single over-inﬂation of the lung spontaneously resealed within ﬁve minutes (19). Others have reported similar partial reversibility of increased epithelial permeability following longer periods of VILI (20,21). Which experimental condition most closely approximates clinical VILI is uncertain; however, lung distension near or exceeding the limits of normal physiology results in a reversible increase in epithelial permeability even in uninjured lungs. Ventilation of previously injured lungs with tidal volumes within a physiologic range can also exacerbate epithelial permeability changes. Because the precise contribution of mechanical ventilation to the underlying lung injury is difﬁcult to determine, this category of lung injury has been termed ventilator-associated lung injury (VALI). In a rat model of acid-induced ALI, ventilation with 6 mL/kg resulted in less alveolar ﬂooding and less alveolar epithelial injury as measured by plasma levels of a type I cell–speciﬁc marker of injury (RTI40) compared with 12 mL/kg and a similar level of PEEP (22). These ﬁnds correlated with histologic and ultrastructural differences in airspace edema and epithelial cell injury. When tidal volume was further reduced to 3 mL/kg, epithelial injury and airspace edema improved even more (Fig. 4). Reducing PEEP during ventilation with a tidal volume of 12 mL/kg such that end inspiratory lung volume and mean airway pressures were similar to the 6-mL/kg group, did not prevent epithelial injury or edema (22). Similar ﬁndings have also been reported following surfactant depletion. In this model, tidal volume reduction prevented airspace edema formation and preserved oxygenation, suggesting preserved epithelial barrier function. Interestingly, when surfactant-depleted animals were ventilated with a high frequency–oscillatory ventilation, edema and histologic injury were further reduced (23,24). Although the beneﬁcial effects of low tidal volume ventilation are clear, the role of lung volume recruitment maneuvers in the prevention of VALI is less well understood. For example, maintenance of lung volume with a frequently applied recruitment maneuver in acid-injured rats decreased endothelial permeability, but had no effect on epithelial permeability or alveolar type I cell injury (25). Additional clinical studies are needed to determine the net effect of recruitment maneuvers on VALI in patients. In Vitro Studies of Epithelial Barrier Function

Because the alveolar basement membrane is distended with lung inﬂation, alveolar epithelial cells may experience mechanical strain. For example, at lung volumes near 80% of total lung capacity, the alveolar epithelial

384

Frank and Matthay

Figure 4 Tidal volume reduction decreases alveolar epithelial type I cell injury in acid-injured rats. Continued reduction in tidal volume from 12 to 6 to 3 mL/kg results in additional protection to the epithelium. p < 0.05 compared with all other groups. Source: From Ref. 22.

basement membrane is stretched in isolated rat lungs (26). Stretching of the basement membrane may result in the transmission of mechanical stress through integrin/cytoskeleton connections to the cell interior, as well as in the stretching of the plasma membrane. Studies of alveolar epithelial type II cells grown on distensible membranes have helped characterize the mechanical properties of epithelial cells and have provided insight into the mechanisms of cell injury in VILI. In one study, increasing the duration, amplitude, or frequency of cyclic strain increased plasma membrane injury, and cell death (27). Much of the cell injury occurred within ﬁve minutes. If small amplitude deformation was superimposed on basal tonic strain, there was less membrane disruption and cell death compared with that from superimposing large amplitude strain to the same peak level. In that study, the rate of cellular deformation during a single strain did not affect plasma membrane injury (27). Vlahakis et al. (28,29) demonstrated that plasma membrane disruption induced by a cyclic mechanical strain in vitro was dependent on the rate of plasma membrane trafﬁcking to the cell surface. The mechanical strain of primary type II pneumocytes resulted in an increased vesicular lipid trafﬁcking to the plasma membrane, and this process both prevented and repaired breaks in the epithelial plasma membrane during stretch, In contrast to plasma membrane remodeling, the resistance of cells to deformation (i.e., their stiffness) was not

Alveolar Epithelial Function in Ventilator-Injured Lungs

385

a determinant of plasma membrane injury (28,29). These ﬁndings have been further corroborated by data from the intact lung. Using intravital microscopy to examine subplueral alveoli, Gajic et al. (30) found that high volume ventilation induced reversible plasma membrane breaks, as was previously reported in the in vitro studies. Although these data do not exclude strain-induced signaling through the cytoskeleton as an important mechanism of VILI, they support the hypothesis that membrane disruption and impaired lipid trafﬁcking may be a major mechanism. Plasma membrane breaks could potentially detrimentally affect epithelial permeability, ion transport, and surfactant function, although the functional effects of plasma membrane disruption in vivo remain to be investigated. Mechanical strain may also affect cell–cell adhesions and the epithelial cell cytoskeleton, resulting in a loss of barrier function. Cyclic strain of alveolar epithelial cells with an amplitude sufﬁcient to induce a surface area change of 37% for one hour resulted in increased paracellular movement of a ﬂuorescent tracer; however, smaller amplitudes of strain did not alter paracellular permeability (31). The increase in permeability with the higher amplitude strain was associated with a more punctuated appearance of the occludin band around cells and a decreased total cell occludin content. Pharmacologic disruption of actin did not reproduce the observed changes in occludin expression and localization; however, the depletion of intracellular adenosine triphosphate (ATP) did reproduce the effects on occludin (31). Interestingly, pretreatment of rats with keratinocyte growth factor (KGF) 48 hours prior to cell isolation and subsequent mechanical strain resulted in a signiﬁcant decrease in stretch-induced cell death (32). The mechanisms of the protective effect of KGF remain to be investigated; however, the rearrangement of F-actin in the cytoskeleton may play a role (32). The relative contributions of plasma membrane disruption and impaired lipid trafﬁcking as compared to changes in the cytoskeleton and cell adhesions to the loss of epithelial barrier function in VILI have not yet been fully elucidated.

III. Alveolar Epithelial Ion and Fluid Transport The regulation of airspace ﬂuid content is a critical function of the alveolar epithelium. During pulmonary edema, ﬂuid is removed from the airspaces down a sodium concentration gradient established by alveolar epithelial cells (Fig. 1) (7). The driving force for the gradient is sodium–potassium ATPase in the basolateral membrane of the polarized epithelium. Sodium enters the apical surface of the cell largely through the epithelial sodium channel (ENaC). The basal rate of sodium and ﬂuid transport across the epithelium can be upregulated by beta-adrenergic agonists via increased intracellular cAMP. Although other mechanisms for the stimulation of

386

Frank and Matthay

sodium transport have been identiﬁed, including KGF, thyroid hormone, and transforming growth factor-a (TGF-a), the beta-adrenergic pathway has been the most studied. It should be noted that epithelial sodium transport is regulated at multiple levels within the cell, including sodium pump and channel transcription, membrane abundance of sodium pump and ENaC, sodium pump activity, and open probability of ENaC (7). Abnormal alveolar epithelial ﬂuid transport may be critical in the pathogenesis of VALI because the presence of edema ﬂuid in the airspaces is both an effect of lung injury and a potential mechanism by which injurious mechanical forces in the lung are ampliﬁed. Edema ﬂuid ﬁlls the alveoli and promotes airspace collapse by inactivating surfactant and ﬁlling airways. This loss of lung volume leads to heterogeneity of the lung, resulting in even greater overdistension of the remaining lung units (33). Therefore, if the active sodium transport–dependent clearance of edema ﬂuid from the distal airspaces is reduced, a vicious cycle of airspace edema leading to greater lung overdistension and shear stress will ensue. For example, ﬂooding of distal lung units of rats with saline was found to act synergistically with high tidal volume ventilation to increase lung permeability to albumin (34). In this study, the authors also found that as respiratory system compliance decreased, permeability to albumin increased, suggesting that as edema worsened, a smaller lung volume was ventilated and greater injury resulted (34). Injurious ventilation results in the loss of the alveolar ﬂuid transport capacity in animals. For example, ventilation with a high tidal volume of 30 mL/kg for one hour reduced basal and maximal cAMP-dependent ﬂuid transport rates in rats. Although basal ﬂuid transport returned to baseline levels when tidal volume was reduced, this ventilation strategy resulted in the sustained (three hours) loss of cAMP-dependent transport by a mechanism involving inducible nitric oxide synthase (iNOS) (20). In this study, high volume ventilation increased the iNOS expression and airspace nitrite accumulation. Airspace nitrite is a marker for reactive nitrogen species formation. Pretreatment with a speciﬁc inhibitor of iNOS prevented the loss of cAMP-dependent ﬂuid transport. During experimental lung injury, tidal volume reduction resulted in a more preserved airspace ﬂuid transport. Also, the frequent application of sustained inﬂation recruitment maneuvers decreased epithelial ﬂuid–transport rates in acid-injured rats (25). Therefore, high strain ventilation decreases epithelial ﬂuid–transport rates, although the precise mechanisms for this decrease are the subject of ongoing investigation. Interestingly, inhibition of TGF-b activity, which is increased in the bronchoalveolar lavage (BAL) from patients with ALI (35), decreased pulmonary edema in a rat model of VALI (36). Previous data have indicated that TGF-b downregulates the ENaC gene and protein expression and reduces epithelial sodium and ﬂuid transport capacity (37).

Alveolar Epithelial Function in Ventilator-Injured Lungs

387

Mechanical strain may directly affect the energy-dependent sodium transport driving the airspace ﬂuid clearance. In a model using alveolar type II cells from rats, Leucona et al. found high volume ventilation decreased sodium-potassium ATPase activity, but did not affect mRNA expression of the alpha1 subunit of this transporter. Lecuona et al. (38) reported that in these rats, the sodium–potassium ATPase activity was reduced, but mRNA for the a-1 subunit of this transporter was not reduced when compared with rats ventilated with lower tidal volume (10 mL/kg). In another study, airspace edema clearance in lungs isolated from rats ventilated for 40 minutes with a tidal volume of 40 mL/kg for one hour was reduced by approximately 50%. Instilling the airspaces of the isolated lungs with a badrenergic agonist restored the rate of airspace edema clearance by increasing the activity and quantity of sodium–potassium ATPase in the basolateral membrane. This effect was blocked by disrupting the microtubule assembly with colchicine, suggesting that it is the translocation of sodium–potassium ATPase from intracellular pools to the plasma membrane that accounts for much of the effect (39). Others have reported that the mechanical strain on alveolar epithelial type II cells in vitro increases the sodium–potassium ATPase activity (40,41). The reasons for the differences in the results of the in vitro, in vivo, and isolated lung studies are yet to be fully explored. Further evidence that impaired alveolar ﬂuid clearance is critical to the pathogenesis of VALI is provided by data from animal models in which sodium–potassium ATPase activity is altered. Using a rat model, Adir et al. (42) found that intratracheal administration of an adenovirus vector containing the b1 subunit of sodium–potassium ATPase, resulting in the overexpression of this protein in the lung epithelium and in higher rates of sodium active transport, signiﬁcantly increased lung liquid clearance following high tidal volume ventilation. This ﬁnding was also associated with a decrease in epithelial permeability to mannitol and a decreased passive movement of sodium across the epithelium. Therefore, the increase in lung liquid clearance in the treated rats likely resulted in decreased epithelial injury with high tidal volume ventilation (Fig. 5). In summary, these data indicate that high volume ventilation and mechanical factors inﬂuence alveolar epithelial sodium and ﬂuid transport. Impaired lung epithelial ﬂuid transport appears to play a role in the pathogenesis of ventilator-induced and -associated lung injury. The therapeutic potential of augmenting epithelial sodium and ﬂuid transport requires additional study.

IV. Effects of Mechanical Strain on Epithelial Inﬂammatory Mediators Mechanical strain induces changes in the expression of hundreds of proteins in alveolar epithelial cells (43), many of which may be important in

388

Frank and Matthay

Figure 5 Expression of the b1 subunit of sodium–potassium ATPase in rats increases lung liquid clearance. Lung liquid clearance in isolated rat lungs was increased in HVT rat b1 Na,K-ATPase subunit cDNA (ad b1) lungs as compared with HVT lungs from uninfected [control (computed tomography)], no cDNA (adnull) and sham-infected rat lungs. Ouabain (5 104 M, black bars) inhibited the adb1-increased clearance to a greater degree as compared with other control groups. Data represent mean SEM. This increase in lung liquid clearance was associated with decreased alveolar epithelial injury. Key: , p < 0.0001HVT adb1 (without ouabain, gray bars) versus all other ventilated groups, and p¼0.03 HVT adb1 (without ouabain) versus nonventilated adb1; , p < 0.001 HVT control (without ouabain) versus nonventilated control. Abbreviations: ATP, adenosine triphosphate; HVT, high tidal volume. Source: From Ref. 42.

the initiation of the inﬂammatory response characteristic of VILI. However, ventilator-induced changes in gene expression have perhaps been more widely reported. Using a macroarray of approximately 1200 genes, Copland et al. (44) recently reported that high tidal volume ventilation induced an increased expression of six gene clusters and a decreased expression of another four gene clusters. Northern blotting conﬁrmed that the greatest increases in expression were observed for interleukin-1b (IL-1b), heat shock protein-70, cJun, and Egr-1. Immunohistochemistry demonstrated that expression of these proteins was increased in the distal lung epithelium (44); however, an increased expression of these genes in other cell types could not be entirely excluded with this technique. In another study, three hours of high tidal volume ventilation increased the expression of IL-1b, IL-6, macrophage inﬂammatory protein (MIP)-2, and IL-10 mRNA in adult rats (45). Interestingly, the expression of these genes was not identical in newborn rats, suggesting an age-related or developmental factor in the inﬂammatory response to high tidal volume ventilation. Li and colleagues (46) reported that high volume ventilation for

Alveolar Epithelial Function in Ventilator-Injured Lungs

389

one hour in mice induced an increase in phorphorylated jun-N-terminal kinase (JNK). Pharmacologic inhibition of JNK resulted in decreased lung MIP-2 expression and decreased recruitment of neutrophils into the lung. Similar results were observed in JNK knock-out mice (46). In another study, Uhlig et al. (47) demonstrated that high tidal volume ventilation induced an increase in nuclear translocation of nuclear factor (NF)-jB by a phosphoinositide 3-kinase (PI3K)-mediated pathway. Inhibition of PI3K signaling reduced the nuclear translocation of NF-jB after high volume ventilation and was associated with a reduced expression of IL-6 and MIP-2. Interestingly, inhibition of PI3K did not block the nuclear translocation of NF-jB after endotoxin exposure, suggesting an alternate pathway in VILI. Mechanical stimulation of epithelial cells, macrophages, endothelial cells, ﬁbroblasts, and smooth muscle cells induces a change in protein phosphorylation and alterations in cytoskeleton proteins. In VILI, most attention has focused on alveolar epithelial cells and lung macrophages as the primary source for inﬂammatory mediators in the initial phase of this injury. Vlahakis et al. (48) found that in cultured A549 cells, mRNA for IL-8 increased fourfold after four hours of cyclic strain, sufﬁcient to change the cell surface area by 30%. Continued strain for up to 48 hours resulted in a nearly 50% increase in IL-8 secretion compared with nonstrained controls. A lesser degree of strain did not affect IL-8 secretion. This ﬁnding was conﬁrmed by Quinn et al. (49), who also found that the increase in IL-8 secretion was associated with the activation of the JNK family mitogen-activated protein kinases (MAPKs). Cyclic strain of 15% for two hours induced a 237% increase in phosphorylation of JNK in A549 cells. Phosphorylation of p38 MAPK increased by 468%, but phosphorylation of extracellular-regulated kinase (ERK1/2) was not changed. In a bronchial epithelial cell line, mechanical stretching for brief periods (5–10 minutes) resulted in the activation of activator protein-1 and cAMP-responsive elements, as well as in the activation of all three MAPK pathways (50). These changes were associated with an increase in IL-8 production that was decreased by inhibition of the p38 MAPK pathway. Using primary alveolar type II cells, others (51) have reported that stretch activates the ERK MAPK pathway via a novel G-protein and the epidermal growth factor receptor–mediated pathway. In that study, ERK activation was not dependent on small guanosine triphosphatases such as ras or strain-gated calcium or sodium channels (51). To identify the source of inﬂammatory cytokines in VILI, Pugin et al. (52) cultured human alveolar macrophages on ﬂexible membranes and exposed the cells to cyclic stretch for up to 32 hours. These authors found that cyclic strain increased secretion of IL-8 and matrix metalloproteinase-9 (gelatinase b), but not that of tumor necrosis factor (TNF)-a or IL-6. When the macrophages were pretreated with lipopolysaccharide (LPS),

390

Frank and Matthay

TNF-a and IL-6 secretion increased to a greater extent in strained cells than in static cultures. Mechanical strain also activated NF-jB in macrophages after 30 minutes. In another study, a variety of cell types including macrophages, A549 cells, a bronchial epithelial cell line, two endothelial cell lines, and primary lung ﬁbroblasts were exposed to the same cyclic strain. Of these cell types, only macrophages and A549 cells secreted IL-8 in response to mechanical distension. The relative amount of IL-8 secreted from macrophages was much greater than the amount secreted from A549 cells. In the absence of LPS stimulation, cytokines were not secreted in appreciable amounts from the other cell types (53). Importantly, IL-8 is present in high levels in the edema ﬂuid of ventilated patients with ALI/ARDS (54,55). Furthermore, the activation of NF-jB in response to intratracheal instillation of immune complexes is dependent upon the presence of alveolar macrophages (56). In experimental VILI, macrophage depletion prevents much of the early increase in epithelial protein permeability and the increase in IL-1, IL-6, and CXC chemokine ligand 1 (an IL-8 analog) (57). Taken together, these data implicate the alveolar macrophage as the initial stretch–responsive cell in the initiation of the inﬂammatory response observed in VILI. Oxidative stress may also increase the expression of proinﬂammatory mediators and epithelial injury during high tidal volume ventilation (58,59). Jafari et al. (59) exposed A549 cells to cyclic strain and found a signiﬁcant decrease in cellular glutathione levels. Supplementation with exogenous glutathione before the mechanical strain resulted in decreased activation of NF-jB. It is clear from other animal studies that alveolar macrophages and alveolar epithelial cells do not account for all of the proinﬂammatory mediators present in the lung during VILI. For example, in one study, neutrophil depletion signiﬁcantly decreased BAL IL-8 levels in rabbits ventilated with a high volume for four hours (60). In another study, Grembowicz et al. (61) reported that sublethal plasma membrane disruption results in increased expression of c-fos, a transcription factor important in cytokine expression and activation of NF-jB in vascular endothelial cells and smooth muscle cells. NF-jB activation via I-jB kinase upregulates the expression of a variety of proinﬂammatory mediators, including IL-6, IL-8, IL-1b, and TNF-a (56,62,63). Tracheal epithelial cells exposed to either magnetic twisting cytometry or static compressive stress were found to upregulate expression of Egr-1, another early response gene that encodes a transcription factor with binding sites in the promoter regions of genes such as TNF-a, platelet derived growth factor, TGF-b, and plasminogen activator inhibitor-1 (PAI) (64–67). Finally, it should be noted that experimental data on the role of inﬂammation in the initiation of VILI have been somewhat discordant (68). Interpretation of experimental studies must be in the context of the speciﬁc model studied, and generalization to the clinical setting should be done cautiously.

Alveolar Epithelial Function in Ventilator-Injured Lungs

391

V. Consequences of the Loss of Epithelial Barrier Function Disruption of the alveolar–capillary barrier appears to be an important mechanism of the permeability edema characteristic of VILI. This loss of compartmentalization combined with the ventilator-induced ampliﬁcation of inﬂammation in ALI may also be an important mechanism of multisystem organ failure (MSOF), one of the most common causes of death in ALI/ ARDS. Several investigators have shown that increased permeability of the alveolar–capillary barrier correlated with increased levels of proinﬂammatory mediators in the systemic circulation. Von Bethmann et al. (69) reported that in an isolated perfused murine lung model, ventilation with a transpulmonary pressure of 25 cmH2O compared with 10 cmH2O led to a signiﬁcant increase in the concentrations of both TNF-a and IL-6 in the perfusate. In patients with ALI/ARDS, concentrations of TNF-a, IL-1b, and IL-6 were higher in the arterial blood (obtained via a wedged pulmonary artery catheter) than in mixed venous blood, suggesting that the lungs were a major source of systemic proinﬂammatory cytokines in these patients (70). Another mechanism whereby mechanical ventilation may contribute to the development of a systemic inﬂammatory response is by promoting bacterial translocation from the air spaces into the circulation (71–73), analogous to the gut bacterial translocation hypothesis of multiple organ failure (74). Two recent studies evaluated the inﬂuence of the mechanical ventilation strategy on the translocation of bacteria from the lung into the bloodstream in dogs (71) and rats (72). After intratracheal instillation of bacteria, these animals were ventilated with a high transpulmonary pressure (around 30 cmH2O) and minimal (0–3 cmH2O) or 10 cmH2O PEEP. Bacteremia seldom occurred in control animals ventilated with low airway pressure, whereas it was found in nearly all animals ventilated with high tidal volume and a low PEEP. In contrast, ventilation with the same transpulmonary pressure but with 10 cmH2O PEEP resulted in rates of bacteremia as low as that in controls. The release of proinﬂammatory cytokines into the systemic circulation may have important consequences. In the NIH-sponsored ARDS Network low tidal volume study (3), plasma levels of IL-6 in the 6-mL/kg tidal volume group were signiﬁcantly lower than in the conventional tidal volume group. This result was associated with a greater number of organ failure–free days, although this outcome variable may not be independent of mortality. Low tidal volume ventilation has also been associated with lower plasma levels of TNF soluble receptor-1 (TNF-RI) in ALI/ARDS patients (75). Higher plasma levels of TNF-RI and -RII were also associated with higher mortality in this patient population (75). In an experimental study, Imai et al. reported that injurious mechanical ventilation led to epithelial cell apoptosis in the kidney and small intestine, and

392

Frank and Matthay

increased plasma creatinine levels (76). Inhibition of fas ligand binding blocked renal cell apoptosis induced by plasma from rabbits ventilated with the injurious strategy. Furthermore, plasma levels of fas ligand correlated with increases in creatinine in patients with ALI/ARDS (76). Previous data indicated that fas and fas ligand are upregulated in pulmonary edema ﬂuid from patients with ALI/ARDS (77). An increase in distal ileal permeability has also been reported in rats ventilated with a tidal volume of 20 mL/kg compared with 10 mL/kg (78). Taken together, these data suggest a role for ventilator-induced alveolar epithelial injury and loss of barrier function in the pathogenesis of MSOF.

VI. Effects of VILI on Surfactants Surfactant lipid and protein turnover is rapid, and surfactant synthesis and secretion are affected by mechanical factors (79–81). Although epithelial cell stretch induces a transient increase in surfactant release (82,83), surfactant activity is consistently impaired in most animal studies of VILI. Accordingly, impairment in surfactant metabolism is a potential mechanism of VILI. Data from even the earliest studies of VILI have supported the conclusion that large lung volume ventilation impairs surfactant function resulting in increased surface tension and reduced lung compliance (84–86). The reduction in lung compliance and associated lung volume loss resulting from injurious ventilation potentially exacerbates the detrimental forces acting on the lung. In an isolated, nonperfused lung model of injurious ventilation, Veldhuizen et al. (87) found that ventilation with a tidal volume of 40 mL/ kg, and without PEEP, for two hours resulted in a decrease in surfactant proteins B and C mRNA, but surfactant protein A mRNA was not different from controls. Surfactant proteins B and C are the major surface-active proteins, but surfactant protein A acts in part to prevent the inactivation of other surfactant proteins by plasma proteins (88–90). A subsequent study using a similar model found that the surface activity of large aggregate surfactants from lungs ventilated with an injurious tidal volume was reduced, but the amount of large aggregate surfactants was not changed (91). These authors concluded that the presence of protein-rich edema ﬂuid in the airway combined with the reduced synthesis of surfactant proteins B and C contributed to the observed decrease in surface activity. Using models of preterm delivery and ALI, others have found that mechanical ventilation with lung volumes in excess of total lung capacity or without PEEP results in the loss of surfactant function (92–94). In one study, rabbits given N-nitroso-N-methylurethane to induce lung injury were ventilated with a tidal volume of either 5 or 10 mL/kg. Ventilation with the higher tidal volume resulted in more rapid inactivation of large aggregate surfactants compared with the lower tidal volume (92). Altering the level of

Alveolar Epithelial Function in Ventilator-Injured Lungs

393

PEEP from 3 and 5 to 12 cmH2O with a tidal volume of 5 mL/kg did not alter the rate of surfactant inactivation. These authors concluded that large tidal excursions in lung volume, and therefore large changes in alveolar surface area, were important in surfactant inactivation. Others have found that excessively low lung volume also inactivates surfactants (95–98). For example, in experimental lung injury due to sepsis, the use of PEEP is associated with greater preservation of large aggregate surfactant pools when compared with a similar tidal volume and no PEEP (93). In the injured lung, there are several potential mechanisms for the observed loss of surface-active properties. First, because the surfactant lipid has little surface activity alone, the loss of surfactant proteins results in a less active surfactant (99). Accordingly, decreased synthesis or secretion of these proteins would result in reduced surfactant function. Second, because plasma proteins bind and inactivate surfactant proteins B and C, the presence of protein-rich edema ﬂuid in the airways reduces lung compliance in part by inactivating surfactants (100). Furthermore, airspace edema can act to wash away the surfactant from the alveoli (84,101). The loss of surfactant function directly results in increased alveolar epithelial injury during mechanical ventilation. Additional evidence for the loss of surfactant function resulting in alveolar epithelial injury comes from a recent study by Steinberg et al. (102). Using a surfactant-depleted pig model and intravital video microscopy, these investigators found that surfactant inactivation by lung lavage caused alveoli to collapse. Areas of alveolar collapse were noted to be injured on histopathology. Furthermore, this form of epithelial injury was independent of airspace neutrophil inﬁltration (102). Therefore, mechanical ventilation with high tidal volume and frequency or without PEEP can reduce surfactant function directly by affecting surfactant protein synthesis and secretion and indirectly by inducing airspace edema formation.

VII. Summary Although mechanical ventilation for ALI/ARDS patients is lifesaving, clinicians now know that high tidal volume, high plateau pressure ventilation exacerbates lung injury. Experimental data have begun to deﬁne the mechanisms of this injurious effect of mechanical ventilation. Central to the development of VILI is the loss of alveolar epithelial barrier function. Speciﬁcally, the abnormal mechanical forces encountered during injurious mechanical ventilation directly increase epithelial paracellular permeability, adversely affect epithelial sodium and net ﬂuid transport, and induce an inﬂammatory environment in the airspaces (Fig. 6). In addition, injurious mechanical ventilation impairs surfactant function, potentially leading to more lung injury. The loss of epithelial barrier function combined

394

Frank and Matthay

Figure 6 Potential mechanisms of ventilator-induced lung injury. Mechanical ventilation induces tensile strain and shear forces in the lung. These forces result in the disruption of alveolar epithelial barrier function. The concentrations of proinﬂammatory mediators (including IL-1, IL-6, and IL-8) increase in the airspaces. The loss of compartmentalization in the lung also results in the release of these mediators and proapoptotic peptides such as fas ligand into the systemic circulation where they may play a role in end organ dysfunction. Mechanical strain also reduces the active sodium transport–dependent clearance of edema ﬂuid from the airspaces. This potentially contributes to increased edema formation, ongoing lung volume loss, and greater ventilator-associated lung injury. Abbreviation: IL, interleukin.

with the formation and release of inﬂammatory and proapoptotic mediators in the airspace may contribute to the multiple organ system failure common in ARDS and ALI patients. Continued study of the speciﬁc effects of mechanical stimuli and inﬂammatory mediators on alveolar epithelial permeability, ion transport, and surfactant metabolism may provide an insight into future therapies for ventilator-associated lung injury in patients. References 1. Matthay MA, Wiener-Kronish JP. Intact epithelial barrier function is critical for the resolution of alveolar edema in humans. Am Rev Respir Dis 1990; 142(6 Pt 1):1250–1257. 2. Ware LB, Matthay MA. Alveolar ﬂuid clearance is impaired in the majority of patients with acute lung injury and the acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163(6):1376–1383.

Alveolar Epithelial Function in Ventilator-Injured Lungs

395

3. ARDSnet. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342(18):1301–1308. 4. Gattinoni L, Caironi P, Pelosi P, Goodman LR. What has computed tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001; 164(9):1701–1711. 5. Frank J, Imai Y, Slutsky A. Pathogenesis of ventilator-induced lung injury. In: Matthay M, ed. Acute Respiratory Distress Syndrome. New York: Marcel Dekker, 2003:201–244. 6. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28(5):596–608. 7. Matthay MA, Folkesson HG, Clerici C. Lung epithelial ﬂuid transport and the resolution of pulmonary edema. Physiol Rev 2002; 82(3):569–600. 8. Webb H, Tierney D. Experimental pulmonary edema due to intermittent positive pressure ventilation. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110:556–565. 9. Dreyfuss D, Soler P, Basset G, Saumon G. High inﬂation pressure pulmonary edema. Respective effects of high, airway pressure, high tidal volume, and positive end-expiratory pressure. Am Rev Respir Dis 1988; 137:1159–1164. 10. West JB, Tsukimoto K, Mathieu-Costello O, Prediletto R. Stress failure in pulmonary capillaries. J Appl Physiol 1991; 70:1731–1742. 11. Parker JC, Ivey CL. Isoproterenol attenuates high vascular pressure-induced permeability increases in isolated rat lungs. J Appl Physiol 1997; 83(6): 1962–1967. 12. Parker JC, Ivey CL, Tucker JA. Gadolinium prevents high airway pressureinduced permeability increases in isolated rat lungs. J Appl Physiol 1998; 84(4):1113–1118. 13. Parker JC, Ivey CL, Tucker A. Phosphotyrosine phosphatase and tyrosine kinase inhibition modulate airway pressure-induced lung injury. J Appl Physiol 1998; 85(5):1753–1761. 14. Parker JC. Inhibitors of myosin light chain kinase and phosphodiesterase reduce ventilator-induced lung injury. J Appl Physiol 2000; 89(6): 2241–2248. 15. Cooper JA, van der Zee H, Line BR, Malik AB. Relationship of endexpiratory pressure, lung volume, and 99mTc-DTPA clearance. J Appl Physiol 1987; 63(4):1586–1590. 16. Marks JD, Luce JM, Lazar NM, Wu JN, Lipavsky A, Murray JF. Effect of increases in lung volume on clearance of aerosolized solute from human lungs. J Appl Physiol 1985; 59(4):1242–1248. 17. Egan EA. Lung inﬂation, lung solute permeability, and alveolar edema. J Appl Physiol 1982; 55:121–125. 18. Egan EA. Response of alveolar epithelial solute permeability to changes in lung inﬂation. J Appl Physiol 1980; 49:1032–1036. 19. Elliott AR, Fu Z, Tsukimoto K, Prediletto R, Mathieu-Costello O, West JB. Short-term reversibility of ultrastructural changes in pulmonary capillaries caused by stress failure. J Appl Physiol 1992; 73(3):1150–1158.

396

Frank and Matthay

20. Frank JA, Pittet JF, Lee H, Godzich M, Matthay MA. High tidal volume ventilation induces NOS2 and impairs cAMP-dependent air space ﬂuid clearance. Am J Physiol Lung Cell Mol Physiol 2003; 284(5):L719–L718. 21. Dreyfuss D, Soler P, Saumon G. Spontaneous resolution of pulmonary edema caused by short periods of cyclic overinﬂation. J Appl Physiol 1992; 72(6): 2081–2089. 22. Frank JA, Gutierrez JA, Jones KD, Allen L, Dobbs L, Matthay MA. Low tidal volume reduces epithelial and endothelial injury in acid-injured rat lungs. Am J Respir Crit Care Med 2002; 165(2):242–249. 23. Imai Y, Nakagawa S, Ito Y, Kawano T, Slutsky AS, Miyasaka K. Comparison of lung protection strategies using conventional and high-frequency oscillatory ventilation. J Appl Physiol 2001; 91(4):1836–1844. 24. Simma B, Luz G, Trawoger R, et al. Comparison of different modes of highfrequency ventilation in surfactant-deﬁcient rabbits. Pediatr Pulmonol 1996; 22(4):263–270. 25. Frank JA, McAuley D, Gutierrez JA, Daniel B, Dobbs L, Matthay M. Differential effects of sustained inﬂation recruitment maneuvers on alveolar epithelial and lung endothelial injury. Crit Care Med. In press. 26. Tschumperlin DJ, Margulies SS. Alveolar epithelial surface area–volume relationship in isolated rat lungs. J Appl Physiol 1999; 86(6):2026–2033. 27. Tschumperlin DJ, Oswari J, Margulies AS. Deformation-induced injury of alveolar epithelial cells. Effect of frequency, duration, and amplitude. Am J Respir Crit Care Med 2000; 162(2 Pt l):357–362. 28. Vlahakis NE, Schroeder MA, Pagano RE, Hubmayr RD. Deformationinduced lipid trafﬁcking in alveolar epithelial cells. Am J Physiol Lung Cell Mol Physiol 2001; 280(5):L938–L946. 29. Vlahakis N, Schroeder MA, Pagano RE, Hubmayr RD. The role of the cytoskeleton in deformation induced alveolar epithelial wounding and repair. Am J Respir Crit Care Med 2002; 165:A379. 30. Gajic O, Lee J, Doerr CH, Berrios JC, Myers JL, Hubmayr RD. Ventilatorinduced cell wounding and repair in the intact lung. Am J Respir Crit Care Med 2003; 167(8):1057–1063. 31. Cavanaugh KJ Jr, Oswari J, Margulies SS. Role of stretch on tight junction structure in alveolar epithelial cells. Am J Respir Cell Mol Biol 2001; 25(5):584–591. 32. Oswari J, Matthay MA, Margulies SS. Keratinocyte growth factor reduces alveolar epithelial susceptibility to in vitro mechanical deformation. Am J Physiol Lung Cell Mol Physiol 2001; 281(5):L1068–L1077. 33. Martynowicz MA, Minor TA, Walters BJ, Hubmayr RD. Regional expansion of oleic acid-injured lungs. Am J Respir Crit Care Med 1999; 160(1): 250–258. 34. Dreyfuss D, Martin-Lefevre L, Saumon G. Hyperinﬂation-induced lung injury during alveolar ﬂooding in rats: effect of perﬂuorocarbon instillation. Am J Respir Crit Care Med 1999; 159(6):1752–1757. 35. Fahy RJ, Lichtenberger F, McKeegan CB, Nuovo GJ, Marsh CB, Wewers MD. The acute respiratory distress syndrome: a role for transforming growth factor-beta 1. Am J Respir Cell Mol Biol 2003; 28(4):499–503.

Alveolar Epithelial Function in Ventilator-Injured Lungs

397

36. Frank JA. TGF-beta is a mediator of experimental ventilator-associated lung injury. FASEB Abstr 2004. 37. Frank J, Roux J, Kawakatsu H, et al. Transforming growth factor-beta 1 decreases expression of the epithelial sodium channel alphaENaC and alveolar epithelial vectorial sodium and ﬂuid transport via an ERK1/2-dependent mechanism. J Biol Chem 2003; 278(45):43939–43950. 38. Lecuona E, Saldias F, Comellas A, Ridge K, Guerrero C, Sznajder JI. Ventilator-associated lung injury decreases lung ability to clear edema in rats. Am J Respir Crit Care Med 1999; 159:603–609. 39. Saldias FJ, Lecuona E, Comellas AP, Ridge KM, Rutschman DH, Sznajder JI. Beta-adrenergic stimulation restores rat lung ability to clear edema in ventilatorassociated lung injury. Am J Respir Crit Care Med 2000; 162(1):282–287. 40. Fisher JL, Margulies SS. Naþ-Kþ-ATPase activity in alveolar epithelial cells increases with cyclic stretch. Am J Physiol Lung Cell Mol Physiol 2002; 283(4):L737–L746. 41. Waters CM, Ridge KM, Sunio G, Venetsanou K, Sznajder JI. Mechanical stretching of alveolar epithelial cells increases Na(þ)–K(þ)–ATPase activity. J Appl Physiol 1999; 87(2):715–721. 42. Adir Y, Factor P, Dumasius V, Ridge KM, Sznajder JI. Na,K-ATPase gene transfer increases liquid clearance during ventilation-induced lung injury. Am J Respir Crit Care Med 2003; 168(12):1445–1448. 43. Hirsch J, Hansen KC, Frank JA, et al. Changes in proteome of rat alveolar type II cells in ventilator induced lung injury [abstract]. FASEB Abstr 2003; 83.1. 44. Copland IB, Kavanagh BP, Engelberts D, McKerlie C, Belik J, Post M. Early changes in lung gene expression due to high tidal volume. Am J Respir Crit Care Med 2003; 168(9):1051–1059. 45. Copland IB, Martinez F, Kavanagh BP, et al. High tidal volume ventilation causes different inﬂammatory responses in newborn versus adult lung. Am J Respir Crit Care Med 2004; 169(6):739–748. 46. Li L-F, Yu L, Quinn DA. Ventilation-induced neutrophil inﬁltration depends on c-Jun N-terminal kinase. Am J Respir Crit Care Med 2004; 169(4):518–524. 47. Uhlig U, Fehrenbach H, Lachmann RA, et al. Phosphoinositide 3-OH kinase inhibition prevents ventilation-induced lung cell activation. Am J Respir Crit Care Med 2004; 169(2):201–208. 48. Vlahakis NE, Schroeder MA, Limper AH, Hubmayr RD. Stretch induces cytokine release by alveolar epithelial cells in vitro. Am J Physiol 1999; 277(1 Pt 1):L167–L173. 49. Quinn D, Tager A, Joseph PM, Bonventre JV, Force T, Hales CA. Stretchinduced mitogen-activated protein kinase activation and interleukin-8 production in type II alveolar cells. Chest 1999; 116(suppl 1):89S–90S. 50. Oudin S, Pugin J. Role of MAP kinase activation in interleukin-8 production by human BEAS-2B bronchial epithelial cells submitted to cyclic stretch. Am J Respir Cell Mol Biol 2002; 27(1):107–114. 51. Correa-Meyer E, Pesce L, Guerrero C, Sznajder JI. Cyclic stretch activates ERK1/2 via G proteins and EGFR in alveolar epithelial cells. Am J Physiol Lung Cell Mol Physiol 2002; 282(5):L883–L891.

398

Frank and Matthay

52. Pugin J, Dunn I, Jolliet P, et al. Activation of human macrophages by mechanical ventilation in vitro. Am J Physiol 1998; 275(6 Pt 1):L1040–L1050. 53. Dunn I, Pugin J. Mechanical ventilation of various human lung cells in vitro: identiﬁcation of the macrophage as the main producer of inﬂammatory mediators. Chest 1999; 116(suppl 1):95S–97S. 54. Kurdowska A, Miller EJ, Noble JM, et al. Anti-IL-8 autoantibodies in alveolar ﬂuid from patients with the adult respiratory distress syndrome. J Immunol 1996; 157(6):2699–2706. 55. Miller EJ, Cohen AB, Matthay MA. Increased interleukin-8 concentrations in the pulmonary edema ﬂuid of patients with acute respiratory distress syndrome from sepsis. Crit Care Med 1996; 24(9):1448–1454. 56. Lentsch AB, Czermak BJ, Bless NM, Van Rooijen N, Ward PA. Essential role of alveolar macrophages in intrapulmonary activation of NF-kappaB. Am J Respir Cell Mol Biol 1999; 20(4):692–698. 57. Frank JA, McAuley DA, Schwendener R, et al. Alveolar macrophages mediate neutrophil inﬁltration and epithelial permeability in experimental ventilator lung injury [abstract]. FASEB Abstr 2004; 684.6. 58. Hammerschmidt S, Sandvoss T, Gessner C, Schauer J, Wirtz H. High in comparison with low tidal volume ventilation aggravates oxidative stress-induced lung injury. Biochim Biophys Acta 2003; 1637(1):75–82. 59. Jafari B, Ouyang B, Li LF, Hales CA, Quinn DA. Intracellular glutathione in stretch-induced cytokine release from alveolar type-2 like cells. Respirology 2004; 9(1):43–53. 60. Kotani M, Kotani T, Ishizaka A, et al. Neutrophil depletion attenuates interleukin-8 production in mild-overstretch ventilated normal rabbit lung. Crit Care Med 2004; 32(2):514–519. 61. Grembowicz KP, Sprague D, McNeil PL. Temporary disruption of the plasma membrane is required for c-fos expression in response to mechanical stress. Mol Biol Cell 1999; 10(4):1247–1257. 62. McRitchie DI, Isowa N, Edelson JD, et al. Production of tumour necrosis factor alpha by primary cultured rat alveolar epithelial cells. Cytokine 2000; 12(6):644–654. 63. Schwartz MD, Moore EE, Moore FA, et al. Nuclear factor-kappa B is activated in alveolar macrophages from patients with acute respiratory distress syndrome. Crit Care Med 1996; 24(8):1285–1292. 64. Tschumperlin DJ, Dai G, Maly IV, et al. Mechanotransduction through growth-factor shedding into the extracellular space. Nature 2004; 429(6987): 83–86. 65. Liu C, Yao I, de Belle I, Huang RP, Adamson E, Mercola D. The transcription factor EGR-1 suppresses transformation of human ﬁbrosarcoma HT1080 cells by coordinated induction of transforming growth factor-beta 1, ﬁbronectin, and plasminogen activator inhibitor-1. J Biol Chem 1999; 274(7):4400–4411. 66. Ressler B, Lee RT, Randell SH, Drazen JM, Kamm RD. Molecular responses of rat tracheal epithelial cells to transmembrane pressure. Am J Physiol Lung Cell Mol Physiol 2000; 278(6):L1264–L1272. 67. Silverman ES, Collins T. Pathways of Egr-1-mediated gene transcription in vascular biology. Am J Pathol 1999; 154(3):665–670.

Alveolar Epithelial Function in Ventilator-Injured Lungs

399

68. Ricard JD, Dreyfuss D, Saumon G. Production of inﬂammatory cytokines in ventilator-induced lung injury: a reappraisal. Am J Respir Crit Care Med 2001; 163:1176–1180. 69. von Bethmann AN, Brasch F, Nusing R, et al. Hyperventilation induces release of cytokines from perfused mouse lung. Am J Respir Crit Care Med 1998; 157:263–272. 70. Douzinas E, Tsidemiadou P, Pitaridis M, et al. The regional production of cytokines and lactate in sepsis-related multiple organ failure. Am J Respir Crit Care Med 1997; 155(1):53–59. 71. Nahum A, Hoyt J, Schmitz L, Moody J, Shapiro R, Marini JJ. Effect of mechanical ventilation strategy on dissemination of intratracheally instilled Escherichia coli in dogs. Crit Care Med 1997; 25:1733–1743. 72. Verbrugge SJ, Sorm V, van’t Veen A, Mouton JW, Gommers D, Lachmann B. Lung overinﬂation without positive end-expiratory pressure promotes bacteremia after experimental Klebsiella pneumoniae inoculation. Intensive Care Med 1998; 24(2):172–177. 73. Savel RH, Yao EC, Gropper MA. Protective effects of low tidal volume ventilation in a rabbit model of Pseudomonas aeruginosa-induced acute lung injury. Crit Care Med 2001; 29(2):392–398. 74. Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL, Sibbald WJ. Multiple organ dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med 1995; 23(10):1638–1652. 75. Parsons PE, Matthay MA, Ware LB, Eisner MD. Elevated plasma levels of soluble TNF receptors are associated with morbidity and mortality in patients with acute lung injury. Am J Physiol Lung Cell Mol Physiol 2005; 33:181–188. 76. Imai Y, Parodo J, Kajikawa O, et al. Injurious mechanical ventilation and endorgan epithelial cell apoptosis and organ dysfunction in an experimental model of acute respiratory distress syndrome. JAMA 2003; 289(16):2104–2112. 77. Albertine KH, MF Soulier, Wang Z, et al. Fas and fas ligand are up-regulated in pulmonary edema ﬂuid and lung tissue of patients with acute lung injury and the acute respiratory distress syndrome. Am J Pathol 2002; 161(5):1783–1796. 78. Guery B, Neviere R, Fialdes P, et al. Mechanical ventilation regimen induces intestinal permeability changes in a rat model. Am J Respir Crit Care Med 2002; 165:A505. 79. Gutierrez JA, Ertsey R, Scavo LM, Collins E, Dobbs LG. Mechanical distension modulates alveolar epithelial cell phenotypic expression by transcriptional regulation. Am J Respir Cell Mol Biol 1999; 21(2):223–229. 80. Nakamura T, Liu M, Mourgeon E, Slutsky A, Post M. Mechanical strain and dexamethasone selectively increase surfactant protein C and tropoelastin gene expression. Am J Physiol Lung Cell Mol Physiol 2000; 278(5):L974–L980. 81. Wirtz HR, Dobbs LG. Calcium mobilization and exocytosis after one mechanical stretch of lung epithelial cells. Science 1990; 250(4985):1266–1269. 82. Torday JS, Rehan VK. Stretch-stimulated surfactant synthesis is coordinated by the paracrine actions of PTHrP and leptin. Am J Physiol Lung Cell Mol Physiol 2002; 283(1):L130–L135.

400

Frank and Matthay

83. Martinez F, Lewis J, Copland I, et al. Mechanical ventilation effect on surfactant content, function, and lung compliance in the newborn rat. Pediatr Res 2004; 56(1):19–25. 84. Faridy EE. Effect of ventilation on movement of surfactant in airways. Respir Physiol 1976; 27(3):323–334. 85. Faridy EE, Permutt S, Riley RL. Effect of ventilation on surface forces in excised dogs’ lungs. J Appl Physiol 1966; 21(5):1453–1462. 86. McClenahan JB, Urtnowski A. Effect of ventilation on surfactant, and its turnover rate. J Appl Physiol 1967; 23(2):215–220. 87. Veldhuizen RA, Tremblay LN, Govindarajan A, van Rozendaal BA, Haagsman HP, Slutsky AS. Pulmonary surfactant is altered during mechanical ventilation of isolated rat lung. Crit Care Med 2000; 28(7):2545–2551. 88. Cockshutt AM, Weitz J, Possmayer F. Pulmonary surfactant-associated protein A enhances the surface activity of lipid extract surfactant and reverses inhibition by blood proteins in vitro. Biochemistry 1990; 29(36):8424–8429. 89. Veldhuizen RA, Welk B, Harbottle R, et al. Mechanical ventilation of isolated rat lungs changes the structure and biophysical properties of surfactant. J Appl Physiol 2002; 92(3):1169–1175. 90. Yukitake K, Brown CL, Schlueter MA, Clements JA, Hawgood S. Surfactant apoprotein A modiﬁes the inhibitory effect of plasma proteins on surfactant activity in vivo. Pediatr Res 1995; 37(1):21–25. 91. Veldhuizen RA, Slutsky AS, Joseph M, McCaig L. Effects of mechanical ventilation of isolated mouse lungs on surfactant and inﬂammatory cytokines. Eur Respir J 2001; 17(3):488–494. 92. Ito Y, Veldhuizen RA, Yao LJ, McCaig LA, Bartlett AJ, Lewis JF. Ventilation strategies affect surfactant aggregate conversion in acute lung injury. Am J Respir Crit Care Med 1997:155. 93. Malloy JL, Veldhuizen RA, Lewis JF. Effects of ventilation on the surfactant system in sepsis-induced lung injury. J Appl Physiol 2000; 88(2):401–408. 94. Ogawa A, Brown CL, Schlueter MA, Benson BJ, Clements JA, Hawgood S. Lung function, surfactant apoprotein content, and level of PEEP in prematurely delivered rabbits. J Appl Physiol 1994; 77(4):1840–1849. 95. Albert RK, Lakshminarayan S, Hildebrandt J, Kirk W, Butler J. Increased surface tension favors pulmonary edema formation in anesthetized dogs’ lungs. J Clin Invest 1979; 63(5):1015–1018. 96. Krause M, Olsson T, Law AB, et al. Effect of volume recruitment on response to surfactant treatment in rabbits with lung injury. Am J Respir Crit Care Med 1997; 156(3 Pt 1):862–866. 97. Krause MF, Jakel C, Haberstroh J, Schulte-Monting J, Leititis JU, OrlowskaVolk M. Alveolar recruitment promotes homogeneous surfactant distribution in a piglet model of lung injury. Pediatr Res 2001; 50(1):34–43. 98. Oyarzun MJ, Stevens P, Clements JA. Effect of lung collapse on alveolar surfactant in rabbits subjected to unilateral pneumothorax. Exp Lung Res 1989; 15(6):909–924. 99. Clements JA. Lung surfactant: a personal perspective. Annu Rev Physiol 1997; 59:1–21.

Alveolar Epithelial Function in Ventilator-Injured Lungs

401

100. Baker CS, Evans TW, Randle BJ, Haslam PL. Damage to surfactant-speciﬁc protein in acute respiratory distress syndrome. Lancet 1999; 353(9160):1232–1237. 101. Wyszogrodski I, Kyei-Aboagye K, Taeusch HW Jr, Avery ME. Surfactant inactivation by hyperventilation: conservation by end-expiratory pressure. J Appl Physiol 1975; 38(3):461–466. 102. Steinberg JM, Schiller HJ, Halter JM, et al. Alveolar instability causes early ventilator-induced lung injury independent of neutrophils. Am J Respir Crit Care Med 2004; 169(1):57–63.

This page intentionally left blank

16 Genomic Insights into Ventilator-Induced Lung Injury

STEPHANIE A. NONAS, JEFFREY R. JACOBSON, and JOE G. N. GARCIA Division of Pulmonary and Critical Care Medicine, Center for Translational Respiratory Medicine, Johns Hopkins University School of Medicine Baltimore, Maryland, U.S.A.

I. Introduction—VALI and Genome Medicine The institution of mechanical ventilation remains a life-saving intervention in patients with acute lung injury (ALI), a devastating clinical syndrome deﬁned by acute respiratory failure with lung inﬂammation, increased vascular permeability, and alveolar edema. ALI represents a ﬁnal common pathway of response to a variety of insults (sepsis, trauma, pneumonia, etc.) and despite recent advances, still carries an annual mortality rate of 30% to 50% (1). Mechanical ventilation, although a necessary part of therapy, is known to worsen existing ALI and may even be a primary cause of ALI (2,3). During mechanical ventilation, a high end-inspiratory lung volume [from large tidal volumes (VTs) or high levels of positive end-expiratory pressure (PEEP)] results in high-permeability pulmonary edema known as ventilator-associated lung injury (VALI). High volume mechanical ventilation is now recognized to be potentially directly harmful to susceptible patients, with the beneﬁt of reducing airway pressures by ventilating with lower VTs in ALI patients [and its most severe form—acute respiratory distress syndrome (ARDS)] ﬁrmly established by the landmark 403

404

Nonas et al.

ARDSNet ﬁndings, which demonstrated a signiﬁcant decrease in mortality from ARDS with a reduction in VT from 12 to 6 mL/kg (4). In addition to directly inducing lung injury, each day on the ventilator prolongs sedation requirements and increases the risk of developing malnutrition and nosocomial- and ventilator-associated pneumonia. These ﬁndings have spurred the research on VALI. High volume ventilation-induced lung injury (VILI) was ﬁrst examined in animal models in 1974, when Webb and Tierney demonstrated that mechanical ventilation of intact lungs from healthy animals induced pulmonary edema in direct proportion to airway pressure and lung volume (5). Follow-up studies by Dreyfuss and Saumon demonstrated that VILI was directly related to lung volume, speciﬁcally end-inspiratory volume, and coined the term ‘‘volutrauma’’ (6). Since then, multiple studies have demonstrated that volutrauma from high volume ventilation results in lung edema and tissue injury in a variety of in vivo animal models (rat, rabbit, lamb, etc.) (7–12). Important studies by Parker and Yoshikawa demonstrated changes in microvascular permeability in both isolated lung and intact animal models exposed to increased airway pressures, implicating the effects of mechanical stimuli on various cell-signaling pathways (10,13–17). Despite advances in care for patients with ALI, the development of sophisticated hemodynamic monitoring technologies, and new insights into the pathogenesis underlying sepsis and ALI, there remains a signiﬁcant gap in the full translation of this progress into increased ALI survival (1,4,18). Increasing information regarding the pathobiology of ALI suggests that the remarkable cellular heterogeneity observed in the integrated response to acute injury may contribute to the translation gap noted. In addition, there is considerable heterogeneity in the response to injurious stimuli among the population, suggesting that genetic susceptibility, in addition to environmental factors, is a key determinant in the development of ALI. To date, little is known about the genetics of ALI and the genetic determinants that render patients susceptible to the adverse effects of mechanical ventilation. Clearly, extensive understanding of VALI is needed, and the identiﬁcation of novel therapeutic targets for ALI is essential if there is to be continued progress in the management of this devastating disorder. New molecular targets must be deduced from human and animal studies identifying genetic susceptibility loci for VALI. This approach will help to reveal pathophysiologic mechanisms of VALI and will accelerate the development of therapies for this devastating human disease.

II. Challenges to Unraveling the Genetics of VALI Although the genetic basis of ALI has not been fully established, increasing evidence derived from association-based studies suggests that genetic

Genomic Insights into Ventilator-Induced Lung Injury

405

variation contributes to ALI susceptibility and severity (19–25). The incidence and clinical course of ALI varies widely among patients even with equivalent pathologic stimuli from known triggers such as sepsis or trauma, strongly suggesting a genetically determined susceptibility to the development of ALI (1,4,26–29). While improved understanding of ALI at both the molecular and population level has not yet reconciled the heterogeneity in patient susceptibility to ALI or multiorgan failure, or substantially impacted upon the mortality rate, insights into the genetics of ALI/VALI hold great promise. Unfortunately, despite promising preliminary results, it remains difﬁcult to study the genetics of ALI in humans for several reasons. First, as with many polygenic diseases, ALI represents a continuum rather than a discrete group of phenotypes. ALI is a clinical syndrome that lacks unique, easily measured markers, but rather represents a ﬁnal common pathway, arising from diverse precipitating factors in a critically ill population and exhibiting a variable clinical course. This large phenotypic variance along with incomplete penetrance, complex gene–environment interactions, and a strong potential for locus heterogeneity further obscure the role of speciﬁc genes in the pathogenesis of ALI. Furthermore, the sporadic nature of ALI precludes the conventional genetic approaches such as heritability studies or linkage mapping (or ‘‘positional cloning’’), strategies which are effective in other lung disorders such as asthma, where large families with both affected and unaffected individuals can be examined for loci linked to the trait of interest (30). The 19th century German mathematician David Hilbert aptly stated, ‘‘signiﬁcant advances require the development of sharper tools for exploration’’ (31). In the past, studies looking at the molecular basis of complex diseases such as ALI, had to rely on a slow, gene-by-gene approach. However, the recent advent of rapid, high-throughput gene expression proﬁling and genotyping has allowed for rapid large-scale analysis of the genome and have opened the door for the development of powerful new approaches, such as large-scale microarray analysis, to identify speciﬁc gene expression patterns associated with disease susceptibility and pathogenesis. The recent availability of the completed genome of the mouse, rat, and canine (in addition to prokaryotic and eukaryotic model organisms) has sparked efforts to identify speciﬁc gene expression patterns that may help diagnose, prognosticate, guide therapy, or otherwise contribute to our overall understanding of human disease. The completion of the sequence of an organism’s genome, however, is only the ﬁrst step in developing a thorough understanding of its biology. New genomic techniques including the identiﬁcation and establishment of single nucleotide polymorphisms (SNPs), new methods of characterizing quantitative trait loci (QTLs), identiﬁcation of new pathways of cell function/regulation, new analysis of mechanisms of transcriptional and posttranscriptional regulation, protein–protein interactions, and cross-talk

406

Nonas et al.

between the signaling cascades as well as parallel developments in computational analysis have heralded the era of molecular medicine and revolutionized the concept of translational biomedical research. These approaches often provide a daunting amount of data, and the challenge facing the current generation of researchers is to collect and integrate these data and to use the available tools and techniques to produce a comprehensive understanding of fundamental biological processes both in health and in disease.

III. Current Status of VALI/VILI Genetics and the Candidate Gene Approach Much of the genetic variation between individuals lies in differences known as SNPs, variant forms of genes that occur in at least 1% of the population. These polymorphisms inﬂuence either the transcriptional regulation of the gene, if the SNP resides within the promoter region, or alter the structure/ function of the gene product (i.e., protein) via amino acid substitutions and processes, which may include posttranslational modiﬁcation (Fig. 1). SNP analysis provides the underpinning of the ‘‘candidate gene approach’’ (Fig. 2) to investigate the molecular basis of speciﬁc diseases. This approach allows for the study of the association between polymorphisms of certain genes and a disease (or disease susceptibility) phenotype by measuring the frequency of the target variant allele in the population of affected patients and comparing it with the frequency of the allele in controls. Ultimately, an association between functional variants of a gene and a clinical phenotype may help to identify key pathophysiologic processes during disease and provide genetic factors and potential therapeutic targets. Moreover, SNPs can be used with respect to their epidemiological associations to test susceptibilities to common diseases as well as explain the diversity of clinical manifestations, outcome, and risk of chronicity among patients with a given disease. A large number of human SNPs are available through public databases, (32) with more than ﬁve million SNPs identiﬁed to date. Several recent clinical studies have used a truncated candidate gene approach to identify genetic polymorphisms linked with susceptibility to ALI. Recent case–control design studies found that polymorphisms in angiotensin-converting enzyme (ACE), inﬂammatory markers [tumor necrosis factor (TNF)-a, interleukin (IL)-6], surfactant proteins (SPs), CD14, and toll-like receptors correlate with susceptibility and outcome in patients with severe ALI (22,33,34). In an allele association study of 19 polymorphisms in the genes of SP-A1, SP-A2, SP-B, and SP-D in ARDS, Lin et al. (22) identiﬁed in SP-B, the SNPs that were associated with poor outcomes in ALI/ARDS, including a SNP at nucleotide 1580 [C/T (1580)] within codon 131 (Thr131Ile) of the SP-B gene, which determines the presence or absence of a potential N-linked glycosylation site. Multivariate analysis

Genomic Insights into Ventilator-Induced Lung Injury

407

Figure 1 Potential role of SNPs in altering the transcription, structure, or function of a gene product. Polymorphisms in the promoter region of a gene may inﬂuence transcriptional regulation of the gene. Other polymorphisms may exist in an intronic site with no effect on amino acid sequence or may exist within exons and alter the structure and/or function of the gene product. Abbreviation: SNPs, single nucleotide polymorphisms.

revealed signiﬁcant differences only for the C/T (1580) polymorphism. When the ARDS population was divided into subgroups based on etiology— idiopathic (e.g., pneumonia) or exogenic (e.g., trauma)—signiﬁcant differences were observed for the C/T (1580) SNP in the idiopathic ARDS group, with an increased frequency of the C/C genotype in this group. Based on the odds ratio, the C allele of the SP-B SNP may be viewed as a susceptibility factor for ARDS, although the expression of both C and T alleles occurs in heterozygous individuals and it is currently not known whether these alleles correspond to similar levels of SP-B protein. Recently, another study reported a positive association of a candidate gene displaying polymorphic variants with the incidence of and outcome from ARDS (34). ACE is a zinc metallopeptidase that converts angiotensin I to angiotensin II and degrades bradykinin and may not only contribute to blood pressure–related diseases, via regulation of vascular tone and cardiac function, but may also mediate inﬂammatory states. ACE is derived mainly from endothelial cells and, because the lung represents the body’s largest endothelial surface, the pulmonary microcirculation may account for half of the variance of ACE plasma levels and affect the incidence and course of many pulmonary diseases, including ALI/ARDS. The allele and genotype

408

Nonas et al.

Figure 2 Overview of the ‘‘candidate gene approach.’’ Candidate genes are identiﬁed by expression arrays from both human disease and animal models of human disease, or by published association or linkage studies. As depicted, linkage studies are not feasible in patients with ALI. The selection of these candidate genes for further genotyping is then validated using knockout and transgenic technologies, and the structure and function of their corresponding gene products evaluated using proteomic methods. Ultimately, genotyping of SNPs in these genes using well-phenotyped patient populations will allow the identiﬁcation of genetic polymorphisms in diseasespeciﬁc predictor genes that are associated with increased risk or severity of disorders such as ALI. Abbreviations: ALI, acute lung injury; SNPs, single nucleotide polymorphisms.

frequencies of a well-known insertion/deletion polymorphism in the ACE gene appears to be increased in patients with ARDS, and ARDS patients homozygous for the deletion (and therefore carriers of the ACE DD genotype) appear to be at the highest risk (34). This insertion/deletion ACE polymorphism appears to be an important one and has been tested in diseases such as hypertension (35), type 2 diabetes (36), and myocardial infarction (37). Despite these studies, which clearly begin to build a case for a genetic basis of susceptibility and severity in ALI, human studies of ALI/VALI are quite limited due to a number of factors including the decreased tissue availability and an overall lack of viable suspected and novel candidate genes in ALI. For these very reasons, comprehensive gene array analysis of tissues derived from both animal models of ALI/VALI have been used as an alternative method of identifying candidate genes (Fig. 2). IV. Gene Expression in Animal Models of VILI As indicated above, due to the challenges inherent in studying critically ill patients, the attention has appropriately focused on animal models of

Genomic Insights into Ventilator-Induced Lung Injury

409

ALI. While no animal model can completely recapitulate the human syndrome of ALI, animal studies remain useful in pinpointing the mechanisms underlying the susceptibility, pathogenesis, progression, and resolution of lung injury, and in identifying potential therapeutic targets. Animal models of complex diseases like ALI have several advantages over human studies, including the ability to exquisitely control experimental stimuli while eliminating genetic heterogeneity. With available inbred animal strains, genetic background can be eliminated as an experimental variable and the effects of an experimental condition, high volume ventilation for example, can be studied in isolation (Fig. 3). As with human studies, early animal studies of ALI focused on a gene (or a few genes) of interest. For example, studying the effects of mechanical ventilation on cytokine and heat shock protein (HSP) levels, Vreugdenhil et al. found increased expression of HSP-70 and IL-1b mRNA in the test lipopolysaccharide (LPS)-treated rats receiving injurious mechanical ventilation compared with the control LPS-only treated animals (38). Likewise, Takata et al. found that TNF-a mRNA levels increased with mechanical ventilation in a surfactant depletion model of ALI in rabbits (39). Using a traditional gene knockout technique, Wainwright et al. demonstrated the importance of myosin light chain kinase (MLCK) in endotoxin-induced ALI (40). MLCK knockout mice were less susceptible to ALI induced by intravenous LPS and had increased survival with subsequent mechanical ventilation. But, as with human studies, all of these studies relied on a preselected gene of interest, whereas newer approaches take advantage of the availability of genomic tools and techniques and focus on the rational generation of candidate genes that can then be studied individually in patients as well as in speciﬁc animal models of disease. One such technique used successfully to identify candidate genes in polygenic diseases such as asthma, arteriosclerosis, and diabetes, is the identiﬁcation of chromosomal regions termed QTL. While these regions contain a signiﬁcant number of genes, QTLs often contain several interacting genes that inﬂuence the phenotypic expression of complex traits. The availability of inbred animals provides an ideal system to study the genetics of ALI, which has several advantages over human studies, including exquisite control over experimental stimuli and environment without genetic heterogeneity between animals. As each animal is genetically identical throughout its genome, genetic background is eliminated as an experimental variable. Using the natural variations between inbred strains, the genomic approach offers the possibility of studying a disease from its phenotype. Once a strain-speciﬁc phenotypic difference is identiﬁed, chromosomal regions (QTLs) encoding genes that regulate a speciﬁc phenotype, for example susceptibility to VALI/VILI, are identiﬁed through the comparison of genetically distinct offspring with differing phenotypes. From there, potential candidate genes may be individually evaluated using traditional

410

Nonas et al.

Figure 3 Differential hierarchical clustering of murine lung gene expression in low and high VT ventilated mice. C57B6 mice 8 to 10 weeks old were exposed to low (77 mL/kg) or high VT (20 mL/kg) mechanical ventilation for two hours. Murine lung tissue was harvested and utilized for U74 Affymetrix microarray expression proﬁling. Data were normalized to control non-ventilated mice. Each column represents a speciﬁc gene. The degree of constrast for each column corresponds to either up- or downregulation fold-change amplitude relative to control lung tissue (fold-change scale shown on the top). As can be seen in the highlighted area, clearly there are differentially expressed genes in the two groups. Abbreviation: VT, tidal volume.

knockout/transgenic techniques in which the gene of interest is deleted or inserted to induce a phenotype change. This strain-survey approach was successfully utilized by Leikauf et al. to investigate the sensitivity of several mouse strains to nickel-induced ALI (26,41,42). With readouts such as survival time, bronchoalveolar lavage (BAL) ﬂuid protein level, BAL neutrophil counts, lung wet-to-dry ratios, and tissue histology, they demonstrated that susceptibility to nickel-induced injury varied between the strains and identiﬁed a sensitive (A/J), and a resistant (C57BL/6J) strain whose mean survival time was double that of the sensitive strain (with resistance inherited as a dominant trait).

Genomic Insights into Ventilator-Induced Lung Injury

411

Chromosomal analysis of multiple strains of backcross mice generated from the two parental strains allowed the identiﬁcation of QTLs on chromosomes 6, 9, 12, and 16 that contributed to the signiﬁcant differences in ALI phenotype (survival time) in the parental strains. Subsequent cDNA microarray analysis identiﬁed signiﬁcant differences in the expression of more than 100 genes involved in cell proliferation, extracellular matrix repair, oxidative stress, hypoxia, and SPs. Combining the results from QTL and microarray analysis, a short list of candidate genes were identiﬁed for nickel-induced ALI including known candidate genes for ALI, such as SP-B, AQP-1, and tumor growth factor-a, as well as several novel genes, such as metallothionein-l (26,41,42). Despite the demonstrated success of this approach, the process of successive mating and backcrossing with the subsequent creation of knockouts and transgenics has traditionally been difﬁcult and time consuming. Fortunately, recent advances in genomic techniques and genetic mapping have markedly simpliﬁed and accelerated the process with high polymorphic markers available for human, rat, and mouse studies. The completion of the genomic sequence of the human, rat, and mouse have negated the need to experimentally build contigs of QTLs and have allowed for the application of comparative genomics for candidate gene analysis. Additionally, the availability of rapid gene array analysis allows the generation of candidate genes by direct comparison of parental strains with phenotypic differences. Taking advantage of these high throughput techniques, Copland et al. used gene expression proﬁling to study temporal changes in early gene expression with high VT ventilation in vivo (43). Inbred rats were ventilated at 25 mg/kg for either 30 or 90 minutes; lung injury was assessed by histologic scoring, gene expression was measured by microarray, and select gene products were conﬁrmed by northern blot analysis. Investigators found that lung injury correlated with increased expression of transcription factors, stress proteins, and inﬂammatory mediators, and decreased expression of metabolic regulatory genes. Copland’s approach allowed the detection of changes in gene expression that preceded histologic lung injury and identiﬁed Egr-1, c-Jun, HSP-70, and IL-1b as candidate genes involved in VALI/VILI (43). More recent gene mapping and linkage studies in mice/rats of speciﬁc genetic backgrounds that display variable responses to experimental lung injury have identiﬁed many potential candidate genes including: ﬁbrinogen A, coagulation factor III, plasminogen activator, urokinase receptor (uPAR), tissue factor, plasminogen activator inhibitor type 1 (PAI-1), IL1, and IL-6 (44,45). Although notable differences exist with respect to the experimental model of the mechanically ventilated rat, there is considerable agreement between these data and the observations of Copland and colleagues (46). Although ﬁve hours of mechanical ventilation with 12 mL/kg VT failed to induce histologic evidence of injury, several novel

412

Nonas et al.

candidate genes were identiﬁed in addition to several known ALI-related genes, including IL-1b and Egr-1.

V. Ortholog Gene Database in VALI and Mechanical Stress Given the paucity of viable ALI candidate genes, we sought to extend the utility of the ‘‘candidate gene approach’’ with extensive gene expression proﬁling studies in animal and human models of ALI (rat, murine, canine and human) to identify potential candidate genes involved in VALI/VILI (Fig. 3). These studies were combined with a novel bioinformatics approach that uses gene ontologies and comparisons of orthologous genes across species to search for candidate genes involved in the response to mechanical stress (44). The use of gene ontologies refers to grouping genes together based on known biological pathways and allows expression data to be examined based on the pathways activated rather than the single genes. This approach is based on the idea of evolutionarily conserved responses to lung injury. Seeking to identify genetic determinants of susceptibility to mechanical stress, array data was generated for rat, murine, and canine models of ventilator-induced ALI and compared with array data from human tissues. As outlined below, by examining and comparing trends in gene expression in response to injurious stimuli such as mechanical ventilation, we were able to identify common genes across species. Two canine models of ALI with mechanical ventilation were utilized: unilateral saline lavage–induced lung injury and intrabronchially delivered endotoxin (LPS) (47,48). Two murine ALI models were utilized: intratracheal LPS and high VT (17 mL/kg) mechanical ventilation as we recently described (47,49). Control groups in each model were spontaneously ventilated. Lung tissue and BAL were collected for microarray and protein analyses. Likewise, human BAL and serum samples were obtained from ALI patients and healthy controls. For gene expression proﬁling and validation, we utilized the Affymetrix GeneChip Microarray System as we described previously (44). Semi-quantitative reverse transcriptase-polymerase chain reaction (RT-PCR), and western blot and real-time PCR were utilized to validate gene expression in animal lung tissues and human BAL, respectively. Using the ortholog database created by our group (44), the responses of four different biological systems (rat, mouse, dog, and human lung endothelial cells) to the various levels of mechanical stretch relevant to ALI were investigated. Figure 4 depicts the ﬁltering approach we utilized and Fig. 5 demonstrates the major mechanical stretch–related ALI biological processes. The blood coagulation ontology was identiﬁed as the major pathway involved in mechanical stress–induced ALI with genes in this ontology including ﬁbrinogen A, coagulation factor III, plasminogen activator, uPAR,

Genomic Insights into Ventilator-Induced Lung Injury

413

Figure 4 Filtering approach using an ortholog database to identify mechanical stress–related genes. Depicted is the rationale for the development of an ortholog database and ﬁltering of the microarray data obtained from > 50 Affymetrix chips involved in the response to mechanical ventilation or cyclic stretch across four species (rat, mouse, canine, and human).

tissue factor, and PAI-1 detailed in Table 1. Importantly, many of these same genes have been implicated elsewhere as potential ALI gene candidates (24,51,52) and in VILI (53,54). In addition to coagulation, other important gene ontologies identiﬁed by this approach included inﬂammation, immune response, and cell motility/chemotaxis (Fig. 5). Based on these reports and on data generated by our cross-species analysis of ALI, we speculate that mechanical stretch directly upregulates coagulation and inﬂammatory gene expression by pulmonary endothelium, accompanied by the activation of the coagulation cascade, inﬂammatory cell recruitment, and the development of ALI, a scenario consistent with clinical reports on the effect of excessive VT ventilation (29). VI. Regional Heterogeneity in Ventilator-Associated Mechanical Stress A substantial amount of information has been derived from the rodent models of ALI. Unfortunately, these experimental models poorly generalize to the human condition because of their short duration and use of supernormal VTs (7,55,56), and do not scale in terms of the effects of gravity, regional interdependence, chest wall/abdominal interactions, hemodynamics, or ﬂuid dynamics. For example, while large VT ventilation alone has been implicated as a cause of ‘‘pure’’ VALI in mice (7,56), it has been difﬁcult

414

Nonas et al.

Figure 5 Lung GOs involved in the response to mechanical stress. As a subsequent level of ﬁltering the ortholog data presented in Fig. 4, we employed MAPPFinder, a program that uses a relational database to link the thousands of genes in an array dataset with corresponding GOs, to identify GOs involved in the response to mechanical ventilation or cyclic stretch across four species (rat, mouse, canine, and human). Ontology analysis shows biological relationships between genes and gene products, giving experimental gene array expression data a biological context. Here, the blood coagulation ontology was identiﬁed as the major mechanical stretch– related ALI biological process, with signiﬁcant representation of the immune response, cell motility/chemotaxis, and inﬂammatory response ontologies. Abbreviations: ALI, acute lung injury; GOs, gene ontologies. Source: From Ref. 50.

to reproduce this phenomenon in large animals (57,58), perhaps because in an intact large animal only some fraction of the lung is subjected to the same mechanical phenomena as is an entire excised mouse lung. Likewise, mechanical heterogeneity is a fundamental property of ALI in man (59,60), and mechanical and biological phenomena contributing to VALI vary widely throughout the lung and with interventions. To establish whether regional cellular responses to local mechanical conditions could be determined as a ﬁrst step toward understanding the mechanism of development of VALI, we chose the canine saline lavage model of ALI because the insult is primarily mechanical, although a secondary inﬂammatory response does ensue (61,62). Previous experience with this model had demonstrated signiﬁcant mechanical heterogeneity not unlike that observed in patients with ARDS, with dependent, basal ﬂooding and collapse of airspaces, intermediate zones of apparent airspace opening and closing, nondependent overdistension with PEEP, and relatively protected apical regions (63). Finally, the canine model is of a sufﬁcient size to allow lung isolation and independent treatment, permitting the use of each animal as its own control for normalization of microarray results and reducing the inherent baseline variation. Hypothesizing that the responses of lung tissue exposed to (and possibly predisposed to injury from) mechanical stress will vary throughout the

Genomic Insights into Ventilator-Induced Lung Injury

415

Table 1 Novel and Suspected ALI Genes Identiﬁed by the Ortholog Approach During Mechanical Stress PubMatrix term Gene Coagulation Tissue factor/thromboplastin Plasminogen activator inhibitor type 1 Plasminogen activator, urokinase receptor Proteinase activated receptor 2 Inﬂammation/immunity Interleukin 1 beta Interleukin 6 Interleukin 13 Interleukin 1 receptor antagonist Cyclooxygenase II Macrophage migration inhibitory factor Cell differentiation antigen 14 Chemotaxis/cell motility Myosin light chain kinase Complement component 5 receptor 1 Complement component 3 Cell chemokine receptor 2 Novel Pre-B-cell colony–enhancing factor Thrombospondin 1 Chemokine (C–C motif) ligand 2 Endothelial differentiation sphingolipid G-protein receptor 1 Guanine nucleotide–binding protein 2 CCAAT/enhancer binding protein Chemokine (C–X–C motif) receptor Annexin 1/lipocortin 1

Gene symbol

Lung injury

Mechanical ventilation

Innate immunity

F3 PAI-1

63 28

27 3

25 7

PLAUR

9

0

2

PAR-2

2

0

3

IL-lb IL-6 IL-13 IL-1RA

339 279 22 22

43 82 0 8

552 329 36 13

COX2 MIF

13 10

2 1

8 100

CD14

1

0

21

MLCK C5AR

9 10

2 0

11 14

C3 CCR-2

57 13

10 0

833 42

PBEF

0

0

0

TSP-1 CCL-2

1 0

0 0

4 0

EDG-1

0

0

0

GBP2

9

0

71

C/EBP

8

0

5

CXCR-4

0

0

26

ANXA1

0

0

6 (Continued)

416

Nonas et al.

Table 1 Novel and Suspected ALI Genes Identified by the Ortholog Approach During Mechanical Stress (Continued ) PubMatrix term Gene Novel (Continued) Actin-related protein complex 4 Aquaporin 1 Heat shock protein 70 Vascular endothelial growth factor

Gene symbol

Lung injury

Mechanical ventilation

Innate immunity

ARPC4

0

0

1

AQP-1 HSP-70 VEGF

13 25 52

2 4 11

0 40 14

Candidate genes were grouped by ontology using MAPPFinder and cross-referenced with the terms lung injury, mechanical ventilation, and innate immunity in PubMatrix, which allows on-the-ﬂy PubMed searches for the identiﬁcation of previously unknown candidates and conﬁrming the presence of many known/reported candidates. Abbreviation: ALI, acute lung injury. Source: From Ref. 50.

heterogeneous injured lung in relation to local mechanical events, we assessed regional cellular responses using genomic microarrays. The microarray-based genomic approach was combined with functional computed tomography (CT) imaging for the noninvasive measurement of regional mechanical stress (62,64). Independent lung ventilation with 100% oxygen was achieved with a double-piston ventilator at 20 breaths/min, with individual lung VT adjusted to produce a baseline end-tidal partial pressure of CO2 of 30 to 35 mmHg for each lung. The left lung was mildly injured by saline lavage (20 mL/kg repeated four times), a primarily mechanical injury with minimal systemic effects, to remove surfactant and reduce the stability of peripheral airspaces. The right lung remained uninjured to be used as a control. At the end of each experiment, the animals were sacriﬁced by exsanguinations under anesthesia and tissue samples were taken from ﬁve corresponding regions in both the lungs (apex—dependent and nondependent, base—nondependent, mid, and dependent). These regions were chosen because, based on our previous CT studies (65–67), they span areas of diverse mechanical stress. The apex tends to be comparatively normal in lavage injury, while the dependent base is mostly nonventilated, collapsed, and ﬂooded. The mid-base undergoes cyclic recruitment/ derecruitment, and the nondependent base remains well expanded and possibly overdistended. Within the injured lungs, the wet/dry ratios were greater in base regions (9.14 1.37) than apex regions (7.01 l.70, p ¼ 0.02). CT imaging revealed increased lung density throughout the lavaged left lung, with preservation of normal aeration in the control right lung with the exception of

Genomic Insights into Ventilator-Induced Lung Injury

417

some mild dependent atelectasis. The distribution of lung density was consistent with patterns seen in humans with ALI/ARDS (68), with the greatest loss of aeration in dependent, base regions as well as with the pattern of long-term injury seen in the lungs of ARDS survivors (69). Cluster analysis of genes signiﬁcantly changed by mechanical ventilation and saline injury revealed groupings of differential regional gene expression changes (70) (Fig. 6). A selection of genes with apex versus base expression changes in opposite directions or greater than twofold difference in the same direction, grouped into ontologies previously shown to be relevant to ALI (44) are presented in Fig. 7. Thus, the particular pattern of gene expression in the injured lung is critically dependent on the sampling location and suggests that sampling approaches that combine differentially-stressed regions may minimize the important changes through the averaging of up- and downregulated genes (64).

Figure 6 Differential gene expression between mechanically stressed apex/base lung tissue. Genes whose expression was signiﬁcantly affected by injury in both apex and base regions were clustered using MeV (71). Each column represents an experimental condition of corresponding sample location and each row a speciﬁc gene. Increased contrast indicates up- or downregulation of gene expression relative to the corresponding region of control lung in the same animal, with contrast intensity corresponding to the fold-change amplitude. As can be seen, there is a dramatic difference in the level of gene expression in the injured lung between the apex and the base. Sample coding: A, apex; B, base; 1,2,3, number of the animal; D, dependent; ND, nondependent.

418

Nonas et al.

Figure 7 Regional apex/base differences in gene expression and gene ontologies of candidate genes. Regional differences of gene expression (fold change, injured/ control) in lung tissues were evaluated for genes with signiﬁcant changes in expression in at least one region (apex or base) and a difference in regional expression 50% or higher. These were linked to gene ontology terms using MAPPFinder (70). Fold changes of genes from identiﬁed gene ontologies represented by horizontal bars SEM. As can be seen, there is a dramatic differences in the level of gene expression in the injured lung between the apex and the base, where within each ontology, expression is in a completely opposite direction.

VII. Pre-B-Cell Colony–Enhancing Factor as an ALI Candidate Gene As noted above, cross-species expression proﬁling revealed extensive expression of genes along speciﬁc ontologies (Fig. 2, Table 1) (44,46,62). A gene with one of the highest levels of gene expression was the gene encoding pre-B-cell colony–enhancing factor (PBEF). The published literature on PBEF is quite sparse (72–74), with our studies providing the ﬁrst observation that PBEF is signiﬁcantly upregulated in the lung in models of lung injury (75,76). This gene encodes for a proinﬂammatory cytokine, originally described for its role in the maturation of B-cell precursors with gene expression upregulated in amniotic membranes from patients undergoing premature labor, especially with amniotic infections (77,78). As this was the ﬁrst demonstration of PBEF expression in lung tissues (75,76), we validated these results by RT-PCR of lung tissue RNA, real-time PCR, and immunohistochemistry studies. PBEF protein levels were signiﬁcantly increased in both BAL ﬂuid and serum of human, murine, and canine ALI models, as well as in cytokine- or cyclic stretch-activated lung

Genomic Insights into Ventilator-Induced Lung Injury

419

microvascular endothelium (Fig. 8) (75). To evaluate the spatial localization of PBEF expression, triple immunohistochemical staining in canine lungs revealed localization of increased PBEF expression in lung endothelium, type II alveolar epithelial cells, and inﬁltrating neutrophils, as well as upregulation of PBEF expression in inﬂammatory cytokine-stimulated human pulmonary microvascular endothelial cells in vitro (75). Finally, instillation of recombinant human PBEF into the trachea of C57B6 mice resulted in signiﬁcant neutrophilic alveolitis. These results indicate that lung vascular endothelial cells, type II alveolar epithelial cells, and inﬁltrating neutrophils express PBEF in the injured lung and support a potentially important role for PBEF in the inﬂammatory lung processes, and also serve as a potential biomarker in animal and human ALI (76). This notion is supported by a recent report (79) that PBEF expression is signiﬁcantly increased in

Figure 8 Analysis of PBEF gene expression and protein levels in canine BAL/ serum, and murine BAL in ALI. Panel (A) demonstrates that in each ALI model (canine or murine) as well as in BAL cell RNA from humans with ALI, PBEF mRNA levels were validated by semi-quantitative duplex RT-PCR. Statistical comparative analysis of integrated density value (mean SEM) between control and ALI groups was performed using the paired t-test and unpaired t-test. Panel (B) reﬂects western blot results, which conﬁrm PBEF protein content in lung tissue, BAL, or serum. For BAL protein determinations, total protein (10 mg) of each sample was separated by 12% SDS-PAGE and immunodetected by western blot using the anti-canine PBEF antibody. Statistical comparative analysis of integrated density value (mean SEM) between control and ALI groups was performed using the unpaired t-test. PBEF protein levels in human ALI BAL and serum were signiﬁcantly increased relative to healthy controls. These results support PBEF as a potential biomarker in ALI and further validate the microarray-based enhanced PBEF expression in animal and human ALI. Abbreviations: ALI, acute lung injury; BAL, bronchoalveolar lavage; HVT, high tidal volume ventilation; LPS, lipopolysaccharide; NVT, normal tidal volume ventilation; PBEF, pre-B-cell colony–enhancing factor; RT-PCR, reverse transcriptase-polymerase chain reaction; SDS-PAGE, sodium dodecyl (lauryl) sulfate-polyacrylamide gel electrophoresis.

420

Nonas et al.

Figure 9 Gene relevance network and bacterial two-hybrid assay evaluation of PBEF. To explore potential mechanisms of PBEF function in cells, we employed two complementary approaches involving a gene relevance network or a bacterial two-hybrid assay. These studies revealed common targets at the gene and protein level, in three prominent pathways: protein degradation, innate immunity, and cell metabolism. Abbreviation: PBEF, pre-B-cell colony–enhancing factor. Source: From Ref. 80.

circulating peripheral blood neutrophils derived from patients with sepsis, including data that convincingly demonstrated PBEF to inhibit neutrophil apoptosis. Figure 9 depicts potential mechanisms by which PBEF may participate in intracellular pathways as detailed by the gene relevance network and the bacterial two-hybrid protein assay, which identiﬁes partners that may be involved in PBEF action. VIII. Preliminary PBEF Genotyping in ALI Patients As our candidate gene approach identiﬁed PBEF as a viable and novel candidate gene and potential biomarker in ALI, we next examined whether common variants in the human PBEF gene might be associated with susceptibility to sepsis-associated ALI. We have preliminarily explored the relevance of ALI candidate genes in deﬁning risk factors and ethnic predilection in patients with ALI, as well as the potential genetic inﬂuences on ALI susceptibility and outcome, a process that has been greatly facilitated by the development of an ALI Genomic DNA Repository. This DNA bank from patients with sepsis and ALI (currently ~550 samples), involves a collaborative enrollment network entitled Consortium to Evaluate Lung

Genomic Insights into Ventilator-Induced Lung Injury

421

Edema Genetics. Preliminary genotyping of patients with sepsis and ALI, and controls revealed an astonishing ethnic-speciﬁc predilection of allelic variants in candidate genes associated with ALI among the African-American patients with ALI and sepsis (23,81,82). Via direct DNA sequencing (76), we identiﬁed 11 PBEF SNPs with two transversion SNPs, T-1001G and T-1543C, in the human PBEF gene immediate promoter (–1 to –3000 base pair) having the highest degree of representation in 12 ALI subjects. The PBEF (T-1001G) SNPs were genotyped in a case–control population of Caucasian subjects with sepsisassociated ALI, sepsis alone, and healthy controls, with additional relevant characteristics of the study population available elsewhere (76). Both the T-1001G and T-1543C SNPs were in Hardy–Weinberg equilibrium (p ¼ 0.50). Subjects with ALI had a signiﬁcantly greater G variant of T-1001G (30%, p < 0.001) with the frequency of the G variant among sepsis subjects (23%) also higher than healthy controls (p ¼ 0.01) (76). The second SNP, C-1543T, was also in Hardy–Weinberg equilibrium ( p ¼ 0.46). The T-allele frequency observed in subjects with ALI (20%) was signiﬁcantly lower than the frequency observed in the healthy control group (31%)(p < 0.05). The frequency of the T variant (C-1543T) among sepsis subjects (24%) was also lower than in healthy controls but this was not statistically signiﬁcant (p ¼ 0.136). In a univariate analysis, carriers of the G allele (T-1001G) had a 2.75-fold increased risk of ALI compared to controls (p ¼ 0.002). Haplotype-weighted analysis of T-1001G and C-1543T SNPs revealed four haplotypes (GC, GT, TC, TT) with the TT haplotype frequency more than twofold lower in ALI patients compared with controls, representing a protective haplotype. In contrast, the frequency of the GC haplotype was more than twofold greater in ALI and sepsis groups compared with controls representing a susceptible haplotype (76). Further, univariate logistic regression analysis revealed that carriers of the haplotype GC from -1001G and -1543C alleles had a 7.71-fold higher risk of ALI (p < 0.001) and 4.84-fold higher risk of sepsis (p ¼ 0.001). Preliminary studies addressing the functionality of the T-1001G variant using the luciferase reporter gene assay did not demonstrate a signiﬁcant role for this variant in gene transcription regulation; however, the T variant in the C-1543T SNP in the PBEF promoter region resulted in a nearly twofold decrease in the PBEF promoter reporter activity. The frequency of the T-allele was signiﬁcantly lower in patients with ALI than in normal controls (p < 0.05). This result is consistent with our observations from animal models of ALI and in vitro cell culture experiments, and implicates increased expression of PBEF in the pathogenesis of ALI. These results further suggest that genetically determined increased PBEF expression contributes to susceptibility to ALI. Although the pathogenic and genetic basis of ALI remains incompletely understood, the identiﬁcation of novel ALI biomarkers holds promise

422

Nonas et al.

for unique insights. Expression proﬁling in animal models of ALI (canine and murine) and human ALI detected signiﬁcant expression of PBEF, a gene not previously associated with lung pathophysiology. These results were validated by real-time PCR and immunohistochemistry studies with PBEF protein levels signiﬁcantly increased in both BAL ﬂuid and serum of ALI models, as well as in cytokine- or cyclic stretch–activated lung microvascular endothelium. SNP screening revealed a T-1001G transversion in the human PBEF gene promoter, and genotyping of a well-characterized cohort of sepsis-associated ALI patients and normal controls revealed that carriers of the variant G allele had a 2.2-fold higher risk of ALI (95% conﬁdence interval, 1.01–4.62). Together, these results strongly indicate that PBEF is a potential novel biomarker in ALI and demonstrate the successful application of robust genomic technologies in the identiﬁcation of novel candidate genes in complex lung diseases. IX. Preliminary IL-6 Genotyping in VALI Unlike PBEF, IL-6 is a well-recognized ALI candidate gene and ALI biomarker. The elevation and persistence of circulating IL-6 has been associated with increased mortality in critically ill patients with ARDS,

Figure 10 Analysis of SNPs of the IL-6 gene. Depicted are eight SNPs spanning the IL-6 gene generated in 98 patients with sepsis-induced ALI. Three SNPs account for the signiﬁcant protective haplotype. Shown are the genotyping data and odds ratio for each SNP. Abbreviations: ALI, acute lung injury; IL, interleukin; SNPs, single nucleotide polymorphisms.

Genomic Insights into Ventilator-Induced Lung Injury

423

sepsis, and trauma (83). Patients with ALI have increased BAL IL-6 and a soluble IL-6 receptor; however, the functional role of IL-6 in ALI pathology is unclear (84,85). Functional polymorphisms exist in the promoter region of the IL-6 gene (-174GC), whereby the presence of a C allele is associated with reduced gene promoter activity, lower circulating IL-6 concentrations, and a lower mortality rate in patients with acute respiratory failure admitted to the intensive care unit (ICU) (86). In the multispecies ALI studies performed, we noted signiﬁcant IL-6 gene expression across all species, as well as differential region-speciﬁc expression in the canine ALI model. In addition we evaluated eight IL-6 SNPs from Caucasian patients with sepsis and ALI, and controls. Figure 10 depicts the presence of a protective ALI haplotype suggesting that the role of IL-6 in ALI is complex and may have a dual role in temporal components of the response to sepsis and mechanical stress. X. Summary With the completion of the Human Genome Project, the availability of highthroughput biology and parallel developments in computational analysis

Figure 11 Genomic schema depicting pathobiologic events in ALI. Mechanical stress activates both alveolar and lung endothelial cells, resulting in induction of coagulation, innate immunity, and inﬂammatory response genes. Gene activation facilitates inﬂammatory pathways and increases vascular permeability through paracellular endothelial gaps and ultimately alveolar ﬂooding. Variation in the expression of these genes may help to explain differences in susceptibility to and severity of ALI in the population. Abbreviation: ALI, acute lung injury.

424

Nonas et al.

have heralded the era of molecular medicine and revolutionized the concept of translational biomedical research. Characterization of genes abnormally expressed in diseased tissues provides promise for the identiﬁcation of novel genes that will serve as biomarkers, diagnostic markers, prognostic indicators, and targets for therapeutic intervention. Signiﬁcant challenges to the exploration of the genetic basis of complex lung disorders exist for diseases such as ALI. We have performed extensive gene expression proﬁling studies in animal models of human lung disease to identify potential ALI/VALI candidate genes. Our preliminary evaluation suggests that the candidate gene approach is a robust strategy to provide novel insights into the genetic basis of this poorly understood trait, and for the identiﬁcation of potentially novel therapeutic targets in ventilator-associated ALI (Fig. 11). Our data strongly indicate that the candidate gene approach, when coupled to creative bioinformatics approaches and extensive expression proﬁling, can yield novel and valuable information delineating genetic factors in ALI. Further analysis of select candidate genes by additional SNP discovery and mid- or high-level throughput genotyping will undoubtedly provide important insights into the genetic basis for ALI susceptibility and severity. This explosion in genomic discovery is generating mechanistic insights into the complex ALI pathobiologic processes with the ultimate goal of translating this knowledge to the bedside. Critical care physicians of the future will be armed with high-throughput technologies and phenotyping protocols, which will personalize care of the ICU patient, improve the survival of patients with critical illness, and herald a new era in critical care medicine. Acknowledgments This work was supported by NIH HL64368, NIH ALI SCCOR HL073994, and the Johns Hopkins HopGene Program in Genomic Applications UO1HL66583. References 1. Ware LB, Matthay MA. The acute respiratory distress syndrome. N Engl J Med 2000; 342(18):1334–1349. 2. Lee PC, Holsmoortel CM, Cohn SM, Fink MP. Are low tidal volumes safe? Chest 1990; 97(2):430–434. 3. Gajic O, Dara SI, Mendez JL, et al. Ventilator-associated lung injury in patients without acute lung injury at the onset of mechanical ventilation. Crit Care Med 2004; 32(9):1817–1824. 4. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342(18):1301–1308.

Genomic Insights into Ventilator-Induced Lung Injury

425

5. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110(5):556–565. 6. Dreyfuss D, Saumon G. Barotrauma is volutrauma, but which volume is the one responsible? Intensive Care Med 1992; 18(3):139–141. 7. Dreyfuss D, et al. High inﬂation pressure pulmonary edema. Respective effects of high airway pressure, high tidal volume, and positive end-expiratory pressure. Am Rev Respir Dis 1988; 137(5):1159–1164. 8. Dreyfuss D, Saumon G. Role of tidal volume, FRC, and end-inspiratory volume in the development of pulmonary edema following mechanical ventilation. Am Rev Respir Dis 1993; 148(5):1194–1203. 9. Wada K, Jobe AH, Ikegami M. Tidal volume effects on surfactant treatment responses with the initiation of ventilation in preterm lambs. J Appl Physiol 1997; 83(4):1054–1061. 10. Hernandez LA, et al. Chest wall restriction limits high airway pressure-induced lung injury in young rabbits. J Appl Physiol 1989; 66(5):2364–2368. 11. Egan EA. Lung inﬂation, lung solute permeability, and alveolar edema. J Appl Physiol 1982; 53(1):121–125. 12. Lecuona E, et al. Ventilator-associated lung injury decreases lung ability to clear edema in rats. Am J Respir Crit Care Med 1999; 159(2):603–609. 13. Yoshikawa S, et al. Acute ventilator-induced vascular permeability and cytokine responses in isolated and in situ mouse lungs. J Appl Physiol 2004; 97(6): 2190–2199. 14. Parker JC, Yoshikawa S. Vascular segmental permeabilities at high peak inﬂation pressure in isolated rat lungs. Am J Physiol Lung Cell Mol Physiol 2002; 283(6):L1203–L1209. 15. Parker JC, et al. Capillary ﬁltration coefﬁcient, vascular resistance, and compliance in isolated mouse lungs. J Appl Physiol 1999; 87(4):1421–1427. 16. Adkins WK, et al. Age effects susceptibility to pulmonary barotrauma in rabbits. Crit Care Med 1991; 19(3):390–393. 17. Hernandez LA, et al. Mechanical ventilation increases microvascular permeability in oleic acid-injured lungs. J Appl Physiol 1990; 69(6):2057–2061. 18. Bernard GR, et al. Efﬁcacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med 2001; 344(10):699–709. 19. Gao L, Grant A, Halder I, et al. Novel polymorphisms in myosine light chain kinase gene confer risk for acute lung injury [accepted]. Am J Respir Cell Mol Biol 2005. 20. Gong MN, et al. Polymorphism in the surfactant protein-B gene, gender, and the risk of direct pulmonary injury and ARDS. Chest 2004; 125(1): 203–211. 21. Maloney J, et al. Variant thrombospondin-1 (TSP-1) SNPs are more common in acute lung injury patients and correlate with increased mortality. Am J Respir Crit Care Med 2003; 167(7):A780. 22. Lin Z, et al. Polymorphisms of human SP-A, SP-B, and SP-D genes: association of SP-B Thr131Ile with ARDS. Clin Genet 2000; 58(3):181–191. 23. King LS et al. Aquaporin-1: a candidate gene in sepsis and lung injury. Am J Respir Crit Care Med 2003; 167(7):A662.

426

Nonas et al.

24. Prabhakaran P, et al. Elevated levels of plasminogen activator inhibitor-1 in pulmonary edema ﬂuid are associated with mortality in acute lung injury. Am J Physiol Lung Cell Mol Physiol 2003; 285(1):L20–L28. 25. Zambelli-Weiner A, et al. Association between the CD14-159 polymorphism and acute lung injury. Am J Respir Crit Care Med 2003; 167(7):A662. 26. Leikauf GD, McDowell SA, Wesselkamper SC, et al. Acute lung injury: functional genomics and genetic susceptibility. Chest 2002; 121(suppl 3): 70S–75S. 27. Parsons PE. Mediators and mechanisms of acute lung injury. Clin Chest Med 2000; 21(3):467–476. 28. Russell JA. Genetics of coagulation factors in acute lung injury. Crit Care Med 2003; 31(suppl 4):S243–S247. 29. Villar J, Flores C, Mendez-Alvarez S. Genetic susceptibility to acute lung injury. Crit Care Med 2003; 31(suppl 4):S272–S275. 30. Matthay MA, Zimmerman GA, Esmon C, et al. Future research directions in acute lung injury. Am J Respir Crit Care Med 2003; 167:1027–1035. 31. Hilbert D. Mathematical problem. Lecture delivered before the International Congress of Mathematicians in Paris, 1900. 32. http://www.ncbi.nlm.nih.gov/SNP/snp summary.cgi. 33. Marshall RP, et al. Genetic polymorphisms associated with susceptibility and outcome in ARDS. Chest 2002; 121(suppl 3):68S–69S. 34. Marshall RP, et al. Angiotensin converting enzyme insertion/deletion polymorphism is associated with susceptibility and outcome in acute respiratory distress syndrome. Am J Respir Crit Care Med 2002; 166(5):646–650. 35. Matsubara M, et al. Angiotensin-converting enzyme I/D polymorphism and hypertension: the Ohasama study. J Hypertens 2002; 20(6):1121–1126. 36. Feng Y, et al. Insertion/deletion polymorphism of the ACE gene is associated with type 2 diabetes. Diabetes 2002; 51(6):1986–1988. 37. Hooper WC, et al. Relationship of venous thromboembolism and myocardial infarction with the renin-angiotensin system in African-Americans. Am J Hematol 2002; 70(1):1–8. 38. Vreugdenhil HA, et al. Ventilator-induced heat shock protein 70 and cytokine mRNA expression in a model of lipopolysaccharide-induced lung inﬂammation. Intensive Care Med 2003; 29(6):915–922. 39. Takata M, et al. Intraalveolar expression of tumor necrosis factor-alpha gene during conventional and high-frequency ventilation. Am J Respir Grit Care Med 1997; 156(1):272–279. 40. Wainwright MS, et al. Protein kinase involved in lung injury susceptibility: evidence from enzyme isoform genetic knockout and in vivo inhibitor treatment. Proc Natl Acad Sci USA 2003; 100(10):6233–6238. 41. Leikauf GD, et al. Pathogenomic mechanisms for particulate matter induction of acute lung injury and inﬂammation in mice. Res Rep Health Eff Inst 2001; 105:5–58; discussion 59–71. 42. Prows DR, et al. Genetic susceptibility to nickel-induced acute lung injury. Chemosphere 2003; 51(10):1139–1148. 43. Copland IB, et al. Early changes in lung gene expression due to high tidal volume. Am J Respir Grit Care Med 2003; 168(9):1051–1059.

Genomic Insights into Ventilator-Induced Lung Injury

427

44. Grigoryev DN, Ma SF, Irizarry RA, Ye SQ, Quackenbush J, Garcia JG. Orthologous gene-expression proﬁling in multi-species models: search for candidate genes. Genome Biol 2004; 5(5):R34. 45. Grigoryev DN, et al. Heterologous gene expression proﬁling of acute lung injury (ALI) models: search for the ALI candidate genes. Am J Respir Crit Care Med 2003; 167(7):A340. 46. Jacobson JR, Garcia JG. Genomics made functional in ventilator-associated lung injury. Am J Respir Crit Care Med 2003; 168(9):1023–1025. 47. McVerry BJ, et al. Sphingosine 1–phosphate reduces vascular leak in murine and canine models of acute lung injury. Am J Respir Crit Care Med 2004; 170(9):987–993. 48. Ye SQ, Lavoie T, Gao L, et al. Molecular cloning of canine pre-B cell colony enhancing factor (PBEF) cDNA [abstr.]. Am J Respir Crit Care Med 2003; 167(7):A779. 49. Peng X et al. Protective effects of sphingosine 1–phosphate in murine endotoxin-induced inﬂammatory lung injury. Am J Respir Crit Care Med 2004; 169(11):1245–1251. 50. www. genMAPP.org. 51. Groeneveld AB, et al. Systemic coagulation and ﬁbrinolysis in patients with or at risk for the adult respiratory distress syndrome. Thromb Haemost 1997; 78(6):1444–1449. 52. Ruf W, Riewald M. Tissue factor-dependent coagulation protease signaling in acute lung injury. Crit Care Med 2003; 31(suppl 4):S231–S237. 53. Schultz M, et al. Local activation of coagulation and inhibition of ﬁbrinolysis in the lung during ventilator-associated pneumonia [abstr.]. Am J Respir Crit Care Med 2003; 167(7):A861. 54. Schmidt B, et al. Thrombin inhibitors reduce intrapulmonary accumulation of ﬁbrinogen and procoagulant activity of bronchoalveolar lavage ﬂuid during acute lung injury induced by pulmonary overdistention in newborn piglets. Pediatr Res 1996; 39(5):798–804. 55. Tremblay L, et al. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997: 944–952. 56. Imanaka H, Shimaoka M, Matsuura N, Nishimura M, Ohta N, Kiyono H. Ventilator-induced lung injury is associated with neutrophil inﬁltration, macrophage activation, and TGF-beta 1 mRNA upregulation in rat lungs. PG Anesth Analg 2001; 92:428–436. 57. Tsuno K, Prato P, Kolobow T. Acute lung injury from mechanical ventilation at moderately high airway pressures. J Appl Physiol 1990; 69(3):956–961. 58. Kolobow T, et al. Severe impairment in lung junction induced by high peak airway pressure during mechanical ventilation. An experimental study. Am Rev Respir Dis 1987; 135(2):312–315. 59. Gattinoni L, et al. What has computed tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001:1701–1711. 60. Pelosi P, D’Onofrio D, Chiumello D, et al. Pulmonary and extrapulmonary acute respiratory distress syndrome are different. Eur Respir J Suppl 2003; 42:48S–56S.

428

Nonas et al.

61. Spragg RG, Smith RM, Harris K, Lewis J, Hafner D, Germann P. Effect of recombinant SP-C surfactant in a porcine lavage model of acute lung injury. J Appl Physiol 2000; 88(2):674–681. 62. Simon BA, Ye SQ, Easley RB, Lavoje T, Gregovey D, Garcia JGN. Microarray analysis of regional cellular responses to local mechanical stress in experimental acute lung injury [in review]. Nature Med 2005. 63. Luecke T, Roth H, Joachim A, et al. Effects of end-inspiratory and end-expiratory pressures on alveolar recruitment and derecruitment in saline-washoutinduced lung injury—a computed tomography study. Acta Anaesthesiol Scand 2004; 48(1):82–92. 64. Simon BA, Ye SQ, Easley RB, Lavoie T, Gregovey D, Garcia JGN. Microarray analysis of regional cellular responses to mechanical stress in canine ventilatorassociated lung injury (VALI) [abstr.]. Am J Respir Crit Care Med 2003; 167:A777. 65. Downie JM, Piper MG, Simon BA. Regional patterns of lung recruitment with PEEP after lung lavage [abstr.]. AJRCCM 1999:A488. 66. Simon BA, et al. CT P-V analysis differentiates ﬂooding vs. collapse in oleic acid (OA) and lavage injury models [abstr.]. Am J Resp Crit Care Med 2000:A486. 67. Downie JM, Nam AJ, Simon BA. Pressure-volume curve does not predict steady-state lung volume in canine lavage lung injury. Am J Respir Crit Care Med 2004:957–962. 68. Rouby JJ, et al. Acute respiratory distress syndrome; lessons from computed tomography of the whole lung. Crit Care Med 2003:S285–S295. 69. Gattinoni L. et al. Lung structure and function in different stages of severe adult respiratory distress syndrome JAMA 1994:1772–1779. 70. Doniger SW, et al. MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression proﬁle from microarray data. Genome Biol 2003; 4(1):R7. 71. Saeed AI, Sharov V, White J, et al. TM4: a free, open-source system for microarray data management and analysis. Biotechniques 2003; 34(2):374–378. 72. Nemeth E, Millar LK, Bryant-Greenwood G. Fetal membrane distention: II. Differentially expressed genes regulated by acute distention in vitro. Am J Obstet Gynecol 2000; 182(1 Pt 1):60–67. 73. Samal B, et al. Cloning and characterization of the cDNA encoding a novel human pre-B-cell colony-enhancing factor. Mol Cell Biol 1994; 14(2):1431–1437. 74. Ognjanovic S, et al. Genomic organization of the gene coding for human pre-Bcell colony enhancing factor and expression in human fetal membranes. J Mol Endocrinol 2001; 26(2):107–117. 75. McGlothlin JR, et al. Molecular cloning and characterization of canine pre-Bcell colony-enhancing factor cDNA. Biochem Genet 2005; 43:3–10. 76. Ye SQ, Simon BA, Maloney JP, et al. Pre-B-cell colony-enhancing factor as a potential novel biomarker in acute lung injury. Am J Respir Crit Care Med 2005; 171(4):361–370. 77. Ognjanovic S, Bryant-Greenwood GD. Pre-B-cell colony-enhancing factor, a novel cytokine of human fetal membranes. Am J Obstet Gynecol 2002; 187(4): 1051–1058.

Genomic Insights into Ventilator-Induced Lung Injury

429

78. Ognjanovic S, Tashima LS, Bryant-Greenwood GD. The effects of pre-B-cell colony-enhancing factor on the human fetal membranes by microarray analysis. Am J Obstet Gynecol 2003; 189(4):1187–1195. 79. Jia SH, et al. Pre-B cell colony-enhancing factor inhibits neutrophil apoptosis in experimental inﬂammation and clinical sepsis. J Clin Invest 2004; 113(9): 1318–1327. 80. Pathlinx, www.hopkins-genomics.org. 81. Gao L, et al. Role of macrophage migration inhibitory factor (MIF) in human and animal models of acute lung injury (ALI) and sepsis: association of a promoter polymorphism and increased gene expression [abstr.]. Am J Respir Crit Care Med 2003; 167(7):A162. 82. Ye SQ, et al. Single nucleotide polymorphisms (SNPs) in the myosin light chain kinase (MLCK) gene are associated with acute lung injury (ALI). Am J Respir Crit Care Med 2003; 167(7):Al62. 83. Meduri GU, et al. Persistent elevation of inﬂammatory cytokines predicts a poor outcome in ARDS. Plasma IL-1 beta and IL-6 levels are consistent and efﬁcient predictors of outcome over time. Chest 1995; 107(4):1062–1073. 84. Park WY, et al. Cytokine balance in the lungs of patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 164(10 Pt 1):1896–1903. 85. Takala A, et al. A prospective study of inﬂammation markers in patients at risk of indirect acute lung injury. Shock 2002; 17(4):252–257. 86. Finigan JH, Ma SF, Garcia JGN. IL-6 in inﬂammatory lung disease: disruption of human pulmonary endothelial barrier function and preliminary SNO analysis. Exp Biol 2004:A467.9.

This page intentionally left blank

Part III: CLINICAL IMPLICATIONS AND TREATMENT OF VILI

17 Lung Imaging of Ventilator-Associated Injury

JEAN-JACQUES ROUBY and QIN LU Re´animation Chirugicale Polyvalente Pierre Viars, Hoˆpital Pitie´-Salpeˆtrie`re, Assistance Publique Hoˆpitaux de Paris, Universite´ Pierre et Marie Curie Paris, France

I. Introduction The concept of ventilator-induced lung injury includes three different pathophysiological entities: high permeability type pulmonary edema resulting from high tidal volume ventilation—‘‘volutrauma’’ (1), lung inﬂammation resulting from repetitive opening and closure of distal bronchioles— ‘‘biotrauma’’ (2), and mechanical distortion/overinﬂation of anatomical lung structures (3). The ﬁrst two have been identiﬁed from experimental studies (4,5) and have found some indirect conﬁrmation in humans (6,7) but remain to date a subject of controversy (8). The latter is a hallmark of human ventilator-induced lung injury. Lung distortion/overinﬂation was discovered on human lung autopsies more than 25 years ago (9–12), was recently demonstrated to result from prolonged mechanical ventilation in experimental animals (13), and is easy to evidence using whole lung computed tomography (CT) (14,15). Surprisingly, pulmonary edema resulting from ‘‘volutrauma’’ and lung inﬂammation resulting from ‘‘biotrauma’’ have been the focus, the attention of most recent reviews on ventilator-induced lung injury despite their uncertain human clinical relevance, whereas the well-demonstrated 431

432

Rouby and Lu

morphological overinﬂation resulting from mechanical ventilation–induced distortion of human lungs is only incidentally reported. The present article is aimed at reviewing the imaging of lung overinﬂation resulting from mechanical ventilation. In critically ill patients with acute lung injury, whole lung CT occupies a central role in evidencing pulmonary lesions resulting from positive pressure ventilation.

II. Histological Evidence of Mechanical Ventilation–Induced Lung Distortion/Overinﬂation The ﬁrst study that demonstrated lung overinﬂation in mechanically ventilated patients was published in 1982 (9). Another source of evidence for mechanical ventilation–induced lung distortion/overinﬂation came from the Extracorporeal Membrane Oxygenation Study. In one of the articles of a volume published in 1976 (16), photos of open lung biopsies performed in patients enrolled in the Extracorporeal Membrane Oxygenation Study and mechanically ventilated with high airway pressures and high tidal volumes for several weeks demonstrated evidence of gross lung overinﬂation (Fig. 1). In a volume of the present collection published in 1985 (17), gross distortion/overinﬂation of subpleural lung regions were reported in a patient who died from a severe acute respiratory distress syndrome (ARDS) (Fig. 2). Surprisingly, although mentioned in the legend of both ﬁgures, these lesions were neither mentioned nor commented on in the texts of the articles. In the following years, several clinical reports were published showing impressive lesions of lung distortion/overinﬂation in patients with

Figure 1 Lung biopsy performed in a patient enrolled in the Artiﬁcial Lung Study performed 19 days following onset of ARDS. On the biopsy sample, there is evidence of gross overinﬂation with intraparenchymal pseudocysts and subpleural blebs. Abbreviation: ARDS, acute respiratory distress syndrome. Source: From Ref. 16.

Lung Imaging of Ventilator-Associated Injury

433

Figure 2 Multiple pseudocysts of the right lower lobe in a patient who died from ARDS caused by an amniotic embolism. The pulmonary circulation is colored light gray (postmortem intravaclular injection of silicone) whereas the alveolar space is colored dark gray (postmortem intrabronchial injection of indocyanine green). There is evidence of lung ischemia caused by pulmonary vascular obstruction. Abbreviation: ARDS, acute respiratory distress syndrome. Source: From Ref. 17.

various causes of ARDS and mechanically ventilated for long periods of time with high peak inspiratory and end-expiratory pressures and high tidal volume: bronchiolectasis (9), bronchioloalveolar overinﬂation (10), pseudocysts, and bronchopulmonary dysplasia (11) were observed following paraquat intoxication, heroin overdose, postoperative acute respiratory failure, and bronchopneumonia. In 1993, an autopsy study performed in a series of 30 critically ill patients younger than 50 years of age, who died in an intensive care unit following a period of mechanical ventilation for ARDS, revealed that alveolar and bronchiolar overinﬂation were histologically present in 28 of the patients (12). In lung regions remaining well aerated, alveolar overinﬂation was termed ‘‘emphysema-like’’ lesions (Fig. 3). In nonaerated parts of the lungs, bronchiolar overinﬂation and pseudocysts were observed (Fig. 4). Interestingly, the size of emphysema-like lesions and pseudocysts, as evaluated by measuring the mean linear intercept, increased with tidal volume,

434

Rouby and Lu

Figure 3 Emphysema-like lesions in lung regions remaining aerated in a 35-yearold patient who died from an ARDS caused by a severe lung contusion. The lung architecture is preserved without evidence of lung inﬂammation. Multiple alveolar ruptures are present around the bronchovascular axis (pulmonary artery and bronchiole) and there is a marked enlargement of alveolar spaces. Abbreviation: ARDS, acute respiratory distress syndrome. Source: From Ref. 12.

Figure 4 Pseudocysts are present in inﬂammatory parts of the lungs in the same patient as in Figure 3. Source: From Ref. 12.

Lung Imaging of Ventilator-Associated Injury

435

peak airway pressure, weight loss, and the number of days during which FiO2 had to be maintained above 0.6 (Fig. 5). Because of the retrospective and descriptive nature of the study, the issue of whether lung overinﬂation was caused by or is the consequence of excessive tidal volumes, very high pressures, long exposure to high FiO2 (18), and malnutrition (19) could not be deﬁnitively answered. However, in a recent experimental study, similar bronchioloalveolar overinﬂation was experimentally produced in anesthetized piglets whose lungs were mechanically ventilated for a severe Escherichia coli bronchopneumonia (13). Mechanical ventilation was delivered using a tidal volume of 15 mL/kg without positive end-expiratory pressure (PEEP) for three days. Animals were sacriﬁced and lung morphometric measurements were performed. In noninfected lung areas, overinﬂation was present as evidenced by a marked increase in mean alveolar and bronchiolar areas. In infected lung regions, pseudocysts and bronchiectasis were observed. Interestingly, lung overinﬂation was inversely correlated with respiratory compliance, suggesting that mechanical ventilation– induced lung overinﬂation was related to the severity of lung injury. This experimental study brings convincing evidence that mechanical ventilation with excessive tidal volumes rapidly induces emphysema-like lesions in noninfected regions and bronchial distortion/overinﬂation in bronchopneumonic parts of the lungs.

Figure 5 Factors inﬂuencing the size of pseudocysts in a series of 26 patients who died from severe ARDS after a prolonged period of mechanical ventilation using uncontrolled tidal volumes and pressures. Pseudocysts with an internal diameter larger than 2 mm are associated with greater peak airway pressures, tidal volumes, weight loss, and longer exposure to FiO2 > 0.6. Abbreviation: ARDS, acute respiratory distress syndrome. Source: From Ref. 12.

436

Rouby and Lu III. CT Evidence of Mechanical Ventilation–Induced Lung Distortion/Overinﬂation A. CT of Lung Overinﬂation: Deﬁnition and Measurement

Lung CT allows a very accurate measurement of lung aeration if a high spatial resolution is used during acquisitions of CT sections (3,20). In healthy subjects at end expiration, there is less than 2% of lung regions whose aeration exceeds 90% (21). During a Valsalva maneuver performed at total lung capacity and aimed at reproducing conditions existing in mechanical ventilation–induced lung overdistension, a substantial part of the lung parenchyma is characterized by an aeration exceeding 90%. Lung emphysema observed in advanced stages of chronic obstructive pulmonary disease is characterized by a lung aeration greater than 90% [CT attenuations < 900 Hounsﬁeld units (HU)] at endexpiration (22). By analogy with these aeration thresholds, it is likely that lung regions whose CT densities measured during mechanical ventilation at endexpiration are below 900 HU are overinﬂated. To date, lung CT is the only accurate means for determining lung overinﬂation (23). However, the human eye does not have enough accuracy for identifying and separating normally inﬂated lung regions from overinﬂated lung regions. If lung regions with CT attenuations below 900 HU are highlighted by a color coding system on each contiguous section (Fig. 6), lung CT becomes a useful qualitative tool enabling the clinician to detect mechanical ventilation– induced lung overinﬂation (24). An accurate quantitative assessment of lung aeration and mechanical ventilation–induced lung overinﬂation requires a manual delineation of lung parenchyma. Because of the very similar CT attenuations characterizing nonaerated consolidated lung parenchyma and mediastinum, an automatic delineation of the lung parenchyma based on radiological densities is impossible and a fastidious manual delineation has to be performed manually on each CT section. As a consequence, quantitative assessment of mechanical ventilation–induced lung overinﬂation is possible but remains limited to research protocols due to the long time required by the manual delineation of the injured lung parenchyma. Of critical importance for the accurate quantitative and qualitative determination of lung overinﬂation is the spatial resolution. A low spatial resolution, as obtained when contiguous 1-cm thick CT sections are selected for assessing the whole lung, may markedly underestimate mechanical ventilation–induced lung overinﬂation (25). B. CT of Lung Overinﬂation at the Early Phase of ARDS and Acute Lung Injury

Two experimental (13,26) and several human studies (21,24,25,27–31) have clearly demonstrated that lung overinﬂation occurs early in the course of experimental and human acute lung injury. In mechanically ventilated patients lying in the supine position, overinﬂation is predominantly observed

Lung Imaging of Ventilator-Associated Injury

437

Figure 6 CT sections obtained at ZEEP and a PEEP of 15 cmH2O in a patient with acute respiratory failure caused by ventilator associated pneumonia (upper row ¼ carinal level, lower row ¼ juxta-diaphragmatic level). Each rough image is colored according to the lung aeration: light gray ¼ nonaerated lung regions characterized by CT attenuations greater than 100 HU, dark gray ¼ normally aerated lung regions characterized by CT attenuations ranging between 900 and –500 HU, and white ¼ overinﬂated lung regions characterized by CT attenuations < 900 HU. PEEP is associated with alveolar recruitment (decrease in lung regions colored in light gray) and lung overinﬂation (appearance of disseminated white lung regions predominating in nondependant areas). Abbreviations: HU, Hounsﬁeld units; CT, computed tomography; PEEP, positive end-expiratory pressure, ZEEP, zero endexpiratory pressure.

in anterior and caudal parts of middle and lower lobes (31). When a single CT section located above the diaphragmatic cupola serves for the analysis of changes in lung aeration resulting from an increase in airway pressure, lung overinﬂation is not detected (32,33). When contiguous CT sections are performed for analyzing the whole lung, signiﬁcant overinﬂation is often detected in caudal and nondependant lung regions (26,31). The better performance of high spatial resolution CT sections for evidencing overinﬂation (25) suggests a heterogeneous distribution of ventilator-induced lung overinﬂation between adjacent alveoli. One of the main implications of the human studies evidencing signiﬁcant lung overinﬂation at end-expiration is that lung overinﬂation can occur even when plateau airway pressures are in the considered ‘‘safe’’ range of 25 to 35 cmH2O. There may not be a safe plateau airway pressure especially in patients with marked inhomogeneity of the loss of lung aeration. Heterogeneity is observed in a majority of patients with acute lung injury (3,34), increasing the risk of regional overinﬂation at relatively low inﬂation or deﬂation pressures. Furthermore, as recently hypothesized (23), the pressure–volume curve is not useful for detecting overinﬂation that occurs simultaneously with lung recruitment in patients with a focal loss of aeration (Fig. 6).

438

Rouby and Lu

Figure 7 CT section at the carinal level obtained in a patient four weeks after the onset of severe ARDS caused by peritonitis. Bronchiectasis of the divisions of right and left upper bronchi are present. There is a coarse reticular pattern with distortion of bronchial architecture evocative of lung ﬁbrosis. Abbreviations: ARDS, acute respiratory distress syndrome; CT, computed tomography.

Several factors may promote mechanical ventilation–induced lung overinﬂation at early stages of acute lung injury: lung morphology (29), high PEEP (31), high peak inspiratory pressure (26), high tidal volume (13), and preexisting emphysema (31). The distribution of the loss of lung aeration appears to be the main contributing factor. In patients with a focal loss of aeration predominantly involving lower lobes, overinﬂation appears in anterior parts of middle lobes at moderate PEEPs (29–31). In contrast, when the loss of aeration involves homogeneously all parts of the lungs, the administration of high PEEP exclusively results in alveolar recruitment, and virtually no overinﬂation can be detected (34,35). Very likely, lung overinﬂation at the early phase of acute lung injury consists of bronchiolar and alveolar distension and might be caused by early ﬁbrosis (36) and/or high tidal volume (13). C. CT of Lung Overinﬂation at the Late Phase of ARDS and Acute Lung Injury

In patients surviving the acute phase of acute lung injury, lung overinﬂation becomes visible as lung distortion on CT sections (15,36–38). Bronchiectasis, pseudocysts, and emphysema have been reported. Bronchial Dilatation and Distortion

Bronchial dilatation is a frequent CT ﬁnding in acute lung injury survivors and nonsurvivors (12). It is rarely isolated (39) and frequently associated with a coarse reticular pattern suggesting ﬁne intralobular ﬁbrosis (14,36). Bronchiectasis are predominantly found in nondependent and caudal lung

Lung Imaging of Ventilator-Associated Injury

439

Figure 8 (A) CT image obtained eight days after initiating mechanical ventilation in a patient with ARDS caused by bacterial pneumonia shows ground glass opacities with multiple bilateral pseudocysts predominating in nondependant lung areas. (B) CT image obtained at a later stage of ARDS showing diffuse ground glass opacities and a large bulla in the left anterior hemithorax. Abbreviations: ARDS, acute respiratory distress syndrome; CT, computed tomography. Source: From Refs. 44, 45.

regions, suggesting that they result from overdistension of nonconsolidated lung regions (14,15). This anatomical distribution is very similar to the regional distribution of positive pressure ventilation–induced lung overinﬂation observed at early stages of the disease (13,31). In other words, early bronchiolar distension is transformed into visible bronchiectasis a few weeks later if the patient survives. A typical example of bronchial dilatation observed at a late stage of acute lung injury is shown in Figure 7.

440

Figure 9

Rouby and Lu

(Caption on facing page)

Lung Imaging of Ventilator-Associated Injury

441

Pseudocysts, Bullae, and Pneumatoceles

The ﬁrst description of pseudocysts and pneumatoceles complicating the course of acute lung injury in mechanically ventilated patients was reported in the 1980s (10,11,40,41). Subsequently, they were anecdotically reported in many studies (15,38,42) as sometimes resulting in impressive destructive lesions of the lungs (43). Although their pathophysiology is incompletely understood, they likely result from a conﬂuence of alveolar ruptures and/ or bronchiolar dilatation in nonconsolidated areas characterized by a persisting aeration (13). A typical example of a large pseudocyst evidenced at a later stage of acute lung injury is shown in Figure 8. Emphysema-Like Lesions

Emphysema-like lesions were histologically evidenced in mechanically ventilated patients who died from severe acute lung injury in the early 1990s (12) and were reproduced in experimental animals less than 10 years later (13). Two studies have reported CT evidence of macroscopic lung emphysema in two patients who survived ARDS (37,46). However, many investigative CT studies have reported in mechanically ventilated patients and experimental animals the existence of lung areas whose radiological density is in the range of densities characterizing lung emphysema (21,24,26–31). Despite the lack of existing data on the pathological aspects of lung areas for which CT attenuations are below 900 HU in mechanically ventilated patients, it is highly likely according to human (12) and experimental studies (13) that they are made of emphysema-like lesions. Interestingly, during mechanical ventilation applied to critically ill patients lying in the supine position, lung overinﬂation, likely corresponding to emphysema-like lesions, is predominantly

Figure 9 (Facing page) Chest radiographs and corresponding CT images taken at different stages of an ARDS caused by a left lower lobe pneumonia. (A) A chest radiograph and a CT image taken the same day three weeks after the onset of ARDS show an abscess of the apical segment of the left lower lobe. There is dense bilateral dependent lung consolidation with patchy ground glass areas in nondependent lung regions. (B) The chest radiograph and the corresponding CT image taken the same day ﬁve months after the onset of ARDS show bilateral loculated pneumothoraces despite the placement of several chest drains on both sides. (C) The CT image taken six months after the onset of ARDS shows diffuse emphysema and patchy areas of lung ﬁbrosis. Abbreviations: ARDS, acute respiratory distress syndrome; CT, computed tomography. Source: From Ref. 46.

442

Rouby and Lu

distributed in anterior and caudal parts of middle lobes, whereas preexisting emphysema resulting from chronic obstructive pulmonary disease predominates in upper lobes (31) (Fig. 9). Risk Factors and Mechanisms

Different risks factors for late bronchioloalveolar distension have been identiﬁed: high tidal volume (12), high peak airway pressure (12,15), number of days at high inspired O2 fraction (12,18), weight loss due to malnutrition (12,47), lung infection (48), and vascular thrombosis resulting in lung ischemia (48). The exact mechanisms by which alveolar ruptures occur in noninﬂammatory and noninfected parts of injured lungs during positive pressure ventilation are not fully understood. In addition to mechanical stretch applied to alveolar walls, biochemical alterations such as elastine breakdown or metalloproteinase alterations may impair the mechanical properties of alveolar walls. Further studies are required to elucidate the exact mechanisms of mechanical ventilation–induced lung overinﬂation. References 1. Ricard JD, Dreyfuss D, Saumon G, et al. Ventilator-induced lung injury. Eur Respir J Suppl 2003; 42:2s–9s. 2. Tremblay LN, Slutsky AS. Ventilator-induced injury: from barotrauma to biotrauma. Proc Assoc Am Physicians 1998; 110:482–488. 3. Rouby JJ, Puybasset L, Nieszkowska A, Lu Q. Acute respiratory distress syndrome: lessons from computed tomography of the whole lung. Crit Care Med 2003; 31:S285–S295. 4. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 5. Dos Santos CC, Slutsky AS. Invited review: mechanisms of ventilator-induced lung injury: a perspective. J Appl Physiol 2000; 89:1645–1655. 6. Ranieri VM, Suter PM, Tortorella C, et al. Effect of mechanical ventilation on inﬂammatory mediators in patients with acute respiratory distress syndrome: a randomized controlled trial. JAMA 1999; 282:54–61. 7. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342:1301–1318. 8. Dreyfuss D, Ricard JD, Saumon G. On the physiologic and clinical relevance of lung-borne cytokines during ventilator-induced lung injury. Am J Respir Crit Care Med 2003; 167:1467–1471. 9. Slavin G, Nunn JF, Crow J, Dore CJ. Bronchiolectasis—a complication of artiﬁcial ventilation. Br Med J (Clin Res Ed) 1982; 285:931–934. 10. Lemaire F, Cerrina J, Lange F, Harf A, Carlet J, Bignon J. PEEP-induced airspace overdistension complicating paraquat lung. Chest 1982; 81:654–657. 11. Churg A, Golden J, Fligiel S, Hogg JC. Bronchopulmonary dysplasia in the adult. Am Rev Respir Dis 1983; 127:117–120.

Lung Imaging of Ventilator-Associated Injury

443

12. Rouby JJ, Lherm T, Martin de Lassale E, et al. Histologic aspects of pulmonary barotrauma in critically ill patients with acute respiratory failure. Intensive Care Med 1993; 19:383–389. 13. Goldstein I, Bughalo MT, Marquette CH, Lenaour G, Lu Q, Rouby JJ. Mechanical ventilation-induced air-space enlargement during experimental pneumonia in piglets. Am J Respir Crit Care Med 2001; 163:958–964. 14. Desai SR, Wells AU, Rubens MB, Evans TW, Hansell DM. Acute respiratory distress syndrome: CT abnormalities at long-term follow-up. Radiology 1999; 210:29–35. 15. Treggiari MM, Romand JA, Martin JB, Suter PM. Air cysts and bronchiectasis prevail in nondependent areas in severe acute respiratory distress syndrome: a computed tomographic study of ventilator-associated changes. Crit Care Med 2002; 30:1747–1752. 16. Fallat RJ, Lamy M, Koeniger E, Hill JD. Use of Physiologic and Pathologic Correlations in Evaluating Adult Respiratory Distress Syndrome, Artiﬁcials Lungs for Acute Respiratory Failure. Theory and Practice. In: Zapol W, Qvist J, eds. Wasington, London: Hemisphere Publishing Corporation, 1976:391–404. 17. Jones RM, Reid LM, Kirton OC, Zapol W, Kobayashi K, Tomashefski JF. Pulmonary Vascular Pathology. Human and Experimental Studies, Acute Respiratory Failure. In: Zapol W, Falke KJ, eds. New York, Basel: Marcel Dekker, Inc., 1985:23–160. 18. Riley DJ, Kramer MJ, Kerr JS, Chae CU, Yu SY, Berg RA. Damage and repair of lung connective tissue in rats exposed to toxic levels of oxygen. Am Rev Respir Dis 1987; 135:441–447. 19. Sahebjami H, Wirman JA. Emphysema-like changes in the lungs of starved rats. Am Rev Respir Dis 1981; 124:619–624. 20. Gattinoni L, Caironi P, Pelosi P, Goodman LR. What has computed tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001; 164:1701–1711. 21. Vieira SR, Puybasset L, Richecoeur J, et al. A lung computed tomographic assessment of positive end-expiratory pressure-induced lung overdistension. Am J Respir Crit Care Med 1998; 158:1571–1577. 22. Gevenois PA, De Vuyst P, Sy M, et al. Pulmonary emphysema: quantitative CT during expiration. Radiology 1996; 199:825–829. 23. Rouby JJ, Lu Q, Vieira S. Pressure/volume curves and lung computed tomography in acute respiratory distress syndrome. Eur Respir J Suppl 2003; 42:27s–36s. 24. Malbouisson LM, Muller JC, Constantin JM, Lu Q, Puybasset L, Rouby JJ. Computed tomography assessment of positive end-expiratory pressure-induced alveolar recruitment in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163:1444–1450. 25. Vieira SR, Nieszkowska A, Lu L, Elman M, Sartorius A, Rouby JJ. Low spatial resolution computed tomography underestimates lung overinﬂation resulting from positive pressure ventilation. Crit Care Med 2005; 33:741–749. 26. Lim CM, Soon Lee S, Seoung Lee J, et al. Morphometric effects of the recruitment maneuver on saline-lavaged canine lungs. A computed tomographic analysis. Anesthesiology 2003; 99:71–80.

444

Rouby and Lu

27. Dambrosio M, Roupie E, Mollet JJ, et al. Effects of positive end-expiratory pressure and different tidal volumes on alveolar recruitment and hyperinﬂation. Anesthesiology 1997; 87:495–503. 28. Puybasset L, Cluzel P, Chao N, Slutsky AS, Coriat P, Rouby JJ. A computed tomography scan assessment of regional lung volume in acute lung injury. The CT Scan ARDS Study Group. Am J Respir Crit Care Med 1998; 158:1644–1655. 29. Vieira SR, Puybasset L, Lu Q, et al. A scanographic assessment of pulmonary morphology in acute lung injury. Signiﬁcance of the lower inﬂection point detected on the lung pressure-volume curve. Am J Respir Crit Care Med 1999; 159:1612–1623. 30. Puybasset L, Gusman P, Muller JC, Cluzel P, Coriat P, Rouby JJ. Regional distribution of gas and tissue in acute respiratory distress syndrome. III. Consequences for the effects of positive end-expiratory pressure. CT Scan ARDS Study Group. Adult Respiratory Distress Syndrome. Intensive Care Med 2000; 26:1215–1227. 31. Nieszkowska A, Lu Q, Vieira S, Elman M, Fetita C, Rouby JJ. Incidence and regional distribution of lung overinﬂation during mechanical ventilation with positive end-expiratory pressure. Crit Care Med 2004; 32:1496–1503. 32. Pelosi P, Goldner M, McKibben A, et al. Recruitment and derecruitment during acute respiratory failure: an experimental study. Am J Respir Crit Care Med 2001; 164:122–130. 33. Albaiceta GM, Taboada F, Parra D, et al. Tomographic study of the inﬂection points of the pressure–volume curve in acute lung injury. Am J Respir Crit Care Med 2004; 170:1066–1072. 34. Rouby JJ, Lu Q, Goldstein I. Selecting the right level of positive end-expiratory pressure in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2002; 165:1182–1186. 35. Rouby JJ, Constantin JM, Roberto De AGC, Zhang M, Lu Q. Mechanical ventilation in patients with acute respiratory distress syndrome. Anesthesiology 2004; 101:228–234. 36. Howling SJ, Evans TW, Hansell DM. The signiﬁcance of bronchial dilatation on CT in patients with adult respiratory distress syndrome. Clin Radiol 1998; 53:105–109. 37. Owens CM, Evans TW, Keogh BF, Hansell DM. Computed tomography in established adult respiratory distress syndrome. Correlation with lung injury score. Chest 1994; 106:1815–1821. 38. Gattinoni L, Bombino M, Pelosi P, et al. Lung structure and function in different stages of severe adult respiratory distress syndrome. JAMA 1994; 271:1772–1779. 39. Finfer S, Rocker G. Alveolar overdistension is an important mechanism of persistent lung damage following severe protracted ARDS. Anaesth Intensive Care 1996; 24:569–573. 40. Johnson TH, Altman AR, McCaffree RD. Radiologic considerations in the adult respiratory distress syndrome treated wih positive end expiratory pressure (PEEP). Clin Chest Med 1982; 3:89–100.

Lung Imaging of Ventilator-Associated Injury

445

41. Stark P, Greene R, Kott MM, Hall T, Vanderslice L. CT-ﬁndings in ARDS. Radiologe 1987; 27:367–369. 42. Wagner PK, Knoch M, Sangmeister C, Muller E, Lennartz H, Rothmund M. Extracorporeal gas exchange in adult respiratory distress syndrome: associated morbidity and its surgical treatment. Br J Surg 1990; 77:1395–1398. 43. Tobin MJ. Advances in mechanical ventilation. N Engl J Med 2001; 344: 1986–1996. 44. Desai SR, Wells AU, Suntharalingam G, Rubens MB, Evans TW, Hansell DM. Acute respiratory distress syndrome caused by pulmonary and extrapulmonary injury: a comparative CT study. Radiology 2001; 218:689–693. 45. Goodman LR. Congestive heart failure and adult respiratory distress syndrome. New insights using computed tomography. Radiol Clin North Am 1996; 34: 33–46. 46. Dakin J, Grifﬁths M. The pulmonary physician in critical care 1: pulmonary investigations for acute respiratory failure. Thorax 2002; 57:79–85. 47. Kerr JS, Riley DJ, Lanza-Jacoby S, et al. Nutritional emphysema in the rat. Inﬂuence of protein depletion and impaired lung growth. Am Rev Respir Dis 1985; 131:644–650. 48. Greene R. Adult respiratory distress syndrome: acute alveolar damage. Radiology 1987; 163:57–66.

This page intentionally left blank

18 Imaging Ventilator-Induced Lung Injury: Present and Future Possibilities

DANIEL P. SCHUSTER Departments of Internal Medicine and Radiology, Washington University School of Medicine St. Louis, Missouri, U.S.A.

I. Introduction The term ‘‘ventilator-induced lung injury’’ (VILI) refers to those aspects of acute respiratory failure that can be attributed to mechanical ventilatory support or management. In some instances, the use of positive airway pressure itself (rather than ‘‘negative’’ airway pressure, as with normal breathing) may be enough to cause or exacerbate injury to lung tissue, for instance in the presence of an inﬂammatory response to the primary cause of respiratory failure. Alternatively, speciﬁc types of ventilator management (or ‘‘mismanagement’’) may confer additional lung injury. Regardless, the ‘‘manifestations’’ of VILI are identical to any other cause of acute lung injury (ALI) (1). It is for this reason that in the clinical setting, VILI is often referred to as ventilator-associated lung injury. The implications of this uncertainty are important when one considers the value of the various methods for studying VILI—the use of imaging is no exception. Imaging has much to offer when studying VILI: as a noninvasive method, it can provide a seamless translation from studies in animals to later studies in humans. In addition, a diverse group of in vivo 447

448

Schuster

imaging methods are now available (2,3) to visualize lung structure, function, and most recently, molecular/cellular events in whole animals and in humans that are germane to VILI, whether they include tissue damage resulting in air leaks, inﬂammatory cell trafﬁcking, or therapeutic gene transfer. Thus, in this chapter, a variety of imaging strategies and platforms that can be used to study VILI, are reviewed. Nevertheless, these same methods and issues are equally applicable to ALI from any cause, and no imaging method can yet quantify what portion of the total clinical or experimental picture is speciﬁcally the result of ventilator support or management per se. Given the nonspeciﬁc nature of VILI in the clinical setting, most of our understanding about the pathogenesis of VILI has come from animal models. Imaging can help deﬁne the pathogenesis of VILI because a new generation of imaging instrumentation now makes it possible to extend anatomic, functional, and molecular imaging methods to VILI models, including those in rodents (2). Even more recent improvements in instrumentation are making it possible to combine different types of measurements into single multimodality imaging sessions. The terms ‘‘anatomic,’’ ‘‘functional,’’ and ‘‘molecular’’ imaging may be unfamiliar to some readers, and indeed, standard deﬁnitions for these terms have yet to be developed. Generally, we use ‘‘anatomic imaging’’ to include those techniques that are used to display structure (e.g., in the case of the lungs, airway diameter), or to make measurements related to structure [e.g., lung volumes such as functional residual capacity, (FRC)]. Functional imaging methods almost always depend on repeated data over time to measure dynamic biologic processes such as ventilation or perfusion. Finally, molecular imaging includes in vivo methods that are used to detect the presence, expression, or activity of speciﬁc molecular targets in tissuesof-interest (e.g., the expression of transgenes). Anatomic imaging involves relatively long time constants, i.e., changes, if any, in the structure being imaged occur over a period of hours, days, or even years. Changes mapped by functional or molecular imaging usually occur over seconds to hours. Obviously, these deﬁnitions and distinctions are not always mutually exclusive of one another.

II. Anatomic Imaging of VILI: Quantifying Edema Accumulation Regardless of cause, pulmonary edema is one of the universal consequences of ALI. We include the quantitation of extravascular lung water (EVLW) as a type of ‘‘anatomic’’ imaging in that such measurements represent a change in the composition of one or more structural compartments of the lungs (the interstitial and alveolar spaces). Measurements of EVLW

Imaging VILI: Present and Future Possibilities

449

are among the most common methods of quantifying ALI. Included in these methods are a number of noninvasive imaging techniques, including chest radiography, X-ray computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET). A. Chest Radiography

Obviously, the chest radiograph can serve multiple purposes, but it can also be used, at least semiquantitatively, to estimate the amount of pulmonary edema. Certain characteristic radiographic ‘‘signs’’ such as pulmonary ‘‘congestion,’’ vascular ‘‘redistribution,’’ peribronchial cufﬁng, perihilar ‘‘haze,’’ the presence of Kerley’s lines, and an ‘‘interstitial’’ pattern to the radiographic densities are associated with relatively modest increases in EVLW (perhaps as little as 30% above normal values) (4). In relative terms, one can also usually determine whether the amount of pulmonary edema has increased or decreased because, as EVLW changes, the radiographic opacities become more or less dense, or occupy greater or lesser fractions of the total lung airspace. Even so, the overall accuracy of chest radiography in measuring EVLW, especially in patients with ALI, is signiﬁcantly limited by technical problems such as the lack of a standardized yet practical acquisition technique that is appropriate for a clinical setting (5–9). B. Computed Tomography

As in the chest radiograph, the intensity of opacities visualized by X-ray CT within the lung parenchyma is determined by differences in tissue density. Because the lungs are air-containing organs, the density of the lungs (on average, approximately 0.3–0.5 g/mL) is considerably less than that of water (1 g/mL). Typically, calibration of CT densitometry is performed against air and water densities [and expressed in arbitrary Hounsﬁeld units (HU) with air ¼ 1000 HU and water ¼ 0 HU]. Accordingly, normal lung parenchyma has a CT density between 900 and 500 HU (10). During ALI, density values of the lung parenchyma are shifted toward that of water (Fig. 1) (11). In experimental animals, CT densitometry can detect modest (approximately 50%) increases in EVLW (12) [for context, EVLW can increase two to six times the normal value during ALI/ARDS (13)]. Obviously, changes in lung density are not speciﬁc for pulmonary edema and cannot be distinguished from increases due to inﬂammatory inﬁltrate, hemorrhage, or atelectasis (Fig. 2). However, this is a general problem for the quantitation of any measurable variable during imaging of the lungs by any method. The so-called ‘‘microCT’’ instruments are now being used to generate anatomic images of the lungs and thorax in exquisite detail. State-of-the-art systems can produce images with a spatial resolution of 50 to 100 mm (14). For example, microCT images of rat lung before and after VILI are

450

Schuster

Figure 1 Lung compartments deﬁned by CT numbers in healthy lungs (solid line) and during ALI/ARDS (dashed line). Abbreviations: ALI, acute lung injury; CT, computed tomography; ARDS, acute respiratory distress syndrome. Source: From Ref. 11.

Figure 2 (See color insert.) Effect of state of lung inﬂation on the measured density of lung parenchyma in each volume element (voxel) of an image, at TLC, FRC during atelectasis (collapse), or in the presence of alveolar edema. Abbreviations: FRC, functional residual capacity; TLC, total lung capacity. Source: From Ref. 11.

Imaging VILI: Present and Future Possibilities

451

Figure 3 Coronal images of a normal mouse and a mouse with OA-induced lung injury obtained with an Imtek microCAT II scanner. Note the increase in lung density overall in the OA-treated mouse. The trachea and major bronchi are also visible (reconstructed at a resolution of 200 mm). Abbreviation: OA, oleic acid.

shown in Figure 3. Relevant issues such as X-ray source, focal spot size, detector element size, system geometry, and X-ray ﬂux are important in determining image spatial and contrast resolution, and are discussed in detail elsewhere (15). An issue of some importance when designing serial studies in rodent models is the radiation dose delivered to the animal during microCT imaging. This may approach 5% of the LD50 in mice, potentially limiting the number of repeat studies that could be performed over time (15). It is estimated that the life of a mouse is shortened by 7.2% per Gy of radiation exposure, and residual effects from sublethal radiation doses can accumulate (16). C. Mechanical Resonance Imaging

Another approach to estimating lung water is based on the ability to align hydrogen nuclei (protons) of water in the direction of an externally applied magnetic ﬁeld (17). The fundamental principle underlying proton MRI (1H MRI) is that when an individual, lying within a magnetic ﬁeld, is subjected to a radio-frequency (rf) pulse applied at the correct (‘‘resonant’’) frequency, the protons within the body absorb energy and generate a detectable signal. The signal ‘‘strength’’ depends upon the number (concentration) of protons; its ‘‘frequency’’ depends on the identity of the nuclide (e.g.,1H and 3He) and the strength of the local magnetic ﬁeld. By applying magnetic ﬁeld gradients along well-deﬁned directions in space, the frequency of the proton signal is

452

Schuster

made dependent upon spatial position. Collection and processing of this signal produces a map of MR signal intensity versus position, i.e., an image. Following excitation by rf pulses, protons (and all other MR-active nuclides) return to equilibrium (‘‘relax’’). This relaxation process can be described by two fundamental rate constants: longitudinal relaxation (T1) and transverse (T2) relaxation. Importantly, different tissues can have different T1 and T2 values, and the intensity of signals in MRI images can be made sensitive to these relaxation parameters. As a result, T1 and T2 become potential sources of contrast between different organs within the body and between healthy and diseased tissues. The timing parameters, TR and TE, are also fundamental to all MR imaging experiments. TE is deﬁned as the time between the excitation rf pulse and the start of data acquisition, while TR is deﬁned as the time between sequential (repeat) acquisitions. T2-weighting is achieved by adjusting TE, while T1-weighting is achieved by changing TR. MRI experiments themselves can be divided into two basic types: gradient-echo experiments, which use a single rf pulse, and spin-echo experiments, which use two rf pulses. Gradient-echo experiments can generally be performed more quickly; spin-echo studies generally produce higher quality images. An advantage of MRI over other methods to quantify lung water is that the measurements can be obtained without the need for ionizing radiation; furthermore, the measurements are completely independent of any disturbance in ventilation or perfusion [unlike techniques which depend upon the administration of indicators or tracers by inhalation or intravenous (i.v.) injection]. However, imaging the lung parenchyma presents some unique challenges for MRI (18), including: (i) relatively low tissue density (and therefore low water content within the lungs), severely limiting signalto-noise; (ii) variations in magnetic ‘‘susceptibility,’’ associated with the many air–tissue interfaces of the alveoli and bronchioles, creating local magnetic ﬁeld inhomogeneities (ﬁeld gradients) that can lead to shortening of some relaxation times; and (iii) respiratory and cardiac motion, leading to signiﬁcant image blurring (true, however, of all lung imaging methods). In response to these challenges, very fast gradient-echo imaging sequences with short TE times or spin-echo sequences with respiratory synchronization are used in pulmonary studies (19,20). In addition, static magnetic ﬁeld strength, gradient strength, and rf coil performance can all affect sensitivity and spatial resolution. For small-animal imaging studies, 4.7 T MR scanners are commonly used (compared with the 1.5 T magnetic ﬁeld typical of clinical scanners). High gradient strengths (up to 60 G/cm, compared with 2 G/cm on most clinical scanners) are also typical. Experiments in small animals are performed with respiratory gating (20,21), because

Imaging VILI: Present and Future Possibilities

453

breath holding is often not an option. Data collection varies from minutes to one or more hours per animal, depending upon the nature of the experiments being performed. Although numerous studies have reported a good ‘‘correlation’’ between MRI-determined estimates of lung water and estimates from the gold-standard gravimetric method (22–30), the normal or mildly edematous lung produces relatively little signal and as a result, MRI methods can underestimate true lung water in absolute terms by as much as 20% to 40% (22,23,30). In the presence of signiﬁcant amounts of pulmonary edema, repeated measures of lung water by MRI vary by only 5% to 10% (23). D. Positron Emission Tomography

To evaluate pulmonary edema with PET, sterile water is labeled with the positron-emitting isotope oxygen-15 (H215O) (t1/2 ¼ 2.06 minutes) and administered intravenously (31). H215O, like all isotopes used in PET, decay by the process of positron emission. The emitted positron is quickly annihilated by its interaction with an ambient electron. Two photons are then emitted after each annihilation event, traveling nearly 180 apart. These photons are counted with detectors placed on opposite sides of the body. Because the annihilation event that produced the photons can be assumed to have occurred somewhere within the tissue volume subtended by the two detectors, the radiation source can be partially located in space. As additional detector pairs are added to the system, intersecting lines further establish spatial location. This type of collimation by electronic coincidence detection is far more efﬁcient than that provided by the lead shielding used in conventional single-photon-emission detection systems because less radiation is ignored. The O-15–labeled water is allowed to equilibrate with tissue water over a three- to four-minute period. Then, the isotope’s activity concentration in the lung tissue is determined with a PET camera. The radioactivity data in the PET image are scaled to radioactivity data simultaneously obtained in blood (blood ‘‘water,’’ as a standard), allowing an image to be developed that is a quantitative regional map of lung water distribution (31). An analogous approach is used to measure the intravascular concentration of water in blood within the lungs. First, carbon monoxide is labeled with a positron-emitting isotope such as O-15 or carbon-11. Trace amounts are then inhaled as a gas, binding immediately to blood hemoglobin. After a few minutes, to allow equilibration within the body’s blood pool, another PET scan is obtained. When again normalized to activity measurements in blood (either from blood sampling or by placing a region-of-interest within a blood pool such as one of the cardiac chambers on the PET image), the resulting image in this case is a regional display of the concentration of blood. With the assumption that 84% of blood is

454

Schuster

water (a reasonable assumption at normal hematocrits), the blood water concentration in a lung region can be subtracted from the total lung water concentration, yielding a derived image of ‘‘extravascular’’ water concentration. The total time required to measure EVLW with PET is about 45 minutes, but repeat measurements can begin in as little as 10 to 15 minutes from the previous one. EVLW measurements by PET correlate well with EVLW measurements obtained by gravimetrics (r ¼ 0.860.93) (32,33). For valid absolute values, a number of corrections are required (e.g., for differences in large and small vessel hematocrit, differences in tissue versus blood density, among others), but the errors involved are only about 10% to 15% compared to a gravimetric gold standard if these are ignored. The measurements are highly reproducible (coefﬁcient-of-variation for repeat measurements <5%), linear (r ¼ 0.99 for changes in lung water over a 20-fold concentration range) (33), and sensitive (as little as 1 mL of additional extravascular water can be detected with PET) (33). Until recently, limitations in spatial resolution with PET made such studies impossible in small animals. For instance, most PET scanners in clinical use have an image spatial resolution of 10 to 15 mm, but the mouse thorax, in vivo, is only 20 to 25 mm wide. Other issues, such as scanner sensitivity (i.e., the fraction of radioactive events actually detected by the device) and the amount of radioactivity that can be injected without causing physiologic disturbances in the system under study (a function of tracer speciﬁc activity), also affect the ability to perform imaging studies in small laboratory animals. Nevertheless, recent advances in the scintillation materials used to make the radiation detectors used in PET devices, and in the ability to transfer the scintillation light from the detectors to photomultiplier tubes (34), now make it possible to manufacture devices with remarkable improvements in spatial resolution. The resolution of currently available microPET scanners may still limit the ability to make measurements of pulmonary edema in small animals, particularly in mice. The problem lies in the positron range of the tracer, O-15, used to label H2O. Positron range is the distance traveled by the positron after nuclear decay before the positron is annihilated by its interaction with an ambient electron (recall it is the annihilation radiation that is actually detected by the PET camera). The positron range for O-15 is on average 2 mm, with a range up to 7 mm (35). Given that the average chest diameter of a mouse is only 20 to 25 mm, the resolution may not be adequate to give accurate measurements of lung water in mice. Additionally, because PET imaging is done on actively breathing animals, so-called partial volume effects due to motion of the diaphragm, chest wall, and heart cause further blurring of the signal at the edges of the lungs, making it difﬁcult to identify regions of interest that do not include these surrounding structures in small animals. With rats, whose chest diameters are

Imaging VILI: Present and Future Possibilities

455

signiﬁcantly larger, the resolution of current microPET scanners should be adequate for making measurements of pulmonary edema. We recently performed PET imaging in an oleic-acid injury model in rats using O-15–labeled water to determine total lung water accumulation that is characteristic of this model (Fig. 4). Images were obtained with a ﬁve-minute scan beginning four minutes after the tail vein injection of approximately 10 mCi of [15O]H2O. Radioactivity measurements were decay-corrected to the time of tracer injection. Lung water concentration is calculated as the activity during the equilibrium [15O]H2O scan in any lung region-of-interest divided by the activity over the heart, with the assumption that the water content of blood or cardiac tissue is approximately 0.84 mL H2O/mL of tissue (31). The lung water concentration increased from 0.48 to 0.76 mL H2O/mL lung, a 57% increase. This increase in lung water concentration matched a comparable increase in postmortem lung weight, compared to normal rat lung weights in our laboratory. Further validation experiments will be necessary to determine the accuracy of these measurements. As noted earlier, to measure EVLW with PET also requires an estimate of blood water, for instance with O-15–labeled carbon monoxide, in the same region-of-interest used to estimate total lung water (32). This particular measurement has not yet been developed for use in small animals, and issues regarding positron range would be as relevant for this radiotracer as with O-15–labeled water.

Figure 4 (See color insert.) Lung water images from a normal rat and a rat treated with i.v. oleic acid obtained using a Concorde R4 microPET scanner after tail vein injection of approximately 10 mCi of O-15–labeled water. Images were referenced to a region over the heart, assumed to have a tissue water concentration of 0.84 gm/mL tissue. Note the increase in signal in the OA-treated rat, indicative of increased pulmonary edema as a result of the lung injury. Abbreviations: OA, oleic acid; O-15, oxygen-15. Source: From Ref. 36.

456

Schuster E. Other Anatomic Imaging During Lung Injury

Measurements of air rather than water compartments of the lungs would represent another type of anatomic imaging. However, there are very few examples of this type of imaging applied to ALI. Recently, Mitzner et al. reported a method of measuring lung volumes such as FRC in rats and mice (37–39). Even though their studies were performed with a clinical CT scanner, the same principles should apply and data accuracy should be better with microCT scanners.

III. Functional Imaging of VILI A. Quantifying the Severity of Lung Injury

The apposition of the type 1 alveolar epithelium with the capillary endothelium serves as a barrier to excess water and protein ﬂux into the alveolar airspace, leaving these units clear for gas exchange. The barrier, however, functions as a semipermeable membrane in which water has greater ‘‘permeability’’ than proteins. Thus, the barrier can be intact, but increased amounts of water in the airspaces can still accumulate because of increased hydrostatic forces. Accordingly, increased EVLW is only one piece of evidence for lung injury [if injury includes evidence for structural damage to the alveolar epithelium, otherwise known as ‘‘diffuse alveolar damage’’ (40)]. A common additional criterion is the evidence of increased permeability to protein across the alveolocapillary barrier. Various imaging methods have been studied as methods of quantifying increases in vascular permeability. These methods usually depend on monitoring the time-dependent behavior of some radioactively labeled substance. Thus, imaging platforms such as X-ray CT or MRI have not been used to quantify lung injury in this fashion. Furthermore, while both imaging and nonimaging methods may be able to document and quantify increased permeability, it should be noted that increased permeability (along with increased EVLW) is an example of a ‘‘necessary but not sufﬁcient’’ criterion for the presence of alveolar damage per se. In other words, a normal permeability measurement (by imaging or any other method) is inconsistent with injury. However, increased permeability is also a universal feature of the inﬂammatory response, even in the absence of actual structural damage to local tissues. Thus, available imaging methods may be sensitive, but are not speciﬁc markers of injury. Radioaerosol Inhalation

One imaging technique for detecting increased permeability of the alveolocapillary barrier involves inhalational administration of an aerosol containing

Imaging VILI: Present and Future Possibilities

457

a radioactively labeled tracer. The clearance of tracer activity from the lung is then measured, usually with some external radiation detection system such as a gamma camera. Because the alveolar epithelium is normally virtually impenetrable to macromolecules, clearance of the tracer from the alveolar space is expected to be quite slow. Any signiﬁcant increase in the rate of clearance would constitute evidence for altered epithelial integrity. Unfortunately, many unsolved problems limit the value of this approach, including inefﬁcient delivery of tracer, inability to ensure tracer delivery to the most injured (and thus, often, the most poorly ventilated) areas, inability to quantify the amount of tracer delivered, instability of some radioactive labels on some tracers, and uncertainty about the site of clearance (e.g., airway vs. alveolus) (13). Most problematic of all, changes in lung inﬂation alone (as with positive end-expiratory pressure), possibly because of changes in exchangeable surface area, can increase clearance rates to levels that are indistinguishable from those observed in ARDS patients. Smoking in otherwise healthy individuals produces a similar result (41). As a result, this approach to quantifying lung injury is rarely used. Radioactive Tracers in Blood

Measuring the time-dependent accumulation of an intravenously administered radioactively labeled protein tracer into lung tissue has been the most common clinical method of determining whether or not the integrity of the capillary endothelial barrier has been compromised (13,42,43). Intuitively, as the microvascular barrier becomes more ‘‘leaky’’ to macromolecules, the rate of accumulation of a radiolabeled protein (e.g., albumin) within the lung should increase. With imaging, the radioactive tracer is injected IV, and then the radioactivity within lung tissue is detected and recorded for minutes to hours with one of several kinds of external radiation detection devices. Time-activity data are simultaneously obtained from the blood, either by measuring activity within the blood pool of a cardiac chamber with the same external detection system, or by measuring the protein radioactivity in separately obtained blood samples (44–46). ‘‘Permeability’’ can then be quantiﬁed as a rate constant for the movement of the tracer between the blood and lung tissue compartments (47). Most (but not all) experimental studies have shown that estimates of vascular permeability by these approaches are insensitive to changes in the hydrostatic pressure, but are very sensitive to changes in the membrane ‘‘porosity’’ (48). Tenfold increases in permeability have been measured in humans with ARDS compared to normals (13). Most clinical studies of these imaging techniques report little or no increases in permeability in patients with heart failure. However, a recent study, (49) in which patients were chosen simply on the basis of the radiographic presence of pulmonary edema and

458

Schuster

not on the basis of clinical criteria, was unable to distinguish between patients with ALI or heart failure, casting doubt on whether these imaging methods can reliably distinguish these two forms of pulmonary edema. B. Ventilation and Ventilation-Perfusion Imaging

Recent advances in MRI using hyperpolarized (HP) gases have allowed experimenters to measure ventilation in rodents. With the development of new techniques for synchronizing MR scan acquisition with ventilation, detailed maps of the lungs can now be created with high resolution (50). One of the challenges in using HP gases for pulmonary imaging, however, is that these gases do not naturally maintain a polarized state for long. Thus, any polarized gas that is subjected to a magnetic ﬁeld whose direction is signiﬁcantly different from the direction of the gas polarization will rapidly depolarize the gas and render it unusable for further study. In the lungs, this phenomenon results in the loss of the MR signal before the HP gas has a chance to reach the periphery. To use HP gases to monitor ventilation throughout the lungs, MR data acquisition protocols must be manipulated to preserve signal intensity long enough to allow the gases to penetrate into the periphery of the lungs while still in a state of hyperpolarization. Chen et al. recently reported such a technique using the radial acquisition cine pulse sequence in conjunction with a skipping scheme in the applied rf pulse sequences, which allowed HP gas time to reach the lung periphery while maintaining its polarization (51). This technique requires imaging over multiple breaths of HP gases in order to generate the ﬁnal images. By using this dynamic scanning technique, they were able to generate images clearly showing the peripheral lung as well as the major airways; in addition, airﬂow in the major airways could be measured by normalizing the MR signal intensity in a region of interest placed over an airway to the diameter of the airway. Regional ventilation could also be calculated because it would be proportional to the signal intensity in the periphery of the lung as long as care was taken to preserve the polarization of the gas in the area of interest (51). Dupuich et al. employed a single-breath technique using a different acquisition technique and data processing protocol, which they called sliding pulmonary imaging for respiratory overview, to assess ventilation in rats (52). Using this technique, pixel-by-pixel parametric maps of gas arrival time, ﬁlling time constant values, average inﬂation rate, and gas volume values using a single breath of HP 3He gas were generated. These values were then used to calculate overall gas ﬂow rates and lung volumes and compared with the actual gas ﬂow delivery rate and the actual injected volume of gas. Ventilation–perfusion studies are also now possible in rats. HP 3He has been used in conjunction with gadolinium diethylene triamine

Imaging VILI: Present and Future Possibilities

459

pentaacetic acid or superparamagnetic iron oxide nanoparticles as i.v. contrast agents to create ventilation–perfusion images in rats (50,53,54). While ventilation–perfusion measurements have been obtained with PET and CT in large animal models (55,56), these techniques have not yet been translated into small animal models. Dynamic image acquisition with respiratory gating should make it possible to visualize the impact of ALI on lung mechanics (an example from a normal mouse is shown in (Fig. 5).

C. Pulmonary Hemodynamics

ALI often has a profound effect on right heart function. Thus, new methods of visualizing right heart function during VILI should be of interest. Recent advances in ultrasound system technology now make it possible to acquire cardiac images with the necessary spatial and temporal resolution for studying pulmonary hypertension in rodents. Using a modiﬁcation to the Bernoulli equation (Prv–ra ¼ 4V2) (where P ¼ pressure in mmHg, rv ¼ right ventricular, ra ¼ right atrial, and V ¼ Doppler ﬂow velocity), changes in pulmonary artery pressure (actually, the peak systolic right ventricular–right atrial pressure gradient) were estimated by Jones et al., using echocardiography in rats and measuring the peak velocity of the tricuspid regurgitation jet that is typically present during pulmonary hypertension. They found, however, that quantitatively useful tricuspid regurgitation was only detected in animals with severe pulmonary hypertension (tricuspid regurgitation velocities of 3 m/sec or higher). Structural changes, such as right ventricular hypertrophy, can also be evaluated in response to increased pulmonary pressure. Depending on the duration and severity, pulmonary hypertension may be associated with right ventricular chamber dilatation, impaired systolic function, and changes in the right ventricular wall thickness, all detectable by high-resolution echocardiography (Fig. 6) (57). The small size of the mouse heart (approximately 10 times smaller than the rat heart) represents a signiﬁcant challenge for imaging right heart structures by echocardiography. Right ventricular wall thickness and chamber dimensions are at the limits of spatial resolution of currently available ultrasound systems. Although echocardiography of the mouse left ventricle has been widely used and validated, there are limited data on imaging of the right heart. MRI with its higher spatial resolution and inherent 3-D analysis capabilities may offer a more accurate alternative to echocardiography for the noninvasive evaluation of right ventricular structure and function in mice (58). These advantages, however, have to be balanced against the expense and relative lack of availability of MRI versus echocardiography.

460

Schuster

Figure 5 Coronal X-ray microCT scans of a mouse, demonstrating the potential use to study lung mechanics. The images were obtained with respiratory gating. (A) End-inspiration. (B) End-expiration. The horizontal white lines show the displacement of the diaphragm. Abbreviation: CT, computed tomography.

Imaging VILI: Present and Future Possibilities

461

Figure 6 Ultrasound images and Doppler recordings obtained in normal rats (left) and rats with mild (center) and severe (right) pulmonary hypertension. (A) Enddiastolic images of the right (crescent region) and left (round region) ventricles along the short-axis view. (B) M-mode images of the inferior portion of the right ventricular free wall. (C) Pulse-wave Doppler recordings of pulmonary outﬂow velocity. Source: From Ref. 36.

IV. Molecular Imaging of VILI A. Inﬂammation Imaging

Once neutrophils are activated, they become sequestered in the lungs, migrate from the microvasculature into the interstitial and alveolar spaces, and ﬁnally increase their production of oxidants and the release of presynthesized proteases. These events require energy, and for neutrophils, glycolysis is the major energy source for cellular functions such as chemotaxis, phagocytosis, and microbial killing. Thus, even though the mechanism(s) and regulation of glucose uptake into the neutrophil have yet to be studied in detail, neutrophil activation is reliably associated with increased glucose uptake.

462

Schuster

In vivo glucose metabolism can be measured with PET imaging after the administration of the glucose analog [18F]ﬂuorodeoxyglucose ([18F]FDG) (59). [18F]FDG moves into cells via the same GLUT family of transporter proteins as glucose (primarily via GLUT-1). There, in the presence of hexokinase, it is phosphorylated to [18F]FDG-6-monophosphate ([18F]FDG-6-P). However, unlike glucose-6-phosphate, [18F]FDG-6-P cannot be metabolized further (60). Therefore, in tissues that lack dephosphorylases (most tissues other than the liver), [18F]FDG is trapped intracellularly. A PET imaging signal can then be generated as [18F]FDG accumulates in the tissue over time (Fig. 7). The estimation of actual glucose metabolic rates from data acquired via FDG-PET imaging requires the use of a correction factor that accounts for differences in the rate of transport and phosphorylation of [18F]FDG when compared with glucose itself (60). Such corrections are important in tissues such as the myocardium, which can switch between oxidative and nonoxidative metabolic pathways to meet energy needs. This phenomenon is not known to occur in neutrophils. Importantly, the pulmonary uptake of [18F]FDG may be a function of the state of neutrophil activation and not just the number of neutrophils present in the tissue (61). This intriguing possibility remains to be validated, however, with other measures of neutrophil activation. The normal value for [18F]FDG uptake in the lungs in humans is reported to be about 0.6

Figure 7 (See color insert.) Example of multimodality inﬂammation imaging in a mouse with experimental right lung pneumonia. (A) X-ray CT image showing increased pulmonary inﬁltrates in the right lung consistent with acute pneumonia. (B) An overlay of the CT and PET images showing anatomic and functional correspondence. (C) PET image showing increased pulmonary uptake of the radiotracer [18F]ﬂuorodeoxyglucose. Abbreviations: CT, computed tomography; PET, positron emission tomography.

Imaging VILI: Present and Future Possibilities

463

to 1.9 mmol/hr/mL (62). In acute pneumonia, in both experimental animals and patients, the rate of [18F]FDG uptake in the lungs increases some 10 to 40 times above values obtained in normal subjects (63). Evidence to date suggests that in states of acute inﬂammation, it is the neutrophil that is the primary cell type responsible for the increase in [18F]FDG tissue uptake (63–66). Thus, FDG-PET imaging appears to be a promising new tool to study the kinetics of neutrophil activation during ALI. Most examples of inﬂammation imaging to date have focused on monitoring neutrophil kinetics and trafﬁcking. However, a number of new tracers are in the process of development for monitoring other types of inﬂammatory cells (e.g., macrophages, lymphocytes, etc.), mediators (e.g., selectins, cytokines, etc.), or processes critical to the regulation of inﬂammation (e.g., apoptosis) (59,67). B. Transgene Expression Imaging: Platforms

One recent and particularly exciting form of molecular imaging seeks to follow the expression of speciﬁc genes via the noninvasive visualization of in vivo gene ‘‘reporters’’ (14,68–70). Three different imaging platforms have been used to image gene expression: MRI, PET, and optical methods. Three general strategies have evolved for optical molecular imaging in vivo: use of endogenous ﬂuorochromes; use of reporter genes that generate internal light from speciﬁc biochemical reactions (bioluminescence and ﬂuorescent proteins); and use of injected optical contrast agents incorporating visible light ﬂuorophores, near-infrared ﬂuorophores, or activatable ﬂuorophores (71–73). To date, most in vivo optical imaging studies have employed bioluminescence methods. Information about the other approaches has been summarized elsewhere (2,74). Bioluminescence speciﬁcally refers to the enzymatic generation of visible light by living organisms (75–77). Measurements with optical imaging methods are relatively easy, quick, and cheap. Available platforms allow multiple animals to be evaluated at the same time, increasing experimental throughput. However, spatial resolution, quantitation, and the imaging of deep tissues are poor or problematic. For instance, as a rule of thumb, there is an approximate 10-fold loss of photon intensity for each centimeter of tissue depth (78). Accordingly, the images are surface-weighted, meaning that light sources closer to the surface of the animal appear brighter compared to deeper sources due to tissue attenuation properties (71). Another disadvantage is the lack of depth information because bioluminescent images have a planar instead of the tomographic or three-dimensional display typical of X-ray CT or PET. Compared to ﬂuorescence imaging (for example, with green ﬂuorescent protein), bioluminescence imaging can be achieved with remarkably high signal-to-noise in vivo because mammalian tissues do not emit signiﬁcant levels of intrinsic light. However, bioluminescence imaging requires

464

Schuster

that the gene encoding the bioluminescent reporter protein be cloned into expression cassettes in cells or tissues of interest. Although the most commonly used bioluminescent reporter for research purposes has been luciferase from the North American ﬁreﬂy (Photinus pyralis; FLuc), useful luciferases have also been cloned from jellyﬁsh (Aequorea), sea pansy (Renilla; RLuc), corals (Tenilia), and several bacterial species (V. ﬁscheri, V. harveyi) (79). FLuc catalyzes transformation of the substrate D-luciferin into oxyluciferin in an adenosine triphosphate (ATP)–dependent process, leading to the emission of photons, while RLuc catalyzes oxidation of the substrate coelenterazine into coelenteramide in an ATP-independent reaction. Moreover, luciferases and their cognate substrates appear to be nontoxic to mammalian cells (80). The sensitivity of detecting these internal light sources is dependent upon many parameters, including the level of luciferase expression, the depth of labeled cells within the body (i.e., the distance that the photons must travel through tissue), and the sensitivity of the detection system. Key advances in detector technology for imaging low levels of light now enable optical imaging in living animals with charged coupled device cameras. With these devices, fewer than 10,000 cells can be visualized in the lungs after IV injection of labeled cells (81). Gene expression imaging methods with MRI have also been reported (71,82,83). However, this approach is probably too insensitive for lung studies because the proton density of the lung parenchyma is too low to generate a suitable MR image (84). Radiotracer imaging on the other hand, with methods such as PET, has a high degree of sensitivity (level of detection approaches 1011 M of tracer) and isotropism (i.e., the ability to detect expression accurately regardless of tissue depth, unlike light-based techniques that are largely limited to detection at the body surface). Thus, they are ideally suited to detect gene expressions in deep organs such as the lungs. However, PET imaging is complex and requires expensive instrumentation and a team of support staff to generate the radiopharmaceuticals, operate the PET cameras, and to analyze the images. Thus, experimental throughput is much slower than with optical imaging approaches. PET is only one radionuclide-based method currently being evaluated as a platform for gene expression imaging. Others include planar gamma scintigraphy and single-photon emission CT (85,86). The advantages of one imaging platform over another are still theoretical because no direct comparisons have yet been made.

C. Transgene Expression Imaging: Strategies

Technically, imaging can be used to monitor both endogenous genes as well as the transfer of exogenous genes (‘‘transgenes’’). The latter, however, have

Imaging VILI: Present and Future Possibilities

465

received considerably more attention. In both cases, ‘‘direct’’ as well as ‘‘indirect’’ imaging strategies can be used to detect gene expression. Direct approaches depend on probes of various types that either accumulate in tissues by binding directly to the DNA, to the RNA message, or the gene product itself (e.g., when that product is a receptor), or are trapped intracellularly as they are modiﬁed by the gene product (e.g., when that product is an enzyme). With indirect approaches, similar strategies are employed to visualize the expression of a suitable ‘‘reporter’’ gene that is simultaneously linked to the target gene-of-interest by a common promoter. The principal advantage of this strategy is that new speciﬁc probes do not have to be synthesized and evaluated for each new gene-of-interest. On the other hand, care must be taken to ensure that the expression of the reporter gene correlates strongly with the expression of the target gene. Endogenous gene expression can be evaluated at both the transcriptional [messenger RNA (mRNA)] and translational (protein) level. The most direct strategy for monitoring mRNA levels with imaging would be to synthesize a radiolabeled antisense oligodeoxynucleotide (RASON) that is directed against a speciﬁc (usually 10–15 base pair) sequence within the mRNA (87,88). This strategy of ‘‘in vivo hybridization’’ is analogous to the commonly employed microscopic method of ‘‘in situ hybridization.’’ Given current techniques, the chemical synthesis per se of such radiolabeled probes is straightforward, but the production of sufﬁcient quantities for whole animal imaging is still a signiﬁcant challenge. In addition, RASON penetration across cell membranes into the appropriate target cell is often quite limited, and the concentration of mRNA copies within a given cell may be quite low (sometime a few hundred to a few hundred thousand copies). Thus, the imaging signal from RASON binding to target mRNA is usually expected to be weak. Altogether, these restrictions and issues have made it difﬁcult to pursue the RASON approach for gene expression imaging so far. Endogenous gene expression can also be monitored using radiolabeled probes that bind directly to the gene product. This strategy has the advantage of amplifying the imaging signal compared to the RASON approach (many protein molecules can be produced from each mRNA). However, as previously noted, the principal disadvantage is that a new speciﬁc probe would have to be synthesized and evaluated for each new target gene. An alternative indirect approach to monitoring endogenous gene expression would require the development of genetically modiﬁed animals (‘‘transgenics’’) to visualize the expression of a suitable ‘‘reporter’’ transgene that has been introduced into the cell and linked to the target gene by a common promoter. In contrast to monitoring endogenous gene expression, imaging can also be used to monitor the therapeutic or biology-modifying transgenes

466

Schuster

using a reporter gene strategy (89). Here, one need only develop assays to measure the expression of a few reporters that can then be linked using standard molecular biology methods to any number of different target genes. The reporter gene strategy can be used to evaluate gene expression in whole animals as well as in in vitro systems, but traditional methods of assaying reporter gene expression (enzymatic, immunohistochemical, or

Figure 8 Schematic illustrating the principles underlying reporter transgene expression imaging with PET. A vector (e.g., a viral vector) is used to introduce a PET reporter gene into a target cell, such as the lung airway epithelium. The gene is transcribed and translated into a functional protein, such as an enzyme capable of metabolizing a radiolabeled reporter probe. Thus, the probe accumulates only in tissues expressing the reporter gene. An imaging signal can be obtained if sufﬁcient radioactivity accumulates in the target tissue. Abbreviation: PET, positron emission tomography.

Imaging VILI: Present and Future Possibilities

467

in situ hybridization) all require tissue retrieval via biopsy or after sacriﬁcing the animal. Thus, to evaluate the location, magnitude, and/or timing of gene expression with these methods, one must perform assays on multiple individual animals at each time point in each tissue-of-interest. The use of imaging to monitor reporter gene expression can theoretically overcome the disadvantages associated with tissue-based techniques. With this approach, a reporter transgene (‘‘PET reporter gene, PRG’’), capable of trapping or binding a suitable positron-emitting radionuclide-labeled tracer (‘‘PET reporter probe, PRP’’), must be introduced into a target tissue via a suitable vector (Fig. 8). A PET imaging signal is generated as the PRP accumulates over time in tissues expressing the PRG (Fig. 9). A number of different PRG–PRP reporter systems have been described (14,90–92), including

Figure 9 Example of transgene expression imaging of the lung, as a threedimensional view in a rat infected with 1 1011 viral particles of the adenoviral vector of AdCMV-mNLS-sr39tk-egfp. Source: From Ref. 99.

468

Schuster

exogenous enzymes (93,94), membrane-bound receptors (95), and cellmembrane transporters (96). Of these, enzyme-based strategies have the theoretical advantage of additional signal ampliﬁcation because each reporter protein can metabolize (and trap) many molecules of the radioactive probe. In contrast, receptor-based systems, while providing no additional signal ampliﬁcation, do have the advantage of possibly not requiring the rate-limiting intracellular transport of the radioactive probe into the cell interior in order to be trapped as a result of reporter enzyme metabolism. To date, the optimal PRG–PRP system is still a matter of debate, and may be different depending on the tissue or process under study. Even so, the use of PET imaging to monitor exogenous gene transfer in pulmonary tissues by measuring the expression of a suitable enzyme is already proving to be a fruitful experimental strategy (97–99). There is no reason to believe that similar imaging strategies cannot be applied to studies of therapeutic or biology-modifying gene transfer in models of VILI, or even, eventually, in humans enrolled in gene therapy trials during ALI. V. Summary While it is true that no imaging technique will identify changes in lung structure or function that are speciﬁc for VILI, a suite of new imaging platforms and a variety of anatomic, functional, and molecular imaging methods are emerging that can be applied to both experimental models of VILI and human studies of ventilator-associated lung injury. Taking advantage of these new capabilities should allow pulmonary scientists to ‘‘see’’ the lungs in ways that previously could not even be imagined. References 1. Frank J, Imai Y, Slutsky AS. Pathogenesis of ventilator-induced lung injury. In: Matthay M, ed. Acute Respiratory Distress Syndrome. New York: MarcelDekker, 2003:201–244. 2. Schuster DP, Kovacs A, Garbow J, Piwnica-Worms D. Recent advances in imaging the lungs of intact small animals. Am J Respir Cell Mol Biol 2004; 30(2):129–138. 3. Schuster D. Assessment of pulmonary function by PET. In: Bailey D, Townsend D, Valk P, Maisey M, eds. Positron Emission Tomography: Principles and Practice. London: Springer-Verlag, 2003:465–477. 4. Snashall PD, Keyes SJ, Morgan BM, et al. The radiographic detection of acute pulmonary oedema. A comparison of radiographic appearances, densitometry and lung water in dogs. Br J Radiol 1981; 54(640):277–288. 5. Thomason J, Ely EW, Chiles C, Ferretti G, Freimanis R, Haponik EF. Appraising pulmonary edema using supine chest roentgenograms in ventilated patients. Am J Respir Crit Care Med 1998; 157:1600–1608.

Imaging VILI: Present and Future Possibilities

469

6. Rubenfeld GD, Caldwell E, Granton J, Hudson LD, Matthay MA. Interobserver variability in applying a radiographic deﬁnition for ARDS. Chest 1999; 116(5):1347–1353. 7. Meade MO, Cook RJ, Guyatt GH, et al. Interobserver variation in interpreting chest radiographs for the diagnosis of acute respiratory distress syndrome. Am J Respir Crit Care Med 2000; 161(1):85–90. 8. Halperin B, Feeley T, Mihm F, Chiles C, Guthaner D, Blank N. Evaluation of the portable chest roentgenogram for quantitating extravascular lung water in critically ill adults. Chest 1985; 88(5):649–652. 9. Wheeler AP, Carroll FE, Bernard GR. Radiographic issues in adult respiratory distress syndrome. New Horizon 1993; 1(4):471–477. 10. Rouby JJ, Puybasset L, Nieszkowska A, Qin L. Acute respiratory distress syndrome: lessons from computed tomography of the whole lung. Crit Care Med 2003; 31:S285–S295. 11. Gattinoni L, Caironi P, Pelosi P, Goodman LR. What has computed tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001; 164:1701–1711. 12. Forster BB, Muller NL, Mayo JR, Okazawa M, Wiggs BJ, Pare PD. High-resolution computed tomography of experimental hydrostatic pulmonary edema. Chest 1992; 101(5):1434–1437. 13. Schuster DP. Evaluation of pulmonary endothelial barrier function: quantifying pulmonary edema and lung injury. In: Matthay M, Ingbar DH, eds. Pulmonary Edema. New York: Marcel-Dekker, 1998:121–161. 14. Massoud T, Gambhir S. Molecular imaging in living subjects: seeing fundamental biological processes in a new light. Gene Dev 2003; 17:545–580. 15. Paulus M, Gleason S, Kennel S, Hunsicker P, Johnson D. High resolution X-ray computed tomography: an emerging tool for small animal cancer research. Neoplasia 2000; 2:62–70. 16. Ford NL, Thornton MM, Holdsworth DW. Fundamental image quality limits for microcomputed tomography in small animals. Med Phys 2003; 30(11):2869–2877. 17. Mayo JR. Magnetic resonance imaging of the chest. Where we stand. Radiol Clin North Am 1994; 32(4):795–809. 18. Hedlund L, Dewalt S, Cofer G, Johnson G. In: Cutillo A, ed. Application of Magnetic Resonance to the Study of the Lung. Armonk, New York: Futura Press, 1996:401–415. 19. Beckman N, Tigani B, Mazzoni L, Fozard J. MRI of lung parenchyma in rats and mice using a gradient-echo sequence. NMR Biomed 2001; 14:297–306. 20. Hedlund L, Johnson G. Mechanical ventilation for imaging the small animal lung. ILAR 2002; 43:159–174. 21. Garbow J, Dugas J, Conradi M. Respiratory gating for MRI and MRS in rodents. IEEE Symp Bioinform Bioeng 2003; 3:126–129. 22. Cutillo A, Goodrich K, Krishnamurthy G, et al. Lung water measurement by nuclear magnetic resonance: correlation with morphometry. J Appl Physiol 1995; 79(6):2163–2168. 23. Cutillo AG, Morris AH, Ailion DC, et al. Assessment of lung water distribution by nuclear magnetic resonance. A new method for quantifying and monitoring experimental lung injury. Am Rev Respir Dis 1988; 137(6):1371–1378.

470

Schuster

24. Cutillo AG, Morris AH, Ailion DC, Durney CH, Case TA. Determination of lung water content and distribution by nuclear magnetic resonance imaging. J Thorac Imaging 1986; 1(3):39–51. 25. Cutillo AG, Morris AH, Aillon DC, Durney CH, Ganesan K. Quantitative assessment of pulmonary edema by nuclear magnetic resonance methods. J Thorac Imaging 1988; 3(3):51–58. 26. Cutillo AG, Morris AH, Blatter DD, et al. Determination of lung water content and distribution by nuclear magnetic resonance. J Appl Physiol Respir Environ Exer Physiol 1984; 57(2):583–588. 27. Mayo JR, MacKay AL, Whittall KP, Baile EM, Pare PD. Measurement of lung water content and pleural pressure gradient with magnetic resonance imaging. J Thorac Imaging 1995; 10(1):73–81. 28. Morris AH, Blatter DD, Case TA, et al. A new nuclear magnetic resonance property of lung. J Appl Physiol 1985; 58(3):759–762. 29. Phillips DM, Allen PS, Man SF. Assessment of temporal changes in pulmonary edema with NMR imaging. J Appl Physiol 1989; 66(3):1197–1208. 30. Rhodes CG. Measurement of lung water using nuclear magnetic resonance imaging [letter]. Br J Radiol 1986; 59(707):1135–1136. 31. Schuster DP, Mintun MA, Green MA, Ter-Pogossian MM. Regional lung water and hematocrit determined by positron emission tomography. J Appl Physiol 1985; 59(3):860–868. 32. Schuster DP, Marklin GF, Mintun MA. Regional changes in extravascular lung water detected by positron emission tomography. J Appl Physiol 1986; 60:1170–1178. 33. Velazquez M, Haller J, Amundsen T, Schuster DP. Regional lung water measurements with PET: accuracy, reproducibility, and linearity. J Nucl Med 1991; 32(4):719–725. 34. Chatziioannou A. Molecular imaging of small animals with dedicated PET tomographs. Eur J Nucl Med 2002; 29:98–114. 35. Laforest R, Rowland DJ, Welch MJ. MicroPET imaging with nonconventional isotopes. IEEE Trans Nucl Sci 2002; 49(5):2119–2126. 36. Chen DL, Schuster DP. Functional imaging of rodent lungs. In: Schuster DP, Blackwell TR, eds. Molecular Imaging of the Lungs. New York: MarcelDekker, 2005. 37. Wiegman EM, Meertens H, Konings AW, Kampinga HH, Coppes RP. Locoregional differences in pulmonary function and density after partial rat lung irradiation. Radiother Oncol 2003; 69(1):11–19. 38. Vujaskovic Z, Down JD, van t’Veld AA, et al. Radiological and functional assessment of radiation-induced lung injury in the rat. Exp Lung Res 1998; 24(2):137–148. 39. Mitzner W, Brown R, Lee W. In vivo measurement of lung volumes in mice. Physiol Genomics 2001; 4(3):215–221. 40. Schuster DP. What is acute lung injury? What is ARDS? Chest 1995; 107(6):1721–1726. 41. Effros RM, Mason GR, Mena I. 99mTc-DTPA aerosol deposition and clearance in COPD, interstitial disease, and smokers. J Thorac Imaging 1986; 1(2):54–60.

Imaging VILI: Present and Future Possibilities

471

42. Schuster DP. The evaluation of lung function with PET. Sem Nucl Med 1998; 28(4):341–351. 43. Schuster DP. Measuring pulmonary vascular permeability. J Nucl Med 1995; 36(8):1442–1444. 44. Roselli RJ, Coy SR, Harris TR. Models of lung transvascular ﬂuid and protein transport. Ann Biomed Eng 1987; 15(2):127–138. 45. Roselli RJ, Riddle WR. Analysis of noninvasive macromolecular transport measurements in the lung. J Appl Physiol 1989; 67(6):2343–2350. 46. Roselli R, Harris T. Lung ﬂuid and macromolecular transport. In: Chang HK, Paiva M, eds. Respiratory Physiology: An Analytical Approach. New York: Marcel Dekker, 1989:633. 47. Mintun MA, Dennis DR, Welch MJ, Mathias CJ, Schuster DP. Measurements of pulmonary vascular permeability with PET and gallium-68 transferrin. J Nucl Med 1987; 28(11):1704–1716. 48. Abernathy VJ, Pou NA, Wilson TL, Roselli RJ. Noninvasive measurements of albumin ﬂux into lung interstitium with increased microvascular pressure. Am J Physiol 1995; 269(1 Pt 2):H288–H296. 49. Schuster DP, Stark T, Stephenson J, Royal H. Detecting lung injury in patients with pulmonary edema. Intensive Care Med 2002; 28(9):1246–1253. 50. Viallon M, Berthezene Y, Decorps M, et al. Laser-polarized (3)He as a probe for dynamic regional measurements of lung perfusion and ventilation using magnetic resonance imaging. Magn Reson Med 2000; 44(1):1–4. 51. Chen BT, Brau AC, Johnson GA. Measurement of regional lung function in rats using hyperpolarized 3helium dynamic MRI. Magn Reson Med 2003; 49(1):78–88. 52. Dupuich D, Berthezene Y, Clouet PL, Stupar V, Canet E, Cremillieux Y. Dynamic 3He imaging for quantiﬁcation of regional lung ventilation parameters. Magn Reson Med 2003; 50(4):777–783. 53. Cremillieux Y, Berthezene Y, Humblot H, et al. A combined 1H perfusion/3He ventilation NMR study in rat lungs. Magn Reson Med 1999; 41(4):645–648. 54. Stupar V, Berthezene Y, Canet E, Tournier H, Dupuich D, Cremillieux Y. Helium3 polarization using spin exchange technique: application to simultaneous pulmonary ventilation/perfusion imaging in small animals. Invest Radiol 2003; 38(6):334–340. 55. Vidal Melo MF, Layﬁeld D, Harris RS, et al. Quantiﬁcation of regional ventilation-perfusion ratios with PET. J Nucl Med 2003; 44(12):1982–1991. 56. Kreck TC, Krueger MA, Altemeier WA, et al. Determination of regional ventilation and perfusion in the lung using xenon and computed tomography. J Appl Physiol 2001; 91(4):1741–1749. 57. Jones J, Mendes L, Rudd A, Russo G, Loscalco J, Zhang Y. Serial noninvasive assessment of progressive pulmonary hypertension in a rat model. Am J Physiol 2002; 283:H364–H371. 58. Wiesmann F, Frydrychowicz A, Rautenberg J, et al. Analysis of right ventricular function in healthy mice and a murine model of heart failure by in vivo MRI. Am J Physiol 2002; 283:H1065–H1071. 59. Jones H. Inﬂammation imaging of the lungs. In: Schuster DP, Blackwell TR, eds. Molecular Imaging of the Lungs. New York: Marcel-Dekker, 2005.

472

Schuster

60. Pauwels E, Sturm E, Bombardieri E, Cleton F, Stokkel M. Positron-emission tomography with [18F]ﬂuorodeoxyglucose. J Cancer Res Clin Oncol 2000; 126:549–559. 61. Chen DL, Schuster DP. Positron emission tomography with [18F]ﬂuorodeoxyglucose to evaluate neutrophil kinetics during acute lung injury. Am J Physiol Lung Cell Mol Physiol 2004; 286(4):L834–L840. 62. Rhodes CG, Hughes JMB. Pulmonary studies using positron emission tomography. Eur Respir J 1995; 8:1001–1017. 63. Jones H, Sriskandan S, Peters A, et al. Dissociation of neutrophil emigration and metabolic activity in lobar pneumonia and branchiectasis. Eur Respir J 1997; 10(4):795–803. 64. Jones HA, Schoﬁeld JB, Krausz T, Boobis AR, Haslett C. Pulmonary ﬁbrosis correlates with duration of tissue neutrophil activation. Am J Respir Crit Care Med 1998; 158(2):620–628. 65. Jones HA, Clark RJ, Rhodes CG, Schoﬁeld JB, Krausz T, Haslett C. In vivo measurement of neutrophil acitvity in experimental lung inﬂammation. Am J Respir Crit Care Med 1994; 149:1635–1639. 66. Kapucu L, Meltzer C, Townsend DV, Keenan R, Luketich J. Fluorine-18-ﬂuorodeoxyglucose uptake in pneumonia. J Nucl Med 1998; 39:1267–1269. 67. Rennen H, Boerman O, Oyen W, Corstens F. Imaging infection/inﬂammation in the new millennium. Eur J Nucl Med 2001; 28:241–252. 68. Blasberg R. Imaging gene expression and endogenous molecular processes: molecular imaging. J Cereb Blood Flow Metab 2002; 22(10):1157–1164. 69. Blasberg R. PET imaging of gene expression. Eur J Cancer 2002; 38(16):2137– 2146. 70. Biasberg RG, Gelovani-Tjuvajev J. In vivo molecular-genetic imaging. J Cell Biochem Suppl 2002; 39:172–183. 71. Weissleder R, Mahmood U. Molecular imaging. Radiology 2001; 219(2): 316–333. 72. Contag C, Bachmann M. Advances in in vivo bioluminescence imaging of gene expression. Ann Rev Biomed Eng 2002; 4:235–260. 73. Weissleder R. Scaling down imaging: molecular mapping of cancer in mice. Nat Rev Cancer 2002; 2:11–18. 74. Piwnica-Worms D, Schuster DP, Garbow JR. Molecular imaging of hostpathogen interactions in intact small animals. Cell Microbiol 2004; 6(4):319– 331. 75. Contag C, Spilman S, Contag P. Visualizing gene expression in living mammals using a bioluminescent reporter. Photochem Photobiol 1997; 66: 523–531. 76. Lipshutz G, Gruber C, Cao Y, Hardy J, Contag C, Gaensler K. In utero delivery of adeno-associated viral vectors: intraperitoneal gene transfer produced long-term expression. Mol Ther 2001; 3:284–292. 77. Zhang W, Feng J, Harri C, Contag P, Stevenson D, Contag C. Rapid in vivo functional analysis of transgenes in mice using whole body imaging of luciferase expression. Transgenic Res 2001; 10:423–434. 78. Contag C, Contag P, Muliins J, Spilman S, Stevenson D, Benaron D. Photonic detection of bacterial pathogens in living hosts. Mol Microbiol 1995; 18:593–603.

Imaging VILI: Present and Future Possibilities

473

79. Hastings J. Chemistries and colors of bioluminescent reactions: a review. Gene 1996; 173:5–11. 80. Sweeney T, Mailander V, Tucker A. Visualizing the kinetics of tumorcell clearance in living animals. Proc Natl Acad Sci USA 1999; 96:12044– 12049. 81. Edinger M, Cao Y, Hornig Y. Advancing animal models of neoplasia through in vivo bioluminescence imaging. Eur J Cancer 2002; 38:2128–2136. 82. Bogdanov A, Weissleder R. In vivo imaging of gene delivery and expression. Trends Biotechnol 2002; 20(8):S11–S18. 83. Moore A, Josephson L, Bhorade RM, Basilion JP, Weissieder R. Human transferrin receptor gene as a marker gene for MR imaging. Radiology 2001; 221(1):244–250. 84. Weissleder R, Moore A, Mahmood U, et al. In vivo magnetic resonance imaging of transgene expression. Nat Med 2000; 6(3):351–355. 85. Reynolds P, Zinn K, Gavrilyuk V, et al. A targetable injectable adenoviral vector for selective gene delivery to pulmonary endothelium in vivo. Mol Ther 2001:1–17. 86. Zinn K, Chaudhuri T. The type 2 human somatostatin receptor as a platform for reporter gene imaging. Eur J Nucl Med 2002; 29:388–399. 87. Piwnica-Worms D. Making sense out of anti-sense: challenges of imaging gene translation with radiolabeled oligonucleotides. J Nucl Med 1994; 35:1064– 1065. 88. Gambhir S, Barrio J, Herschman H, Phelps M. Imaging gene expression: principles and assays. J Nucl Cardiol 1999; 6:219–233. 89. Naylor L. Reporter gene technology: the future looks bright. Biochem Pharmacol 1999; 58:749–757. 90. MacLaren DC, Toyokuni T, Cherry SR, et al. PET imaging of transgene expression. Biol Psychiat 2000; 48(5):337–348. 91. Gambhir SS, Bauer E, Black ME, et al. A mutant herpes simplex virus type 1 thymidine kinase reporter gene shows improved sensitivity for imaging reporter gene expression with positron emission tomography. Proc Natl Acad Sci USA 2000; 97(6):2785–2790. 92. Blasberg RG, Tjuvajev JG. Molecular-genetic imaging: current and future perspectives. J Clin Invest 2003; 111(11):1620–1629. 93. Haberkorn U, Oberdorfer F, Gebert J, et al. Monitoring gene therapy with cytosine deaminase: in vitro studies using tritiated-5-ﬂuorocytosine. J Nucl Med 1996; 37(1):87–94. 94. Gambhir SS, Barrio JR, Wu L, et al. Imaging of adenoviral-directed herpes simplex virus type 1 thymidine kinase reporter gene expression in mice with radiolabeled ganciclovir. J Nucl Med 1998; 39(11):2003–2011. 95. MacLaren DC, Gambhir SS, Satyamurthy N, et al. Repetitive, non-invasive imaging of the dopamine D2 receptor as a reporter gene in living animals. Gene Ther 1999; 6(5):785–791. 96. Groot-Wassink T, Aboagye EO, Glaser M, Lemoine NR, Vassaux G. Adenovirus biodistribution and noninvasive imaging of gene expression in vivo by positron emission tomography using human sodium/iodide symporter as reporter gene. Hum Gene Ther 2002; 13:1723–1735.

474

Schuster

97. Richard JC, Factor P, Welch LC, Schuster DP. Imaging the spatial distribution of transgene expression in the lungs with positron emission tomography. Gene Ther 2003; 10(25):2074–2080. 98. Richard J, Factor P, Ferkol T, Ponde D, Zhou Z, Schuster D. Repetitive imaging of reporter gene expression in the lungs. Mol Imaging 2003; 2:1–8. 99. Richard JC, Zhou Z, Ponde DE, et al. Imaging pulmonary gene expression with positron emission tomography. Am J Respir Crit Care Med 2003; 167:1257–1263.

19 Modulation of the Cytokine Network by Lung-Protective Mechanical Ventilation Strategies

VITO FANELLI, KAREN J. BOSMA, and V. MARCO RANIERI Dipartimento di Anestesiologia e Rianimazione, Ospedale S. Giovanni Battista-Molinette, Universita´ di Torino Torino, Italy

ARTHUR S. SLUTSKY Interdepartmental Division of Critical Care Medicine and Division of Respirology, Department of Medicine, University of Toronto, and Department of Critical Care Medicine, St. Michael’s Hospital Toronto, Ontario, Canada

I. Introduction The acute respiratory distress syndrome (ARDS) is associated with a hospital mortality rate of 25% to 31% in recent large clinical trials (1,2), and 32% to 46% in the latest observational epidemiologic studies (3,4). Comparison to historical controls is often problematic, but it certainly appears that the mortality rate for ARDS has decreased over the past 15 to 20 years from rates of approximately 60%. Abel et al. examined the change in mortality from ARDS in a single center, which used a standard deﬁnition of ARDS, and found that mortality decreased from 66% to 34% during the last decade (5). The most likely explanation for the reduction in ARDS mortality is the use of lung-protective ventilatory strategies with smaller tidal volumes (VT) and the general enhanced quality of standard care in the intensive care unit. That a reduction in VT could improve survival is somewhat surprising, because the most common cause of death in ARDS is not pulmonary failure per se, but rather the multiple organ dysfunction syndrome (MODS) (6). Once MODS develops, it is often irreversible and has an associated 475

476

Fanelli et al.

mortality rate of 60% to 98%, when three or more organs are involved for a period of more than seven days (7,8). Thus, improving the survival in ARDS must be linked to decreasing the incidence or severity of MODS. The ARDS network trial demonstrated that patients treated with lower VTs had improved survival and fewer days of nonpulmonary organ or system failure than patients treated with traditional VTs (1). There is increasing evidence that the link between ARDS and MODS is mechanical ventilation (MV) and its impact on the local and systemic mediator network. In this chapter we will (i) describe the cytokine network activated by mechanical stretch; (ii) discuss the mechanisms by which injurious MV can induce an inﬂammatory response; (iii) provide evidence that a lung-protective strategy can attenuate the inﬂammatory response and its impact on organ function; and (iv) examine evidence addressing whether the beneﬁts of lung-protective ventilation extend beyond patients with ARDS.

II. MV and the Cytokine Network Under most circumstances, inﬂammation is self-limiting and beneﬁcial. However, if there is a perpetuating injurious stimulus, the inﬂammatory response may spiral out of control, resulting in harm. When generalized activation of the immune system occurs, it may lead to diffuse epithelial and endothelial injury and result in MODS. ARDS often occurs in the setting of systemic inﬂammation. However, MV can itself produce an increase in circulating inﬂammatory mediators, and this is termed ‘‘ventilatorassociated systemic inﬂammation’’ (9). The principal feature of systemic inﬂammation is the interplay between anti-inﬂammatory mediators and proinﬂammatory mediators, and their effects on endothelial cells, epithelial cells, and leukocytes. An important group of such mediators are the cytokines: low-molecular-weight glycoproteins that are produced by a variety of cell types and act as messengers to cell-surface receptors to promote or diminish the inﬂammatory cascade. Prototypic cytokines include interleukin (IL)-1b and tumor necrosis factor (TNF)-a, which promote inﬂammation in the early phase of injury; IL-8 is a chemokine responsible for the recruitment of immune cells and IL-6 activates immune cells such as macrophages and polymorphonuclear cells. Other cytokines have anti-inﬂammatory roles, such as IL-10, or the naturally occurring anticytokine, IL-1 receptor antagonist (IL-1ra). A detailed description of cytokine biology is beyond the scope of this chapter. The reader is referred to Chapters 8 and 9 and several excellent reviews (10,11). Macrophages are one of the major sources of cytokines, but other cells such as mast cells and endothelial cells can also contribute to cytokine

Modulation of the Cytokine Network

477

release. When a noxious or injurious stimulus is present, many host defense systems may be activated, leading to release of preformed or newly generated factors which then modulate the production of cytokines. Once generated, cytokines are able to perpetuate their own production. For example, activation of the coagulation cascade and complement system leads to the appearance of factors such as thrombin, factor Xa, and anaphylatoxins, which in turn activate macrophages to release the proinﬂammatory cytokines IL-1 and TNF-a. IL-I and TNF-a then target endothelial cells, which respond by releasing free radicals, platelet activating factor, and cytokines including IL-6 and IL-8, and by synthesizing cyclo-oxygenase and phospholipase, which lead to the production of prostaglandins. In addition, endothelial cells express surface adhesion molecules that are capable of binding circulating leukocytes. Once leukocytes have adhered to the endothelial cell surface, they migrate in response to chemoattractant signals from chemokines such as IL-8 (Fig. 1). A more indepth review of this topic is provided in Chapter 9. MV can be the perpetuating noxious stimulus that leads to imbalance between the proinﬂammatory and anti-inﬂammatory cytokines. MV can promote the synthesis and secretion of cytokines from cells in the lungs by a number of mechanisms. First, cells may respond to the forces generated by MV in a process known as mechanotransduction, that is, the conversion of physical forces on the cell membrane/receptors into activation of intracellular signal transduction pathways resulting in increased production and release of cytokines (12,13). Second, excessive stretch may result in cell necrosis, with the release of preformed cytokines and stimulants for the production of more cytokines (12,14). Third, MV may increase pressure in the pulmonary microvasculature, resulting in an inﬂammatory response from the vascular endothelial cells (12). These mechanisms may lead to a local inﬂammatory response. This localized response can translate into a systemic response when decompartmentalization of local mediators occurs, that is, when stress failure and necrosis of the endothelial–epithelial barrier allows cytokines, bacteria, and endotoxins to quickly spread from the alveoli into the systemic circulation (15–17). This ‘‘spillover’’ of cytokines from the pulmonary to the systemic compartment may contribute to MODS. This topic is further discussed in Chapters 9 and 10. The cellular stretch and shear stress described above occurs when high VTs and inadequate or zero positive end-expiratory pressure (PEEP) are used to ventilate patients with ARDS. Patients with ARDS have a reduced functional residual capacity due to atelectasis of the dependent lung regions and alveolar edema, and so only a portion of the lung volume is available for ventilation. In the presence of this inhomogeneity in lung aeration, alveolar overdistension (volutrauma) can occur with traditional VTs of 10 to 12 mL/kg, especially with plateau pressures above 30 cmH2O

478

Fanelli et al.

Figure 1 The cytokine network. Cytokine production during an inﬂammatory process. See text for details. Abbreviations: MCP-1, monocyte chemotactic protein 1; CSFs, colony stimulating factors; DIC, disseminated intravascular coagulation.

(1,18–20). Similarly, the absence of PEEP is associated with the cyclic opening and closing of collapsing alveoli (atelectrauma), exposing vulnerable lung units to high shear stresses (21). In summary, MV using high VTs and low PEEP, the so-called ‘‘injurious MV,’’ can induce an inﬂammatory response in patients with ARDS, both locally in the lungs and systemically, affecting various end organs. This evidence then begs the question: Is all MV harmful? Clearly, high VTs can be harmful, but do low VTs also incite an inﬂammatory response, just to a

Modulation of the Cytokine Network

479

lesser degree? If this is so, then ‘‘protective’’ MV is really a misnomer. Must we simply accept that all MV exacerbates lung injury, and resign ourselves to ﬁnding the least harmful strategy, or can MV be noninjurious or neutral in the previously inﬂamed lung? The conceptual framework for ‘‘protective ventilation’’ in ARDS is based on using lower VTs to avoid volutrauma and adequate levels of PEEP to avoid atelectrauma. But, does the use of lower VTs and higher PEEP actually protect the patient with ARDS from a local and systemic inﬂammatory response? To answer this question, we must examine studies analyzing the effects of the so-called ‘‘protective MV strategies’’ on the cytokine system in the setting of acute lung injury (ALI).

III. Modulation of the Cytokine Network in ALI: Evidence from Studies A study by Ranieri et al. (22) examined the effects of a lung-protective strategy (treatment group) versus a conventional strategy of ventilation (the control group) on inﬂammatory mediators measured locally and systemically in patients with ARDS. Patients with ARDS who had been receiving MV for less than eight hours were enrolled; 19 patients were randomized and completed the protocol in the control group and 18 in the treatment group. In both groups, the mode of ventilation, respiratory rate, and inspiratory to expiratory (I:E) ratios were equivalent (controlled mode, rate of 10– 15/min, I:E ratio of 1:2). The difference between the two groups involved the method of setting the VT and the PEEP level. In patients in the treatment group, PEEP and VT were set based on the pressure volume curve. This resulted in the control group being ventilated with a mean VT of 11.1 mL/ kg and a PEEP of 6.5 cmH2O, and the treatment group with a mean VT of 7.6 mL/kg and a PEEP of 14.8 cmH2O. Both groups had the fraction of inspired oxygen (FiO2) set based on identical oxygenation saturation criteria; the control group required an FiO2 of 0.9 versus 0.7 in the lung-protective group. Cytokines were measured in the bronchoalveolar lavage (BAL) ﬂuid at study entry, time 1 (24 to 30 hours after study entry), and time 2 (36 to 40 hours after study entry). Cytokine levels were initially high and comparable in both groups at study entry. In the control arm, BAL levels of IL-b, IL-6, and TNF-a increased signiﬁcantly after 36 hours of ventilation, as did plasma levels of IL-6 and TNF-a. TNF-a receptors (TNF-asR55 and TNF-asR75) also increased in both BAL and plasma, and IL-1ra increased in BAL. In the lung-protective strategy arm, however, BAL levels of IL-1b, IL-6, IL-8, TNF-a, TNF-asR55, TNF-asR75, and IL-1ra decreased over time, as did plasma levels of IL-6, TNF-asR75, and IL-1ra. Plasma TNF-a did not change signiﬁcantly (Fig. 2). This study has two important ﬁndings. First, it showed that MV per se is an important factor in determining the pulmonary and systemic cytokine

480

Fanelli et al.

Figure 2 Levels of inﬂammatory mediators in the BAL ﬂuid and plasma for patients in the control and lung-protective strategy groups. Individual trends of TNF-a, IL-8, and IL-6 in BAL and plasma for patients in the control group and the lung-protective strategy group in the study by Ranieri et al. (18). Time 1 indicates 24 to 30 hours after study entry and time 2 indicates 36 to 40 hours after study entry. Horizontal bars indicate mean values. N is the number of patients for whom measurements were available. P values are for repeated measures analysis of variance for time 2 versus entry. Note that levels of cytokines in the lung-protective strategy group are lower relative to baseline and relative to the control group at time 2. Abbreviations: BAL, bronchoalveolar lavage; IL, interleukin; TNF, tumor necrosis factor. Source: From Ref. 18.

response in patients with ARDS. Second, it showed that the ‘‘lungprotective strategy’’ used in this study was, in fact, protective against ventilator-induced local and systemic inﬂammation. Not only did the protective strategy incite less inﬂammatory response than the conventional strategy, but also BAL and plasma cytokine levels were either lower or equivalent after 36 hours of ventilation than at study entry. Thus, the local pulmonary and systemic inﬂammation present at baseline in these patients with ARDS was partially ameliorated by ventilating the lungs with low VTs and high PEEP (in this case, VT of 7.6 mL/kg and PEEP of 14.8 cmH2O).

Modulation of the Cytokine Network

481

The patients in this study were treated with the lung-protective strategy soon after initiating MV (within eight hours), and the effect on cytokine levels was seen 24 to 36 hours later. A study by Stu¨ber et al. (23) examined the early inﬂammatory response incited by conventional MV and the reversibility of this inﬂammatory response with protective lung ventilation. Twelve patients with ALI who had been receiving MV for at least 24 hours were enrolled. They received a six-hour ‘‘equilibration period’’ of lungprotective MV, after which baseline levels of cytokines were measured in the BAL of six patients, and in the plasma of all 12 patients. All patients were then switched to a conventional strategy with VT of 12 mL/kg and PEEP of 5 cmH2O for six hours, and then switched back to protective MV with VT of 5 mL/kg and PEEP of 15 cmH2O. Repeat cytokine levels measurements were made in the plasma at one and six hours after conventional MV, and at one and six hours after changing back to protective MV, and in BAL in six patients at six hours after initiating each ventilatory strategy. They found that all cytokines measured in BAL (TNF, IL-6, IL-1b, IL-10, and IL-1ra) increased after six hours of conventional MV, and increased further after six hours of protective MV. However, in plasma, TNF-a, IL-6, IL-10, and IL-1ra were moderately elevated at baseline, increased within one hour of starting conventional MV, remained high after six hours of MV, and decreased to values comparable to the baseline levels within six hours after switching back to the lung-protective strategy (Fig. 3). This study by Stu¨ber et al. demonstrates that a conventional strategy of high VT and low PEEP induces the release of pro- and anti-inﬂammatory cytokines into the alveolar space and blood after only one hour, and a lungprotective strategy of low VT and high PEEP reverses the systemic response within six hours, but does not impact the local pulmonary response. The authors hypothesized that after alveolar epithelial cells are damaged or macrophages activated by injurious MV, the release of cytokines may be ongoing, even after changing to a noninjurious strategy. However, no further measurements were made after six hours to determine if the local inﬂammatory response would attenuate after a longer duration of lung-protective MV. In the study by Ranieri et al. previously described (22), after 24 to 30 hours of protective MV, BAL cytokine levels were starting to decline, but it was only after 36 to 40 hours of protective MV that the decrease in BAL cytokine levels was signiﬁcant. No measurements were made prior to 24 hours to determine the early course of cytokine levels with a protective ventilation strategy. Thus, the ﬁndings of these two studies regarding the local pulmonary response to protective MV are not mutually exclusive; it may be that attenuation of the local pulmonary inﬂammatory response takes longer than 24 hours, and may in part be dependent on the degree of lung injury and severity of inﬂammation at baseline. The most intriguing ﬁnding in the study by Stu¨ber et al. is that even while intra-alveolar cytokines were increasing, plasma levels of cytokines

482

Figure 3

Fanelli et al.

(Caption on facing page)

Modulation of the Cytokine Network

483

had signiﬁcantly decreased within six hours of initiating low VT and high PEEP. Although this study did not attempt to identify the distinct cellular sources of the cytokines found in the plasma, previous studies have attributed the systemic inﬂammatory response seen with injurious MV to the release of cytokines from the lung (24), with subsequent transfer of cytokines into the blood (25–27). The concept of decompartmentalization, that disruption of the alveolar epithelial–endothelial barrier may lead to translocation of cytokines from the alveolar compartment to the systemic circulation, is well supported by direct evidence from ex vivo studies (28), in vivo animal studies (27,29), and indirect evidence from human studies (30). Stu¨ber et al. hypothesized that decompartmentalization may have occurred in their patients with ALI during the six hours of ventilation with high VT and low PEEP, resulting in increased plasma levels of cytokines. Because plasma cytokines fell to baseline levels within six hours of switching to a protective strategy, even though intra-alveolar cytokines were continually released, one could also hypothesize that low VT and high PEEP prevented the spillover of cytokines from the alveoli into the blood. Although they did not attempt to prove the mechanism, certainly they demonstrated an immediate reversal of the systemic inﬂammation induced by high VT and low PEEP upon institution of a lung-protective strategy. If MV strategies designed to minimize overdistension and atelectrauma can in fact prevent decompartmentalization of cytokines, then this ﬁnding has important clinical implications. There is a growing body of evidence supporting the concept that inﬂammatory mediators entering

Figure 3 (Facing page) Plasma and BAL levels of cytokines for patients ventilated consecutively with a conventional strategy and lung-protective strategy. (A) Plasma levels of IL-6, TNF, IL-1b, IL-10, and IL-1ra measured in patients who also underwent mini-BALs; (B) Plasma levels of IL-6, TNF, IL-1b IL-10, and IL-1ra measured in patients who did not have mini-BALs performed; (C) BAL levels of IL-6, TNF, IL-1b, IL-10, and IL-lra. Prior to time 0, patients were ventilated with a lung-protective strategy (VT 5 mL/kg PBW and PEEP 15 cmH2O); from time 0 to 6 hours, patients were ventilated with a low PEEP and high VT (VT 12 mL/kg PBW and PEEP of 5 cmH2O), and from time 6 to 12 hours, patients were switched back to the lungprotective strategy. Note that serum levels of cytokines did not differ between patients with and without mini-BAL, indicating that the BAL itself did not cause a signiﬁcant systemic inﬂammatory response. Also, while BAL levels of cytokines continued to increase after conventional MV, the systemic response was attenuated by revision to the lung-protective strategy. Key: , P < 0.05 for all cytokines when compared to baseline cytokine levels at zero hour, except plasma IL-1b (Friedman test and post hoc analysis). Abbreviations: BAL, bronchoalveolar lavage; IL, interleukin; TNF, tumor necrosis factor; IL-1ra, IL-1 receptor antagonist; MV, mechanical ventilation; PBW, predicted body weight; PEEP, positive end-expiratory pressure; VT, tidal volume. Source: From Ref. 19.

484

Fanelli et al.

the blood from the pulmonary system may induce dysfunction and failure in distal organs. Imai et al. (31) investigated the effects of injurious and protective MV on end-organ epithelial cell apoptosis and organ dysfunction in a rabbit model of ARDS. Following acid aspiration, 12 rabbits in the ‘‘injurious’’ group were ventilated with high VTs (15–17 mL/kg) and low PEEP (0–3 cmH2O), and 12 rabbits in the ‘‘noninjurious’’ group were ventilated with low VT (5–7 mL/kg) and high PEEP (9–12 cmH2O). Chemokines, namely monocyte chemotactic protein 1, IL-8, and growth-related oncogene were measured in plasma at baseline, four hours, and eight hours, and in the pulmonary aspirate at eight hours. Markers of liver and renal function, namely aspartate aminotransferase (AST), alanine aminotransferase (ALT), lactate dehydrogenase (LDH), creatinine, and urea nitrogen were also measured in plasma at baseline, four, and eight hours. After eight hours of ventilation, animals were sacriﬁced and tissue fragments from lung, liver, kidney, and small intestine were examined for apoptosis by ﬂuorescent staining, and for apoptosis and necrosis by electron microscopy (EM). Although plasma levels of chemokines increased in both the injurious and noninjurious groups compared to baseline, levels in the injurious group were approximately double that in the noninjurious group (p < 0.01 for all) at four and eight hours. Pulmonary aspirate chemokine levels had no baseline or four-hour measurement for comparison, but at eight hours were signiﬁcantly higher in the injurious than noninjurious group. Correlating with the increase in plasma cytokines, plasma levels of AST, LDH, and creatinine increased over time, but were signiﬁcantly higher in the injurious than the noninjurious group at eight hours. Levels of ALT and urea nitrogen did not change appreciably from baseline and did not differ between the two groups. The injurious ventilatory strategy led to increased epithelial cell apoptosis in the kidney and in the villi of the small intestine compared to the noninjurious strategy, as assessed by both staining and EM (Fig. 4). No differences in the apoptotic index were found in the liver and crypts of the small intestine between ventilatory strategies. In addition, no necrotic changes were found in the kidney or small intestine of either group. In the lung, the ﬁndings for apoptosis were reversed: there were more apoptotic cells in the lung in the noninjurious group, detected by both staining and EM. However, EM revealed that although there was less apoptosis, there was increased necrosis of type II epithelial cells in the injuriously ventilated group, suggesting more severe lung damage, likely related to the combined injury due to the acid aspiration and the injurious ventilatory strategy. In a second part of the study, epithelial cells from the kidneys of healthy rabbits were incubated with plasma from rabbits in both the injurious and noninjurious ventilation groups and examined 12 hours later for apoptosis and necrosis. There was increased apoptosis in cells incubated with plasma obtained from rabbits in the injurious group compared

Modulation of the Cytokine Network

485

Figure 4 Apoptotic index percentages in the lung, kidney, and small intestine after exposure to plasma from rabbits ventilated with noninjurious and injurious strategies. The injurious ventilatory strategy led to increased epithelial cell apoptosis in the kidney (injurious 10.9 0.88% vs. noninjurious 1.86 þ 0.17%; p < 0.001) and small intestine villi (injurious 6.7 0.66% vs. noninjurious 0.97 0.14%; p < 0.001). Values are mean apoptotic index percentage standard deviation, where apoptotic index is the number of apoptotic cells relative to the total number of cells, expressed as a percentage. Note that although the apoptotic index for lung tissue was greater in the noninjurious group, cell necrosis was greater in the injurious group (data not shown). Source: From Ref. 25.

to the noninjurious group. Thus, the study by Imai et al. demonstrated that injurious ventilation can induce apoptosis in distal organs, and suggested that increased plasma mediator levels might be one factor mediating this effect. Furthermore, this study showed that increased apoptosis in end organs correlated with the elevation of biochemical markers, indicating organ dysfunction. It is not known whether apoptosis is the cause or the consequence of cell injury, but certainly strong correlations exist between injurious ventilation, apoptosis, and end-organ damage, indicating a common pathway. Most importantly, protective MV was associated with much lower levels of plasma cytokines, very little apoptosis, and only minimal

486

Fanelli et al.

changes in biochemical markers, suggesting that protective MV can in fact protect distal organs from ventilator-induced end organ dysfunction. The authors suggested that this mechanism might explain the decrease in mortality observed in the ARDS Network trial of low VT ventilation. Further evidence that modulation of the systemic cytokine network by lung-protective strategies may attenuate multiorgan dysfunction comes from a small, post hoc analysis, and a large clinical trial by the ARDS Network. Ranieri et al. (30) performed a post hoc, retrospective analysis of data from their randomized controlled trial of 44 patients with ARDS, looking for a relationship between cytokine concentrations and organ failure. They found that multisystem organ failure scores increased signiﬁcantly 72 hours after admission in patients treated with conventional MV, whereas the number of patients with organ failure was equal or lower in the group randomized to a lung-protective strategy. Furthermore, changes in the overall multisystem organ failure score correlated signiﬁcantly with changes in plasma concentrations of IL-6, TNF-a, IL-1b, and IL-8 (Fig. 5). Corroborating this ﬁnding, the large clinical trial by the ARDS Network, described below, also demonstrated a correlation between protective MV, lower cytokine levels, and fewer days of organ failure (1). The studies by Ranieri et al. (22), Stu¨ber et al. (23), and Imai et al. (31) described earlier combined low VT with high PEEP in their protective ventilatory strategies, and thus it is difﬁcult to ascertain the individual beneﬁts derived from each. Two large clinical trials conducted by the ARDS Network separately examined the effects of low VT and high PEEP on mortality in patients with ARDS, and included an analysis of cytokine levels and measures of organ dysfunction. Published in 2000, the ﬁrst trial (1) randomized 861 patients with ARDS to a traditional VT of 12 mL/kg predicted body weight (PBW) (control arm) or a lower VT of 6 mL/kg PBW with a plateau pressure less than 30 cmH2O (treatment arm). PEEP and FiO2 were set according to a detailed PEEP : FiO2 table and were similar in the two groups (average 8.8 cmH2O). Plasma levels of cytokines were measured on day 0 and day 3. The trial was stopped after the fourth interim analysis because the lower VT group had a reduced mortality, lesser number of days on the ventilator, and lesser number of days of nonpulmonary organ failure (1). In 2005, Parsons et al. published the results of the plasma measurements of key cytokines of patients enrolled in this trial (32). The authors tested two hypotheses: ﬁrst, that elevated levels of IL-6, IL-8, and IL-10 at presentation of ARDS would be independent predictors of mortality, and second, that after three days of ventilation, the group treated with lower VTs would have lower levels of inﬂammatory cytokines than those ventilated with higher VTs. In fact, patients with elevated plasma levels of IL-6, IL-8, and IL-10 at baseline had a worse prognosis in terms of risk of death, and increased IL-6 and IL-8 levels were also associated with fewer ventilatorfree days and organ failure-free days. It is important to note that the highest

Modulation of the Cytokine Network

487

Figure 5 Changes in the total number of failing organs and in plasma concentration of IL-6 in patients ventilated with conventional and lung-protective strategies. Changes in the number of failing organ systems over a 72-hour period correlate with changes in IL-6 level, and both parameters differ between patients treated with a conventional strategy and a lung-protective strategy. Key: , the number of failing organ systems at 72 hours minus the number on study entry; y, change in IL-6 (level at 36 hours minus level at study entry). Abbreviation: IL, interleukin. Source: From Ref. 24.

levels of baseline cytokines were in patients with sepsis and pneumonia rather than patients with ARDS secondary to trauma, denoting the key role of these cytokines in the overwhelming systemic inﬂammatory process associated with infection. The second important result is that all inﬂammatory markers decreased over time in both the groups, but the decrease was signiﬁcantly greater in the group treated with a protective ventilatory strategy, correlating with better outcomes in this group. This provides further evidence that the preexisting inﬂammatory process present at diagnosis of ARDS can be modulated by the early institution of low VT ventilation. The PEEP level set in this trial (average 8.8 cmH2O) may have offered some protection against atelectrauma, but because it was similar in the two groups, the greater reduction of cytokines must be solely attributed to the reduction in VT. Unfortunately, BAL was not performed, so no conclusions

488

Fanelli et al.

can be drawn regarding the pulmonary inﬂammatory response to the low VT strategy employed, or correlations made between pulmonary and systemic levels of cytokines. Following the important results of the low VT trial, the ARDS Network performed a second large clinical trial comparing lower versus higher levels of PEEP in conjunction with low VTs [the Assessment of Low Tidal Volume and Elevated End-Expiratory Volume to Obviate Lung Injury study (ALVEOLI)] (2). In the treatment arm, higher PEEP was paired with a lower FiO2, to minimize both atelectrauma and the atelectasis associated with high FiO2 levels, and compared to the control arm of lower PEEP and higher FiO2. The data and safety monitoring board stopped the trial at the second interim analysis, after enrollment of 549 patients, on the basis of the stopping rule for futility. The authors concluded there were no signiﬁcant differences in mortality, ventilator-free days, or organ failure-free days between the lower and higher PEEP groups. They also found that changes in plasma levels of IL-6, surfactant protein D, and intercellular adhesion molecule from day 0 to day 3 of the study period did not differ signiﬁcantly between the study groups. Based on this study, there is no evidence that the addition of higher PEEP adds a survival beneﬁt or further attenuates inﬂammation beyond the beneﬁt provided by low VTs. However, this study has two limitations. First, there were imbalances in baseline characteristics of the two groups that occurred by chance, but may have inﬂuenced mortality rates. The mean age was higher (54 17) in the higher PEEP group versus the lower PEEP group (49 17, p < 0.05), and the mean PaO2/FiO2 was lower in the higher PEEP group (151 67) versus the lower PEEP group (165 77, p < 0.05); there was a trend to higher Acute Physiology and Chronic Health Evaluation III scores in the higher PEEP group. The authors used a model to obtain adjusted mortality rates to take into account these imbalances, and still found no signiﬁcant difference in the outcomes between the two groups. A second limitation is that the use of a ﬁxed table approach for setting the PEEP in no way guarantees recruitment and stabilization of alveolar units during tidal inﬂation, and recruitment maneuvers were abandoned after enrollment of the ﬁrst 80 patients, based on ineffectiveness in improving arterial oxygenation. The ﬁxed table method of setting PEEP used in the ARDSnet trial differed from a previous clinical trial in which PEEP was set according to each patient’s pressure–volume curve, and recruitment maneuvers were used, and in which the use of high PEEP combined with low VT produced a reduction in mortality (33). Although the mean PEEP values used were similar in these two trials, there was no means of assessing the efﬁcacy of PEEP in preventing tidal derecruitment or overinﬁlation (34). The following questions therefore remain unanswered: What is the optimal method of setting PEEP? And, if PEEP is objectively shown to provide recruitment and alveolar stabilization during tidal inﬂation, thereby preventing

Modulation of the Cytokine Network

489

atelectrauma, does it provide additional beneﬁt to low VTs in protecting against inﬂammation and reducing mortality? Future clinical trials will shed further light on these issues.

IV. Impact of MV on the Cytokine Network in Healthy Lungs In addition to increasing inﬂammation and mortality in ARDS, large VTs also increase the risk of developing ALI in patients ventilated for more than 48 hours for reasons other than ARDS (odds ratio 1.3 for each mL above 6 mL/kg PBW) (35). Because large VTs can induce a systemic inﬂammatory response in patients with ARDS, and protective ventilatory strategies attenuate this response, should lung-protective ventilation be used in all mechanically ventilated patients? To answer this question, we must examine the effect of MV on the cytokine network in patients without ALI. We will examine evidence from two groups of such patients: those with and without systemic inﬂammation, representing patients with and without risk factors for the development of ARDS. Wrigge et al. (36) randomized 39 patients with normal pulmonary functions, low risk of anesthesia-related morbidity, and no signs of systemic infection, undergoing elective, extrathoracic surgery to one of three ventilator strategies: VT 15 mL/kg and no PEEP, VT 6 mL/kg and no PEEP, or VT 6 mL/kg and PEEP 10 cmH2O. Plasma levels of TNF, IL-6, IL-10, and IL-1ra were measured before and after one hour of ventilation with each of the above strategies. IL-10 was below the detection level in 35 of 39 patients. Plasma levels of the other cytokines were low and did not change signiﬁcantly after one hour of ventilation. There were no differences between the groups. BAL cytokines were not measured, and plasma measurements were not made beyond the one-hour time frame. Thus, it appears that short courses of MV do not impact the systemic cytokine network appreciably in anesthetized patients with healthy lungs and without preexisting systemic inﬂammation. Results of this study were reproduced in a further study by the same group in which cytokine responses were examined during a longer duration of MV in the setting of major thoracic or abdominal surgery (37). In this study, 64 patients without preexisting inﬂammation were randomized to either low VT and high PEEP or high VT and zero PEEP, and plasma cytokines were measured at baseline (immediately after induction of anesthesia), and after one, two, and three hours of MV; tracheal aspirate cytokines were measured after three hours. Although plasma cytokines increased over time during surgery, levels of all cytokines except IL-8 remained relatively low and did not differ between the two ventilatory groups. Likewise, tracheal aspirate cytokine levels did not differ between

490

Fanelli et al.

the groups. The response to MV in healthy patients undergoing surgery contrasts with the response of patients with ALI in the study by Stu¨ber et al. (23), where plasma cytokines were moderately elevated at baseline and increased within one hour of MV with a conventional strategy. Thus, although conventional MV may induce cytokine release in preinjured or infected lungs, this does not occur in healthy lungs during major noncardiac surgery where the surgery-induced systemic inﬂammatory response is relatively small. Recent studies suggest that MV does modulate the cytokine network in the presence of a more signiﬁcant primary inﬂammatory stimulus such as cardiopulmonary bypass (CPB). CPB and aortic cross-clamping may induce lung ischemia and lead to pulmonary endothelial dysfunction (38); activation of complement, neutrophils, and cytokines (39–41); platelet accumulation (42); and an increase in myeloperoxidase activity (43). For these reasons, CPB is associated with a local inﬂammatory response in the lung capillary bed and a generalized systemic inﬂammatory response, with increased risk of postoperative coagulopathy, ALI/ARDS, and MODS (44). In the light of these considerations, Zupancich et al. studied the inﬂuence of MV on pulmonary and systemic concentrations of inﬂammatory mediators in patients undergoing CPB for cardiac surgery (45). They tested the hypothesis that a traditional ventilatory strategy would worsen the inﬂammatory response when primed by CPB, and a protective strategy would minimize this response. Forty patients were randomized to be ventilated after CPB disconnection with either a high VT/low PEEP (10–12 mL/ kg and 2–3 cmH2O) or a low VT/high PEEP (8 mL/kg and 10 cmH2O). IL-6 and IL-8 were measured in the BAL and plasma; samples were obtained before sternotomy (baseline), immediately after CPB disconnection (T1), and after six hours of MV (T2). They found that in both groups IL-6 and IL-8 concentrations increased after CPB, and increased further after six hours of MV with a high VT and a low PEEP, but remained stable in the group treated with a low VT and a high PEEP in both BAL and plasma. Levels of IL-6 and IL-8 were signiﬁcantly higher in the high VT/low PEEP group than the low VT/high PEEP group after six hours of MV. These data suggest that conventional MV can exacerbate the systemic inﬂammation induced by CPB, while protective MV avoids this response, thereby minimizing risk of an overwhelming inﬂammatory response that can lead to ALI/ARDS and MODS (7). Two additional studies have examined the effect of ventilation strategies on the cytokine network post-CPB, with some contrasting results. Koner et al. found no differences in serum concentrations of TNF-a and IL6 between groups of patients ventilated with VT of 6 mL/kg and PEEP of 5 cmH2O, 10 mL/kg and PEEP 5 cmH2O, and 10 mL/kg and zero PEEP, undergoing CPB (46). However, there are several possible explanations for these conﬂicting results: Koner et al. found considerable variation in the

Modulation of the Cytokine Network

491

levels of serum cytokines, making it difﬁcult to detect a signal among the noise, and measured cytokine levels only up to two hours post-CPB. The authors suggest that a larger patient population, and/or more frequent sampling for a longer period would have yielded more reliable cytokine results. In addition, this study did not measure cytokines in BAL, which may have allowed the authors to detect differences in the pulmonary inﬂammatory response with different strategies of MV. Furthermore, Koner et al. measured cytokine levels only two hours after the end of CPB, while the inﬂammatory response was still increasing, whereas Zupancich et al. sampled plasma after six hours. It may be that a longer duration of MV is necessary before signiﬁcant differences between ventilatory strategies with respect to cytokine modulation can be detected. Finally, the level of PEEP used in the low VT/high PEEP group (5 cmH2O) was lower than that used by Zupancich et al. (10 cmH2O), and the ARDS Network low VT trial (8.8 cmH2O), in their respective protective ventilatory strategies. The level of PEEP used in the study by Koner et al. may have been inadequate to prevent atelectrauma. A third study was conducted by Wrigge et al. (to date, abstract published only) in which 44 patients were randomized to protective (VT 6 mL/kg ideal body weight) versus conventional (VT 12 mL/kg ideal body weight) MV for six hours following CPB (47). TNF, IL-6, and IL-8 levels were measured in the serum at entry, and after two, four, and six hours of MV, and in BAL after six hours of MV. Six-hour BAL measurements revealed signiﬁcantly lower TNF levels in the lower VT group than the higher VT group ( p ¼ 0.01), with the same trend for IL-6 ( p ¼ 0.08) but not for IL-8. No differences between groups were detected for plasma cytokine levels, but a subgroup analysis of patients with elevated TNF levels at entry showed that serum levels of TNF decreased faster in patients treated with a lower VT. Because this study has only been published in abstract form, insufﬁcient information is available at present to compare and contrast these results with the studies by Zupancich et al. (45) and Koner et al. (46). The studies by Wrigge et al. (36,37,47) and Zupancich et al. (45) provide indirect evidence for the two-hit hypothesis, suggesting that injurious MV (the ‘‘second hit’’) can produce an inﬂammatory response only in the setting of preexisting inﬂammation from an initial insult, or ‘‘ﬁrst hit.’’ Evidence from the above trials would suggest that CPB provides sufﬁcient inﬂammation to sensitize the lungs to the harmful effects of MV, while major thoracic or abdominal surgery alone does not. In the former case, protective MV is likely useful for protecting against an overwhelming systemic inﬂammatory response. The theory that protective MV may be useful in modulating the cytokine response in cases where there is a sufﬁcient primary inﬂammatory stimulus has widespread application, but has yet to be examined outside the context of CPB. Potential applications would include multiple trauma, sepsis, severe brain injury, and brain death. In these

492

Fanelli et al.

contexts, patients require MV, have high levels of systemic inﬂammatory mediators, may or may not have direct pulmonary insults, and have a signiﬁcant risk of developing ARDS. V. Conclusion In conclusion, a protective ventilatory strategy with VT of 6 mL/kg has been shown to decrease mortality and the number of days with organ failure. The studies described above provided evidence linking protective MV with the reduction of local and systemic cytokines, and a decreased incidence of end-organ damage. Multiple organ dysfunctions have previously been shown to correlate strongly with mortality in ARDS (6). Although combined strategies of a low VT and a high PEEP have been shown to protect against systemic inﬂammation (22) and incur a survival beneﬁt (33), there is currently no evidence from clinical trials that PEEP exerts a mortality beneﬁt above and beyond that of low VT ventilation. However, currently there is also no consensus on the optimal PEEP level or the best method of setting PEEP. Ongoing clinical trials may shed further light on this topic. In addition to modulating the inﬂammatory response in ARDS, protective MV has also been shown to reduce cytokine levels compared to conventional MV in patients undergoing CPB (45). Whether protective MV also modulates the inﬂammatory response in other contexts, and the clinical signiﬁcance of this impact, has yet to be proven. Acknowledgment This work was supported by MURST COFIN 2002–2005. References 1. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342: 1301–1308. 2. The National Heart, Lung, and Blood Institute ARDS Clinical Trials Network. Higher versus lower positive end expiratory pressures in patients with acute respiratory distress syndrome. N Engl J Med 2004; 351:327–336. 3. Brun-Buisson C, Minelli C, Bertolini G, et al. ALIVE Study Group. Epidemiology and outcome of acute lung injury in European intensive care units. Results from the ALIVE study. Intensive Care Med 2004; 30:51–61. 4. Bersten AD, Edibam C, Hunt T, Moran J. Australian and New Zealand Intensive Care Society Clinical Trials Group. Incidence and mortality of acute lung

Modulation of the Cytokine Network

5.

6.

7.

8.

9. 10.

11.

12. 13.

14.

15. 16. 17. 18.

19.

20.

493

injury and the acute respiratory distress syndrome in three Australian states. Am J Respir Crit Care Med 2002; 165:443–448. Abel SJ, Finney SJ, Brett SJ, Keogh BF, Morgan CJ, Evans TW. Reduced mortality in association with the acute respiratory distress syndrome (ARDS). Thorax 1998; 53:292–294. Esteban A, Anzueto A, Frutos F, et al. Mechanical Ventilation International Study Group. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study. JAMA 2002; 287:345–355. Papathanassoglou ED, Moynihan JA, Ackennan MH. Does programmed cell death (apoptosis) play a role in the development of multiple organ dysfunction in critically ill patients? A review and a theoretical framework. Crit Care Med 2000; 28:537–549. Marshall JC, Cook DJ, Christou NV, et al. Multiple organ dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med 1995; 23:1638–1652. Putensen C, Wrigge H. Ventilator-associated systemic inﬂammation in acute lung injury. Intensive Care Med 2000; 26:1411–1413. Cytokines in pulmonary disease. Infection and inﬂammation. In: Nelson S, Martin TR, eds. Lung Biology in Health and Disease. Vol. 141. New York, U.S.A.: Marcel Dekker Inc., 2000. Cavaillon JM, Adib-Conquy M. The pro-inﬂammatory cytokine cascade. In: Marshall JC, Cohen J, eds. Immune Response in the Critically Ill. Series Update in Intensive Care Medicine. Berlin, Germany: Springer, 2002. Uhlig S. Ventilation-induced lung injury and mechanotransduction: stretching it too far? Am J Physiol Lung Cell Mol Physiol 2002; 282:L892–L896. Vlaharis NE, Hubmayr RD. Cellular responses to capillary stress: plasma membrane stress failure in alveolar epithelial cells. J Appl Physiol 2000; 89:2490–2496. Mourgeon E, Isowa N, Keshavjee S, Zhang X, Slutsky AS, Liu M. Mechanical stretch stimulates macrophage inﬂammatory protein-2 secretion from fetal rat lung cells. Am J Physiol Lung Cell Mol Physiol 2000; 279:L699–L706. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. Slutsky AS, Tremblay LN. Multiple system organ failure: is mechanical ventilation a contributing factor? Am J Respir Crit Care Med 1998; 157:1721–1725. Tremblay LN, Slutsky AS. Ventilator-induced injury: from barotrauma to biotrauma. Proc Assoc Am Physicians 1998; 110:482–488. Roupie E, Dambrosio M, Servillo G, et al. Titration of tidal volume and induced hypercapnia in acute respiratory distress syndrome. Am J Respir Crit care Med 1995; 152:121–128. Brower R, Matthay MA, Schoenfeld DA. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials. Am J Respir Crit Care Med 2002; 166:1515–1516. Brower R, Krishnan J, Thompson BT, et al. Effects of tidal volume reduction (VtR) in acute lung injury (All) patients with inspiratory plateau pressures (PPLAT) < 32 cmH2O before VtR [abstr]. Am J Respir Crit Care Med 2003; 167:A616.

494

Fanelli et al.

21. Slutsky AS. Lung injury caused by mechanical ventilation. Chest 1999; 196: 9S–15S. 22. Ranieri VM, Suter PM, Tortorella C, et al. Effect of mechanical ventilation on inﬂammatory mediators in patients with acute respiratory distress syndrome: a randomized controlled trial. JAMA 1999; 282:54–61. 23. Stu¨ber F, Wrigge H, Schroeder S, et al. Kinetic and reversibility of mechanical ventilation-associated pulmonary and systemic inﬂammatory response in patients with acute lung injury. Intensive Care Med 2002; 28:834–841. 24. Tremblay LN, Miatto D, Hamid Q, Govindarajan A, Slutsky AS. Injurious ventilation induces widespread pulmonary epithelial expression of tumor necrosis factor-alpha and interleukin-6 messenger RNA. Crit Care Med 2002; 30: 1693–1700. 25. Tremblay L, Valenza F, Ribeiro SP, Jingfang L, Slutsky AS. Injurious ventilation strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99:944–952. 26. von Bethmann AN, Brasch F, Nusing R, et al. Hyperventilation induces release of cytokines from perfused mouse lung. Am J Respir Crit Care Med 1998; 157:263–272. 27. Haitsma JJ, Uhlig S, Goggel R, Verbrugge SJ, Lachmann U, Lachmann B. Ventilator-induced lung injury leads to loss of alveolar and systemic compartmentalization of tumor necrosis factor-alpha. Intensive Care Med 2000; 26: 1515–1522. 28. Tutor JD, Mason CM, Dobard E, Beckerman RC, Summer WR, Nelson S. Loss of compartmentalization of alveolar tumor necrosis factor after lung injury. Am J Respir Crit Care Med 1994; 149:1107–1111. 29. Nahum A, Hoyt J, Schmitz L, Moody J, Shapiro R, Marini JJ. Effect of mechanical ventilation strategy on dissemination of intratracheally instilled Escherichia coli in dogs. Crit Care Med 1997; 25:1733–1743. 30. Ranieri VM, Suter PM, Slutsky AS. Mechanical ventilation as a mediator of multisystem organ failure in acute respiratory distress syndrome. JAMA 2000; 284:43–44. 31. Imai Y, Parodo J, Kajikawa O, et al. Injurious mechanical ventilation and end-organ epithelial cell apoptosis and organ dysfunction in an experimental model of acute respiratory distress syndrome. JAMA 2003; 289:2104–2112. 32. Parsons PE, Eisner MD, Thompson BT, et al. and the NHLBI Acute Respiratory Distress Syndrome Clinical Trials Network. Lower tidal volume ventilation and plasma cytokine markers of inﬂammation in patients with acute lung injury. Crit Care Med 2005; 33:1–6. 33. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338:347–354. 34. Grasso S, Fanelli V, Cafarelli A, et al. Effects of high versus low positive end-expiratory pressures in acute respiratory distress syndrome. Am J Respir Crit Care Med 2005; 171(9):1002–1008. 35. Gajic O, Dara SI, Mendez JL, et al. Ventilator-associated lung injury in patients without acute lung injury at the onset of mechanical ventilation. Crit Care Med 2004; 32:1817–1824.

Modulation of the Cytokine Network

495

36. Wrigge H, Zinserling J, Stu¨ber F, et al. Effects of mechanical ventilation on release of cytokines into systemic circulation in patients with normal pulmonary function. Anesthesiology 2000; 93:1413–1417. 37. Wrigge H, Uhlig U, Zinserling J, et al. The effects of different ventilatory settings on pulmonary and systemic inﬂammatory responses during major surgery. Anesth Analg 2004; 98:775–781. 38. Chai PJ, Williamson JA, Lodge AJ, et al. Effects of ischemia on pulmonary dysfunction after cardiopulmonary bypass. Ann Thorac Surg 1999; 67: 731–735. 39. Butler J, Rocker GM, Westaby S. Inﬂammatory response to cardiopulmonary bypass. Ann Thorac Surg 1993; 55:552–559. 40. Wan S, LeClerc JL, Vincent JL. Inﬂammatory response to cardiopulmonary bypass: mechanisms involved and possible therapeutic strategies. Chest 1997; 112:676–692. 41. Friedman M, Wang SY, Sellke FW, Cohn WE, Weintraub RM, Johnson RG. Neutrophil adhesion blockade with NPC 15669 decreases pulmonary injury after total cardiopulmonary bypass. J Thorac Cardiovasc Surg 1996; 111:460–468. 42. Brix-Christiansen V, Tonnesen E, Hjortdal VE, et al. Neutrophils and platelets accumulate in the heart, lungs, and kidneys after cardiopulmonary bypass in neonatal piglets. Crit Care Med 2002; 30:670–676. 43. Bando K, Pillai R, Cameron D, et al. Leukocyte depletion ameliorates free radical-mediated lung injury after cardiopulmonary bypass. J Thorac Cardiovasc Surg 1990; 99:873–877. 44. Paparella D, Yau TM, Young E. Cardiopulmonary bypass induced inﬂammation: pathophysiology and treatment. An update. Eur J Cardio-thorac Surg 2002; 21:232–244. 45. Zupancich E, Paparella D, Turani F, et al. Mechanical ventilation affects inﬂammatory mediators in patients undergoing cardiopulmonary bypass for cardiac surgery: a randomized clinical trial. J Thorac Cardiovasc Surg 2005; 130(2):378–383. 46. Koner O, Celebi S, Balci H, Cetin G, Karaoglu K, Cakar N. Effects of protective and conventional mechanical ventilation on pulmonary function and systemic cytokine release after cardiopulmonary bypass. Intensive Care Med 2004; 30:620–626. 47. Wrigge H, Uhlig U, Zinserling J, Menzenbach J, Uhlig S, Putensen C. Inﬂammatory effects of conventional and lower tidal volume ventilation after cardiac surgery [abstr.]. Intensive Care Med 2003; 29:307.

This page intentionally left blank

20 Role of Tidal Volume and PEEP in the Reduction of VILI

DAVID N. HAGER and ROY G. BROWER Pulmonary and Critical Care Medicine, Johns Hopkins University School of Medicine Baltimore, Maryland, U.S.A.

I. Introduction Acute lung injury and acute respiratory distress syndrome (ALI/ARDS) are life-threatening disorders associated with severe impairments of respiratory gas exchange. Without mechanical ventilation (MV), many patients with ALI/ARDS would die from hypoxemic and hypercarbic respiratory failure. MV allows time for treatment of the underlying conditions that cause ALI/ARDS, such as pneumonia and sepsis, and for natural healing processes. However, MV can also cause ALI [ventilator-induced lung injury (VILI)], which may delay or prevent recovery. Modiﬁcations of traditional MV techniques can reduce VILI and improve the likelihood of recovery. These modiﬁcations include the use of relatively small tidal volumes with low inspiratory pressures (volume-and-pressure limitation) and relatively high levels of positive end-expiratory pressure (higher PEEP). In this chapter, we review the traditional approach to MV, the experimental data that motivated investigators to study lung-protective modiﬁcations of the traditional approach, and the results of clinical studies that compare different MV approaches. 497

498

Hager and Brower II. Traditional Approach to MV in ALI/ARDS

Positive pressure ventilation techniques were developed during the early 20th century to support patients requiring general anesthesia for surgery (1). These techniques were later used to support increasing numbers of patients with acute respiratory failure, including those with lung injury from severe pneumonia, trauma, and sepsis who would have met current criteria for ALI/ARDS (2). The traditional approach to MV in ALI/ARDS that evolved in the mid–late 20th century (Table 1) was recommended by experts in the use of MV and in the management of acute respiratory failure (8–10). A. Tidal Volumes

Control of tidal volumes was an important aspect of earlier approaches to MV in ALI/ARDS. Pressure-cycled modes frequently result in relatively small or variable tidal volumes in patients with severe parenchymal lung disease such as ALI/ARDS. Therefore, volume-cycled modes were utilized more frequently. Generous tidal volumes of 10 to 15 mL/kg were recommended, because they are useful for maintaining normal arterial PCO2 and pH (11) in patients with increased dead space, as in ALI/ARDS. Even higher tidal volumes were sometimes used in the most severe cases (11). Table 1 Traditional and Lung-Protective Approaches to MV in ALI/ARDS Mode

Tidal volume

Traditional approach

Volume cycled

10–15 mL/kg; no explicit guidance regarding use of measured body weight vs. lean or dry body weight

Lungprotective approach

Either volume or pressure cycled

6–8 mL/kg lean or dry body weight

a

Inspiratory pressure limits No explicit guidance for when to decrease tidal volume.a Inspiratory pressures > 40 cmH2O raised concerns of barotrauma Inspiratory plateau pressure of 30–35 cmH2O

PEEP and FiO2 PEEPs of 5–12 cmH2O adequate in most patients to support arterial oxygenation while avoiding FiO2s >0.65 Higher PEEPs (approximately 12–20 cmH2O) to achieve greater lung recruitment

In four clinical trials, the following inspiratory pressure limits were used to represent the traditional MV approach: peak inspiratory pressures of 60 cmH2O and 50 cmH2O; plateau pressures of 50 and of 45 to 55 cmH2O. Abbreviations: ALI, acute lung injury; ARDS, acute respiratory distress syndrome; MV, mechanical ventilation; PEEP, positive end-expiratory pressure; FiO2, fraction of inspired oxygen. Source: From Refs. 3–7.

Role of Tidal Volume and PEEP in the Reduction of VILI

499

B. Support for Arterial Oxygenation

The primary causes of hypoxemia in ALI/ARDS are increased intrapulmonary shunt and ventilation–perfusion imbalances. Increasing the fraction of inspired oxygen (FiO2) can correct hypoxemia from lung units with low ventilation–perfusion ratios. However, increasing FiO2 is often insufﬁcient to maintain clinically acceptable arterial oxygenation when shunt fractions are elevated substantially. Moreover, sustained exposure to a high FiO2 may cause oxygen toxicity (12–15), which may exacerbate lung injury. Safe levels of FiO2 in humans with ALI/ARDS have not been clearly established, but most clinicians attempt to avoid prolonged exposures to FiO2s greater than 0.6 to 0.7. Intrapulmonary shunt can be reduced and arterial oxygenation improved by applying PEEP, which reverses or prevents atelectasis of some unstable lung units and redistributes ﬂuid from alveolar to interstitial compartments (16,17). However, PEEP may cause circulatory depression (18–23), which, despite improved arterial oxygenation, could decrease oxygen delivery to systemic tissues. PEEP also tends to cause higher airway pressures during inspiration, raising the potential for barotrauma and lung injury from overdistention (vide infra). There is little information to guide clinicians in the difﬁcult task of balancing the beneﬁcial and detrimental effects of elevated FiO2 and PEEP. Early clinical practices were guided by case series reports in which arterial oxygenation goals were achieved in most ALI/ARDS patients by applying PEEPs of 5 to 12 cmH2O and FiO2s equivalent to 0.65 at sea level. This approach is still used by most clinicians. Higher PEEP levels have generally not been used unless a high FiO2 was necessary to achieve acceptable arterial oxygenation. III. Mechanisms of VILI Mechanisms of VILI have been elucidated in experimental models and are reviewed in detail in several chapters of this volume and elsewhere (21,22). VILI is initiated by excessive mechanical forces in the lung parenchyma. These forces cause physical damage to endothelial and epithelial cells (23–25) and trigger inﬂammation in the lung parenchyma through mechanotransduction pathways or by exposure and release of cellular components (21,26,27). Inﬂammation is marked by the inﬂux of inﬂammatory cells into the interstitium and alveolar spaces and by production and release of mediators that propagate the inﬂammatory cascade. These events may exacerbate or perpetuate the lung injury caused by treatable or self-limited conditions such as pneumonia, sepsis, and trauma. There is good evidence that at least two aspects of the traditional MV approach may cause VILI in ALI/ARDS patients.

500

Hager and Brower A. High Lung Volume and End-Inspiratory Pressure (High Volume/Pressure VILI)

Studies in experimental models demonstrated that overdistention of the normal lung can cause VILI. A substantial portion of the lung in ALI/ ARDS is not available for ventilation because of alveolar ﬂooding, consolidation, and atelectasis. The use of generous tidal volumes is therefore likely to cause overdistention injury in the remaining aerated lung regions (28–30). Moreover, ALI from other causes may be exacerbated by overdistention during inspiration (31). B. Low Lung Volume and End-Expiratory Pressure (Low Volume/Pressure VILI)

Some unstable lung units may open with each inspiration and close during expiration. Mechanical forces associated with this repeated opening and closing may deplete surfactants and injure small bronchioles and alveoli (32). Lung injury may also result from excessive stress and strain in the parenchymal connections between aerated and nonaerated lung units (33). Finally, some alveoli and small bronchioles may be ﬁlled with ﬂuid and foam at low lung volumes (34). Under these conditions, tidal volumes may only be delivered to the remaining aerated lung units and cause overdistention injury, even when the tidal volumes are not large. Animal models have demonstrated that these forms of VILI can be attenuated by ventilating with some level of PEEP, which raises lung volume at end-expiration and reduces the proportion of lung that is atelectatic or ﬂuid ﬁlled (23,32,35,36). IV. Lung-Protective Ventilation A better understanding of the mechanisms of VILI led to the development of ‘‘lung-protective’’ MV strategies to limit high volume/pressure VILI and low volume/pressure VILI (Table 2). One component of lung-protective MV is the use of lower tidal volumes and inspiratory pressures (volume-andpressure limited MV) than are used in the traditional MV approach (3,37). However, volume-and-pressure limited MV may result in hypercapnia and acute acidosis. In some patients, this causes tachycardia and hypertension, impaired myocardial contractility, decreased systemic vascular tone and responsiveness to catecholamines, and increased dyspnea and agitation (38–41). Furthermore, requirements for oxygenation support (FiO2 and PEEP) may increase when lower tidal volumes and inspiratory pressures are used (42–45). Thus, volume-and-pressure limited MV represents a change from the traditional MV scheme. In the traditional MV approach, maintenance of normal acid–base homeostasis and reduction of intrapulmonary shunt have higher priority than prevention of high volume/pressure VILI. With volume-and-pressure limited MV, this prioritization scheme is inverted.

Role of Tidal Volume and PEEP in the Reduction of VILI

501

Table 2 Prioritization Schemes for Traditional Tidal Volume and Volume-and-Pressure Limited MV Strategies Traditional tidal volume strategy Higher priority objectives Maintain normal acid–base balance Improve arterial oxygenation Lower priority objectives Prevent high volume and high pressure VILI

Volume-and-pressure limited strategy Higher priority objectives Prevent high volume and high pressure VILI Lower priority objectives Maintain normal acid–base balance Improve arterial oxygenation

Abbreviations: MV, mechanical ventilation; VILI, ventilator-induced lung injury.

A second component of lung-protective MV is the use of higher levels of PEEP than are used in the traditional MV approach (46,47). Higher PEEP limits the amount of cyclic closing and opening of small bronchioles and alveoli and promotes a more homogeneous distribution of tidal volumes. This may reduce low volume/pressure VILI by increasing the proportion of aerated lung during tidal ventilation (lung recruitment). Moreover, increased lung recruitment usually decreases intrapulmonary shunt, allowing acceptable arterial oxygenation at lower FiO2s, thus reducing the potential for oxygen toxicity. However, as mentioned earlier, MV with higher PEEP can cause circulatory depression, which may adversely affect the functioning of other organs and systems. The use of higher PEEP typically results in higher end-expiratory lung volumes. If applied in combination with generous tidal volumes, high lung volumes and pressures during inspiration may cause high volume/pressure VILI (48). This effect of higher PEEP can be attenuated by decreasing tidal volumes to avoid inspiratory pressures that exceed limits that are considered safe. However, when higher PEEPs are combined with lower inspiratory pressure limits, the resulting tidal volumes may be very small, which could cause severe hypercapnia and acidosis. Thus, the use of higher PEEP represents another change in the traditional scheme for prioritizing clinical objectives. In the traditional MV approach with relatively low levels of PEEP, prevention of circulatory depression and lung damage from high inspiratory pressures and volumes, and maintenance of normal acid–base homeostasis have a higher priority than prevention of lung damage from low volume/pressure VILI and oxygen toxicity. As with the use of lower tidal volumes, the higher PEEP approach represents a change in priorities (Table 3). The recommendations to use lung-protective MV strategies are based on ample evidence from experimental models (21,22). However, none of the experimental models provided a very accurate representation of lung mechanics and inﬂammatory processes in ALI or ARDS patients. Another

502

Hager and Brower

Table 3 Prioritization Schemes for Traditional PEEP and Higher PEEP Mechanical Ventilation Strategies Traditional PEEP strategy Higher priority objectives Prevent circulatory depression Prevent high volume and high pressure VILI Maintain normal acid–base balance Lower priority objectives Prevent low volume and low pressure VILI

Higher PEEP strategy Higher priority objectives Prevent low volume and low pressure VILI

Lower priority objectives Prevent circulatory depression Prevent high volume and high pressure VILI Maintain normal acid–base balance

Abbreviations: PEEP, positive end-expiratory pressure; VILI, ventilator-induced lung injury.

limitation of the experimental models is that they were not designed to assess the potentially detrimental effects of lung-protective MV strategies on nonpulmonary organ and system function. Most importantly, the experimental models did not assess the ‘‘balance’’ between the potentially beneﬁcial and detrimental effects of lung-protective MV strategies. Therefore, results of lung-protective MV strategies in experimental models could not be directly applied to change clinical practice. Clinical studies were necessary to assess this balance by demonstrating the effects of lung-protective MV strategies on important clinical outcomes such as duration of MV and mortality. In two case series reports, each with approximately 50 ARDS patients, mortality was considerably lower than expected in patients who received volume-and-pressure limited MV, despite substantial increases in arterial pCO2 and decreases in pH (37,49). However, these studies did not compare outcomes of patients who received volume-and-pressure limited MV to those who received traditional MV concurrently in the same intensive care units. Some studies have strongly suggested that clinical outcomes from critical illness have improved over time, independent of speciﬁc efforts to utilize lung-protective MV strategies (50). Therefore, these case series reports were not conclusive. In another study conducted earlier, 103 patients with acute respiratory failure from various causes were randomized to either a traditional MV strategy or a volume-and-pressure limited strategy (51). This study also suggested that volume-and-pressure limited MV was safe and could improve clinical outcomes. However, because of the diverse group of patients included in the study and the modest size of the study groups, this randomized trial was not conclusive.

Role of Tidal Volume and PEEP in the Reduction of VILI

503

V. Clinical Trials of Lung-Protective MV Strategies Amato et al. randomized 53 patients with severe ARDS to study groups that received either a traditional MV strategy or a comprehensive lungprotective MV strategy designed to prevent both high volume/high pressure VILI and low volume/low pressure VILI (47). As expected, patients who received the lung-protective MV strategy experienced respiratory acidosis and required more sedation than was typically used in other patients. However, the lung-protective MV strategy was associated with signiﬁcant improvements in survival and weaning during the 28 days following randomization. This clinical trial was notable because it provided strong evidence that lung-protective MV strategies could improve clinical outcomes in patients with ALI/ARDS. However, the rate of survival in the traditional MV study group was less favorable than in some contemporary reports in which ARDS patients were treated with traditional MV approaches (52,53). This suggested that imbalances between the study groups at baseline could have favored the lung-protective MV study group. Also, it was not clear if the beneﬁcial effects of the lung-protective strategy were attributable to the effects of volume-andpressure limitation (to prevent high volume/pressure VILI), higher PEEP and lung recruitment maneuvers (to reduce low volume/pressure VILI), or both. Therefore, the results of the trial of Amato et al. were provocative but required conﬁrmation and further elaboration. A. Clinical Trials of Volume-and-Pressure Limited MV

Four clinical trials conducted in the mid to late 1990s were designed speciﬁcally to assess the clinical value of volume-and-pressure limited MV strategies in patients with or at high risk for ALI/ARDS (Table 4) (4–7). In each of the four trials, levels of PEEP were similar to those used in traditional MV strategies. In three of these trials, the volume-and-pressure limited MV strategies were not associated with either a lower mortality or improvements in other important clinical outcomes (5–7). However, in the fourth trial (the ARDS Network trial), mortality and other clinical outcomes were signiﬁcantly better in the study group that received the volumeand-pressure limited approach (4). There are several possible explanations for the different results in this trial compared to the other trials. Traditional Tidal Volumes and Inspiratory Pressures

There were some differences in the tidal volumes and the resulting inspiratory airway pressures in the study groups that received traditional MV strategies. In the ARDS Network trial, the mean inspiratory plateau pressure in the traditional MV group during the ﬁrst several days after randomization was approximately 34 cmH2O. In the other three trials, the mean inspiratory plateau pressures in the traditional MV groups were 28 to

504

Hager and Brower

Table 4 Clinical Trials of Lung-Protective Mechanical Ventilation Strategies in ALI or ARDS Tidal volumes as reported

Amato et al. (47) NIHARDS Network (4) Brochard et al. (5) Stewart et al. (6) Brower et al. (7)

Mortality %

Traditional

Lower

Traditional

Lower

12a 11.8b 10.3c 10.8 10.2b

6a 6.2b 7.1c 7.2d 7.3b

71 40 38 47 46

38 31 47 50 50

a

Tidal volumes in mL/kg measured body weight. Tidal volumes in mL/kg PBW: male PBW (kg) ¼ 50 þ 2.3 [(height in inches) 60]; female PBW (kg) ¼ 45.5 þ 2.3 [(height in inches) 60]. c Tidal volumes in mL/kg dry body weight (measured weight minus estimated weight gain from water and salt retention). d Tidal volumes in mL/kg IBW ¼ 25 (height in meters)2. Abbreviations: ALI, acute lung injury; ARDS, acute respiratory distress syndrome; IBW, ideal body weight; PBW, predicted body weight. b

32 cmH2O (5–7). Thus, there could have been more high volume/pressure VILI in the traditional tidal volume group of the ARDS Network trial than in the other three trials. Volume-and-Pressure Limited Tidal Volumes and Inspiratory Pressures

Each of the volume-and-pressure limited MV study groups utilized lower tidal volumes and inspiratory pressure limits than were used in traditional MV strategies. However, the tidal volumes used in the volume-and-pressure limited study group of the ARDS Network trial were lower than in the other three trials. Thus, there could have been less high volume/pressure VILI in the volume-and-pressure limited group of the ARDS Network trial than in the other three trials. Power

The power of a clinical trial represents the conﬁdence we should have if the results of the trial do not reject the null hypothesis. Several aspects of a trial can affect its power. These include the probability that the two study groups will be well balanced at baseline, before the study interventions. Many factors intrinsic to each patient, such as age, cause of ALI/ARDS, and the presence of nonpulmonary organ dysfunction, can inﬂuence the

Role of Tidal Volume and PEEP in the Reduction of VILI

505

outcome of a patient, independent of study intervention effects. Randomization is intended to prevent ‘‘systematic’’ differences between study groups in these variables. However, even with randomization, the study groups may be unbalanced in one or more of these factors. The probability of such imbalances is greater in trials that enroll fewer patients. The three trials in which volume-and-pressure limited MV was not associated with improved outcomes were relatively modest in size. In each of these three trials, baseline differences between the study groups in arterial pH, Acute Physiology and Chronic Health Evaluation scores, and PaO2/FiO2 ratios favored the traditional MV study groups, suggesting that the volume-and-pressure limited study groups were at a higher risk of death. Imbalances in baseline variables such as these could have obscured beneﬁcial effects of the volume-and-pressure limited strategies utilized in these trials (54). Enrollment in the ARDS Network trial was approximately three times as great as the combined enrollment in the other three trials, and the two study groups were better matched at baseline in known predictors of mortality. Differences in the Subjects Enrolled in the Trials

In the trial by Stewart et al. (6), patients could be enrolled if they were at risk for ALI/ARDS, but it was not necessary for all of the criteria for ALI/ARDS to be present at the time of enrollment or at any time after randomization. The tidal volumes used in the traditional study group of this trial were as high or higher than those used in the other three trials of volume-and-pressure limited MV, but the resulting inspiratory pressures were lower. This suggests that lung injury was mild or absent in some patients in this trial. The balance between beneﬁcial and detrimental effects of volume-and-pressure limited MV (Table 2) may be less favorable in these patients because there was less risk of high volume/ pressure VILI. Management of Acidosis

In the ARDS Network trial (4), MV respiratory rates were increased to a maximum of 35 breaths per minute, to reduce hypercapnia and acidosis in the lower tidal volume study group. Moreover, infusions of sodium bicarbonate were allowed at clinicians’ discretion (neither encouraged nor discouraged), if the arterial pH was lower than 7.30. These measures to prevent acidosis were not included in the protocols of the other three trials, and acidosis was more severe in the volume-and-pressure limited study groups of these trials. This difference between the protocols could have contributed to the more favorable outcomes in the volume-and-pressure limited group of the ARDS Network trial.

506

Hager and Brower Bias Introduced by ‘‘Early Stopping Rules’’

Enrollment in the ARDS Network trial was terminated before the maximum planned enrollment, because an interim data analysis demonstrated convincing evidence for the superiority of the volume-and-pressure limited approach relative to the traditional MV approach. Enrollments in the trials by Brochard et al. (5) and Brower et al. (7) were also terminated before the maximum planned enrollments, because interim analyses indicated that the probability of demonstrating the efﬁcacy of the volume-and-pressure limited groups was very small. The results of clinical trials that stop before their maximum planned enrollments tend to be biased in favor of the early stopping rules. If the ARDS Network trial had continued to its maximum planned enrollment, it might have demonstrated a less favorable effect of the volume-and-pressure limited strategy. In contrast, if the trials by Brochard et al. and Brower et al. had continued to their maximum planned enrollments, they might have demonstrated favorable effects.

B. Clinical Trials of MV with Higher PEEP

In a second trial of MV strategies, the ARDS Network randomized 549 ALI/ARDS patients to receive MV with either traditional or higher levels of PEEP (Table 5) (55). Both study groups received the same volumeand-pressure limited strategy as in the previous ARDS Network trial of volume-and-pressure limited MV. In all patients, PEEP and FiO2 were adjusted in discrete steps according to tables of PEEP/FiO2 combinations (PEEP/FiO2-steps) to achieve the same arterial oxygenation goals. The table used in the traditional PEEP study group represented a consensus of how the investigators and clinical colleagues used PEEP and FiO2 in 1995, when there was little consideration for the potential lung-protective effects of PEEP. The table used in the higher PEEP study group was designed to apply PEEPs that were approximately 5 to 7 cmH2O higher than in the traditional PEEP group, as in the trial by Amato et al. (47). In the higher PEEP group, mean PEEP levels were approximately 6 cmH2O higher on the ﬁrst day after randomization and approximately 5 cmH2O higher on subsequent days. With higher levels of PEEP, lower FiO2s were necessary to achieve the arterial oxygenation goal, indicating that the higher PEEPs induced some lung recruitment. However, mortality rates were similar in the two study groups, and there were no signiﬁcant differences in ventilator- or intensive care unit–free days (55). The results of this single randomized study suggest that when a volume-and-pressure limited strategy is used, there is little or no value (or detriment) in raising PEEP to higher levels than are used traditionally. However, there are several reasons why the results of this single trial should

Lower PEEP/higher FiO2 strategy 0.3 0.4 FiO2 PEEP 5 5 Higher PEEP/lower FiO2 strategy FiO2 0.3 0.3

0.4 8

0.5 8

0.5 10

0.6 10

0.7 10

0.7 12

0.7 14

0.8 14

0.4

0.4

0.5

0.5

0.5– 0.8 20

0.8

0.9

1.0

22

22

22–24

PEEP 12 14 14 16 16 18 Arterial oxygenation goal 55 PaO2 80 mmHg or 88% oxyhemoglobin saturation 95%

0.9 14

Abbreviations: PEEP, positive end-expiratory pressure; FIO2, fraction of inspired oxygen; PaO2, partial pressure of oxygen. Source: From Ref. 55.

0.9 16

0.9 18

1.0 18–24

Role of Tidal Volume and PEEP in the Reduction of VILI

Table 5 Traditional (Lower) PEEP/Higher FiO2 and Higher PEEP/Lower FiO2 Strategies Used in a Clinical Trial

507

508

Hager and Brower

not be considered conclusive. (i) The trial was stopped for futility, not because there was convincing evidence that MV with higher PEEP was not useful. (ii) The trial by Amato et al. and the trial by the ARDS Network set the initial PEEP levels according to respiratory system characteristics that were speciﬁc to each patient, but the methods were different. There is no evidence for the superiority of either approach, and the resulting mean PEEP levels were similar in the two studies. However, the PEEP levels set according to the ARDS Network higher PEEP/lower FiO2 table could have deviated substantially in some patients from levels that would have been set according to pressure–volume characteristics, as in the trial by Amato et al. (47). It is possible that such deviations, if any, resulted in suboptimal outcomes. (iii) Recruitment maneuvers were performed occasionally in the study group that received higher PEEP in the trial by Amato et al. These consisted of sustained inﬂations of the lungs to higher levels than occurred during tidal ventilation, to reverse atelectasis in some unstable lung units. Recruitment maneuvers were not conducted in most patients in the ARDS Network trial of MV with higher PEEP. (iv) The most common cause of ALI/ARDS was pneumonia. Some investigators have suggested that higher PEEP may be less effective for recruitment in patients with pneumonia- and aspiration-induced ALI/ARDS (direct lung injury) than in patients in whom ALI/ARDS is caused by indirect lung injury such as sepsis, pancreatitis, and trauma (56). Moreover, higher PEEP in patients with direct lung injury may cause overdistention in aerated lung units (48). Thus, higher PEEP could have been beneﬁcial in some patients in the ARDS Network trial but detrimental in other patients. (v) Despite the relatively large number of patients enrolled in the ARDS Network trial of higher PEEP, there were some imbalances between the study groups in baseline predictors of mortality that favored the lower PEEP study group. After statistical adjustments for these imbalances, the mortality rates for the traditional and higher PEEP study groups were still similar. However, it is possible that some imbalances in baseline variables were not adequately represented in the statistical adjustment model. Two additional trials of traditional versus higher PEEP were enrolling patients in 2004. In both of these trials, all patients were to receive volumeand-pressure limited MV, similar to the strategy used in the ARDS Network trial of traditional versus lower tidal volumes (4). These trials will provide additional, clinically valuable information regarding the potential value of lung-protective MV with higher PEEP. Overall mortality for both study groups of the ARDS Network trial of higher PEEP was 26%. This was as low or lower than the mortality rate in the volume-and-pressure limited group of the previous trial of volume-andpressure limited MV. This reinforced the clinical value of the volumeand-pressure limited approach relative to the traditional approach.

Role of Tidal Volume and PEEP in the Reduction of VILI

509

VI. Controversies A. Safe Limit for Inspiratory Plateau Pressures

Some investigators have suggested that inspiratory plateau pressures lower than 30 to 35 cmH2O are safe, and that there is no value in reducing tidal volumes in patients whose plateau pressures are below this level (3,57,58). Several considerations lead to this suggestion. First, VILI was not apparent in some experimental models when inspiratory pressures were lower than 30 cmH2O (59,60). However, lung injury was present in other experimental models when inspiratory pressures were lower than 30 cmH2O (23,25,61,62). One of the reasons for the different results between studies could be the sensitivity of the methods used to detect VILI. Also, the duration of MV differed among the studies. VILI was not detected after 20 minutes of MV with peak inspiratory pressures less than 30 cmH2O (59,60), but it was apparent after longer periods of MV (25,63). Another reason for the different results could be that in some studies there was a second experimental cause of ALI. This could predispose to VILI at airway pressures and lung volumes that could be safe in the absence of another cause of lung injury (31). Second, normal humans can inspire voluntarily to a total lung capacity at which the inspiratory plateau pressure would be approximately 35 cmH2O (64). This suggests that MV with plateau pressures as high as 35 cmH2O could be safe. However, a typical patient with ALI and/or ARDS receives 20,000 to 30,000 breaths per day. We know of no studies that indicate the safety of MV with plateau pressures consistently greater than 30 cmH2O for hours and days, as required for support of patients with ALI/ARDS. On the other hand, when previously normal sheep were induced to ventilate spontaneously with large tidal volumes for up to 12 hours, lung water increased and surfactant function decreased (65). In other experiments, VILI was apparent in previously normal sheep after 48 hours of MV with peak inspiratory pressures of 30 cmH2O (63). Third, in the ﬁve clinical trials of lung-protective ventilation strategies (Table 4), mortality was higher in the traditional study groups when mean plateau pressures exceeded 32 cmH2O, but not when they were less than 32 cmH2O. This suggested that MV with higher tidal volumes may be safe when plateau pressures are below 32 cmH2O (57,58). However, there was substantial variation in the plateau pressures in each of the study groups in all of the trials. In each of the three trials in which mean plateau pressures in the traditional group were lower than 32 cmH2O, the proportion of subjects in the traditional groups with plateau pressures greater than 32 cmH2O exceeded the proportion in the volume-and-pressure limited MV groups (Fig. 1). If 32 cmH2O was a safe threshold, then mortality should have been lower in the volume-and-pressure limited study groups. The absence of such trends suggests that the study groups were not well

510

Hager and Brower

Figure 1 Plateau pressures on day 1 after randomization in the three nonbeneﬁcial trials of traditional versus volume-and-pressure limited mechanical ventilation. Each pair of bars represents mean and two standard deviations for plateau pressures in the traditional (taller bar) and volume-and-pressure limited study groups. In each of the nonbeneﬁcial trials, the proportion of traditional group patients whose plateau pressures were greater than 32 cmH2O exceeded the proportion in the volume-andpressure limited study group. Source: From Refs. 5–7.

balanced at baseline (54), or that there were unrecognized differences in other aspects of patient management that affected clinical outcomes. To determine if a safe plateau pressure could be identiﬁed, the ARDS Network investigators constructed the relationship of mortality to plateau pressure on the ﬁrst day after enrollment in the trial of volume-andpressured limited MV (Fig. 2) (66). This relationship has a positive slope because one of the determinants of plateau pressure is respiratory system elastance, which tends to be higher in patients with more severe lung injury. The positive slope continues in the region of the relationship in which day 1 plateau pressures were lower than 32 cmH2O. If plateau pressures lower than 32 cmH2O were safe, the slope of the relationship should decrease toward zero at the lower plateau pressure levels. To further deﬁne the value of volume-and-pressure limitation in patients with relatively low plateau pressures, the ARDS Network investigators ranked the patients in the traditional and lower tidal volume study groups separately, according to plateau pressures on day 1 after randomization (67). In addition to respiratory system elastance, plateau pressure is determined by tidal volume and end-expiratory alveolar pressure. Plateau pressures were lower in the volume-and-pressure limited group primarily because tidal volumes were lower. However, PEEP was used according to the same PEEP/FiO2 table in both study groups of this trial. Therefore, the corresponding plateau pressure ranks in each study group represent corresponding strata according to respiratory system elastance. On the ﬁrst day

Role of Tidal Volume and PEEP in the Reduction of VILI

511

Figure 2 Mortality versus day 1 plateau pressure in the ARDS Network trial of traditional versus volume-and-pressure limited mechanical ventilation. Abbreviation: ARDS, acute respiratory distress syndrome. Source: From Ref. 4.

after randomization, plateau pressures were lower than 32 cmH2O (21–36) in approximately 50% of the patients in the traditional tidal volume study group. Plateau pressures were lower in the corresponding ranks in the volume-and-pressure limited study group (10–25 cmH2O), and there was a trend toward a lower mortality rate in these patients (33% vs. 26%). This effect of tidal volume reduction was similar to the effect in patients whose plateau pressures exceeded 32 cmH2O. This analysis suggested that there was a beneﬁcial effect of tidal volume reduction in patients whose plateau pressures would have been lower than 32 cmH2O while receiving the traditional MV strategy. B. Lung-Protective Effects of Auto-PEEP

Levels of PEEP necessary to maintain the arterial oxygenation goal were slightly higher in the volume-and-pressure limited study group of the ARDS Network trial of traditional versus volume-and-pressure limited MV (4). Moreover, respiratory rates were higher in the volume-and-pressure limited MV group, which could have caused greater auto-PEEP (68–70). Some investigators have suggested that the improved clinical outcomes in the volume-and-pressure limited study group were attributable to lung-protective effects of alveolar recruitment from higher levels of end-expiratory alveolar pressure (71). However, minute ventilation, which is a better predictor of auto-PEEP, was virtually the same in the two study groups of the ARDS Network trial. Some auto-PEEP may occur with the volumeand-pressure limited approach, but the magnitude of this effect is small (72). Moreover, if increases in end-expiratory alveolar pressure in the volume-and-pressure limited study group caused substantial lung

512

Hager and Brower

recruitment, it should have also improved arterial oxygenation. However, PaO2/FiO2 ratios were signiﬁcantly ‘‘lower’’ after randomization in this study group, probably because MV with lower tidal volumes and inspiratory pressures was associated with more atelectasis. This suggests that the improved clinical outcomes with volume-and-pressure limited MV occurred despite more atelectasis, not because of less atelectasis. C. Beneﬁcial vs. Adverse Effects of Respiratory Acidosis

Evidence from experimental models indicates that respiratory acidosis can attenuate ALI from oxidant stress and possibly other causes of inﬂammation, and that the beneﬁcial effects of respiratory acidosis may be negated when acidosis is buffered (73–75). Thus, acute respiratory acidosis could be beneﬁcial rather than detrimental in the clinical setting. Moreover, these experimental studies suggest that clinical management of ALI/ARDS should not include high respiratory rates or infusions of buffer solutions to prevent respiratory acidosis. However, the studies in experimental models did not monitor for adverse effects of acute respiratory acidosis, and they did not assess the balance between beneﬁcial and adverse effects on clinical outcomes. The role of intentional respiratory acidosis, with or without buffering, remains undeﬁned in the clinical management of ALI/ARDS. D. Changing Clinical Practice

Before completion of the clinical trials summarized in Table 4, there was little clinically useful evidence to guide clinicians in the difﬁcult decisions they must make to prioritize important clinical objectives (Tables 2 and 3). Therefore, usual care practices were highly variable, representing broad ranges of clinical experience and opinion. This was especially evident in a 1992 survey of intensivists’ approaches to MV in ARDS patients (76), in which usual care practices included initial tidal volumes as low as 5 mL/kg and as high as 17 mL/kg of measured body weight. This was also apparent from the broad range of tidal volumes prescribed by intensivists before their patients were enrolled in the ARDS Network trial of traditional versus volume-and-pressure limited MV in ALI/ARDS (77). The tidal volumes used in the traditional MV group of this trial were at the 80th percentile of this range of usual care tidal volumes. Therefore, one ﬁrm recommendation for changing clinical practice is that the volume-and-pressure limited approach used in the ARDS Network trial is preferable to at least 20% of usual care approaches that used tidal volumes and inspiratory pressures as high or higher than those used in the traditional MV study group. However, some investigators contend that there is insufﬁcient evidence to modify usual care practices that utilize intermediate tidal volumes

Role of Tidal Volume and PEEP in the Reduction of VILI

513

and inspiratory pressures. They contend further that the volume-andpressure limited MV strategies used in the trials conducted by the ARDS Network and Amato et al. could have been detrimental relative to the intermediate range of usual care practices (58,78). This position is based on the assumption that there is a safe plateau pressure limit of approximately 30 to 35 cmH2O. As explained previously, the considerations that lead to this assumption represent a limited view of the pertinent data. A more complete consideration of pertinent data suggests that the volume-and-pressure limited approach may reduce mortality in patients whose plateau pressures are lower than 30 cmH2O before tidal volume reduction.

VII. Summary MV is a necessary bridge to survival for most patients with ALI/ARDS. Traditional MV approaches that used generous tidal volumes were designed to achieve acceptable gas exchange, utilizing readily available measures such as arterial PaO2, partial pressure of carbon dioxide (PaCO2), and pH to guide ventilator management. Experimental models demonstrated that this approach can cause ALI from overdistention, even in previously uninjured lungs. If the lungs are acutely injured before implementation of MV, the traditional approach can exacerbate lung injury and prevent recovery from otherwise treatable or self-limited conditions. The use of a volumeand-pressure limited approach can reduce VILI from overdistention and improve clinical outcomes, including mortality, in ALI/ARDS patients. Safe upper limits for inspiratory pressures have not been established. Analyses of existing datasets suggest that such a safe upper limit is lower than had been suggested previously. Traditional MV approaches utilized modest levels of PEEP. Abundant evidence from experimental models indicates that PEEP can reduce low volume/pressure VILI, leading many investigators to recommend the use of higher levels of PEEP than were used in traditional approaches. One clinical trial of MV with traditional versus higher levels of PEEP did not demonstrate signiﬁcant beneﬁcial effects from the higher PEEP approach in patients who simultaneously received volume-and-pressure limited MV. Other clinical trials may subsequently demonstrate better results utilizing alternative approaches, or in different patient populations.

Acknowledgments Supported by NIH NHLBI Contract NOl-HR-46063.

514

Hager and Brower References

1. Colice GL. Historical perspective on the development of mechanical ventilation. In: Tobin M, ed. Principles and Practice of Mechanical Ventilation. New York: McGraw-Hill, Inc., 1994:1–35. 2. Bernard GR, Artigas A, Brigham KL, et al. The American–European Consensus Conference on ARDS. Am J Respir Crit Care Med 1994; 149:818–824. 3. Slutsky AS. Mechanical ventilation. Chest 1993; 104:1833–1859. 4. Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342:1301–1308. 5. Brochard L, Roudot-Thoraval F, Roupie E, et al. Tidal volume reduction for prevention of ventilator-induced lung injury in the acute respiratory distress syndrome. Am J Respir Crit Care Med 1998; 158:1831–1838. 6. Stewart TE, Meade MO, Cook DJ, et al. Evaluation of a ventilation strategy to prevent barotrauma in patients at high risk for acute respiratory-distress syndrome. N Engl J Med 1998; 338(6):355–361. 7. Brower RG, Shanholtz CB, Fessler HE, et al. Prospective randomized, controlled clinical trial comparing traditional vs. reduced tidal volume ventilation in ARDS patients. Crit Care Med 1999; 27:1492–1498. 8. Petty TL. Acute respiratory distress syndrome (ARDS). Dis Mon 1990; 36(1):1–58. 9. Pontoppidan H, Gefﬁn B, Lowenstein E. Acute Respiratory Failure in the Adult. 78th ed. Boston: Little, Brown, & Co., 1972. 10. Bendixen HH, Egbert LD, Hedley-Whyte J, Laver MD, Pontoppidan H. Respiratory Care. 149th ed. St. Louis: CV Mosby Co., 1965. 11. Brochard L, Lemaire F. Tidal volume, positive end-expiratory pressure, and mortality in acute respiratory distress syndrome. Crit Care Med 1999; 27(8):1661–1663. 12. Davis WB, Rennard SI, Bitterman PB, Crystal RG. Pulmonary oxygen toxicity. N Engl J Med 1983; 32(7):878–883. 13. Sackner MA, Landa J, Hirsch J, Zapata A. Pulmonary effects of oxygen breathing. Ann Intern Med 1975; 82:40–43. 14. Lodato RF. Oxygen toxicity. In: Tobin MJ, ed. Principles and Practice of Mechanical Ventilation.: McGraw-Hill, Inc., 1994:837. 15. Fracica PJ, Knapp MJ, Piantadosi A, et al. Responses of baboons to prolonged hyperoxia: physiology and qualitative pathology. J Appl Physiol 1991; 71(6): 2352–2362. 16. Shapiro BA, Cane RD, Harrison RA. Positive end-expiratory pressure therapy in adults with special reference to acute lung injury: a review of the literature and suggested clinical correlations. Crit Care Med 1984; 12:127–141. 17. Malo J, Ali J, Wood LDH. How does positive end-expiratory pressure reduce intrapulmonary shunt in canine pulmonary edema? J Appl Physiol 1984; 57:1002–1010. 18. Scharf SM. Mechanical cardiopulmonary interactions in critical care. In: Dantzker DR, Scharf SM, eds. Cardiopulmonary Critical Care. Philadelphia: W.B. Saunders Company, 1998:75–91.

Role of Tidal Volume and PEEP in the Reduction of VILI

515

19. Jardin F, Delorme G, Hardy A, Auvert B, Beauchet A, Bourdarias JP. Re-evaluation of hemodynamic consequences of positive pressure ventilation emphasis on cyclic right ventricular afterloading by mechanical lung inﬂation. Anesthesiology 1990; 72:966–970. 20. Pinsky MR. The effects of mechanical ventilation on the cardiovascular system. Crit Care Clin 1990; 6:663–678. 21. Dreyfuss D, Saumon G. State of the art: ventilator-induced lung injury; lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 22. Parker JC, Hernandez LA, Peevy KJ. Mechanisms of ventilator-induced lung injury. Crit Care Med 1993; 21(1):131–143. 23. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high pressures. Am Rev Respir Dis 1974; 110:556. 24. Dreyfuss D, Basset G, Soler P, Saumon G. Intermittent positive-pressure hyperventilation with high inﬂation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 1985; 132:880–884. 25. Frank JA, Gutierrez JA, Jones KD, Allen L, Dobbs L, Matthay MA. Low tidal volume reduces epithelial and endothelial injury in acid-injured rat lungs. Am J Respir Crit Care Med 2002; 165(2):242–249. 26. Uhlig S. Ventilation-induced lung injury and mechanotransduction: stretching it too far? Am J Physiol Lung Cell Mol Physiol 2002; 282:L892–L896. 27. Slutsky AS. Basic science in ventilator-induced lung injury. Am J Respir Crit Care Med 2001; 163:599–600. 28. Roupie E, Dambrosio M, Servillo G, et al. Titration of tidal volume and induced hypercapnia in acute respiratory distress syndrome. Am J Respir Crit Care Med 1995; 152:121–128. 29. Gattinoni L, Pesenti A, Avalli L, Ross F, Bomino M. Pressure-volume curve of total respiratory system in acute respiratory failure: computed tomographic scan study. Am Rev Respir Dis 1987; 136:730–736. 30. Gattinoni L, Pesenti A, Caspani ML, et al. The role of total static lung compliance in the management of severe ARDS unresponsive to conventional treatment. Intensive Care Med 1984; 10:121–126. 31. Dreyfuss D, Soler P, Saumon G. Mechanical ventilation-induced pulmonary edema: interaction with previous lung alterations. Am J Respir Crit Care Med 1995; 151:1568–1575. 32. Muscedere JG, Mullen JBM, Gan K, Slutsky AS. Tidal ventilation at low airway pressures can augment lung injury. Am J Respir Crit Care Med 1994; 149:1327–1334. 33. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of pulmonary elasticity. J Appl Physiol 1970; 28:596–608. 34. Martynowicz MA, Walters BJ, Hubmayr RD. Mechanisms of recruitment in oleic acid-injured lungs. J Appl Physiol 2001; 90(5):1744–1753. 35. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99(5):944–952. 36. Corbridge TC, Wood LDH, Crawford GP, Chudoba MJ, Yanos J, Sznajder JI. Adverse effects of large tidal volume and low PEEP in canine acid aspiration. Am Rev Respir Dis 1990; 142:311–315.

516

Hager and Brower

37. Hickling KG, Walsh J, Henderson S, Jackson R. Low mortality rate in adult respiratory distress syndrome using low-volume, pressure-limited ventilation with permissive hypercapnia: a prospective study. Crit Care Med 1994; 22(10):1568–1578. 38. Manley ES, Nash CB, Woodbury RA. Cardiovascular responses to severe hypercapnia of short duration. Am J Physiol 1964; 207(3):634–640. 39. Steinhart CR, Permutt S, Gurtner GH, Traystman RJ. Beta adrenergic activity and cardiovascular response to severe respiratory acidosis. Am J Physiol 1983; 244:H46–H54. 40. Tang W, Weil MH, Gazmuri RJ, Bisera J, Rackow EC. Reversible impairment of myocardial contractility due to hypercarbic acidosis in the isolated perfused rat heart. Crit Care Med 1991; 19:218. 41. Walley KR, Lewis TH, Wood LDH. Acute respiratory acidosis decreases left ventricular contractility but increases cardiac output in dogs. Circ Res 1990; 67:628–635. 42. Cheney FW, Burnham SC. Effect of ventilatory pattern on oxygenation in pulmonary edema. J Appl Physiol 1971; 31(6):909–912. 43. Benedixen HH, Hedley-Whyte J, Laver MB. Impaired oxygenation in surgical patients during general anesthesia with controlled ventilation. N Engl J Med 1963; 269:991–997. 44. Hedley-Whyte J, Laver MB, Benedixen HH. Effect of changes in tidal ventilation on physiologic shunting. Am J Physiol 1964; 206:891–897. 45. Hedley-Whyte J, Pontoppidan H, Morris MJ. The response of patients with respiratory failure and cardiopulmonary disease to different levels of constant volume ventilation. J Clin Invest 1966; 45(10):1543–1554. 46. Lachman B. Open up the lung and keep the lung open. Intensive Care Med 1992; 18:319–321. 47. Amato MBP, Barbas CSV, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338(6):347–354. 48. Rouby JX, Lu Q, Goldstein I. Selecting the right level of positive end-expiratory pressure in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2002; 165(8):1182–1186. 49. Hickling KG, Henderson SJ, Jackson R. Low mortality associated with low volume pressure limited ventilation with permissive hypercapnia in severe adult respiratory distress syndrome. Intensive Care Med 1990; 16:372–377. 50. Steinberg KP, Hudson L. Acute lung injury and acute respiratory distress syndrome. Clin Chest Med 2000; 21(3):401–417. 51. Lee PC, Helsmoortel CM, Cohn SM, Fink MP. Are low tidal volumes safe? Chest 1990; 97:430–434. 52. Anzueto A, Baughman RP, Guntupalli KK, et al. Aerosolized surfactant in adults with sepsis-induced acute respiratory distress syndrome. N Engl J Med 1996; 334(22):1417–1421. 53. Dellinger RP, Zimmerman JL, Taylor RW, et al. Effects of inhaled nitric oxide in patients with acute respiratory distress syndrome: results of a randomized phase II trial. Crit Care Med 1998; 26:15–23. 54. Amato M, Brochard L, Stewart T, Brower R. Metaanalysis of tidal volume in ARDS. Am J Respir Crit Care Med 2003; 168(5):612–613.

Role of Tidal Volume and PEEP in the Reduction of VILI

517

55. Acute Respiratory Distress Syndrome Network. Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351(4):327–336. 56. Gattinorii L, Pelosi P, Suter P, Pedoto A, Vercesi P, Lissoni A. Acute respiratory distress syndrome caused by pulmonary and extrapulmonary disease. Am J Respir Crit Care Med 1998; 158:3–11. 57. Tobin M. Culmination of an era in research on the acute respiratory distress syndrome. N Engl J Med 2000; 342:1360–1361. 58. Eichacker PQ, Gerstenberger EP, Banks SM, Cui X, Natanson C. A metaanalysis of ALI and ARDS trials testing low tidal volumes. Am J Respir Crit Care Med 2002; 166:1510–1514. 59. Parker JC, TownsIey MI, Rippe B, Taylor AE, Thigpen J. Increased microvascular permeability in dog lungs due to high peak airway pressures. J Appl Physiol 1984; 57:1809–1816. 60. Cariton DP, Cummings JJ, Scheerer RG, Poulain FR, Bland RD. Lung overexpansion increases pulmonary microvascular protein permeability in young lambs. J Appl Physiol 1990; 69(2):577–583. 61. Omlor G, Niehaus GD, Maron MB. Effect of peak inspiratory pressure on the ﬁltration coefﬁcient in the isolated perfused rat lung. J Appl Physiol 1993; 74(6):3068–3072. 62. Carr C, Ling KDE, Boulos P, Singer M. Randomised trial of safety and efﬁcacy of immediate postoperative enteral feeding in patients undergoing gastrointestinal resection. BMJ 1996; 312:869–871. 63. Tsuno K, Prato P, Kolobow T. Acute lung injury from mechanical ventilation at moderately high airway pressures. J Appl Physiol 1990; 69(3):956–961. 64. Otis AB, Fenn W, Rahn H. Mechanics of breathing in man. J Appl Physiol 1950; 2:592–607. 65. Klein S, Kinney J, Jeejeebhoy K, et al. Nutrition support in clinical practice: review of published data and recommendations for future research directions. JPEN J Parenter Enteral Nutr 1997; 21:133–156. 66. Brower R, Matthay MA, Schoenfeld DA. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials. Am J Respir Crit Care Med 2002; 166:1515–1516. 67. Brower R, Krishnan J, Thompson BT, et al. Effects of tidal volume reduction (VtR) in acute lung injury (ALI) patients with inspiratory plateau pressures (PPLAT) <32 cmH2O before VTR [abstr.]. Am J Respir Crit Care Med 2003; 167:A616. 68. Pepe PE, Marini IT. Occult positive end-expiratory pressure in mechanically ventilated patients with airﬂow obstruction—the auto-PEEP effect. Am Rev Respir Dis 1982; 126:166–170. 69. Broseghini C, Brandolese R, Poggi R, et al. Respiratory mechanics during the ﬁrst day of mechanical ventilation in patients with pulmonary edema and chronic airway obstruction. Am Rev Respir Dis 1988; 138:355–361. 70. Wright PE, Bernard GR. The role of airﬂow resistance in patients with the adult respiratory distress syndrome. Am Rev Respir Dis 1989; 139(5):1169–1174. 71. De Durante G, Del Turco M, Rustichini L, et al. ARDSNet lower tidal volume ventilatory strategy may generate intrinsic positive end-expiratory pressure in

518

72.

73. 74.

75.

76.

77.

78.

Hager and Brower patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2002; 165:1271–1274. Hough CL, Kallet RH, Ranieri VM, Rubenfeld GD, Luce JM, Hudson L. Intrinsic positive end-expiratory pressure in ARDS Network subjects. Crit Care Med 2004. Laffey JG, Engelberts D, Kavanaugh BP. Injurious effects of hypocapnic alkalosis in the isolated lung. Am J Respir Crit Care Med 2000; 162:399–405. Laffey JG, Tanaka M, Engelberts D, et al. Therapeutic hypercapnia reduces pulmonary and systemic injury following in vivo lung reperfusion. Am J Respir Crit Care Med 2000; 162:2287–2294. Sinclair SE, Kregenow DA, Lamm WJ, Starr IR, Chi EY, Hlastala MP. Hypercapnic acidosis is protective in an in vivo model of ventilator-induced lung injury. Am J Respir Crit Care Med 2002; 166(3):403–408. Carmichael LC, Dorinsky PM, Higgins SB, et al. Diagnosis and therapy of acute respiratory distress syndrome in adults: an international survey. J Crit Care 1996; 11(1):9–18. Thompson BT, Hayden D, Matthay MA, Brower R, Parsons PE. The NIH ARDS Network. Clinicians’ approaches to mechanical ventilation in acute lung injury and ARDS. Chest 2001; 120:1622–1627. Eichacker PQ, Banks SM, Natanson C. Meta-analysis of tidal volumes in ARDS (letter). Am J Respir Crit Care Med 2003; 167:798–799.

21 A Critical Review of RCTs of Tidal Volume Reduction in Patients with ARDS and Their Impact on Practice

PETER C. MINNECI and KATHERINE J. DEANS Critical Care Medicine Department, Warren G. Magnuson Clinical Center, National Institutes of Health Bethesda, Maryland, U.S.A. Department of Surgery, Massachusetts General Hospital Boston, Massachusetts, U.S.A.

STEVEN M. BANKS, CHARLES NATANSON, and PETER Q. EICHACKER Critical Care Medicine Department, Warren G. Magnuson Clinical Center, National Institutes of Health Bethesda, Maryland, U.S.A.

I. Introduction Critically ill patients requiring mechanical ventilation frequently develop the acute respiratory distress syndrome (ARDS) (1,2). Over the past 15 years, improvements in ventilatory management and other supportive care measures have led to a decreased mortality related to ARDS. However, the mortality rate in patients with ARDS remains high at 30% to 40% (3). This devastating syndrome results in decreased lung compliance, requiring ventilation with a higher airway pressure to maintain tidal volume (4). Patients with ARDS are at risk for developing further lung injury secondary to the potentially harmful effects of mechanical ventilation. Ventilator-induced lung injury (VILI) can develop secondary to increased pressure and/or volume in the alveoli (5,6). Barotrauma secondary to increased pressure within the alveoli can lead to alveolar rupture, while volutrauma secondary to increased volume within the alveoli can lead to alveolar overdistention. Animal studies of mechanical ventilation have demonstrated that increased pressure and overdistention of the alveoli lead to increases in membrane permeability, edema, and inﬂammation (7–9). These processes may then 519

520

Minneci et al.

lead to a worsening of the lung injury and deterioration of the lung function in patients with ARDS. Concerns over these harmful effects of mechanical ventilation have led to the development of lung-protective ventilation strategies, which attempt to minimize VILI by lowering the tidal volume and limiting airway pressure. Observational studies performed in the early 1990s demonstrated the potential beneﬁcial effects of ventilating ARDS patients with lower tidal volumes and airway pressures (10,11). An initial study by Hickling et al. reported on a series of 50 patients with ARDS who were managed with synchronized intermittent mandatory ventilation during which peak inspiratory pressures were reduced by limiting the tidal volume and allowing hypercapnia to develop (11). These authors reported a reduction in hospital mortality for these patients with ARDS compared to the mortality predicted by their Acute Physiology and Chronic Health Evaluation II (APACHE II) scores (16% vs. 39%, p < 0.001). In a subsequent prospective observational study, Hickling et al. reported on the treatment of 53 patients with ARDS who were managed with synchronized intermittent mandatory ventilation (SIMV) with limitation of peak inspiratory pressures to 30 to 40 cmH2O, low tidal volumes (4–7 mL/kg), and permissive hypercapnia. The hospital mortality in these ARDS patients was also lower than that predicted by their APACHE II scores (26% vs. 53%, p ¼ 0.008) (10). The historical improvement in survival in these studies provided a rationale to further investigate the effects of low tidal volume ventilation to limit airway pressure and prevent VILI in patients with ARDS. Based on increasing evidence that higher airway pressures may be harmful, recommendations were published to avoid plateau airway pressures greater than 35 cmH2O (12). In addition, intensivists began to change the patterns of mechanical ventilation in patients with ARDS. In a survey among physicians performed in 1992, nearly half of the respondents reported using lower tidal volumes (5–9 mL/kg) and more than 96% indicated that their choice of tidal volume was dependent on the level of airway pressure (13). Another survey, performed from 1994 to 2001, found that physicians caring for patients with ARDS consistently reduced the tidal volume as lung injury worsened and maintained airway pressures close to 30 cmH2O (14). A third survey of physician practice during 1998 also found that physicians caring for patients with ARDS maintained airway pressures close to 30 cmH2O (15). In addition, after adoption of ventilation strategies that limited plateau airway pressure to less than 30 cmH2O and allowed permissive hypercapnia, Jardin et al. demonstrated a signiﬁcant decrease in mortality in patients with ARDS compared to historical controls (16). Despite these changing practice patterns, randomized clinical trials testing low tidal volume ventilation strategies were warranted because of the potential adverse effects associated with hypercapnia, including acidosis, increased intracranial pressure, pulmonary hypertension, and depressed

RCTs of Tidal Volume Reduction in Patients with ARDS

521

myocardial contractility (17). To date there have been ﬁve published randomized controlled clinical trials that have examined the effects of lungprotective ventilation strategies on survival in patients with ARDS. This chapter will review these individual trials as well as the meta-analyses that have been performed on them, and subsequently describe the impact of these investigations on clinical practice.

II. Randomized, Controlled Trials of Tidal Volume Reduction in ARDS Three randomized controlled trials (RCTs) of lower tidal ventilation in patients with ARDS were published in 1998 (Table 1) (18–20). The ﬁrst was a Brazilian multicenter study of 53 patients, performed by Amato et al. from 1990 to 1995 (18). In this trial, the enrollment criteria included a lung injury score of 2.5 or higher and a pulmonary capillary wedge pressure (PCWP) less than 16 mmHg. Exclusion criteria included mechanical ventilation for more than one week, neuromuscular disease, previous barotrauma, lung biopsy or resection, uncontrollable acidosis, age more than 70 years, intracranial hypertension, and coronary insufﬁciency. All patients underwent a standardized regimen of ventilatory-hemodynamic procedures during their initial clinical evaluation, including the determination of a pressure–volume curve. Patients were subsequently randomized using sealed envelopes and a 1:1 assignment scheme. The conventional group received volume-controlled ventilation with a tidal volume of 12 mL/kg measured body weight, a target arterial pCO2 of 35 to 38 mmHg, and a goal for fractional concentration of inspired oxygen of 0.6. The protective group received pressure-limited ventilation with a tidal volume of less than 6 mL/kg measured body weight and a respiratory rate less than 30 breaths/min. Intravenous sodium bicarbonate infusions were administered if pH was less than 7.2. In the protective group, the driving pressures [Pplat positive endexpiratory pressure (PEEP)] and peak airway pressures were kept below 20 and 40 cmH2O, respectively. In addition, in the protective group, the PEEP level was preset at 2 cmH2O above the lower inﬂection point on the pressure–volume curve, and recruitment maneuvers were utilized (continuous positive airway pressure of 35 to 40 cmH2O for 40 seconds). The primary end point was 28-day survival. Secondary endpoints included survival to hospital discharge, occurrence of barotrauma, and weaning rate. There were 29 patients enrolled in the protective group and 24 in the conventional group. There were no signiﬁcant differences in the base-line characteristics of the study groups. Thirty-six hours after randomization, the traditional and protective ventilation groups had signiﬁcantly different mean tidal volumes (768 mL vs. 368 mL, p < 0.001) and mean plateau pressures (Pplat) (36.8 cmH2O vs. 30.1 cmH2O, p < 0.001).

522

Table 1 Selected Study Characteristics of the Randomized Trials of Lower Tidal Volume Ventilation in Patients with ARDS Author Amato et al. (18) Years of study Treatment strategy

Comparison group

Brochard et al. (20)

Brower et al. (21)

ARDS Network (22)

1990–1995 1995–1996 1994–1996 1994–1996 1996–1999 6–10 mL/kg titrated <8 mL/kg titrated to 6 mL/kg with <6 mL/kg titrated for <8 mL/kg with PIP <30 cmH2O keep plateau plateau pressure to plateau pressure (plateau pressure<30 cmH2O pressure <25 cmH2O PEEP) <20 cmH2O <30 cmH2O 12 mL/kg with 12 mL/kg measured 10–15 mL/kg with > 10 mL/kg with PIP 10–12 mL/kg with plateau pressure <60 cmH2O target plateau body weight PIP <50 cmH2O <50 cmH2O pressure <45– 55 cmH2O Measured Ideal Dry weight Predicted Predicted

Controls: volume control; treatment: pressure-limited

Assist control with Volume assistcontrol decelerating waveform

Volume assistcontrol

Volume assistcontrol

Minneci et al.

Method of body weight determination Mode of ventilation

Stewart et al. (19)

None

Response to acidosis

Bicarbonate if pH < 7.20

Endpoints

28-day mortality; hospital mortality, barotrauma, weaning rate

>2 hr of exposure None to PIP above 30 cmH2O Bicarbonate if Bicarbonate if pH < 7.00 pH < 7.05

None

None

Bicarbonate if pH < 7.30

Increased ventilation if pH < 7.15 60-day mortality; Hospital Hospital mortality, Hospital mortality, time to reversal of barotrauma, organ mortality; ventilator-free respiratory failure, failure, duration of barotrauma, days, organ dyspnea, agitation, ventilation, ICU organ barotrauma failure–free length of stay dysfunction, days, duration of barotrauma ventilation, ICU and hospital stay

Abbreviations: ARDS, acute respiratory distress syndrome; ICU, intensive care unit; PEEP, positive end expiratory pressure; PIP, peak inspiratory pressure.

RCTs of Tidal Volume Reduction in Patients with ARDS

Unique exclusion criteria

523

524

Minneci et al.

In comparison to the traditional group, in the protective group there was a signiﬁcant decrease in 28-day mortality (38% vs. 71%, p < 0.00l) and the rate of clinical barotrauma (7% vs. 42%, p ¼ 0.02), and a signiﬁcant improvement in weaning rates (66% vs. 29%, p ¼ 0.005). There was no difference in survival to hospital discharge between the protective and conventional groups (55% vs. 29%, p ¼ 0.37). The objectives of each ventilatory support strategy were achieved in 48 of the 53 patients. The authors of this trial concluded that the protective ventilation strategy improved the 28-day survival and weaning rates, but did not improve survival to hospital discharge. The second randomized study of lower tidal volume ventilation was a multicenter Canadian trial of 120 patients performed by Stewart et al. from 1995 to 1996 (19). Inclusion criteria were intubation for less than 24 hours, a high risk for the development of ARDS, and an initial arterial oxygen tension (PaO2) to the fraction of inspired oxygen (FiO2) ratio less than 250 at a PEEP of 5 cmH2O. All intubated patients with sepsis and burns were eligible regardless of PaO2/FiO2 ratio. Exclusion criteria included more than two hours of exposure to peak inspiratory pressures above 30 cmH2O prior to enrollment, a high likelihood of death or extubation within 48 hours, cardiogenic pulmonary edema, risk of coronary ischemia or arrhythmia, intracranial lesion, or pregnancy. Patients were randomly assigned using random number tables to a limited ventilation group or a conventional ventilation group. In the conventional group, peak inspiratory pressure was limited to less than 50 cmH2O and tidal volume was maintained between 10 and 15 mL/kg ideal body weight (IBW) [IBW ¼ 25 (height in meters)2]. In the limited ventilation group, the peak inspiratory pressure was limited to less than 30 cmH2O and tidal volume to less than 8 mL/kg IBW. Sodium bicarbonate could be administered if pH was less than 7.0. In both groups, assist-control ventilation with a decelerating waveform was used. The primary end point was in-hospital mortality. Secondary outcomes included barotrauma, multiple organ dysfunction, arrhythmia, need for dialysis, duration of mechanical ventilation, and lengths of intensive care unit (ICU) and hospital stay. Sixty patients were enrolled in each of the treatment groups. Baseline characteristics were similar between the two groups with the exception of a higher PaO2:FiO2 ratio in the conventional group compared to the limited-ventilation group (145 vs. 123). Over the ﬁrst three days after randomization, there were signiﬁcant differences between the control group and the limited-ventilation group in the mean tidal volume (10.8 mL/kg vs. 7.2 mL/kg, p < 0.001) and plateau pressures (28.5 cmH2O vs. 22.2 cmH2O, p < 0.01). There were no signiﬁcant differences between the limited ventilation group and the conventional group in mortality (50% vs. 47%), barotrauma (10% vs. 7%), organ dysfunction (2.3 vs. 2.1), arrhythmias (28% vs. 33%), duration of mechanical ventilation (16.6 days vs. 9.7 days), ICU stay (19.9 days vs. 13.7 days), or hospital stay (33.7 days vs. 27.4 days). In the limited-ventilation group compared to the

RCTs of Tidal Volume Reduction in Patients with ARDS

525

conventional group, more patients required dialysis (22% vs. 8%, p ¼ 0.04) and more patients required paralytic drugs (38% vs. 22%, p ¼ 0.05). Based on their ﬁndings, Stewart et al. concluded, ‘‘a strategy of mechanical ventilation that limits peak inspiratory pressure and tidal volume does not appear to reduce mortality and may increase morbidity’’ (19). The third randomized trial of lower tidal volume ventilation was an international multicenter trial of 116 patients performed by Brochard et al. from 1994 to 1996 (20). Inclusion criteria included diffuse bilateral inﬁltrates on the chest radiograph, mechanical ventilation with an FiO2 > 50% for at least 24 hours, and a lung injury score above 2.5 for more than 72 hours. Exclusion criteria included cardiogenic edema, PCWP > 18, presence of severe organ failure other than lung, need for high levels of vasopressors, presence of chronic disease, a moribund state, AIDS, morbid obesity, chest wall abnormalities, chest tube with air leak, bone marrow transplant, head injury, and intracranial hypertension. Patients were stratiﬁed into three groups (multiple trauma, immunosuppressive therapy, and other) and then randomized to plateau-pressure limited ventilation or standard treatment using the sealed envelope method. Patients in the standard treatment group received volume assist-control ventilation with a tidal volume greater than 10 mL/kg dry body weight (dry weight ¼ measured weightestimated weight gain from salt and water retention) and a maximum peak airway pressure of 60 cmH2O. The pressure-limited group received volume assist-control ventilation with tidal volumes between 6 and 10 mL/kg dry body weight titrated to maintain an end-inspiratory plateau airway pressure of less than 25 cmH2O. Sodium bicarbonate was recommended when pH was less than 7.05. The primary end point was 60-day mortality. Secondary end points included incidence of pneumothorax requiring chest tube placement, incidence of secondary organ system failure, duration of mechanical ventilation, and length of ICU stay. There were 58 patients enrolled in each group. There were no signiﬁcant differences in the baseline characteristics between the two groups. Twenty-four hours after randomization, there were signiﬁcant differences between the standard group and the pressure-limited group in the mean tidal volume (10.3 mL/kg vs. 7.1 mL/kg, p < 0.001) and plateau pressures (31.7 cmH2O vs. 25.7 cmH2O, p < 0.001). There were no signiﬁcant differences between the pressure-limited group and the standard group in 60-day mortality (46.6% vs. 37.9%), incidence of multiple organ failure (41% vs. 41%), incidence of pneumothorax (14% vs. 12%), duration of mechanical ventilation (23.1 days vs. 21.4 days), and ICU length of stay (33.5 days vs. 29.7 days). These authors concluded that no beneﬁt could be observed with tidal volume reduction to maintain a plateau pressure less than 25 cmH2O compared to conventional ventilation with plateau pressures less than 35 cmH2O. Brower et al. enrolled 52 patients in a randomized, controlled multicenter trial of lower tidal volume ventilation, performed in the United States

526

Minneci et al.

from 1994 to 1996 (21). Inclusion criteria were a PaO2:FiO2 ratio less than 200, bilateral radiographic inﬁltrates, no suspicion of congestive heart failure, and requirement for mechanical ventilation. Exclusion criteria included pregnancy, an acute neurological process, life expectancy less than three months, severe chronic obstructive or restrictive respiratory disease, sickle cell disease, or history of lobectomy/pneumonectomy during the current hospitalization. Patients were randomized in equal blocks to a traditional ventilation group or a small tidal volume–ventilation group. In the traditional ventilation group, the tidal volume was 10 to 12 mL/kg predicted body weight (PBW) [PBW ¼ 50 (for males) or 45.5 (for females) þ 2.3 (height in inches) 60)] with a target plateau pressure less than 45 to 55 cmH2O. In the small tidal volume group, the tidal volume was less than 8 mL/kg PBW to keep plateau pressure less than 30 cmH2O. Sodium bicarbonate could be administered when pH was less than 7.3, and had to be administered when pH was less than 7.2. In both groups, a volume assist-control ventilation mode was used. End points included mortality, time to reversal of respiratory failure, circulatory effects (ﬂuid balance, vasopressor requirements), effects on dyspnea and agitation, and effects on pulmonary gas exchange. Twenty-six patients were enrolled in each group. There were no signiﬁcant differences in the baseline characteristics between the two groups. Over the ﬁrst ﬁve days after randomization, there were signiﬁcant differences between the traditional group and the small tidal volume group in tidal volumes (10.2 mL/kg vs. 7.3 mL/kg, p < 0.001) and plateau pressures (30.6 cmH2O vs. 24.9 cmH2O, p < 0.001). There were no signiﬁcant differences between the small tidal volume group and the traditional group in PEEP requirements, oxygen requirements, ﬂuid balance, vasopressor requirements, sedative or paralytic use, ventilator days (11.3 days vs. 11.9 days), or mortality (50% vs. 46%). Brower et al. concluded that the failure to observe beneﬁcial effects with small tidal volume ventilation could be explained by a sample size that was too small, only modest differences in tidal volumes and plateau pressures between the two groups, or because reduced tidal volume ventilation was not beneﬁcial. The largest randomized trial of lower tidal volume ventilation was performed by the ARDS Network. This multicenter trial enrolled 861 patients from 1996 to 1999 (22). Inclusion criteria were patients receiving mechanical ventilation with a PaO2:FiO2 ratio less than 300, bilateral pulmonary inﬁltrates on chest radiograph, and PCWP < 18 mmHg. Exclusion criteria included meeting inclusion criteria for more than 36 hours, pregnancy, participation in another trial within 30 days, increased intracranial pressure, sickle cell disease, neuromuscular disease, severe chronic respiratory disease, severe obesity, greater than 50% chance of death within six months, greater than 30% total body surface area burns, liver disease, or bone marrow or lung transplantation. Patients were randomly assigned using a

RCTs of Tidal Volume Reduction in Patients with ARDS

527

centralized system, to either a lower tidal volume group or a traditional tidal volume group. In the traditional tidal volume group, patients received volume assist-control ventilation with tidal volumes of 12 mL/kg PBW with a plateau pressure limit of less than 50 cmH2O. In the lower tidal volume group, patients received volume assist-control ventilation with tidal volumes of 6 mL/kg PBW with a plateau pressure limit of less than 30 cmH2O. The primary end point was death before discharge home or breathing without assistance. Secondary end points were ventilator-free days, number of days without organ failure, and occurrence of barotrauma. There were 432 patients enrolled in the lower tidal volume group and 429 patients in the traditional group. The baseline characteristics between the two groups were similar with the exception of a higher minute ventilation in the lower tidal volume group (13.4 L/min vs. 12.7 L/min, p ¼ 0.01). Over the ﬁrst three days after randomization, there were signiﬁcant differences between the traditional group and the lower tidal volume group in the mean tidal volume (11.8 mL/kg vs. 6.2 mL/kg, p < 0.001) and plateau pressure (34 cmH2O vs. 26 cmH2O, p < 0.001). There was a signiﬁcant decrease in mortality in the lower tidal volume group compared to the traditional tidal volume group (31% vs. 40%, p ¼ 0.007). There were signiﬁcant differences between the lower tidal volume group and the traditional tidal volume group in the number of ventilator-free days (12 days vs. 10 days, p ¼ 0.007), and the number of days without organ failure (15 days vs. 12 days, p ¼ 0.006). There was no difference in the incidence of barotrauma (10% vs. 11%, p ¼ 0.43). The ARDS Network concluded that ‘‘in patients with acute lung injury and ARDS mechanical ventilation with a lower tidal volume than is traditionally used results in decreased mortality and increases the number of days without ventilator use’’ (22). These ﬁve trials demonstrated variable effects of lung-protective ventilation strategies on survival in patients with ARDS (Table 2). By design, randomization to the two treatment groups in each of these ﬁve trials created a difference in tidal volume and consequently plateau airway pressure. Despite this similarity, two of these trials demonstrated beneﬁt with a lungprotective ventilation strategy, and the other three did not. Several metaanalyses have now been performed to determine a basis for the differences in the results of these trials.

III. Meta-Analyses of the RCTs of Tidal Volume Reduction During ARDS An initial meta-analysis of these ﬁve trials demonstrated that there was heterogeneity among the ﬁve trials ( p ¼ 0.06, Breslow-Day test) that could be explained by separating the trials into two homogenous groups that were signiﬁcantly different from each other (p ¼ 0.017).

528

Minneci et al.

Table 2 Results of the Randomized Trials of Lower Tidal Volume Ventilation in Patients with ARDS Amato et al. (18)

Stewart et al. (19)

Brochard et al. (20)

Treatment Comparison Treatment Comparison Treatment Comparison group group group group group group Number of patients 29 24 Mean tidal volume 6.1 11.9 (mL/kg) Method of body Measured Measured weight determination Plateau pressure (cm) 30.1 36.8 112 134 PaO2:FiO2 Age (years) 33 36 Severity of illness: Scale APACHE II Score 28 27 Barotrauma 7% 42% Mechanical ventilation NR NR (days) 28-day mortality 38% 71% Hospital mortality 45% 71%

60 7.2

60 10.8

58 7.1

58 10.3

Ideal

Ideal

22.2 123 59

28.5 145 58

Dry weight 25.7 144 57

Dry weight 31.7 155 57

APACHE II 22 22 10% 7% 16.6 9.7

APACHE II 18 17 14% 12% 23.1 21.4

NR 50%

NR 47%

Brower et al. (21)

Number of patients Mean tidal volume (mL/kg) Method of body weight determination Plateau pressure (cm) PaO2:FiO2 Age (years) Severity of illness: Scale Score Barotrauma Mechanical ventilation (days) 28-day mortality Hospital mortality

NR 47%

NR 38%

ARDS Network (22)

Treatment group

Comparison group

Treatment group

Comparison group

26 7.3

26 10.2

432 6.2

429 11.8

Predicted

Predicted

Predicted

Predicted

24.9 129 47

30.6 150 50

26 138 51

34 134 52

91 4% 11.3

85 4% 11.9

81 10% 16

84 11% 18

NR 50%

NR 46%

NR 31%

NR 40%

APACHE III

APACHE III

Abbreviations: ARDS, acute respiratory distress syndrome; APACHE, acute physiology and chronic health evaluation; PaO2, arterial oxygen pressure; FiO2, fraction inspired oxygen; NR, not reported.

RCTs of Tidal Volume Reduction in Patients with ARDS

529

Figure 1 Plateau airway pressures in the randomized trials of low tidal volume ventilation. Mean plateau airway pressures (SE) over time after randomization in the high (A) and low (B) tidal volume groups of each of the ﬁve trials are represented by the clear line (gray bars). The individual mean plateau pressure values reported within each trial are represented by the closed circles. Prerandomization plateau airway pressures (mean SE) were reported in two studies and are displayed on the left side of each panel. Abbreviation: SE, standard error. Source: From Ref. 25.

One group of two ‘‘beneﬁcial’’ trials demonstrated a signiﬁcant improvement in the odds ratio of survival with the lower tidal volume ventilation (18,22). The other group of three ‘‘nonbeneﬁcial’’ trials demonstrated a nonsigniﬁcant decrease in the odds ratio of survival with a lower tidal volume ventilation (19–21). Because four different methods of setting the tidal volume were used in the ﬁve trials, differences in tidal volumes could not be used to explain the inconsistent results of the trials. However, plateau pressure, a surrogate marker for the effect of changing tidal volumes in patients with ARDS, was available in all ﬁve trials (Fig. 1). In the two beneﬁcial trials, the prerandomization tidal volumes (665 and 646 mL) and plateau pressures (29.5 and 30.3 cmH2O) in control patients were similar (18,22). In comparison to these prerandomization values, after randomization to 12 mL/kg in each of these trials, tidal volumes signiﬁcantly increased ( p < 0.001), which resulted in signiﬁcant increases in plateau pressures over the ﬁrst seven days of the study (36.3 and 34.1 cmH2O, p < 0.001). In contrast, in the three nonbeneﬁcial trials, mean tidal volumes in control patients after randomization resulted in

530

Minneci et al.

plateau pressures over the ﬁrst ﬁve to seven days of study of 31.6 cmH2O (20), 27.8 cmH2O (19), and 30.6 cmH2O (21). These plateau pressures are signiﬁcantly lower than the plateau pressures in the two beneﬁcial trials over the ﬁrst week of study (p < 0.001). Further analysis of the low tidal volume groups demonstrated similar plateau airway pressures over the ﬁrst week among all ﬁve trials [28.8 cmH2O (18) and 25.6 cmH2O (22) in the two beneﬁcial trials; 21.8 cmH2O (20), 25.1 cmH2O (19), and 24.9 cmH2O (21) in the nonbeneﬁcial trials]. Therefore, one explanation for the differing effects of low tidal volume ventilation in these trials is the use of signiﬁcantly different comparison groups. In the three nonbeneﬁcial trials, the selected tidal volumes for the control groups maintained plateau airway pressures in a range (28–32 cmH2O) that was similar to the prerandomization values in the two beneﬁcial trials. In contrast, in the two beneﬁcial trials, the selected tidal volumes for the control group resulted in signiﬁcant increases in plateau pressures compared to both the prerandomization levels in these two trials and to the levels reported in the nonbeneﬁcial trials. Thus, the higher plateau pressures in the control groups of the beneﬁcial trials may have led to increased mortality in these groups. Therefore, the reported improvement in survival with low tidal volume ventilation in these two trials may not be generalizable to current practice. In these two trials, the comparison groups were randomized to a ventilation strategy that increased the tidal volumes to a level where plateau pressures approached or exceeded 35 cmH2O. In correspondence regarding this meta-analysis, Amato et al. presented an analysis of the three ‘‘nonbeneﬁcial trials’’ of low tidal volume ventilation. They argued that the best-adjusted analysis demonstrated ‘‘that there is no evidence for harm of the lower tidal volume strategies in these studies’’ (24). However, the authors of the meta-analysis responded that ‘‘even the best adjusted 95% conﬁdence intervals suggest that there is a one in three chance that low tidal volumes produce an increase in mortality rates. Therefore this practice also has not been proven to be safe’’ (24). Recently, Drs. Petrucci and Iacovelli performed a second metaanalysis of the randomized trials of lung-protective ventilation in patients with ARDS (23). These authors expressed concerns over the possible inﬂuence of selection bias in the initial meta-analysis and thought further examination of the clinical heterogeneity of the ﬁve trials of lower tidal volume ventilation was warranted. The goals of their analysis were to determine if lung-protective ventilation strategies reduced morbidity and mortality in patients with ARDS, and to determine if the effects of these strategies were different if a plateau pressure greater than 30 to 35 cmH2O was used. Based on the authors’ goal to combine the results of these trials, the relative risk values reported are based on a ﬁxed-effects statistical model when there was minimal heterogeneity, and a random-effects statistical model when there was moderate heterogeneity.

RCTs of Tidal Volume Reduction in Patients with ARDS

531

When combining the mortality results at the end of the follow-up period for each of the trials, the authors reported moderate heterogeneity (I2 ¼ 45.9%) with a nonsigniﬁcant decrease in the relative risk of death with low tidal volume ventilation [relative risk (RR) ¼ 0.91, 95% conﬁdence interval (CI): 0.72–1.14, p ¼ not signiﬁcant (NS)]. The trials were then stratiﬁed into groups that reported comparable outcomes (i.e., hospital mortality and 28-day mortality). The three trials that reported 28-day mortality rates (18,20,22) demonstrated a consistent (I2 ¼ 0%) and overall signiﬁcant decrease in the relative risk of death with low tidal volume ventilation (RR ¼ 0.74, 95% CI: 0.61–0.88, p < 0.05). However, the four trials that provided hospital mortality data (18,19,21,22) demonstrated moderate heterogeneity (I2 ¼ 31.8%) with an overall nonsigniﬁcant decrease in the relative risk of death with low tidal volume ventilation (RR ¼ 0.84, 95% CI: 0.68– 1.05, p ¼ NS). In addition, there were four trials that provided data on the duration of mechanical ventilation (19–22). In these four trials, low tidal volume ventilation demonstrated variable effects on the duration of mechanical ventilation (I2 ¼ 33.1%) without an overall beneﬁcial effect (mean difference in duration of mechanical ventilation ¼ 0.38 ventilator days, 95% CI: 3.06 to 3.82 ventilator days, p ¼ NS). Drs. Petrucci and Iacovelli also performed a subgroup analysis in their meta-analysis to compare the treatment effects of low tidal volume ventilation in trials with a ‘‘low pressure’’ control group (plateau pressure 31 cmH2O) to trials with a ‘‘high pressure’’ control group (plateau pressure >31 cmH2O). This partition divided the ﬁve trials into the same two sets of trials as the prior meta-analysis (25). The effects of lower tidal volume ventilation were signiﬁcantly different in the trials with ‘‘high pressure’’ control groups compared to ‘‘low pressure’’ control groups ( p ¼ 0.004) (Fig. 2). In the two trials with a ‘‘high pressure’’ control group (18,22), there was a consistent (I2 ¼ 0%) and signiﬁcant decrease in the relative risk of death with lower tidal volume ventilation (RR ¼ 0.76, 95% CI: 0.64–0.91, p < 0.05). However, in the three trials with a ‘‘low pressure’’ control group (19–21), there was a consistent (I2 ¼ 0%) and nonsigniﬁcant increase in the relative risk of death with lower tidal volume ventilation (RR ¼ 1.13, 95% CI: 0.88–1.45, p ¼ NS). The authors of this meta-analysis concluded that ‘‘(the) study (by Eichacker et al.) suggests that as long as tidal volumes produce airway pressures considered safe (31 cmH2O or less), there is no beneﬁt from using lower tidal volumes. The results of subgroup analysis of our review largely conﬁrm that ﬁnding’’ (26). Finally, in a non–peer reviewed analysis of the ﬁve trials, Brower and Rubenfeld reported an overall signiﬁcant decrease in the relative risk of death with low tidal volume ventilation (27). However, this result is difﬁcult to interpret because the heterogeneity in the effect of low tidal volume ventilation in these trials ( p ¼ 0.06) suggests that the results of these trials should not be combined.

532

Minneci et al.

Figure 2 Effects of low tidal volume ventilation on survival. The relative risk of death and 95% conﬁdence interval (open triangle; solid lines) with low tidal volume ventilation in each of the ﬁve randomized clinical trials and overall for the trials with ‘‘low pressure’’ control groups (mean plateau pressure 31 cmH2O) and ‘‘high pressure’’ control groups (mean plateau pressure > 31 cmH2O) trials (23). The effects of low tidal volume ventilation on survival were signiﬁcantly different in trials with ‘‘low pressure’’ control groups compared to ‘‘high pressure’’ control groups (p ¼ 0.004) (23). In trials with ‘‘high pressure’’ control groups, low tidal volume ventilation was signiﬁcantly beneﬁcial. In contrast, in trials with ‘‘low pressure’’ control groups, low tidal volume ventilation had a nonsigniﬁcant harmful effect.

IV. Impact of the Low Tidal Volume Trials on Practice Patterns The implications of the results of these clinical trials and meta-analyses for patient care continue to be debated amongst intensivists. Experts in the ﬁeld of critical care have expressed opinions on what should be taken away from the results of these trials. Drs. Dreyfuss and Saumon state that these trials ‘‘ . . . conﬁrm that excessively high tidal volumes are unsafe. These studies do not tell us whether ARDS patients should be ventilated with a tidal volume of 6 mL/kg body weight or simply less than 12 mL/kg’’ (28). They go on to state that ‘‘ . . . the optimal ventilator strategy is still unknown. Reduction in tidal volume is mandatory, but no one knows whether reducing it to 8 to 9 mL/kg body weight is sufﬁcient or whether a goal of

RCTs of Tidal Volume Reduction in Patients with ARDS

533

Table 3 Impact of the ARDS Network Low Tidal Volume Trial on Practice Patterns

Study Sakr et al. (34) Brower et al. (32) Brower et al. (35) Young et al. (33) Weinert et al. (14)

Tidal volume preARDS Network trial (mL/kg)

Tidal volume postARDS Network trial (mL/kg)

Method of body weight determination

NR 10.4 1.9 NR 12.3 2.7 11.2a

9.9 2.4 7.9 1.9 8.0 2.0 10.6 2.4 10.1 1.9

NR PBW PBW PBW IBW

a Conﬁdence intervals not reported. Abbreviations: ARDS, acute respiratory distress syndrome; NR, not reported; PBW, predicted body weight; IBW, ideal body weight.

6 mL/kg is desirable’’ (28). Similarly, Dr. Ricard writes ‘‘no doubt that volumes >12 mL/kg are detrimental, but the unresolved question . . . is whether or not it is prudent to reduce tidal volume below 8 to 10 mL/kg. In this matter as in others, clinicians should continue to apply . . . the principle of precaution and maintain plateau pressure below 30 to 32 cmH2O rather than clinging to determination of tidal volume dictated on the basis of body weight’’ (29). Marini and Gattinoni express that ‘‘ . . . as a general rule, the desired goal is to use the least PEEP and tidal volume necessary to achieve acceptable gas exchange while avoiding tidal collapse and reopening of unstable lung units’’ (30). Since the publication of the ARDS Network low tidal volume trial, several studies have reported on the impact of the low tidal volume trials on ventilation patterns in patients with ARDS. One study from an ARDS Network center reported no increase in the percentage of patients receiving lungprotective ventilation (tidal volumes of <6 mL/kg PBW and plateau pressures 30 cmH2O) since the publication of the ARDS Network low tidal volume trial (31). Others studies, including a subsequent trial by the ARDS Network, have demonstrated that patients with ARDS are being ventilated with tidal volumes in the 8 to 10 mL/kg range (Table 3) (14,32–35). In addition, three studies comparing tidal volumes before and after the ARDS Network trial demonstrate only conservative tidal volume reductions [12.3–10.6 mL/kg PBW (33), 10.4–7.9 mL/kg PBW (32), and 11.2–10.1 mL/kg IBW (14)]. Hence, intensivists are lowering tidal volumes, but not to the level of 6 mL/kg studied by the ARDS Network investigators. V. Conclusions To date, there have been ﬁve published RCTs of lower tidal volume ventilation in patients with ARDS. These trials did not use the same treatment or

534

Minneci et al.

comparison groups and they demonstrated varying effects of low tidal volume ventilation on survival. Subsequent meta-analyses of these trials have demonstrated moderate levels of heterogeneity, which can be explained, in part, by differences in the control groups of the individual trials. The ability to titrate care in the control groups may account for the different treatment effects in these trials. Standard management of patients with ARDS usually involves titration of tidal volumes to minimize airway pressures. In the two beneﬁcial trials, the control group tidal volume was ﬁxed at a level of 12 mL/kg (18,22). In contrast, in the three nonbeneﬁcial trials, physicians could titrate the tidal volume in the control group within a given range [10–15 mL/kg (19); >10 mL/kg (20); 10–12 mL/kg (21)]. Thus, in these three trials, physicians could, and did, decrease tidal volume to minimize airway pressures [mean tidal volume, mean plateau pressure: 10.8 mL/kg, 28.2 cmH2O (19); 10.3 mL/kg, 31.7 cmH2O (20); 10.2 mL/kg, 30.6 cmH2O (21)]. Therefore, the omission of titrated care may explain the different results from these trials. Based on the currently available data, it has not been demonstrated that low tidal volume ventilation in patients with ARDS improves survival compared to current practice. It has been demonstrated that ventilation with high tidal volumes and high airway pressures is harmful. It would appear at this time that patients should continue to be managed with tidal volumes titrated to produce safe plateau pressures (< 32 cmH2O) until further clinical trials demonstrate a ventilation strategy that improves survival compared to the current standard of care. References 1. Lewandowski K. Epidemiological data challenge ARDS/ALI deﬁnition. Intensive Care Med 1999; 25(9):884–886. 2. Roupie E, Lepage E, Wysocki M, et al. Prevalence, etiologies and outcome of the acute respiratory distress syndrome among hypoxemic ventilated patients. SRLF Collaborative Group on mechanical ventilation, Societe de Reanimation de Langue Francaise. Intensive Care Med 1999; 25(9):920–929. 3. Steinberg KP, Hudson LD. Acute lung injury and acute respiratory distress syndrome. The clinical syndrome. Clin Chest Med 2000; 21(3):401–417. 4. Marini JJ. Lung mechanics in the adult respiratory distress syndrome. Recent conceptual advances and implications for management. Clin Chest Med 1990; 11(4):673–690. 5. Dreyfuss D, Soler P, Basset G, et al. High inﬂation pressure pulmonary edema. Respective effects of high airway pressure, high tidal volume, and positive endexpiratory pressure. Am Rev Respir Dis 1988; 137(5):1159–1164. 6. Parker JC, Hernandez LA, Peevy KJ. Mechanisms of ventilator-induced lung injury. Crit Care Med 1993; 21(1):131–143. 7. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157(1):294–323.

RCTs of Tidal Volume Reduction in Patients with ARDS

535

8. Corbridge TC, Wood LD, Crawford GP, et al. Adverse effects of large tidal volume and low PEEP in canine acid aspiration. Am Rev Respir Dis 1990; 142(2):311–315. 9. Webb HH, Tierney DF. Experimental pulmonary edema due to intermittent positive pressure ventilation with high inﬂation pressures. Protection by positive end-expiratory pressure. Am Rev Respir Dis 1974; 110(5):556–565. 10. Hickling KG, Walsh J, Henderson S, et al. Low mortality rate in adult respiratory distress syndrome using low-volume, pressure-limited ventilation with permissive hypercapnia: a prospective study. Crit Care Med 1994; 22(10): 1568–1578. 11. Hickling KG, Henderson SJ, Jackson R. Low mortality associated with low volume pressure limited ventilation with permissive hypercapnia in severe adult respiratory distress syndrome. Intensive Care Med 1990; 16(6):372–377. 12. Slutsky AS. Mechanical ventilation. American College of Chest Physicians’ Consensus Conference. Chest 1993; 104(6):1833–1859. 13. Carmichael LC, Dorinsky PM, Higgins SB, et al. Diagnosis and therapy of acute respiratory distress syndrome in adults: an international survey. J Crit Care 1996; 11(1):9–18. 14. Weinert CR, Gross CR, Marinelli WA. Impact of randomized trial results, on acute lung injury ventilator therapy in teaching hospitals. Am J Respir Crit Care Med 2003; 167(10):1304–1309. 15. Esteban A, Anzueto A, Frutos F, et al. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study. JAMA 2002; 287(3):345–355. 16. Jardin F, Fellahi JL, Beauchet A, et al. Improved prognosis of acute respiratory distress syndrome 15 years on. Intensive Care Med 1999; 25(9):936–941. 17. Feihl F, Perret C. Permissive hypercapnia. How permissive should we be? Am J Respir Crit Care Med 1994; 150(6 Pt 1):1722–1737. 18. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338(6):347–354. 19. Stewart TE, Meade MO, Cook DJ, et al. Evaluation of a ventilation strategy to prevent barotrauma in patients at high risk for acute respiratory distress syndrome. Pressure- and Volume-Limited Ventilation Strategy Group. N Engl J Med 1998; 338(6):355–361. 20. Brochard L, Roudot-Thoraval F, Roupie E, et al. Tidal volume reduction for prevention of ventilator-induced lung injury in acute respiratory distress syndrome. The Multicenter Trail Group on tidal volume reduction in ARDS. Am J Respir Crit Care Med 1998; 158(6):1831–1838. 21. Brower RG, Shanholtz CB, Fessler HE, et al. Prospective, randomized, controlled clinical trial comparing traditional versus reduced tidal volume ventilation in acute respiratory distress syndrome patients. Crit Care Med 1999; 27(8):1492–1498. 22. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342(18): 1301–1308.

536

Minneci et al.

23. Petrucci N, Iacovelli W. Ventilation with smaller tidal volumes: a quantitative systematic review of randomized controlled trials. Anesth Analg 2004; 99(1):193–200. 24. Amato M, Brochard L, Stewart T, et al. Metaanalysis of tidal volume in ARDS. Am J Respir Crit Care Med 2003; 168(5):612; author reply 612–613. 25. Eichacker PQ, Gerstenberger EP, Banks SM, et al. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials testing low tidal volumes. Am J Respir Crit Care Med 2002; 166(11):1510–1514. 26. Petrucci N, Iacovelli W. Ventilation with lower tidal volumes versus traditional tidal volumes in adults for acute lung injury and acute respiratory distress syndrome. Cochrane Database Syst Rev 2003(3):CD003844. 27. Brower RG, Rubenfeld GD. Lung-protective ventilation strategies in acute lung injury. Crit Care Med 2003; 31(suppl 4):S312–S316. 28. Dreyfuss D, Saumon G. Evidence-based medicine or fuzzy logic: what is best for ARDS management? Intensive Care Med 2002; 28(3):230–234. 29. Ricard JD. Are we really reducing tidal volume and should we? Am J Respir Crit Care Med 2003; 167(10):1297–1298. 30. Marini JJ, Gattinoni L. Ventilatory management of acute respiratory distress syndrome: a consensus of two. Crit Care Med 2004; 32(1):250–255. 31. Rubenfeld GD, Caldwell E, Hudson LD. Publication of study results does not increase use of lung protective ventilation in patients with acute lung injury. Am J Respir Crit Care Med 2001; 163(5):A295. 32. Brower RG, Thompson BT, Ancukiewicz M, et al. Clinical trial of mechanical ventilation with traditional versus lower tidal volumes in acute lung injury and acute respiratory distress syndrome: effect on physicians’ practices. Am J Respir Crit Care Med 2004; 169(7):A256. 33. Young MP, Manning HL, Wilson DL, et al. Ventilation of patients with acute lung injury and acute respiratory distress syndrome: has new evidence changed clinical practice? Crit Care Med 2004; 32(6):1260–1265. 34. Sakr Y, Vincent JL, Le Gall JR, et al. High tidal volume and positive ﬂuid balance in acute lung injury are associated with worse outcome. Chest 2003; 124:180Sa. 35. Brower RG, Lanken PN, MacIntyre N, et al. Higher versus lower positive endexpiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351(4):327–336.

22 The Importance of Protocol-Directed Patient Management for Research on Lung-Protective Ventilation

ALAN H. MORRIS Pulmonary and Critical Care Divisions, Department of Medicine, LDS Hospital and University of Utah School of Medicine Salt Lake City, Utah, U.S.A.

I. Introduction Nothing in this chapter should be interpreted as a condemnation of clinicians or clinical practice. I have unlimited respect for clinicians who I believe do a remarkably good job under trying circumstances. My focus in the following comments is on methods that could enhance the performance of clinicians and clinical investigators. Unnecessary variation in clinical practice was formally brought to the health care community’s attention in the 1970s (1). Unnecessary variation in clinical care appears to be an unavoidable feature of modern medicine (1–3) and has likely played a role in our failure to resolve multiple important problems in critical care. Many clinicians think variability is desirable because of the importance of individualizing treatment to a patient’s speciﬁc needs. While this is an intuitively attractive argument, it incorporates two assumptions. First, we assume that clinicians can effectively tailor treatment, particularly when reliable evidence for preferable therapies is absent, and that the resultant nonuniformity of treatment decisions is desirable at a 537

538

Morris

community-wide scale. Clinicians can, of course, tailor treatment to individual patients successfully. However, they can also fail to deal correctly with the individualized needs of patients and can thereby cause harm (4–18). Clinicians’ decision-making is limited (9,18,19) and clinicians cannot easily predict who will respond to a speciﬁc intervention (14,15). For many, if not most, medical interventions, the medical community and the community of patients can only draw conclusions about the balance between potential good and potential harm through examination of the results of systematic investigations. Second, we assume that nonuniformity is itself desirable because it fosters insight and innovation. It is true that many advances in medicine have been introduced through observations by bright and clever clinicians. However, the many questions addressed in modern medicine frequently involve small improvements (odds ratios of 3 or less) that will escape the attention of most observers if not examined within systematic studies (20). This evokes the success of those who recognized the importance of standardization of processes in nonmedical domains as a means of stabilizing systems so that small improvements could be observed when the system was changed (21–25). Medical workers in the clinical process improvement movement have adopted this approach successfully (10,26–29). Some opponents might argue that standardization is ﬁne in cases where we have compelling evidence to standardize (lung-protective ventilation, insulin, sedation). However, studies indicate that clinicians do not adopt such compelling evidence quickly or uniformly (30,31). Opponents might also argue about clinical domains in which we do not have compelling evidence, such as intravenous (IV) ﬂuid management in patients with acute lung injury. Is standardization of clinician decisions under such clinical uncertainty a good idea? The health care quality improvement success with stabilization of process through standardization of decisions indicates that the response to this question should be ‘‘yes.’’ Without standardization, our chances of detecting promising elements of clinical management are reduced and frequently low. In general, variability is fostered by incorrect perceptions (32–35) and is associated with unwanted and widespread error (6,7,11,36–40). The mismatch between human decision-making ability (9,18,19) and the excess information clinicians routinely encounter probably contributes to both the variability of performance and to the high error rate of clinical decisionmakers (41–52). The recent U.S. Institute of Medicine (National Academy of Science) publications have brought to wide notice the long known importance of medical error and its deleterious impact on patient outcome (6,7). Guidelines and protocols can reduce variation and increase compliance with evidence-based interventions, can effectively support clinical decision-making (53), and can inﬂuence clinician performance and patient outcome favorably (54–59). They likely reduce error (60), but this has not been formally studied. Simple protocols such as physician reminders for

Protocol-Directed Patient Management

539

a serum potassium measurement when a diuretic is prescribed are commonly employed (61). Similar protocols for serum potassium evaluation can be stratiﬁed for renal function level and have an intuitive appeal to clinicians, in part because of their simplicity. More complex protocols have the same potential to aid clinicians and reduce error, but they are more difﬁcult to comprehend and can appear abstruse and intimidating. Such complex protocols are discussed below and include those for mechanical ventilation, IV ﬂuid and hemodynamic support, and blood glucose control with insulin. Justiﬁable clinical interventions must produce more good than harm. This requires evidence of efﬁcacy, effectiveness, and safety. Evidence-based medicine is an outgrowth of efforts to meet these requirements. Unfortunately, much of clinical care does not meet this evidence-based standard (5). This includes intensive care unit (ICU) delivery of mechanical ventilation and IV ﬂuid therapy, both of which are characterized by large variations (62–64). The history of medicine is littered with therapies once widely disseminated and enthusiastically embraced by experts, only to be abandoned when subjected to systematic evaluation and demonstrated to be harmful (65). The Cardiac Arrhythmia Suppression Trial (CAST) and the Ischemic Optic Neuropathy Decompression Trial provide examples of commonly delivered therapies that proved to be harmful or useless after systematic evaluation (66–69). Medicine, like social science, likely enjoys an ‘‘ecology of science . . . in which there are available many more wrong responses than correct ones . . . ’’ (70). In complex clinical circumstances, clinician decision-making when unsupported by outcome data is likely to be both variable and incorrect. Because patients are nonlinear complex biologic systems (71–73), one frequently cannot easily anticipate the changes that might occur following a decision. Furthermore, the number of combinations of the complex variables involved is staggering (71). Human error and injury are inevitable (44,52,74,75). Clinical error rates vary from about 1% to 50% (5,6,44,45,47–52,76–93). This is due in part to the inaccuracy with which even well trained and highly skilled physicians perceive physiologic data while making clinical decisions (94). The use of ill-deﬁned terms or statements, such as ‘‘ . . . caution should be exercised when pulmonary artery balloon occlusion pressure (PAOP) becomes increased to the extent that pulmonary edema is a risk’’ (95,96) must contribute to this inaccuracy. Patients can be harmed when clinicians do not comply with standard practice (6,7,97–99). Errors are common (6,74). Errors with antibiotic administration (100,101) include prescribing the wrong drug or wrong dose (102), failing to correct the dose for renal failure, failing to comply with ICU admission policies, and failing to use recommended deep venous thrombosis (DVT) prophylaxis. Sites of error documentation include hospital wards (103), surgical units (92), and critical care units (44,45,48,49,51,93). Critical care errors include failure to use

540

Morris

recommended tidal volumes for mechanical ventilation of acute respiratory distress syndrome (ARDS) patients (30,31). Even when error rates in a carefully managed academic ICU were only 1%, every patient was subjected to an error that constituted a major threat to life or limb every other day (93). Persons in this ICU performing correctly 99% of the time cannot improve much with education programs. Much more improvement in performance can be realized through systems approaches (21–23,25,74). The reduction in quality and safety of care associated with clinical error is a major concern for the health care community (6,7). Evidence-based therapy protocols are decision-support tools that likely can reduce clinical error (60). If clinicians wish to pursue the best evidence, accurate literature search techniques for systematic reviews (evidence-based information) are available (104). However, compliance of physicians with evidence-based treatments or guidelines is low across a broad range of health care topics (101,105–109) and persists even when guidelines based on reputable evidence are available (54,110). Many factors, including cultural issues and health beliefs inﬂuence compliance (111,112). Widespread distribution of evidence-based guidelines (113,114) and education programs (115–119) have had only limited impact on this low compliance. On a more positive note, both paper-based and computerized decision-support tools that provide explicit, point-of-care (point-of-decision-making) instructions to clinicians have overcome many of these problems and have achieved clinician compliance rates of 90% to 95% (35,58,120). Our understanding of clinical management of ARDS has not kept pace with our understanding of the mechanisms of ventilator-associated lung injury. This is also true of the closely related problem of sepsis. Although occasional successes are highly touted (121), many promising therapeutic agents have failed to be established as therapeutic advances (122–127). Among these are cyclooxygenase inhibitors (corticosteroids and ibuprofen), a platelet-activating factor antagonist (BN 52021), an antioxidant (N-acetylcysteine), an opiate antagonist (naloxone), a bradykinin antagonist(CP-0127), a cyclic-guanosine monophosphate stimulant (inhaled NO), antiendotoxins (E5 and HA1A), anticytokines [interleukin-1 receptor antagonist and anti-tumor necrosis factor (TNF)], and extracorporeal gas exchange (low-frequency positive pressure ventilation-extracorporeal CO2 removal). The absence of a clear beneﬁt from this broad spectrum of tested interventions leads to a compelling question. Are these clinical problems insoluble, have the needed interventions not yet been tested, or is our clinical investigative strategy ﬂawed? The signiﬁcant reduction in mortality of ARDS patients between the 1970s (128,129) and recent years (120,130) suggests that these clinical problems are, at least in part, soluble and that effective interventions currently do exist. I develop the argument that our clinical investigative strategy is seriously ﬂawed. I explore the reasons for the absence of compelling clinical outcome data

Protocol-Directed Patient Management

541

(131), and I propose a solution based on the use of bedside computerized protocols that aid clinical decision-makers (3). I use examples from other scientiﬁc domains because of the importance of interdisciplinary communication and collaboration in science (132) and to support the argument that clinical experimental requirements are not different from those in other scientiﬁc domains. Currently conducted clinical trials, especially nonblinded trials, have serious limitations. This may explain in part why many critical care clinical trials have failed to produce evidence of clinical beneﬁt in spite of large investments of resources (133). The disappointingly low quality of critical care clinical trials (124,127,134) could, in part, be due to the widespread use of suboptimal methods. Meta-analyses cannot overcome this low clinical trial quality because meta-analyses can only generate credible conclusions if the analyzed clinical trial results are credible (135,136). Meta-analyses focus on methodology at the trial design scale (e.g., were true randomization and effective blinding employed) but do not deal with the methodologic details of the patient–clinician encounter for either outpatient (137) or critical care (124,127,134) clinical trials. The medical community is thus challenged to develop new approaches to experimental clinical trials and to use them to produce more rigorous clinical experiments and results. Unfortunately, the medical community lacks the tools that might aid clinicians in making more consistent and appropriate decisions and thus produce more rigorous clinical trials. These deﬁciencies contribute to unnecessary variation in clinical care (7,60). They are barriers to both the consistent delivery of safe and high-quality clinical care and the conduct of rigorous clinical research at the holistic clinical trial scale (34). These deﬁciencies therefore limit clinical research on lung-protective ventilation and impede the implementation of changes requested by the National Institutes of Health (NIH) in its Roadmap program (138). II. Experimental Scientiﬁc Principles A. Aim of Science

The principal aim of all science is generally considered to be the ordering of the complex appearances detected by our senses or by instruments that extend our senses (139). Scientists explore the world around us to explain its behavior. Scientists extend their senses in this endeavor by using instruments that enable observers to detect otherwise hidden behaviors. Among these instruments one must include statistical analytic tools. Experimental Replicability

New scientiﬁc advances that seem to correctly represent the behavior of the world eventually become incorporated in the body of knowledge of the

542

Morris

discipline and appear in reference works such as textbooks. The key requirement, here, is the phrase ‘‘correctly representing the behavior of the world.’’ It does not appear possible to ever know with absolute certainty that this requirement has been fulﬁlled (73,140) and many philosophers including Kant have explained and explored this impossibility. Nevertheless, the lack of absolute certainty does not prevent application of new knowledge and of the advances that follow therefrom (73,140). Rothman and Greenland have nicely summarized the philosophical contributions of many including Popper who has clariﬁed some fundamental aspects of the scientiﬁc method (140). Human experimental (clinical trial) results are analogous to the probabilistic predictions of quantum mechanics. The Schro¨dinger wave function does not predict the exact position (outcome) of a particle but rather gives its location as a probability distribution that predicts the pattern of outcomes in repeated measurements [p. 198 in Ref. (141)]. Replicability of experimental results is the fundamental criterion by which general acceptance of new knowledge is gained in scientiﬁc circles (70, p. 196 in 141, 142–147). Scientists generally believe that an observation that accurately reﬂects the way the world behaves should be replicable by other investigators. Both in physical science (139,141) and in social science (70,148), two scientiﬁc domains that bracket the scale of clinical trials, the requirements for scientiﬁc rigor and for experimental replicability, are well accepted. Actual or potential replicability of results is a basic requirement of all rigorous scientiﬁc investigation, regardless of scale (70,139,148–154). Replication of an experimental result requires, of course, a detailed knowledge of the experimental method. This is a major challenge for the holistic human experimentation in clinical trials for two reasons. First and foremost, most clinical trials are not conducted with adequately explicit methods (see Section ‘‘Adequately Explicit Methodology: Protocols vs. Guidelines’’). Second, editorial policies severely restrict the methodologic detail in medical publications. Ethical Considerations for Replicability

Some clinicians with strongly held opinions have raised ethical arguments against several features of clinical research. These include arguments for scientiﬁcally questionable ‘‘rescue therapy’’ in clinical trials, arguments against or resistance to randomization of subjects, or arguments against adherence to protocol rules. These arguments lack merit. Investigators have two responsibilities to their experimental subjects: Their ﬁrst responsibility is to maximize the quality of clinical care delivered to the subjects. Their second important responsibility is to conduct credible clinical research. Reasonable people should not be expected to agree to participate as experimental subjects if the experiments are unlikely to lead to credible results. Investigators must seriously consider both of these responsibilities in the

Protocol-Directed Patient Management

543

resolution of perceived conﬂicts between the delivery of the best clinical care and the achievement of the most credible clinical trial result. The complexity of ethical issues and the conﬂict between these two investigator responsibilities are brought into striking relief when treatments being evaluated in the intervention group of a randomized clinical trial are used as ‘‘rescue therapy’’ for deteriorating control group subjects (‘‘rescue therapy’’ is a euphemism for ‘‘desperation therapy,’’ a more accurate but repellant appellation not likely to garner approval in the informed consent process). Such use of unproved treatments may violate study protocols. They may also be included within study protocols. In either case, ‘‘rescue therapy’’ use can destroy clinical trial result credibility. Administration of a study intervention (an unproved treatment) as ‘‘rescue therapy’’ requires the presumption that the intervention is efﬁcacious before the results of the trial are known. This is logically difﬁcult to defend. If the intervention is being evaluated because its effect is unknown, how can the intervention be offered as a beneﬁt to a failing subject? For example, a clinical trial of extracorporeal membrane oxygenation (ECMO) allowed the application of ECMO, the test intervention, as ‘‘rescue therapy’’ when control subjects seemed without hope of recovery and satisﬁed what were termed de jure death criteria (128). In the ECMO trial, conditions that deﬁned ‘‘control failures’’ and allowed the use of ECMO as rescue therapy for control patients were PaO2 < 45 mmHg for more than 12 hours with FiO2 ¼ 1.0 and with maximum tolerated positive end-expiratory pressure (PEEP), or PaO2 < 35 mmHg for more than six hours (128,129). Since then, it has become clear that such predictions are quite uncertain. Patients who met these criteria have survived in the control treatment arm of a more recent randomized controlled clinical trial of extracorporeal therapy (122). However, ECMO was a dangerous and unproved treatment with uncertain clinical effects. If ECMO were given as ‘‘rescue therapy’’ to a failing control patient, it would violate principles of sound experimental design by introducing a crossover effect (20,149). This violation of experimental design principles would risk the sacriﬁcing of one of the investigator’s two obligations to the subject (the obligation to produce credible clinical trial results) for the other (the obligation to maximize the quality of clinical care). This sacriﬁce would be unquestioned and unassailable if it involved the administration of therapy known to be beneﬁcial. However, it should be seriously questioned when it involves a potentially harmful intervention of unproved effect. Sacriﬁce of the credibility of the clinical trial results for a questionable and unproved anticipated beneﬁt for the subject could be interpreted as exploitation of enrolled clinical trial subjects and therefore as a violation of the requirements of ethical research (146). This sacriﬁce appears to be too easily accepted by the clinical trial community, both by investigators delivering ‘‘rescue therapy’’ and by clinicians either providing clinical care in violation of protocol rules or

544

Morris

withdrawing subjects from clinical trials. In so doing, the clinical trial community risks exploiting subjects. It ultimately compromises the clinical care quality available to the health care community at large and thereby reduces societal beneﬁt. Interestingly, this ethical implication of such questionable clinical care and clinical experimentation does not ﬁgure prominently in discussions of error, even though the reduction in quality and safety of care associated with clinical error is a major community concern (6,7). Clinician Resistance to Adequately Explicit Methods

Some clinicians claim that anything short of total freedom to make clinical decisions threatens the traditional expert approach to making and individualizing clinical decisions. Two paradigms of clinical decision-making— the expert or authoritarian paradigm and the actuarial or numerical paradigm—have been the subjects of an historical debate in medicine (155,156). Medicine has, since ancient times, traditionally employed the expert or authoritarian paradigm of decision-making. This relies on expert clinical judgment. The ‘‘best decision’’ for the individual patient at a particular time in this paradigm is based upon the expert’s background, experience, and training (clinician expertise). This clinician expertise has been described as intuitive, and therefore difﬁcult to articulate and describe. In addition, it is also likely not reproducible because the expert’s background and experience are always changing. Clinicians may challenge an adequately explicit decision-support tool such as a computerized protocol by asking how they can be assured the rules are ‘‘right.’’ This is, however, an inappropriate question. We generally do not know what is right in any absolute sense. We only know, when we have good evidence, what works better than other approaches. All that should be required of a decision-support tool to justify its use or evaluation is that it be both a reasonable strategy of response and that it be safe. Generating the rules of such a decisionsupport tool requires effort and it requires attention to disagreements between clinicians (35). First, the best evidence is extracted from the literature. Next, domain experts contribute rules for which no published data exist. Finally, a consensus among the clinicians, followed by an iterative reﬁnement process leads to robust protocol (35). Adequately explicit methods are those that elicit the same clinical decision from different clinicians when they are faced with the same clinical information (see Section ‘‘Experimental Group Equation’’). In complex clinical environments such as those in critical care, adequately explicit methods require sufﬁciently detailed protocols that provide clinician decision support (34,131). The use of decision-support tools such as adequately explicit computerized protocols incorporates an actuarial paradigm, one based on numerical analysis and on outcome data (157). The tension between these two paradigms in the clinical community dates back to at least the early 19th century (34,158). Then, it involved important clinicians

Protocol-Directed Patient Management

545

and scientists, for example, Poisson and Pinel (155,156). However, these two approaches to clinical decision-making are complementary, rather than mutually exclusive. Notably, the bedside clinician is always in control when adequately explicit methods are used in an open-loop servocontrol manner (159). The protocols in the Utah Clinical Trial Toolbox generate adequately explicit instructions (Table 1) and lead different clinicians to the same action. We employ open-loop servocontrol with clinicians always examining the computerized protocol instructions before they are carried out. This forces the clinician to examine the patient and to evaluate if the patient still belongs to the set of subjects for which the protocol rules were developed [an evaluation of external validity or generalizability (153,160)]. The clinician always has the opportunity to judge whether the patient has changed and no longer belongs to the group for which the protocol was intended. This external validity–directed clinician judgment is appropriate. However, the tendency to reject a validated protocol instruction because of a clinician’s opinion (an opinion often not founded on evidence) is frequently inappropriate and can threaten the internal validity of clinical trials (160). This error is fostered by the well-recognized overconﬁdence of physicians in the correctness of their beliefs and opinions (32,33,161). Computerization forces attention to detail at a level not humanly possible with paper-based protocol development alone (34,35). Consequently, the depth of understanding of clinician decision-making is strikingly increased among the developers. The current Utah Clinical Trial Toolbox electronic tools include bedside screens for manual clinician data entry or for automatic data capture from electronic medical records (EMRs) (see section ‘‘Current Utah Clinical Trial Toolbox Electronic Tools’’). The use of computerized protocols to standardize clinical decision-making in complex clinical settings has a sound ethical foundation (see above) (162,163). There seems to be a widespread reluctance to accept rules for standardizing clinician decisions. (Many of us may be willing to acknowledge that explicit methods may produce better outcomes for ‘‘less expert’’ physicians,

Table 1 One Iteration (Protocol Run) of Computerized Protocol Instructions for Mechanical Ventilation Arterial oxygenation instructions Reduce inspired O2 by 10% from 90% to 80% Reassess oxygenation in 15 min Ventilation and arterial pH instructions Maintain tidal volume at 540 mL Increase ventilatory rate by three breaths from 22 to 25 per minute Sample arterial blood in 15 min at 15:40 hrs

546

Morris

but not for ourselves.) Some authorities conclude that control with computerized protocols will not be widely distributable (160). Their association of detailed protocols with ‘‘complexity and rigidity’’ is a common expression of resistance to use of adequately explicit methods (160). It suggests acceptance of the expert paradigm as the ultimate reference for proper medical response (155,156). More importantly, it suggests an association of decision-support tools with ‘‘cookbook’’ (patient invariant) care. This is in fact an error, as described below (see section ‘‘Individualized, PatientSpeciﬁc Instructions’’). The reluctance of clinicians to accept rules (decision-support) for standardizing clinician decisions is an interesting challenge to a larger view in science that regards reductions of the possible number of system states to be the foundation of stability in natural systems. Both at the quantum level of atomic behavior [pp. 198–199 in Ref. (141)] and at the system statistical process control level (21–25), restriction of the innumerable possible states by application of rules results in stabilization of the system. Quantum restrictions provide the foundation for atomic and molecular stability and thus are fundamental to the generations of biologically important molecules [pp. 198–199 in Ref. (141)]. It seems ironic that, given the repeated criticism of unnecessary clinical variation (1,2) and its harmful consequences in medicine (6), many clinicians continue to defend the ultimate decisionmaking freedom (and reject efforts to standardize decisions) that contributes to this variation. Clinicians may forget to do intended actions and may make different decisions when faced with identical problems. Clinicians frequently seem to believe that the complexity of clinical problems cannot be captured by man-made sets of rules. This has an intuitive appeal, because clinicians understand how overwhelming clinical information can be. However, complex outcomes can be produced by simple rules. This appears to be the case in nature, when viewed from a cosmologist’s perspective [pp. 235–236 in Ref. (141)]. A similar relationship between complex and patient-speciﬁc treatment strategies and an underlying simple set of protocol rules has been observed for adequately explicit computerized critical care protocols (34,35,60,131,164,165). This reluctance to embrace adequately explicit methods is made more puzzling by the evidence of favorable beneﬁt following their use.

Impact of Protocols

The impact of protocols including computer-based clinical decisionsupport systems on cost and efﬁcacy appears to be favorable (56,57,166). Standardization of aspects of the clinical encounter appears to avoid vexing problems in many areas of clinical medicine (167,168). Computerized protocols have favorably impacted hospital pharmacy and infectious disease departments (100,101,169–171). Both outpatient and inpatient computerized

Protocol-Directed Patient Management

547

protocol use has favorable consequences (43,77,100,157,172–182). Computerized protocols for mechanical ventilation have controlled the intensity of care of patients with ARDS in both treatment arms of a randomized clinical trial (122). Three beneﬁts follow the use of such speciﬁc computerized protocols: (i) precise description of the method (process) of patient care (the rules and logic for clinical decision making); (ii) assurance of equal intensity of care (experimental group equation or equivalence); and (iii) common intermediate (surrogate) end points (e.g., therapy regulated to produce the same PaO2 and pHa). Because management of mechanical ventilation in severe ARDS is perceived as a complicated and intellectually demanding process, it is likely that many other facets of critical care can be successfully addressed with computerized protocols. Patients with ARDS supported with computerized protocols experienced a higher survival than expected from historical control data (122). The impact of computerized protocols on patient outcome is favorable, when compared to the outcome of patients with ARDS treated without protocols (58). Limited data support a causal association between a computerized protocol and more favorable patient outcome, when compared with a paper-based protocol using the same decision-support rules for mechanical ventilation weaning (183). Other randomized clinical trials, using less detailed and manually applied paper-based protocols, have demonstrated clearly that protocol-guided care favorably affects the outcome of patients with thromboembolic disease (184–186). B. Conceptual Paradigms

Clinicians and clinical investigators think about problems in ways that depend on their perceptions of experiences and conceptual frames of reference. On an individual scale, these perceptions set the bounds for thinking and for posing questions. These perceptions are inﬂuenced by prevailing conceptual paradigms. Thomas Kuhn has argued that paradigms are a prerequisite for any science. He proposed that science is impossible in the absence of such paradigms (187). New paradigms introduce rapid changes in human intellectual behavior. This is analogous to both punctuated equilibrium in evolutionary biology contemplated by paleobiologists (188) and to changes in the basic forces of nature contemplated by cosmologists [p. 264 in Ref. (141)]. One of the most profound conceptual (paradigm) changes occurred as the middle ages were ending. Western societies adopted the numerical scheme introduced by Hindu mathematicians in India and brought to Europe through Spain by Arabic mathematicians. This decimal system that incorporated both the zero cipher and the concept of nullity has been described as ‘‘ . . . one of the greatest discoveries that humanity has ever made (141).’’ It enabled major advances in thinking, calculation, and

548

Morris

recording. This, interestingly, is claimed to have led to separation of theological and scientiﬁc scholarship and to have opened the door to careful observation and experimentation (141). These advances fostered unparalleled developments in science and engineering with incalculable consequences on worldwide civilization. On a professional community–wide scale, medical paradigms inﬂuence medical thinking, medical questions, and medical investigations. Galen’s patently groundless view of the body held sway for 1400 years before William Harvey revolutionized medical thinking when he published ‘‘Exercitatio Anatomica de Motu Cordis et Sanguinis in Animalibus’’ in 1628. Several anatomists had for two centuries failed to ﬁnd the interventricular septal communications demanded by Galen’s theory. The Galenic view dominated so long, not because it was intellectually compelling, but, some argue, because there were no alternative conceptual paradigms available for organizing information about the world (189). Sir Isaac Newton, following in Sir Francis Bacon’s footsteps, advanced a reductionist conceptual paradigm of science (71,190). This paradigm proposed that the behavior of complex systems could be constructed and fully understood from an understanding of the behavior of its component parts. This reductionist paradigm was fostered by Cartesian dualism (190). More recently, science has embraced a different and contrasting conceptual paradigm of ‘‘emergent properties’’ (71,72,189–200). In this recent conceptual paradigm, new properties emerge as parts or subsystems aggregate to form more complex systems. Consequently, laws or regularities that emerge at a higher scale cannot be expected to appear at lower scales. For example, the behavior of large and complex aggregates of elementary particles such as electrons cannot be understood in terms of simple extrapolation of the properties of a few particles. At each scale of complexity, entirely new properties appear [Philip Anderson, quoted in Ref. (192)]. Common sense supports this notion if one thinks of the behavior of a classroom of ﬁve year olds, or of a group of teenagers. The emergence, with increasing scale, of new properties important to medicine is common (192,193,199,200). Medically pertinent examples abound. The behavior of two bacterial species in a bacterial bioﬁlm cannot be predicted from an understanding of each of the bacteria independent of the other (195). Cardiac or circulatory function cannot be understood in terms of the heart and blood vessels alone (193). Adaptive biochemical responses of the heart to stress are usually interpreted to be favorable at the physiologic scale. However, when chronic they lead, at the medical scale, to dysregulation of cardiac intracellular molecules and ultimately to congestive heart failure (194). Other examples are discussed below (see section ‘‘Clinical Care Examples of the Importance of Scaling’’). Interestingly, the emergence of new properties at higher scales of organization occurs even though common functional states and tasks can be identiﬁed

Protocol-Directed Patient Management

549

at different scales of organization (72). These two important concepts related to scale (emergent properties and common functional states or tasks) should not be confounded. C. Scale of Investigation

The scales of scientiﬁc inquiry vary from greater than 10þ20 m, the scales of astronomical studies, to less than 1020 m, the scales of atomic and subatomic particle studies [p. 241 in Ref. (141)]. This large range of scales of inquiry requires many different instruments, techniques, laboratories, and conceptual paradigms that must be matched to the scale of the item under study. Studies cannot be readily conducted with instruments that are matched to a different scale. One would not use a telescope to study bacteria, or an IR spectrophotometer to study gamma ray bursts. Mandelbrot illustrates this issue with a series of provocative answers to the question ‘‘what is the length of the coastline of Great Britain?’’ The answers form a set of responses, each of which is correct for the scale at which it was generated—but incorrect for other scales of inquiry (201). Penrose and Penrose, two eminent psychologists, discuss an interesting reﬂection of scaling in the ‘‘Waterfall’’ of the graphical artist Maurits Escher (202). Similar scale-speciﬁc answers are encountered in medical research (see section ‘‘Clinical Care Examples of the Importance of Scaling’’). The medical scales of inquiry (34), like those of the physical sciences (201), vary. They range from the reductionist focus on the behavior of the parts of a system to the holistic focus on the integrated behavior of the intact system. The parts of the patient system include biochemical, cellular, organ, and physiologic elements. For many medical questions, the intact system consists of the patient within the clinical environment, with all of the interactions and foibles that occur during the patient–clinician encounter (203– 205). For medical decision-making, the concept of the scale of inquiry is important both to clinician decision-makers and to clinical researchers. Clinical Care Examples of the Importance of Scaling

Three clinical examples illustrate the medical importance of the scale of inquiry: the use of vitamin C for the treatment and prevention of scurvy, the use of sodium channel–blocking agents for the prevention of sudden death following myocardial infarction, and a genetic deﬁciency of elastin (206). Scurvy

The absence of a crucial enzyme (L-glucuronolactone oxidase) in humans (the result, apparently, of a genetic accident), which leads to an inability of cells to produce L-ascorbic acid, provides the foundation for scurvy. A deﬁciency of L-ascorbic acid leads to decreased peptidyl hydroxylation of procollagen. This leads to a failure to produce the normal triple collagen

550

Morris

helix. The deﬁciency of normal triple collagen helix leads to abnormal connective tissue. The abnormal connective tissue leads to clinical manifestations, including death. This fascinating story was delineated centuries after the cardinal clinical observation and the crucial medical intervention had been made. In the 18th century, long before any of the interesting reductionist information about enzymes or cell biology was available, Dr. James Lind, a Royal Navy surgeon, performed the ﬁrst clinical trial of which I am aware (207). It was a cohort-controlled study of multiple agents including citrus fruit ingestion aboard ship, with only two subjects in each group (149). The signal-to-noise (S/N) ratio was so high, however, that the results were deﬁnitive even though the number of patients was small. Sudden Death Following Myocardial Infarction

For decades, agents that suppress premature ventricular contractions (PVCs) in patients following myocardial infarction were commonly prescribed based on pathophysiologic and other reductionist research evidence (66,67). The rationale behind this therapy was well founded on eletrophysiologic and cardiac physiologic data from animals and from humans. It was expected that suppression of PVCs would lead to increased survival, through interruption of the sequence: PVCs ! ventricular tachycardia ! ventricular ﬁbrillation ! sudden death. Subsequently, the results of the CAST, a large multicenter clinical trial, revealed that the agents—encainide, ﬂecainide, and moricizine—effectively suppressed PVCs following myocardial infarction. They were, therefore, effective at the pathophysiologic scale. However, the death rate in the treatment group exceeded the death rate in the control group. The disparity between the results at the pathophysiologic scale (effective suppression of PVCs) and those at the holistic medical scale (reduced patient survival) is a sobering example of emergent properties of complex systems. This excess death rate following therapy that is effective at the pathophysiologic scale is a striking reminder of the need for holistic clinical outcome data from rigorously conducted clinical studies. Elastin Deﬁciency and Supravalvular Aortic Stenosis

Some animals and some humans are hemizygous for the elastin gene. This genetic deﬁciency information was obtained from research at the reductionist biochemical scale (Fig. 1) (206). This genetic deﬁciency leads to the anticipated reduction in elastin production at the cell physiology scale, with reduced elastin in each of the elastin lamellae of the aorta. However, this knowledge obtained at the biochemical and cell physiology scales does not permit one to anticipate the emergent property that appears at the organ physiology scale. The pulsatile stress sustained by the aorta in the intact organism leads to an emergent property, the unexpected production of an increased number of elastin lamellae, each one of which contains less

Protocol-Directed Patient Management

551

Figure 1 Animals and humans can be hemizygous for the elastin gene. This condition provides a good example of the importance of scale of inquiry (level of interest). At the reductionist scale, the chemical knowledge of the presence of only half of the elastin gene complement allows accurate prediction of the production of elastin (half of that predicted for each elastin lamellum) at the cell physiology level. However, this reductionist-scale knowledge does not allow predictions of the emergent properties that appear at higher scales. At the organ physiology scale, an unexpected increase in elastin lamellae number is encountered. At the medical scale, an unexpected problem emerges: supravalvular aortic stenosis (SVAS) due to the thickened aortic wall that results from the increased number of elastin lamellae. Source: Modiﬁed from Ref. 198.

elastin than normal. More importantly for clinical decision-makers at the medical scale, the foreknowledge of biochemical and cell physiological information does not lead to anticipation of an emergent medical disease, supravalvular aortic stenosis, that results from increased aortic wall thickness due to the increased number of elastin lamellae (206). Human Experimental Outcomes

Human experiments (clinical trials) provide many choices of outcome variables. Ultimately, patients, subjects, clinicians, and clinical investigators are most interested in the outcomes that should drive clinical decision making, such as survival, quality of life, and related variables. These are reasonably termed ‘‘ultimate outcomes.’’ For monitoring and experimental purposes, clinicians and investigators measure many other variables that reﬂect

552

Morris

intermediate physiologic states that precede, and may be linked mechanistically to, the ultimate outcome. These variables include body temperature, white blood cell count, serum sodium level, blood pressure, cardiac output, minute ventilation, urine output, creatinine clearance, sleep character, electrocardiographic variables, and many others. These are reasonably termed ‘‘intermediate outcomes’’ (also called physiologic variables, and surrogate outcomes). The different experimental outcomes that can be assessed belong to different scales (or subscales) within the medical scale of inquiry. The precise deﬁnition of outcome of interest is a crucial step in identifying the kinds of data required and the necessary tools that must be incorporated in study design at the medical scale. Surrogate Outcomes

The settings appropriate for cell or organ physiology studies provide results that are frequently not directly applicable to clinical decision-making. The experimental results required for clinician decision-making in complex settings are best provided by rigorous studies at the holistic medical scale (34). Surrogate end points (intermediate outcomes) are variables that reﬂect outcomes at a reductionist scale, rather than at the holisitic medical scale at which ultimate outcomes like survival and quality of life are sought. Surrogate end points may not reﬂect the ultimate outcomes and therefore can mislead clinicians and investigators (208,209). High mechanical ventilator pressure applications to experimental subjects with acute lung injury have been evaluated with surrogate outcomes. Kirby et al. proposed that the care of patients with ARDS had become an easy exercise with the application of high PEEP (super PEEP, PEEP up to 50 cmH2O) (210). Lachmann and colleagues proposed ‘‘opening’’ the damaged lungs of ARDS patients, with a lung recruitment technique that was not, to my knowledge, ever adequately explicit (211–213). Both of these approaches focused on physiologic considerations and surrogate outcomes, principally levels of arterial oxygenation. They were subjected to a systematic evaluation at the scale of the intact patient in the clinical environment, using ultimate clinical outcomes (see below). Subsequent clinical trials, using different experimental strategies, failed to reveal beneﬁt (120,130,214) from these approaches although they seemed appealing when physiologic scale (surrogate) outcomes were used. Reductionist Animal vs. Holistic Human Experiments

I believe it informative to contrast the imperatives of reductionist animal experiments with those of human experiments. The arguments apply equally well to other reductionist research (e.g., organ physiologic, cell biologic, biochemical, genetic, etc.). I think it helpful to distinguish between those experimental attributes that focus on the experimental subjects from

Protocol-Directed Patient Management

553

those that focus on the experimental design. Both animals and humans are sentient beings that require attention to and reduction of both pain and suffering during experimentation. However, self-determination is an attribute only assigned to humans. Therefore humans cannot, because of the ethical principle of autonomy (also called respect for persons), be asked to sacriﬁce their well-being for the good of others (other humans—also called the greater good). Animals, in contrast, are routinely forced to sacriﬁce their well being for the good of others. The primary outcome focus for animal experiments is usually an intermediate outcome variable, because the goals of animal experiments usually are directed at uncovering mechanisms of injury or disease. In contrast, the primary outcome focus for holistic human experiments is usually an ultimate outcome. For human experiments, intermediate outcomes are usually secondary and frequently used to identify surrogate outcome variables that might substitute for ultimate outcomes, or to identify potential mechanisms compatible with the experimental results. The level of experimental control for both investigator behavior and for experimental subjects is high in reductionist animal experiments, but low in holistic human experiments. Human subjects are usually more variable than animal subjects. Animals may be chosen with similar genetic background, identical age and comorbidities, and identical injuries. This conjunction of conditions is not possible in holistic human experiments. Consequently, the standardization both of interventions and of investigator decisions that characterize reductionist animal experiments cannot be expected in holistic human experiments. Clinical investigators and clinicians must tailor their decisions to the individual human subject’s needs (identiﬁed by the subject’s expressions of illness). Both the intervention and the cointerventions that the human subject receives are therefore difﬁcult to standardize. Only the clinician decisions (responses to the subject’s speciﬁc expressions of the disease) can be standardized. Transactional Unit of Analysis

The appropriate unit of analysis in clinical experiments (clinical trials) is neither the patient nor the clinician (or the clinical environment). It is, rather, the combination of patient and clinician (or clinical environment). Environmental psychologists (203,204) identify the interacting patient and clinical environment, together, as a ‘‘transactional unit’’ (Fig. 2). For example, it is meaningless to inquire about the outcome of an infant born with cystic ﬁbrosis, or of an adult with human immunodeﬁciency viral infection, without identifying their location. Both of these diseases will have different natural histories for patients in San Francisco and patients in rural sub-Saharan Africa. The same need to identify the clinical environment is present in ICU patient encounters. It makes little sense to ask about the

554

Morris

Figure 2 The patient and the clinical caregiver (clinical environment) form a transactional unit that determines clinical outcomes. The patient expresses the disease or clinical problem with unique expressions and values of various quantiﬁed measurements. The clinical caregiver responds to these unique expressions. The iterative interaction of the patient and the clinical caregiver ultimately produce an outcome that can be examined by investigators. For example, the response of patients to PEEP is determined in part by the ﬂuid administration chosen by the caregiver. One cannot deﬁne PEEP responsiveness unless the clinical setting and its therapies are identiﬁed. Likewise, one cannot describe the natural history of cystic ﬁbrosis without identifying the clinical setting. In a rural area of a poor country, cystic ﬁbrosis is a lethal gastrointestinal (GI) disease of infants. In a wealthy country with modern medicine, it is a pulmonary disorder of children and adults. Abbreviation: PEEP, positive end-expiratory pressure.

natural history of a patient with sepsis or acute lung injury without inquiring about the clinical response provided to patients with these conditions. The patient and the ICU clinicians, interacting iteratively and reciprocally, constitute the unit of analysis for which clinical outcome measures become meaningful. Therefore critical care researchers must be sensitive to the clinical care environment and the therapy it delivers when considering a patient or a group of patients with a critical care problem. Both the treatments and the decision-making of clinicians delivering these treatments to subjects are important elements in a rigorous critical care experiment. Interestingly, this transactional unit concept is reminiscent of interactive systems at multiple scales. At the subatomic level, the Heisenberg uncertainty principle leads physicists to include the observer (and associated instrumentation) as part of the system, along with the particle being observed [p. 209 in Ref. (141)]. In philosophical discourse, the Hegelian dialectic is an accepted conceptual model. In biology, the new ﬁeld of epigenetics formally acknowledges the interaction of the genome with its

Protocol-Directed Patient Management

555

environment, both local and macroscopic, as interactive determinants of phenotypic expression (215–219). The interaction of genotype and environment is also well illustrated by studies in microgravity. The accumulation of phosphate ions at the osteoblast surface under simulated microgravity conditions is higher by as much as a factor of three. This may explain the increased sensitivity of osteoblasts to apoptogens and provide a partial explanation for bone loss in the absence of gravity (220,221). The importance of gravity can only be explored, of course, when technological advances allow its elimination. The gastrointestinal tract is also inﬂuenced unfavorably by microgravity (222). An interesting extension of the transactional unit concept might include experimental design and data collection as important components, much like the instrumentation used to ‘‘observe’’ subatomic phenomena. Then more than the patient, clinician and clinical environment need to be considered. Perhaps the experimental environment itself contributes to the outcomes and may lead to divergent results in seemingly similar clinical environments. This is one form of argument that might be raised against efﬁcacy research and in favor of effectiveness research (160). However, in general, effectiveness research should only be conducted after the efﬁcacy of the intervention has been established (see section ‘‘Efﬁcacy vs. Effectiveness Clinical Trials’’) (140,160). Multinational Critical Care Experimentation

The requirement for hundreds of patients in many clinical trials of critical care issues frequently necessitates multicenter trials lasting several years. The multiple-year duration of many clinical trials introduces two serious limiting logistic problems: (i) unavoidable ‘‘secular’’ changes (changing cointerventions that occur as a result of the passage of time); and (ii) faltering enthusiasm and interest among participating clinicians (MD, RN, RRT, etc.). The use of bedside (point-of-care) computerized protocols to standardize clinician decision-making between institutions makes possible the conduct of clinical trials with explicit methodology in large numbers of institutions. The acquisition of the required number of patients could then be realized in a short period of time, perhaps a few months. For example, a consortium of 1000 hospitals conﬁgured to conduct clinical studies with computerized protocol standardization of clinical decision-making could likely enroll 8000 ARDS patients within six months. Other clinical trial tasks, in addition to bedside decision-support with computerized protocols, would also require electronic tool support and are discussed below (see section ‘‘Scalable Electronic Tools’’). Critical care medicine needs a robust experimental clinical outcomes laboratory in which to conduct rigorous human experiments (clinical trials). This laboratory must produce results that can be replicated. An adequately

556

Morris

explicit method is required for replication of results (34). Actual or potential replicability of results is a basic requirement of all scientiﬁc investigation (70,139,148–154). Standardization of clinician decision-making with computerized protocols can be a key to the development of the required adequately explicit clinical methodology. Cointerventions (see section ‘‘Confounders and Cointerventions’’) in nonblinded randomized clinical trials are not well managed by randomization. Rather, adequately explicit protocols are the important tools for minimizing cointervention-induced differential bias in critical care randomized clinical trials. This is analogous to the use of protocols to minimize information bias in observational epidemiological studies [p. 279 in Ref. (223)]. Widespread efforts to introduce separate, institution-speciﬁc guideline or protocol decision support are under way in many institutions. While this may further standardization within each institution, the development of separate and institution-speciﬁc decision-support tools will have little impact on the interinstitutional variation in health care delivery. The potential of computerized protocols to effectively address the problems of secular trends and faltering enthusiasm of participating clinicians, through the conduct of large multicenter (including multinational) trials, will not be realized by such institution-speciﬁc protocols. The beneﬁts of standardized explicit methodology in clinical trials and the experimental rigor that they will convey on clinical trials will only be realized by the standardization of clinician decision-making between large numbers of institutions. The development of separate, institution-speciﬁc, decision support tools is, in this regard, counterproductive and will merely formalize much of the unnecessary variation in medicine (1,2). While large simple trials can be done with paper-based tools (224) and large simple critical care trial in subjects with sepsis has been successfully executed (121), paper-based tools do not seem attractive. Critical care clinical trials will be improved if we can enroll large numbers of subjects quickly by engaging large numbers of hospitals, by employing the same adequately explicit method in all participating sites, and by acquiring a comprehensive set of data for each subject. Electronic tools should enable such clinical trials. Large-Scale Critical Care Clinical Trials

Scalable Electronic Tools. Operational electronic tools are scalable and can catalyze the formation of a large-scale human clinical trial laboratory, linked through the Internet. This could support clinical trials with adequately explicit methodology and enable completion in months rather than years. Obstacles have been identiﬁed (35,225,226) but these have been overcome in the research clinical trial setting (58). The large enrolled subject number could allow multiple experimental groups and could produce dose–response curves (see section ‘‘Limitations of the Two-ExperimentalGroup Design’’).

Protocol-Directed Patient Management

557

Human Technical Support. However, the human technical support staff are, in contrast to the electronic tools themselves, not scalable. The number of staff is limited and therefore they have a limited time to ﬁeld questions from users. Each question or consultation is likely to take about half-an-hour of support staff time, because it is frequently necessary to acquire information about experimental subjects and clinical conditions to provide useful responses to queries. For example, consider the situation in which support staff could ﬁeld six questions each day because they have only a ﬁxed time of three hours available for consultation. If the support staff are limited to six questions daily, the fraction of electronic tool interactions that are unclear to clinicians and that generate clinician questions that require support must decrease hyperbolically as the number of clinical trial sites increase. According to Figure 3: % unclear interactions ¼ ½six questions=ðnumber of electronic tool interactionsÞ 100% In a large-scale multicenter human laboratory, fewer than 500 enrolled patients could easily lead to 6000 protocol or other electronic tool interactions per day. If only 0.1% of these 6000 electronic tool interactions/day generated questions, the limit of six questions/day would be reached (Fig. 3). These considerations do not include the challenge of study start-up. Start-up should be

Figure 3 Hyperbolic relationship between the percentage of electronic tool interactions that generate user questions that require support. Data are expressed as a function of the total daily number of electronic tool interactions for a human support team with three available hours daily to respond to questions.

558

Morris

expedited as well by carefully crafted electronic tools including tutorials and decision-support tools with embedded context-sensitive help. In addition, local clinical coordinators and the clinical coordinating center (data monitoring center) coordinators are not scalable. If we are to conduct large-scale clinical trials efﬁciently and without excessive cost, current levels of coordinator staff must be able to conduct large-scale trials. Tools that assist local and clinical coordinating center coordinators and web-based tools that expedite communication and allow ‘‘virtual site visits’’ by the clinical coordinating center coordinators are part of the effort my colleagues and I have engaged in during the past ﬁve years (see section ‘‘Computerized Protocol Experience’’). Clarity of Electronic Tools for the User. As the scale of the human laboratory increases, the clarity with which the protocols or other electronic tools communicate with clinician users must increase. The human computer interface must be clear and the messages displayed by the computerized tool must be unambiguous. If more than 0.1% of 6000 electronic tool interactions generated questions, they would overload the clinical trial support staff. With appropriate clarity of the electronic tools, their scalability could be realized without requiring a costly increase in the number of humans supporting the electronic tools, and their associated and difﬁcult education and maintenance. The achievement of electronic tool interactions that are clear to the clinician 99.9% of the time and do not generate questions is a challenge—but it is achievable with proper attention to computer screen button labels, instruction texts, protocol instruction explanations, etc. D. Human Experiments (Clinical Trials)

Among the several clinical study types described in epidemiology textbooks, the clinical experiment (clinical trial) is the source of the most credible information. Clinical trials are special cases of epidemiologic cohort studies. The fundamental difference between clinical experiments and other epidemiologic studies is that in clinical experiments the investigator determines the exposure of subjects to the test or experimental intervention. Clinical experiments (clinical trials) produce estimates or measurements of the effect size of the experimental intervention. Clinical experiments produce their most credible results when the experimental groups are comparable and differ only by chance, save for the exposure of subjects in the test group to the experimental intervention (140,227). The two major attributes of clinical experiments that assure, or at least increase conﬁdence in, achievement of comparability between experimental groups are randomization and blinding (73,140,149,151,153,223,228,229). Two sources of error can reduce the credibility of the measurement of effect size. Random error leads to imprecision. Systematic error leads to bias. Increasing the number of subjects in a clinical trial increases precision.

Protocol-Directed Patient Management

559

Bias, in contrast, is a more challenging source of error and requires careful attention to experimental design. The foundation of experimental design was derived from agricultural trials (228). They were designed so systematic error (bias) played little role. The randomized experimental design was believed to both assure random distribution of confounders to all experimental groups and allow statistical analytic correction for confounder imbalance between experimental groups (140). Randomized experiments stimulated development of statistical analysis tools (140,228). Unfortunately, evidence indicates that bias cannot be addressed and controlled by randomization alone for many critical care clinical experiments. Consequently, the conclusion that systematic error plays little role in clinical trials (140,228) is incorrect for many critical care experiments. In any case, the statistical approach to unmasking statistical or biological interactions is not satisfactory (140). The systematic error produced by postexperimental group allocation (postrandomization) bias due to confounders (better called cointerventions) can be antagonistic to, or synergistic with, the effect of the experimental intervention under study. Anticipation of either the direction or magnitude of such cointervention effects is hazardous (140,228). Bias is expected and unavoidable in all clinical experiments because humans conduct these experiments on human subjects. Bias per se is not necessarily a problem if it appears equally in all experimental groups. Bias exerts its most pernicious effect when it is not equally expressed in the results of the different experimental groups. For example, IV nitroglycerine therapy, instead of a beta-blocking agent, for hypertension applied equally to both experimental groups might not eliminate the effect of a preferable mechanical ventilation–weaning strategy. Unequal administration of IV nitroglycerine will introduce a differential (between group) bias and might completely obscure the effect (230). Confounders that inﬂuence the different experimental groups unequally may produce a differential bias (Fig. 4). Differential bias threatens, and may invalidate, the assumptions of experimental group equation [Ref. (231) in Ref. (70)] or equivalence [Ref. (232) in Ref. (70)], (both early terms for comparability) necessary for the internal validity of a clinical trial (Fig. 4) (153). Internal validity represents the extent to which the study results represent the true effect in the study subjects (see section ‘‘Efﬁcacy vs. Effectiveness Clinical Trials’’). External validity represents the extent to which the study results represent the true effect in the population of interest (140,153,160). Internal validity is emphasized in efﬁcacy clinical trials and external validity in effectiveness clinical trials (153,160). Internal validity is a prerequisite for external validity. Consequently, efﬁcacy clinical trials should precede effectiveness clinical trials (see section ‘‘Efﬁcacy vs. Effectiveness Clinical Trials’’) (140,160,223). Unfortunately, this requirement is frequently violated in critical care trials (131). Clinical trial design always involves striking a balance between the needs of internal and external

560

Morris

Figure 4 The validity of clinical trial results can be assigned to two categories— internal and external validity. Internal validity is usually the focus of efﬁcacy clinical trials and represents the validity with which the study results reﬂect the true effect in the experimental subjects. External validity is usually the focus of efﬁcacy clinical trials and represents the validity with which the study results reﬂect the true effect in the population of interest (from which the study subjects were selected). The outcome ‘‘signal’’ that leads to inferences about the effect of the experimental intervention is the difference between the outcomes of the different experimental groups.

validity. The medical community has difﬁculty meeting the challenge of transferring the results of internally valid efﬁcacy trials to the practicing clinical care community (7,233). Electronic tools have the potential of providing a replicable method that is used in efﬁcacy trials but can be exported to practicing physicians and clinical care institutions (see section ‘‘Computerized Protocol Experience’’). Noise and Signal-to-Noise (S/N) Ratio

The detection of an association between an input signal of interest (Signal, S) and an outcome measure requires that the signal of interest be capable of separation from other unwanted signals (Noise, N) with which it may be confused or by which it may be obscured. A common measure of this capability is the S/N ratio (87). Unless the S/N ratio exceeds 1, the signal will be undetectable. The two major elements of the transactional patient–clinician unit (203), the patient and the clinical caregiver, determine the intensity of care and the outcome. Both the patient and the clinical caregiver are sources of random noise and of systematic noise (bias) (Fig. 5). The patient contributes noise because of uncontrollable host factors and because of variation in disease etiology, severity, extent, and duration. Local factors inﬂuence the patient’s disease and spectrum of clinical problems. The patient identiﬁcation and selection process is quite imperfect and may incorporate much local bias due to the prejudices of individual clinicians and clinical

Protocol-Directed Patient Management

561

Figure 5 Both random and systematic noise are introduced by both major elements of the transactional unit—the patient and the clinical caregiver (clinical environment). This noise decreases the S/N ratio and reduces the ability of investigators to detect the effect of experimental interventions. Abbreviation: S/N, signal-to-noise.

investigators. This bias is the result of many factors, among which are characteristics of local clinical environments and failure of the medical community to establish broadly accepted adequately explicit deﬁnitions of many diseases, including ARDS (234,235). The clinical caregiver contributes noise because of strongly held opinions based on many factors that inﬂuence behavior, including general and local cultural factors, local technical abilities, background, training, and experience. The effects of many experimental interventions on the outcome in complex clinical environments are likely to be small (20). They can easily be obscured by noise (Fig. 5). The S/N ratio for random noise varies inversely with N1/2. The impact of random noise can therefore be reduced by increasing the number (N) of observations (224). In contrast, increasing N has no effect on systematic noise (bias). Techniques to minimize bias in randomized trials include allocation concealment (236), restriction of randomized subjects (140), blinding of patients, caregivers, and outcome assessors (236), using objective, reproducible criteria for assessing outcomes (143,144,154), assiduously screening all potential subjects, using adequately explicit inclusion and exclusion criteria, and assuring clinician and investigator compliance with all clinical trial protocol rules. Clinical investigators have long recognized the importance of the uniform application of clinical interventions to comparable patients in clinical research (237). Existing large national databases lack this uniformity and they contain mixed groups of variable treatments labeled with the same procedure or treatment code. Variations of 60% to more than 400% for common process of care steps for Intermountain Health Care, Inc. patients with comparable presentation and outcomes were observed in a study of

562

Morris

practice pattern variation (238–242). Median procedure times for 16 surgeons performing transurethral prostate resection ranged from 40 to 95 minutes, while prostatic tissue removed ranged from an average of 11 to 42 g/patient (238). These variations had a strong statistical association with the occurrence of urethral strictures within one year of operation (the primary complication). Therefore, outcomes reported for treatments labeled using insurance data leave much uncertainty about the actual procedures employed (243–246). Similar variability due to noise in critical care databases is likely. For example, balloon ﬂotation catheter use and IV ﬂuid and electrolyte therapy (247) vary widely, with some physicians committed to conservative and others to liberal approaches. Clinical Trial Design Principles

Advances at lower scales of inquiry cannot replace the study of integrated systems, such as sick patients in the clinical environment (34,192). Clinical trials will remain an important and the most credible source of information guiding clinical decision making. Attention to sources of nonuniformity between experimental groups is an essential part of experimental design (140,149,151,153,223,228,229). An epidemiologic investigative strategy requires investigators to assess the validity of a statistical association between a variable and an outcome of interest by excluding possible alternative explanations. These include chance, systematic errors in data collection or interpretation (bias), or effects of other variables (confounders) on the outcome (223). While this framework has been successful for observational epidemiological studies, the deﬁnition of confounders does not accommodate the needs of many nonblinded critical care clinical trials. For such clinical trials, it is important to distinguish between two categories of nonexperimental variables that can inﬂuence outcomes. Confounders and Cointerventions

While the determinants of patient outcome may include the intervention under study, these determinants are multiple and complex. Among the multiple variables that may determine outcome (e.g., survival) are variables, called confounders. A confounder (confounding variable) is a variable that is associated with a predictor (input or test variable) and is also a cause of the outcome variable (153). For example, IV ﬂuid can be a confounder in trials of mechanical ventilation strategies. Excessive ﬂuid and salt infusions appear to harm patients (248,249) and may inﬂuence mediator levels in clinical trials (250,251). Conversely, PEEP can be a confounder in clinical trials of hemodynamic support strategies. Confounders can inﬂuence, alter, or reverse the results of clinical trials (140,149,153,223,252). Many experts claim that confounders can be addressed by randomization (see section ‘‘Experimental Group Equation’’). However, confounders can be present

Protocol-Directed Patient Management

563

before or after random allocation of subjects (randomization) to the experimental groups of a clinical trial (140). Those confounders present before allocation are dealt with by randomization and by restriction of randomized subjects (140). They are commonly recognized and discussed in epidemiology texts (140). In critical care research, these variables should be called confounders because they correspond well to the traditional deﬁnition of confounding identiﬁed in epidemiology (223). However, when confounders are introduced after subject assignment to the experimental groups, they are better termed ‘‘cointerventions’’ (153,154,252). This will distinguish them from those confounders present before subject allocation to the experimental groups. Cointervention is a good term because, like the experimental intervention, cointerventions are introduced after randomization and result from the interaction of the subject with the clinical environment (e.g., mechanical ventilation strategy, drug therapy for hypotension, IV ﬂuid therapy, diagnostic strategies for suspected infection, monitoring intervals, laboratory tests, antibiotic therapy, sedation, etc.). Like the experimental intervention, they can alter the results of clinical trials. They can even invalidate the results of clinical trials. In addition, clinicians and investigators can easily overlook cointerventions. Unlike the experimental intervention, cointerventions rarely receive adequate attention in the protocols of randomized clinical trials. The Cochrane Collaboration proposed the following deﬁnition of cointerventions: ‘‘In a randomised controlled trial, the application of additional diagnostic or therapeutic procedures to members of either or both the experimental and the control groups’’ (252). Cointervention is, however, not a widely accepted term. Horwitz et al. called cointerventions cotherapies in an analysis of clinical trial results that suggested that cointerventions reversed the effect of beta-blocking drugs following myocardial infarction (8). This observation was vigorously criticized in a vitriolic exchange because of the post hoc nature of their analysis (253–256). These arguments concern a longstanding debate among statisticians and epidemiologists. These publications focus attention on the importance of identifying and controlling, if possible, important cointerventions before a clinical trial is conducted. They emphasize the importance of including cointervention considerations in study design of efﬁcacy trials and not, after the study is completed, in post hoc analysis (253–256). Most texts do not mention cointervention and one reference deﬁnes cointerventions differently (as simultaneously applied experimental interventions) (257), a deﬁnition at odds with the use here and elsewhere (153,154,252). Even those who mention with authority the importance of cointerventions may not discuss the adequately explicit methods that could reduce the differential bias due to these cointerventions (153,154). Finally, a recent publication dealing with critical care clinical trials did not address the methodolgic details needed to deal with cointerventions (160).

564

Morris

Experimental Group Equation

In the 1920s, experimental group equations [Ref. (231) in Ref. (70)] or equivalence [Ref. (232) in Ref. (70)] were identiﬁed as essential requirements for internal validity of an experiment (Fig. 4) (73,140,149,151,153,223 228,229). Two major attributes of clinical trials, randomization and blinding, reduce differences between groups and help assure experimental group equation. [Restriction of randomized subjects is also used to reduce differences due to confounders present before subject allocation to experimental groups (140).] Randomization. Randomization, properly applied, can provide assurance that confounders, present among potential subjects before experimental group allocation, will be randomly (therefore almost uniformly) distributed among the experimental groups (140,149,151,153,229,258,259). Randomization assures that the experimental groups will differ only by chance. This allows application of statistical analysis in the interpretation of experimental outcome group differences (140,227). This technique is effective, especially when the number of subjects is large, and permits statistical analytic correction for such confounder imbalance. Randomization permits blinding as a tool to reduce unwanted differences between experimental groups (227). Blinding. Blinding is a fundamental experimental mechanism for reducing inequality of care in experimental groups (20,143,149,260). Double-blind drug trials with a placebo or a routinely administered therapy control arm have been conducted successfully without protocols and have provided an important foundation for thinking about clinical trial design (140,149,151,153,223,228,229). Cointerventions are therefore traditionally dealt with by double blinding, as in drug studies (73,140,149,151,153,223 228,229). Differential bias will be reduced in double-blinded trials, but even with double blinding, differential bias may not be eliminated. If the experimental treatment has an effect that, on the average, changes the clinical expression of the disease between the experimental groups, clinicians may perceive (correctly or not) the group to which the subject had been allocated and thus violate the blinding (140,149,223,228). Clinicians might then apply cointerventions differently to the treatment groups. Unfortunately the traditional double-blind drug trial (without protocols for cointerventions) has been frequently used as the model for randomized nonblinded critical care clinical trials. This presumes that all clinical trials can be managed with the same approach. This presumption is an error. Many interventional critical care studies, including those that incorporate mechanical ventilation techniques, cannot be double blinded. Comparability of experimental groups must be achieved by other means. Non-double-blinded clinical trial publications should be scrutinized and the clinical care carefully assessed for comparability of the cointerventions

Protocol-Directed Patient Management

565

Figure 6 Confounders can be introduced into clinical experiments either before or after subject allocation to the experimental groups. Confounders present before allocation are addressed, and can usually be controlled, by true randomization that commonly distributes confounders equally among experimental groups and avoids differential bias. Confounders introduced after subject allocation to the experimental groups are called cointerventions. Cointerventions have been traditionally addressed by double blinding of clinical experiments. In nonblinded (open) clinical trials, blinding is not possible. Adequately explicit protocols can control or at least partially control cointerventions and reduce the chance of introducing differential bias.

in the experimental treatment arms. Cointerventions are frequently neither controlled nor measured and this deﬁciency threatens the internal validity of critical care clinical trials. In nonblinded (open) critical care clinical trials, all experimental arms require well-deﬁned and detailed protocols (Fig. 6) (34,131,135). Adequately Explicit Methodology: Protocols vs. Guidelines

Decision-support tools such as guidelines and protocols have been functionally categorized as reminders, consultants, or as educational (261). They are intended to enable clinicians to deliver evidence-based care consistently by standardizing some aspect of clinical care, thereby helping lead to uniform implementation of clinical interventions (182,262–265). However, many guidelines and protocols lack speciﬁc instructions for commonly encountered clinical practice scenarios, and are useful only in a conceptual sense (182,263,266–270). They neither standardize clinical decisions nor lead to uniform implementation of clinical interventions. The medical subject headings in Ovid1 deﬁne guideline as ‘‘A systematic statement of policy rules or principles’’ and protocol as ‘‘Precise and detailed plans for the study of a medical or biomedical problem and/or

566

Morris

for a regimen of therapy.’’ Guidelines are general statements with little instruction for making speciﬁc decisions (95). For example, ‘‘If the ﬁrst drug is not tolerated, substitute a different drug from another class,’’ from a National Heart Lung and Blood Institute (NHLBI)/Hoescht Marion Roussel1 guideline for controlling hypertension in older women, does not standardize speciﬁc clinical decisions. In contrast, protocols are more detailed and can provide speciﬁc instructions. The explicitness of protocols varies continuously. An adequately explicit protocol is one that can generate speciﬁc instructions (patient-speciﬁc orders) without requiring judgments by the clinician and can elicit the same decision from different clinicians when they are faced with the same clinical information. There is no threshold beyond which protocols are adequately and below which they are inadequately explicit. My colleagues and I have concluded that our protocols are adequately explicit when clinicians accept and carry out over 90% of protocol instructions. This is clearly our choice and represents our judgment of the required level of acceptance of instructions by different clinicians. Adequately explicit computerized protocols can contain the greatest detail (271) and may lead to the upper limit of achievable uniformity of clinician decision-making with open-loop control (58,122,164,272) [closed-loop controllers automate processes and eliminate humans from the decision-making process (159,273–275)]. Paper-based versions can also contain enough detail to be adequately explicit (120,276). Inadequately explicit protocols omit important details (277–279) and elicit different clinical decisions from different clinicians. Clinician decision-makers must ﬁll in the gaps in protocol logic. Judgment, background, and experience vary among clinicians and so will their choices of the rules and variables they use to ﬁll in the gaps of inadequately explicit guidelines and protocols. In addition, because humans are inconsistent, any single clinician may produce different choices at different times, even though faced with the same patient data. Unfortunately, even systematic and scholarly collections of ﬂow diagrams commonly lack the necessary detail and cannot standardize clinical decisions (277–279). Protocols and ﬂow diagrams are also called algorithms but this is an inappropriate use of the term (277,279). An algorithm in mathematics or engineering is a precise solution (71) although its deﬁnition allows the more liberal use common in medicine [‘‘a set of rules for solving a problem in a ﬁnite number of steps’’ (280)]. The distinction between guidelines and protocols, particularly adequately explicit protocols, is crucial (34,35,281). Failure to make this distinction fosters confusion (131 265,281). The health care delivery community’s penchant to use guideline, protocol, and algorithm interchangeably (265,277–279,281–285) is more than a taxonomic error. It is a serious conceptual error that obscures the important distinction between adequately explicit and other methods. Adequately explicit protocols can serve as physician orders, and can function as

Protocol-Directed Patient Management

567

dynamic standing orders because they respond to changes in patient state, but guidelines cannot do so. I chose the term ‘‘adequately explicit’’ with care. Any articulation of a method is explicit. For example, consider ‘‘ . . . caution should be exercised when PAOP becomes increased to the extent that pulmonary edema is a risk’’ (95,96). While statements such as this are explicit, they do not elicit the same clinician decision when multiple clinicians are faced with the same clinical data. From the human experimentation perspective they do not establish a replicable experimental method. Most clinical investigators do not seem to recognize the difference between common ordinary protocols and guidelines and the uncommon adequately explicit methods that satisfy the scientiﬁc requirement of replication of experimental clinical trial results (20,34,35,131,149,237). For example, recent studies of high-frequency ventilation in neonates (286,287) have been described (288) as ‘‘ . . . rigorously controlled conditions with well-deﬁned protocols . . . ’’ (286) even though the method includes the following statements: ‘‘ . . . aggressive weaning if blood gases . . . remained . . . in range . . . ’’. This method will lead to different actions by different clinicians as they interpret ‘‘aggressive’’ and ‘‘remained.’’ These adult and pediatric critical care methods are not adequately explicit. Adequately explicit methods include rules such as ‘‘if [(last PaO2—current PaO2) > 10, and (current PaO2 time—last PaO2 time) < 2 hours, and FiO2 > 0.8, and PEEP < 15 cmH2O], then increase PEEP by 2 cmH2O.’’ A rule such as this can lead to the same decision by multiple clinicians. A protocol composed of such rules can establish a replicable method that permits large-scale critical care trials within a network of many institutions. In continuous quality improvement terms, an adequately explicit method is part of the ‘‘stabilization of process’’ necessary to improve quality (21–23,25). Protocols assure a reproducible level of care. For example, a brain edema protocol can support the performance of experienced as well as inexperienced colleagues such as new medical residents or nurses. Finally, it is important to point out that an adequately explicit method in the form of a computerized protocol has the potential to link efﬁcacy trials with clinical practice, although this is a challenge. An adequately explicit protocol is a model of the clinician decision-maker. The method (computerized protocol) is exportable and could be used to replicate a study or to transfer a decision-support tool to clinical practice. It could provide a needed link between clinical research and clinical care. Individualized, Patient-Speciﬁc Instructions. Clinicians and patients expect therapy to be tailored to the patient’s speciﬁc needs. All demand that clinicians respond to the patient’s individualized expression of disease. Therefore, adequately explicit decision-support tools must accommodate relevant variations among patients. If not, clinicians will reject them. Patients express their response to diseases in individual and variable ways,

568

Morris

including in their clinical data. These patient data lead to the instructions produced by adequately explicit protocols. Every single protocol instruction is based on patient data and leads to a standardized clinician decision according to the protocol rules. Every single protocol iteration yields at least one standardized decision (adequately explicit instruction) from the patient data. However, each patient expresses unique, patient-speciﬁc, clinical data that change over time. The patient’s therapy regimen is the sum of all of the standardized decisions from the patient’s unique clinical data from all protocol iterations over time. Therefore, the patient’s therapy regimen is individualized (tailored to the patient). Preservation of individualized therapy with standardized decisions is a crucial attribute of adequately explicit protocols and a central element of their ethical foundation (34,35,131). Ethical obligations to deliver individualized clinical care to the research subject are thus discharged satisfactorily as the scientiﬁc requirement to reduce differential bias in the clinical trial experiment is met (34,162,163). Limitations of the Two-Experimental-Group Design

The traditional common two-experimental-group limitation seriously limits the interpretation of clinical trial results, because two outcome estimates (e.g., event rates in the two groups) can fall on a straight line, a curvilinear function like a parabola, or a more complex function. While one group may have a more favorable outcome than the other, investigators cannot conclude that either experimental group intervention is an ideal therapy because what is frequently needed is a dose–response curve of outcome against multiple levels of the intervention. Such a dose–response curve is de rigueur in reductionist research such as in biochemistry, pharmacology, organ physiology, or in Phase II safety and dose-ranging human studies of new drugs. The limitation of the common two-experimental-group-design clinical trial was the foundation for the criticism by Eichacker et al. (289) of the NIH/NHLBI ARDS Network study of mechanical ventilation (120). This criticism precipitated the recent controversy between the Ofﬁce of Human Research Protection and the NIH (282–284,290–293). The NIH/ NHLBI ARDS Network investigators adopted the common practice of treating the two-experimental-group outcomes as if they belonged to a linear set (see linear curve in Fig. 7). Eichacker et al. argued that while the mortality of the higher tidal volume strategy (solid gray point in Fig. 7) was greater than that of the lower tidal volume strategy (open point in Fig. 7), an even lower mortality was possible at an intermediate, but not tested, tidal volume strategy (minimum of parabola in Fig. 7). They were articulating a systematic deﬁciency within the clinical trial community— the common use of a two-experimental-group design. A dose–response curve, constructed from multiple study group interventions (e.g., ﬁve groups), would have answered this appropriate criticism directly. A generic

Protocol-Directed Patient Management

569

Figure 7 Theoretical dose-response curve for human experiments (clinical trials). Most clinical trials implicitly accept a linear curve. However, other curves are possible, two of which are depicted. Multiple group clinical trials (e.g., ﬁve) might enable investigators to deﬁne a curvilinear dose response, such as is commonly encountered in biochemical and pharmacologic studies.

drug dose–response curve is represented in Figure 7; however the true shape might be different and would require a Phase III clinical trial for its deﬁnition. It cannot be inferred with conﬁdence from reductionist research results (e.g., cell biological, organ physiological, or animal safety study results). While dose-response curves are generated in many traditional Phase II clinical trials of drugs, these Phase II trials use intermediate (or surrogate) outcome variables such as physiologic measures (e.g., blood pressure), inﬂammatory mediators (e.g., TNFa), or routine laboratory analytes (e.g., serum creatinine). Such intermediate outcome variables are not necessarily linearly linked to the ultimate outcome variables (e.g., survival, days free of organ failure, recovery of function, quality of life) that are the important determinants of clinical decision making at the individual or societal level. The CAST results revealed an increase in mortality, even though the desired intermediate outcome (reduction of the number of PVCs) was achieved (66,294). In the NIH/NHLBI ARDS Network mechanical ventilation randomized clinical trial, mortality was decreased, even though the anticipated and desired increase in intermediate outcome, PaO2/FiO2 ratio, was not achieved and PaO2/FiO2 actually decreased (120). These two examples indicate that intermediate outcome changes in Phase II dose-ranging trials cannot be mapped linearly to the important ultimate clinical outcomes in Phase III randomized clinical trials. The difﬁculty of establishing a doseresponse curve in intact human subjects in the clinical environment of traditional Phase III trials has been so daunting that the health care community has become accustomed to abandoning this fundamental requirement of good scientiﬁc experiments when conducting many, if not most, Phase III

570

Morris

clinical trials. We should be able to achieve this goal with a new strategy based on electronic tools that enable large-scale participation of academic and community hospitals in studies using adequately explicit methods (computerized protocols) to generate Phase III randomized clinical trial dose-response curves. Protection of Subjects

The claims by some that ‘‘rigid’’ protocols (160,264,265) expose patients and experimental subjects to risk and that computerized protocols will not be widely distributable because they are ‘‘rigid’’ (160) deserve a response. A common perception is that ‘‘Protocols should not represent rigid rules but, rather, guides to patient care’’ (264). This is a misperception propagated by confounding ‘‘adequately explicit protocols’’ with ‘‘guidelines,’’ a common mistake (see section ‘‘Adequately Explicit Methodology: Protocols vs. Guidelines’’). The arguments above (see section ‘‘Individualized, Patient-Speciﬁc Instructions’’) that adequately explicit protocols can generate therapy that is tailored to the individual patient or subject’s needs, while standardizing clinician decision making challenge the characterization of these protocols as ‘‘rigid.’’ One cannot argue with the claim that protocol use is associated with risks. Virtually nothing we do in health care delivery is risk free. However, the ample evidence that protocol use leads to more favorable patient and clinician behavior outcomes belies the claim that the balance of risks and beneﬁts from the use of adequately explicit protocol is unfavorable to patients and experimental subjects (53–59). The ethical principles of beneﬁcence and nonmaleﬁcence (295) are thus respected when adequately explicit protocols are used for patients and experimental subjects. Finally, the claim that protocols endanger subjects (120,293) when they are used in clinical trials without an ordinary care (the misleading terms ‘‘usual’’ or ‘‘standard’’ care are more commonly used) experimental group is not an issue speciﬁc to adequately explicit protocols (see section ‘‘Adequately Explicit Methodology: Protocols vs. Guidelines’’), but rather one that relates to experimental design (282–284,290–292) (see section ‘‘Limitations of the Two-ExperimentalGroup Design’’).

Efﬁcacy vs. Effectiveness Clinical Trials

Investigators should clearly identify the clinical trial goal before study design proceeds. Our a priori level of conﬁdence in the ability of a clinical intervention to favorably alter patient outcome determines whether an intervention, after study results are available, will be accepted as part of routine clinical practice, or alternatively will be questioned (34,140,227). Efﬁcacy and effectiveness studies have different goals (20,34,153,160).

Protocol-Directed Patient Management

571

‘‘Efﬁcacy’’ is the term assigned to human experiments conducted under most favorable circumstances such as those in an academic center with sizable resources committed to the experiment (160). Efﬁcacy clinical trials provide the most compelling evidence and they should, whenever possible, precede the conduct of effectiveness trials (140,160). ‘‘Effectiveness’’ is the term assigned to a study conducted under more routine, and less favorable, clinical circumstances such as in community hospitals with more limited resources (20,153,160). Effectiveness studies can include experimental clinical trials such as the large simple trials advocated by Peto (224,296) or observational studies of the results of clinical practice. Some authors propose a hybrid approach, combining properties of efﬁcacy and effectiveness clinical trials for critical care experiments (160). When planning clinical trials, investigators are always faced with challenges that require compromise between these two categories. The electronic tools discussed in this chapter may allow convergence of efﬁcacy and effectiveness efforts (see sections ‘‘Clinical Care Examples of the Importance of Scaling,’’ ‘‘Multinational Critical Care Experimentation,’’ and ‘‘Computerized Protocol Experience’’). [In contrast with efﬁcacy and effectiveness research, continuous quality improvement (process improvement) studies focus on the efﬁciency with which one delivers a clinical intervention that is already accepted as part of ordinary clinical practice.] Comparability of Experimental Groups in Critical Care Clinical Trials

In the 1920s, experimental group comparability was identiﬁed as a goal for the internal validity of an experiment (Fig. 4). Comparability was termed ‘‘experimental group equation’’ [Ref. (231) in Ref. (70)] or equivalence [Ref. (232) in Ref. (70)]. Ideally, scientiﬁcally rigorous clinical trials strive for the delivery, to all patients in all experimental groups, of care that differs only by the random play of chance in every respect except for the experimental intervention itself (20,140,149,151,153,223,228,229,231,232,260,297). An inﬁnite number of repetitions of a trial would achieve identical experimental groups, but any single trial can only approximate such comparability (227). Randomization assures the applicability of statistical analysis that enables the evaluation of experimental group outcome differences relative to those expected by chance alone (140,227). Comparability of study groups is less likely with nonexperimental studies or when experimental studies use historical controls, when patients are selected according to an individual’s judgment, or sequential subject assignment. The importance and the difﬁculty of establishing comparability of the experimental groups are widely acknowledged (140,149,151,153,223,228,229,237,258,259). Two major attributes of clinical trials have been commonly used to achieve the experimental group equation or equivalence: randomization and blinding (140,149,151,153,223,228,229,258,259). These have been most

572

Morris

commonly used in double-blind drug trials, beginning with the British Medical Research Council clinical trials of streptomycin for treatment of tuberculosis in 1949 and of antihistamines for treatment of the common cold in 1950. Experts commonly recognize that confounders may be unevenly distributed among experimental groups both before and after allocation to experimental groups (140,149,151,153,223,228,229,258,259). The inﬂuence of drug trials with their emphasis on double blinding and on the outpatient setting has established common patterns of thinking about experimental design and its limitations. Most workers seem to have emphasized blinding to reduce, and analytic approaches to assess, the effect of confounders. While Friedman et al. discuss concomitant and compensatory treatment in non-double-blinded experiments (228), they and other workers have not carried this argument further. They did not identify these concomitant and compensatory treatments as postrandomization confounders (cointerventions)—a potent source of differential bias in critical care experiments. The ﬁelds of epidemiology and clinical trials do not yet seem to have adopted an approach matched to the needs of the critical care environment. Critical care involves such frequent and such diverse measures of bodily function that postrandomization confounders (cointerventions) loom large as sources of important differential bias. Even with double-blind experiments, the perception by clinicians that they know the subject allocation group is made more likely by the rich and recurrent source of data describing subject function in critical care trials (153). The fact that some of these numerous measurements will be accurate while some will be false-positive and false-negative results only compounds the problem. In addition, many critical care experiments cannot be double blinded, opening the door to the introduction of more postrandomization differential bias. Friedman et al., like many authors, have accepted the common wisdom that precludes control over the variable treatments that characterize clinical care and clinical research (160,228). Some of these experts indicate that one can merely encourage clinicians to uniformly apply treatments, even though the widely recognized unnecessary variation in practice (1,2,96,298,299) makes success with this approach unlikely (30,31,300– 312). In contrast, available tools have enabled clinical investigators to achieve advances in uniform application of treatments through clinician decision support (34,35,77,131,173,265,313–315). Confounders in Critical Care Clinical Trials

Critically ill patients receive multiple interventions in the form of therapies, diagnostic tests, and monitoring. Patients who become research subjects are associated with confounders both before and after allocation to experimental groups. True randomization generally distributes uniformly those confounders present before experimental group allocation (140). However,

Protocol-Directed Patient Management

573

many therapeutic, diagnostic, and monitoring interventions continue to be applied to clinical trial subjects after allocation to the experimental groups. These postallocation interventions are not inﬂuenced by randomization and are potential confounders. Double blinding has been commonly used to assure uniform distribution of postexperimental-group-allocation confounders, especially in the double-blind drug trials that have constituted a major force in statistical development since 1949. Several workers recognize that even double-blind drug studies may not be able to preserve blinding. Therefore, postallocation confounders may not be evenly distributed between experimental groups (140,149,223,228). Blinding is not possible in many critical care trials (e.g., extracorporeal support, IV ﬂuid and hemodynamic support, mechanical ventilation support, sedation and paralysis, positioning, blood sugar control, etc.). Postallocation confounders may strongly inﬂuence experimental outcomes in critical care clinical trials. If confounding variable values are obtained during the clinical trial, an attempt can be made to account for their effect through stratiﬁed analysis. However, this is a complex issue that involves, among its several problems, the assumption of uniformity (that the effect of the confounder is homogeneous across strata) (140). I do not believe this assumption can be defended for IV ﬂuid and hemodynamic support, antibiotic therapy, patient positioning, or other potential postallocation confounders (cointerventions). The reasons for which clinicians regulate or titrate these therapies may be unknown. In addition, patients are complex nonlinear biologic systems (71–73). It would be difﬁcult to account for the sequential changes induced by any given confounders. Analytic attempts to correct confounder effects cannot adjust for the sequential dependence of each confounder on the previous confounder’s effects. This is similar to the difﬁculty of separating, by statistical analysis, the effects of participation (selection bias) from those of disease determinants (140). The three examples below are only a sample of the many postrandomization clinical interventions that could function as confounders (cointerventions) in critical care clinical trials. They serve to illustrate both the subtlety and the potential importance of confounders on experimental outcome. I am not aware of any mechanical ventilation experiment that attempted to control the differential bias that might result from these potential cointerventions. For all these cointerventions, a uniform clinician decision-making approach, using a protocol that embraces a reasonable set of rules, can minimize differential bias between experimental groups. Antihypertensive Drugs (230)

Drugs can inﬂuence the duration of mechanical ventilation. While this seems almost axiomatic, the subtlety and magnitude of this effect is not widely appreciated. Some patients develop hypertension following thoracic surgery. Clinicians attempted to withdraw mechanical ventilation support

574

Morris

from these hypertensive patients when they met certain rules (a protocol) concerning arterial oxygenation (230). Thirty hypertensive patients, when treated with IV sodium nitroprusside or IV nitroglycerine failed to meet the criteria for initiating a trial of withdrawal of mechanical ventilation. When the therapy was changed to a beta receptor blocker, labetalol, to treat the hypertension, 28 of the 30 patients immediately met the criteria for a trial of withdrawal of mechanical ventilation. Should the type of antihypertensive drug employed in a mechanical ventilation clinical experiment be unevenly applied to the experimental groups, a large differential bias that could dramatically inﬂuence the duration of mechanical ventilation, or the number of mechanical ventilation–free days, could occur. Antihypertensive drug therapy is a cointervention. IV Fluids and Hemodynamic Support

Fluid and electrolyte therapy can inﬂuence patient outcome and obscure the effects of therapeutic interventions in clinical trials. Reliable evidence indicates that clinical management of the circulation with IV ﬂuid and hemodynamic support can modulate lung injury (316–319), can alter mediator levels (250,251), and can harm patients (248,249). Although high intravascular pressures may worsen patient outcome (320) and induce alveolar collapse (321), low intravascular pressure may increase lung injury as well (318,319,322–325). The intravascular pressure is therefore likely an important determinant of the evolution of lung injury. This conclusion is supported by isolated lung experimental results (326). Variations in management of IV ﬂuid and hemodynamic support can, if not uniformly applied to all patients in an experiment, alter the experimental outcome in unpredictable ways. Uncertainty about proper IV ﬂuid therapy comes at least in part from equivocal terminology. The terms used to evaluate body ﬂuid status and its treatment are enveloped in confusion (327–337). This confusion precludes development of a systematic evidence-based practice (327–334). Contradictory terminology (96) probably contributes to the uncertainty surrounding ﬂuid and electrolyte therapy for sepsis (338), shock (339–341), and acute lung injury and ARDS (320). Intravenous ﬂuid and hemodynamic support therapy are cointerventions. Body Temperature

The data of Figure 8 led experienced critical care clinicians to conclude that the mechanical ventilation support of a man with ARDS was better provided with a Bird1 than with a Siemens1 ventilator. In fact, the lower PaCO2 measured during patient support with the Bird1 ventilator was explained largely by a decrease in body temperature (Fig. 9), a variable frequently ignored in mechanical ventilation studies. Temperature, in this example, with its attendant reduction in metabolic rate, is a cointervention.

Protocol-Directed Patient Management

575

Figure 8 PaCO2 in a young man with ARDS while supported with a Siemens Servo 300 mechanical ventilator and thereafter with a Bird pressure-limited mechanical ventilator. Abbreviation: ARDS, acute respiratory distress syndrome.

III. Computerized Protocol Experience A. Experience with Computerized Protocols for Adults with Acute Lung Injury

The computerized protocols for mechanical ventilation of ARDS patients developed at the LDS Hospital (35,122,271,342) were exported in standalone PC versions (on Unix and Qnix platforms) to 10 other hospitals in ﬁve states (58). These 10 external hospitals were not involved in the development of the protocol or its rules. Clinician compliance with 38,546 protocol instructions in these 10 hospitals was 95% (58) and indistinguishable from that at the LDS Hospital (272); physicians objected to only 0.3% of the 38,546 instructions (58). The NIH/NHLBI ARDS Network clinician

Figure 9 PaCO2 in a young man with ARDS, as a function of body temperature. Abbreviation: ARDS, acute respiratory distress syndrome.

576

Morris

compliance with a paper protocol based in part on the LDS Hospital computerized protocol was 91% (120). These levels of clinician compliance with paper protocols far exceed the 20% to 50% characteristically associated with guidelines (101,105,115,314,343), but are lower than the 95% compliance experienced with our bedside computerized protocols (272). By the year 2000, this mechanical ventilation protocol had been used in computerized and paper-based versions for over 600,000 hours in over 1164 patients in 35 hospitals (Table 2). The NIH/NHLBI ARDS Network subsequently extended this explicit method approach to other successful randomized controlled clinical trials (120,276,344,345). It was associated with a 23% reduction in mortality of acute lung injury patients (120). These results indicate that an adequately explicit method of care can be effectively transferred for clinical trial use to many different geographically dispersed clinical settings, can reduce variation in clinical decisions, and can signiﬁcantly reduce patient morbidity (58). B. Current Utah Clinical Trial Toolbox Electronic Tools

My colleagues and I found our original rule-based ﬂow-diagram protocols difﬁcult to develop (164,346). We have now developed a more capable, efﬁcient, and user-friendly software tool set built around a ‘‘frame-based’’ system for expressing clinical trial decisions (frames are lists of items necessary to make a decision with an associated logic statement relating the items). These current frame-based adequately explicit computerized protocols, used for bedside decision-support, are now part of the Utah Clinical Trial Toolbox, a comprehensive electronic tools system that brings many Table 2 Adequately Explicit Protocol Experience with Computerized and Paper-Based Mechanical Ventilation Protocol Rules Through Year 2000 Site Computerized version LDS hospital (32,118) 10 other hospitals (55) Paper-based version 24 ARDS Network hospitals (116)a Total (35 hospitals)

Protocol use (hr)

Patients

>200,000 38,528

>200 103

350,880 >600,000

861 >1164

Other hospitals are hospitals to which the LDS Hospital protocol was exported for a clinical trial. These other hospitals had no experience with or part in the development of the computerized mechanical ventilation protocol. a Subsequent NIH/NHLBI ARDS Network clinical trials [higher vs. lower PEEP (352) and FACTT studies] have extended this use to more than another 1000 patients in about 44 participating adult hospitals. Abbreviations: NIH, National Institutes of Health; FACTT, Fluid and Catheter Treatment Trial; ARDS, acute respiratory distress syndrome; NHLBI, National Heart Lung and Blood Institute.

Protocol-Directed Patient Management

577

advantages to clinical trials (347–351). We have also developed both paper bedside instruments (96,164) and computerized tools with which to capture the data necessary to calculate provider response (acceptance or rejection of protocol instructions). This is accomplished in part by software that captures provider compliance as the bedside protocol generates treatment instructions (164,347). The Utah Toolbox is comprised of several desktop applications and a web application. The desktop portion was created with Microsoft (MS) Visual Basic 6.0 using MS Access# as the database (future versions will use the Microsoft.Net development environment or Java development tools). However the Visual Basic# applications will work with any relational DB via ODBC or OLE DB drivers (e.g., Oracle#, SqlServer#, and MySql#). The MS Access# Application itself does not need to be present on client machines. There are two basic Utah Toolbox development tools: the FormBuilder for creating clinical coordinator applications such as data collection forms and the FrameBuilder for capturing the clinical decision-making logic and creating the bedside clinical decision-support applications. After an application is created with a FormBuilder or FrameBuilder template, the development functions are simply turned off, thereby creating the end-user application. Each application has its own MS Access# database. We currently use both a clinical coordinator application, built by the NIH/NHLBI ARDS Network Clinical Coordinating Center at the Massachusetts General Hospital, and a bedside clinical decision-support protocol for the NIH/NHLBI ARDS Network Fluid and Catheter Treatment Trial (FACTT) clinical trial. We have also developed bedside clinical decision-support applications for blood glucose/IV insulin titration, and for the 6 mL/kg (predicted body weight) mechanical ventilation strategy used by the NIH/NHLBI ARDS Network for its mechanical ventilation and FACTT clinical trials (120,276 344,345). We use two additional small applications for HL7 data capture from clinical laboratory computers: (i) A ‘‘socket application’’ (coded with Cþþ), which listens for HL7 lab data at a speciﬁc IP address and saves it in a text ﬁle; (ii) A Visual Basic# Application that opens the ﬁle, ﬁlters only the patients of interest, translates the HL7, and puts the data into an observations table in the coordinator application database. The web application is comprised of secure php pages that are served by an Apache Web server. Encrypted data are transferred to and accessed from MySql# and Oracle# databases on the server. Electronic tools make possible the incorporation of much more complicated knowledge and logic (necessary to generate adequately explicit instructions) at the bedside. Time and effort is necessary to capture, represent, and process the logic. In the past, this has been an arduous process (and has, no doubt, contributed to the rarity of decision-support tools in clinical care) (34,35). The ﬁrst phase involves a consensus process by

578

Morris

recognized experts in the ﬁeld determining the decisions to be made, the component ﬁndings, and the logic relating the ﬁndings. The second phase involves computerization of the rules (in a database or even in ASCII or binary ﬁles). The third phase involves creating a computer application that will process the rules and generate actions in the clinical environment based on clinical data. The FrameAuthor tool of the Utah Clinical Trial Toolbox merges these phases into a single process by automating the second and third phases. The Utah Clinical Trial Toolbox uses methods developed at the LDS Hospital since 1985 and a validated frame-based knowledge engineering tool (164,271,347,353,354). In addition, clinicians can create the knowledge ‘‘frames’’ themselves. A data entry form is automatically generated when a knowledge frame is developed. Frames can be tested at any time during the process with simulated or stored patient data (data can be accumulated and automatically rerun when changes are made in the frame structure). Database queries are created automatically as part of the frame-building process. When the clinicians are satisﬁed, they can turn off the development features and the data entry form becomes the decision-support application for the bedside clinician. The FormBuilder tool allows rapid construction of the administrative electronic forms (with built-in error checking), thereby generating a protocol-speciﬁc clinical coordinator tool. It is also integrated with the bedside tool so that it can automatically retrieve data from the bedside tool. Moreover, the coordinator tool can retrieve data directly from an electronic medical record (EMR). The Coordinator Application can provide automatic error checking and auditing of changed data and can facilitate data transfer to a Clinical Coordinating Center via our web-based communication application. The clinical trial toolbox provides the foundation for an information exchange infrastructure. It is a hybrid system that can function as a stand-alone bedside decision-support or clinical coordinator tool, linking intermittently with a web server, or it can be attached to an EMR system. It provides common electronic tools for linking sites within a research network. We have mapped the knowledge-based dictionary to Logical Observation Identiﬁers and Codes (LOINC) for our current protocols and will also map to Systematized Nomenclature of Medicine (SNOMED) and International Classiﬁcation of Diseases, Ninth Edition (ICD9) terms. In this way, these tools will be as compliant as possible with the evolving national medical informatics standards. IV Fluid and Hemodynamic Support Protocol for FACTT Patients

The data entry screen for the NIH/NHLBI ARDS Network FACTT study requires bedside clinicians to enter only a few mandatory data elements (Fig. 10). The bedside clinician can, if needed, click on the instruction in the white box to read an explanation of the logic. If the bedside clinician

Protocol-Directed Patient Management

579

Figure 10 Computerized FACTT protocol bedside screen. The shaded ﬁelds indicate mandatory data required for a protocol instruction to be generated. Abbreviation: FACTT, Fluid and Catheter Treatment Trial.

Figure 11 Computerized FACTT protocol bedside screen display of the paperbased protocol with the appropriate cell circled. Abbreviation: FACTT, Fluid and Catheter Treatment Trial.

580

Morris

wants a visual display of the paper protocol used by the NIH/NHLBI ARDS Network for the FACTT study, a button click displays the paperbased protocol table with the cell in which the patient currently falls circled in red (Fig. 11). If the bedside clinician wants a display of the footnotes for that speciﬁc cell, a click on the cell reveals the cell-speciﬁc footnotes (Fig. 12). A clinician or knowledge engineer can examine the logic frames used in the generation of a speciﬁc instruction (Fig. 13). The logic on which any protocol table cell (Fig. 14) is based can also be viewed in the logic tree display, either as an overview (Fig. 15) or in detail (Fig. 16). Once the instructions are accepted by the bedside clinician, the screen displays a countdown timer indicating the time remaining until the next protocolmandated subject evaluation (four hours in Fig. 17). Computerized protocols avoid many bedside user errors of interpretation commonly made with paper-based protocols. For example, the ﬁrst two or three FACTT clinical trial patients enrolled at each of 15 NIH/NHLBI ARDS Network hospitals received clinical interventions that deviated from the protocol instructed action 30% of the time (315 paper-based bedside protocol instructions in 20 patients). Ninety-three percent of these errors were due to misinterpretations of the complex protocol footnotes (Fig. 12), not to misidentiﬁcation of the correct protocol table cell (Fig. 11). This was likely due to the complexity of the protocol instruction details. The footnote for Dobutamine administration in the FACTT protocol for Cell 3 (Fig. 11)

Figure 12 Computerized FACTT protocol bedside screen display of the paperbased protocol with the appropriate cell footnote details after the bedside clinician taps on the circled cell. Abbreviation: FACTT, Fluid and Catheter Treatment Trial.

Protocol-Directed Patient Management

581

Figure 13 Logic frame display that aids clinicians and knowledge engineers in evaluating and modifying protocol rules. Clinicians and engineers can examine the basis for a speciﬁc protocol instruction. Each logic frame used in the generation of the speciﬁc protocol instruction is highlighted.

Figure 14 Logic tree partial detailed display for cell 15 of the FACTT protocol. Items with þ sign have not been made more detailed for clarity. Abbreviation: FACTT, Fluid and Catheter Treatment Trial.

582

Morris

Figure 15 Logic tree overview display for Cell 15 of the FACTT protocol. Abbreviation: FACTT, Fluid and Catheter Treatment Trial.

provides an example of the complexity that leads easily to incorrect bedside interpretations of the paper-based protocol. (The detailed footnotes are hidden and automatically interpreted correctly in the computerized protocol. They are available to view on command by the bedside clinician.

Figure 16 Logic tree partial detailed display for cell 15 of the FACTT protocol. Items with þ sign have not been made more detailed for clarity. Abbreviation: FACTT, Fluid and Catheter Treatment Trial.

Protocol-Directed Patient Management

583

Figure 17 Computerized FACTT protocol bedside screen display after the bedside clinician has conﬁrmed acceptance of the protocol instructions. The countdown timer shows four hours remaining until the next FACTT protocol mandated patient assessment. This mandated assessment deﬁnes the longest allowed period between patient observations. Bedside evaluations are otherwise performed according to clinical need. If the patient is reassessed before the mandated maximum four-hour interval, the new data are entered in the bedside computer and the protocol rerun to obtain new instructions. Abbreviation: FACTT, Fluid and Catheter Treatment Trial.

Having captured this decision-making detail in an easily reviewed form at the bedside is an educational asset.) ‘‘Inotrope: If heart rate (HR) < 150/min for >12 hours and right atrial pressure (Pra) < 18 mmHg or echocardiography fails to reveal septal shift or right ventricle (RV) dysfunction, then give Dobutamine: start at 5 m/kg/min. " by 5 q 30 to 60 minutes to 20 mg/kg/min max. If peripheral perfusion adequate for >4 hours #2 mg/ kg/min q 1 to 2 hours as tolerated. Otherwise give Milrinone: loading dose 50 m/kg undiluted IV over 10 min, then 0.375 m/kg/min. " by 0.125 q 30 to 60 minutes to 0.75 mg/kg/min max.’’ The beneﬁts of computerized protocols, compared with paper-based protocols, are clear. The computerized protocol incorporates all of this footnote complexity plus additional detail but it remains transparent to the bedside user (although the underlying logic and details are available upon demand). The user only enters a few data elements and then reads

584

Morris

the computer-generated patient-speciﬁc therapy instruction (see example in Table 1). The bedside clinician can always decline the instruction if there is a compelling reason to do so. The bedside computerized protocols can standardize clinical decisions and can reduce noise associated with both the experimental intervention and with cointerventions (34,35,131,135). The use of explicit detailed methods for the experimental therapies and for the general care of the patient will increase the probability of ﬁnding clinically signiﬁcant and relevant differences between the experimental groups. Blood Glucose/IV Insulin Protocol

A computerized blood glucose/insulin protocol was developed, reﬁned, and implemented using the Utah Clinical Trials Toolbox. The bedside computer screen (Fig. 18) is simpler than that for the FACTT trial (Fig. 10). The blood sugar and insulin IV infusion values are displayed with the resulting treatment instructions and a countdown timer that indicates the remaining time until the next scheduled blood sugar evaluation. If desired, the clinician can display the protocol instructions, whether accepted or declined, and the blood glucose values (Fig. 19). The clinician can choose a graphical display as well (Fig. 20). The computerized protocols compute the insulin dose based on the deviation of measured blood glucose from the center of the target range, the rate of change of blood glucose, and the current insulin dose. The computerized blood glucose protocol has been used as a stand-alone bedside tool in a laptop PC. Through 2003, LDS Hospital clinicians encountered no blood glucose values less than 40 mg/dL in either higher (target range 121–180 mg/dL) or lower (target range 81–115 mg/dL) glucose protocols. Clinician compliance with protocol instructions was 85% with a paperbased and 97% with the computerized protocols (see clinician compliance in Table 3). Web-Based Tools

Our current Web site functions (Fig. 21) include the Web screen for transfer of clinical coordinator data to the study monitors at the clinical coordinating center (Fig. 22). This is only one of the Web pages within a Web tool that facilitates (i) data transfer from the local site clinical coordinator to the Web server; (ii) data-ﬁeld–speciﬁc query and answer exchanges between the local site clinical coordinator and the central Clinical Coordinating Center monitor; (iii) administrative functions that include document version tracking and archiving, meeting and conference call scheduling, etc.; and (iv) questionnaire functions that link the clinical sites with the central Clinical Coordinating Center (Massachusetts General Hospital, for our ARDS Network clinical trial).

Protocol-Directed Patient Management

Figure 18

585

Computerized glucose/insulin protocol bedside screen.

Education and Training Tools

These tools can minimize the training and technical skill needed to conduct a rigorous clinical study by using common electronic tools (Utah Clinical Trial Toolbox) to provide an interoperable information interchange infrastructure. We plan to expand the current ‘‘help-on-demand’’ and step-by-step

586

Morris

Figure 19

Computerized glucose/insulin protocol instruction history.

Figure 20

Computerized glucose/insulin protocol graphical data display.

Blood glucose (mg/dL) Year

Number of Number of blood glucose Protocol type patients measurements

1994 1999–2002 2002–2003 2002–2003

None Paper Computer Computer

450 1600 34 26

> 4,000 > 16,000 658 1,105

Target range – 121–180 121–180 81–115

Mean SD

% Blood glucose < 60 mg/dL

% Instructions accepted by clinicians

180 68 145 40 149 31 120 34

0.8 0.5 0 1

– 85 97 97

Protocol-Directed Patient Management

Table 3 Blood Glucose Experience Without and With Both Paper-Based and Computerized Protocols at LDS Hospital Shock-Trauma ICU from 1994 Through 2003

587

588

Morris

Figure 21

Utah Clinical Trial Toolbox Web site home page.

animated tutorials built with the Flash MX# application. We have developed a clinical coordinator tool step-by-step tutorial that replaces a personal educator (Fig. 23). Current Application of Bedside Protocols

Computerized protocols are currently routinely used for clinical trials in the hospitals of the Utah Critical Care Treatment Group of the NIH/NHLBI

Figure 22 Utah Clinical Trial Toolbox clinical coordinator page for uploading data to the Web site server.

Protocol-Directed Patient Management

589

Figure 23 Computerized FACTT clinical coordinator application tutorial. Abbreviation: FACTT, Fluid and Catheter Treatment Trial.

ARDS Network. Application of three computerized protocols in ICU patients and in subjects enrolled in NIH/NHLBI ARDS Network clinical trials from 15 June to 30 September 2004 is summarized in Table 4.

IV. Summary Intensive care accounts for 20% of the total hospital health care expenditures in the United States. Although the majority of care occurs in adult ICUs, pediatric critical illness is a source of signiﬁcant short- and long-term morbidity, and care of these children consumes signiﬁcant health care resources. Currently, well-designed adequately powered clinical trials are uncommon in adult and rare in pediatric critical care. Currently operational integrated electronic tools such as the Utah Clinical Trial Toolbox can expedite the conduct, improve the data and research quality, and increase the efﬁciency of ICU clinical research. This requires the medical community to adopt a new ICU clinical investigative strategy that utilizes electronic tools to link many different clinical sites into an extended human experimental research laboratory. Currently available

590 Table 4 Computerized Bedside Protocol Experience with the Utah Clinical Trial Toolbox for FACTT (NIH/NHLBI FACTT) Subjects at the Utah Clinical Site from 15th June Through 30th September 2004

Hospital

Hospital care type

Total number of hospital beds

Alta View Cottonwood

Primary Secondary

72 180

LDS

Tertiary

467

Total

Protocol FACTT ﬂuid FACTT ﬂuid FACTT mechanical ventilation Glucose FACTT ﬂuid FACTT mechanical ventilation

# Patients

# Hrs protocol use

# Instructions

% Instructions followed by clinicians

1 2 1

10 212 279

7 65 190

100 91 92

70 3 2

8,823 493 311

3,695 222 234

94 85 94

79

10,128

4,413

93

Abbreviations: FACTT, Fluid and Catheter Treatment Trial; NIH, National Institutes of Health; NHLBI, National Heart Lung and Blood Institute.

Morris

Protocol-Directed Patient Management

591

decision-support tools can provide the adequately explicit methods needed for this extended laboratory. These adequately explicit methods can elicit identical responses from different clinicians when faced with identical clinical data. Importantly, the adequately explicit methods preserve individualized, patient-speciﬁc treatment. This large-scale clinical laboratory could more efﬁciently conduct clinical ICU studies and, with the same common electronic tools, could more rapidly extend ICU research results to clinical ICU practice than is currently possible. The use of adequately explicit detailed computerized bedside protocols should increase the probability of ﬁnding differences between experimental groups. This new strategy should enable investigators to uniformly implement and distribute knowledge-based ICU care and to address ICU clinical problems such as lungprotective ventilation that have deﬁed resolution with traditional clinical investigation approaches. Acknowledgments I am indebted to Drs. Robert Crapo, Gordon Rubenfeld, Jason Christie, and Michael Dean for critical review and helpful suggestions. They and Mr. Thomas Graydon and Ms. Louise Hague have my thanks for reviewing the manuscript. References 1. Wennberg JE, Gittelsohn A. Small area variation analysis in health care delivery. Science 1973; 142:1102–1108. 2. Wennberg JE. Unwarranted variations in healthcare delivery: implications for academic medical centres. BMJ 2002; 325(7370):961–964. 3. Morris AH. Iatrogenic illness: a call for decision support tools to reduce unnecessary variation. Qual Saf Health Care 2004; 13(1):80–81. 4. Silverman W. Doing more good than harm. Ann NY Acad Sci 1993; 703:5–11. 5. Warren K, Mosteller F, eds. Doing more good than harm: the evaluation of health care interventions. New York: The New York Academy of Sciences, 1993. 6. Kohn L, Corrigan J, Donaldson M, eds. To err is human—building a safer health system. Washington, D.C.: National Academy Press, 1999. 7. CoQoHCi America. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, D.C.: National Academy Press, 2001. 8. Horwitz RI, Singer BH, Makuch RW, Viscoli CM. Can treatment that is helpful on average be harmful to some patients? A study of the conﬂicting information needs of clinical inquiry and drug regulation. J Clin Epidemiol 1996; 49(4):395–400. 9. Redelmeier DA, Ferris LE, Tu JV, Hux JE, Schull MJ. Problems for clinical judgement: introducing cognitive psychology as one more basic science. CMAJ 2001; 164(3):358–360.

592

Morris

10. Berwick DM. Disseminating innovations in health care. JAMA 2003; 289(15):1969–1975. 11. Runciman WB, Merry A, Tito F. Error, blame and the law in health care—an antipodean perspective. Ann Intern Med 2003; 138(12):974–979. 12. Barach P, Berwick DM. Patient safety and the reliability of health care systems. Ann Intern Med 2003; 138(12):997–998. 13. Sox HC. Improving patient care. Ann Intern Med 2003; 138(12):996. 14. Senn S. Individual response to treatment: is it a valid assumption? BMJ 2004; 329(7472):966–968. 15. Lake APJ. Every prescription is a clinical trial. BMJ 2004; 329(7478):1346. 16. Corke CF, Stow PJ, Green DT, Agar JW, Henry MJ. How doctors discuss major interventions with high risk patients: an observational study. BMJ 2005; 330(7484):182. 17. Berwick DM. My right knee. Ann Intern Med 2005; 142(2):121–125. 18. Redelmeier DA. The cognitive psychology of missed diagnoses. Ann Intern Med 2005; 142(2):115–120. 19. Kahneman D, Slovik P, Tversky A. Judgment Under Uncertainty: Heuristics and Biases. Cambridge: Cambridge University Press, 1982. 20. Hulley S, Cummings S. Designing Clinical Research. Baltimore: Williams and Wilkins, 1988. 21. Shewart W. Economic Control of Quality of Manufactured Product. New York, NY: D. Van Nostrand Co., Inc, 1931 (republished in 1980, American Society for Quality Control, 230 W. Wells St, Milwaukee, Wisconsin 53203). 22. Deming W. Quality, Productivity, and Competitive Position. Cambridge, MA: Center for Advanced Engineering Study, Massachusetts Institute of Technology, 1982. 23. Deming W. Out of the Crisis. Cambridge, MA: Center for Advanced Engineering Study, Massachusetts Institute of Technology, 1986. 24. Imai M. Kaizen—The Key to Japan’s Competitive Success. New York, NY: McGraw-Hill Publishing Company, 1986. 25. Walton M. The Deming Management Method. New York: Putnam Publishing Group (Perigee Book, 1986. 26. James B, Horn S, Stephenson R. Management by fact: What is CPI and how is it used? In: Horn S, Hopkins D, eds. Clinical Practice Improvement: A New Technology for Developing Cost-Effective Quality Health Care. New York: Faulker & Gray, Inc., 1994:39–54. 27. Horn S, Hopkins D, eds. Clinical Practice Improvement: a New Technology for Developing Cost-Effective Quality Health Care. New York: Faulker & Gray, Inc., 1994. 28. Berwick DM. Eleven worthy aims for clinical leadership of health system reform. JAMA 1994; 272(10):797–802. 29. James BC, Hammond ME. The challenge of variation in medical practice. Arch Pathol Lab Med 2000; 124(7):1001–1003. 30. Akhtar SR, Weaver J, Pierson DJ, Rubenfeld GD. Practice variation in respiratory therapy documentation during mechanical ventilation. Chest 2003; 124(6):2275–2282.

Protocol-Directed Patient Management

593

31. Rubenfeld G, Cooper C, Carter G, Thompson T, Hudson L. Barriers to providing lung-protective ventilation to patients with acute lung injury. Crit Care Med 2004; 32(6):1289–1293. 32. Arkes H, Hammond K. Judgment and Decision Making: An Interdisciplinary Reader. Cambridge: Cambridge University Press, 1986. 33. Arkes H. Impediments to accurate clinical judgment and possible ways to minimized their impact. In: Arkes H, Hammond K, eds. Judgment and Decision Making: An Interdisciplinary Reader. Cambridge: Cambridge University Press, 1986:582–592. 34. Morris A. Algorithm-based decision making. In: Tobin M, ed. Principles and Practice of Intensive Care Monitoring. New York: McGraw-Hill, Inc., 1998:1355–1381. 35. Morris A. Developing and implementing computerized protocols for standardization of clinical decisions. Ann Intern Med 2000; 132:373–383. 36. Leape LL, Berwick DM, Bates DW. What practices will most improve safety? Evidence-based medicine meets patient safety [comment]. JAMA 2002; 288(4):501–507. 37. Kozer E, Seto W, Verjee Z, et al. Prospective observational study on the incidence of medication errors during simulated resuscitation in a paediatric emergency department. BMJ 2004; 329(7478):1321. 38. Schiff GD, Klass D, Peterson J, Shah G, Bates DW. Linking laboratory and pharmacy: opportunities for reducing errors and improving care. Arch Intern Med 2003; 163(8):893–900. 39. Lamb RM, Studdert DM, Bohmer RM, Berwick DM, Brennan TA. Hospital disclosure practices: results of a national survey. Health Affairs 2003; 22(2):73–83. 40. Zhang J, Patel VL, Johnson TR. Medical error: is the solution medical or cognitive? J Am Med Inform Assoc 2002; 9(6 suppl):S75–S77. 41. Tversky A, Kahneman D. Availability: a heuristic for judging frequency and probability. In: Kahneman D, Slovic P, Tversky A, eds. Judgment Under Uncertainty: Heuristics and Biases. Cambridge, UK: Cambridge University Press, 1982:163–178. 42. Jennings D, Amabile T, Ross L. Informal covariation assessment: Data-based versus theory-based judgments. In: Kahneman D, Slovic P, Tversky A, eds. Judgment Under Uncertainty: Heuristics and Biases. Cambridge, UK: Cambridge University Press, 1982:211–230. 43. McDonald CJ. Protocol-based computer reminders, the quality of care and the non-perfectability of man. N Engl J Med 1976; 295:1351–1355. 44. Abramson NS, Wald KS, Grenvik ANA, Robinson D, Snyder JV. Adverse occurrences in intensive care units. JAMA 1980; 244:1582–1584. 45. Morris AH, Chapman RH, Gardner RM. Frequency of technical problems encountered in the measurement of pulmonary artery wedge pressure. Crit Care Med 1984; 12(3):164–170. 46. Morris AH, Chapman RH, Gardner RM. Frequency of wedge pressure errors in the ICU. Crit Care Med 1985; 13:705–708. 47. Morris AH. Elimination of pulmonary wedge pressure errors commonly encountered in the ICU. Cardiologia (Italy) 1985; 30(10):941–943.

594

Morris

48. Iberti T, Fischer E, Leibowitz M, et al. A multicenter study of physician’s knowledge of the pulmonary artery catheter. JAMA 1990; 264:2928–2932. 49. Iberti TJ, Daily EK, Leibowitz AB, Schecter CB, Fischer EP, Silverstein JH. Assessment of critical care nurses’ knowledge of the pulmonary artery catheter. The Pulmonary Artery Catheter Study Group. Crit Care Med 1994; 22:1674–1678. 50. Leape L. Error in medicine. JAMA 1994; 272:1851–1857. 51. Gnaegi A, Feihl F, Perret C. Intensive care physicians’ insufﬁcient knowledge of right-heart catheterization at the bedside: time to act? [see comments]. Crit Care Med 1997; 25(2):213–220. 52. Wu A, Folkman S, McPhee S, Lo B. Do house ofﬁcers learn from their mistakes? JAMA 1991; 265:2089–2094. 53. Grimshaw J, Russell I. Effect of clinical guidelines on medical practice: a systematic review of rigorous evaluations. Lancet 1993; 342:1317–1322. 54. Safran C, Rind D, Davis R, et al. Effects of a knowledge-based electronic patient record on adherence to practice guidelines. MD Comput 1996; 13(1):55–63. 55. Grimm RH, Shimoni K, Harlan WR, Estes EH. Evaluation of patient-care protocol use by various providers. N Engl J Med 1975; 292:507–511. 56. Wirtschafter DD, Scalise M, Henke C, Gams RA. Do information systems improve the quality of clinical research? Results of a randomized trial in a cooperative multi-institutional cancer group. Comput Biomed Res 1981; 14:78–90. 57. Johnston M, Langton K, Haynes B, Mathieu A. Effects of computer-based clinical decision support systems on clinician performance and patient outcome. Ann Intern Med 1994; 120:135–142. 58. East T, Heermann L, Bradshaw R, et al. Efﬁcacy of computerized decision support for mechanical ventilation: results of a prospective multi-center randomized trial. Proc AMIA Symp 1999:251–255. 59. Mullett CJ, Evans RS, Christenson JC, Dean JM. Development and impact of a computerized pediatric antiinfective decision support program. Pediatrics 2001; 108(4):E75. 60. Morris AH. Decision support and safety of clinical environments. Qual Saf Health Care 2002; 11:69–75. 61. Hoch I, Heymann AD, Kurman I, Valinsky LJ, Chodick G, Shalev V. Countrywide computer alerts to community physicians improve potassium testing in patients receiving diuretics. J Am Med Inform Assoc 2003; 10(6): 541–546. 62. Carmichael L, Dorinsky P, Bernard G, Wheeler A. Survey of diagnosis and therapy of adult respiratory distress syndrome (ARDS). Am Rev Resp Dis 1993; 147(4 Pt 2):A347. 63. Marinelli W, Ingbar D. Diagnosis and management of acute lung injury. Clin Chest Med 1994; 15(3):517–546. 64. Michard F, Teboul J-L. Predicting ﬂuid responsiveness in ICU patients: a critical analysis of the evidence. Chest 2002; 121(6):2000–2008. 65. Levin L. Clarifying the treatment of nonarteritic anterior ischemic optic neuropathy. JAMA 1995; 273(8):666–667.

Protocol-Directed Patient Management

595

66. The Cardiac Arrhythmia Suppression Trial (CAST) Investigators. Preliminary report: effect of encainide and ﬂecainide on mortality in a randomized trial of arrhythmia suppression after myocardial infarction. N Engl J Med 1989; 321:406–412. 67. Greene H, Roden D, Katz R, et al. The Cardiac Arrythmia Suppression Trial: ﬁrst cast . . . then cast-II. J Am Col Cardiol 1992; 19:894–898. 68. Group IONDTR. Optic nerve decompression surgery for nonarteritic anterior ischemic optic neuropathy (NAION) is not effective and may be harmful. The Ischemic Optic Neuropathy Decompression Trial Research Group. JAMA 1995; 273(8):625–632. 69. Group IONDTR. Ischemic Optic Neuropathy Decompression Trial: twentyfour-month update. Arch Ophthalmol 2000; 118(6):793–797. 70. Campbell D, Stanley J. Experimental and Quasi-experimental Designs for Research (reprinted from Handbook of Research on Teaching, 1963). Boston: Houghton Mifﬂin Co., 1966. 71. von Bertalanffy L. General System Theory. New York: George Braziller, 1968. 72. Miller J. Living Systems. New York: McGraw-Hill Book Company, 1978. 73. Pearl J. Causality: Models, Reasoning, and Inference. Cambridge, UK: Cambridge, 2000. 74. Reason J. Human Error. Cambridge, UK: Cambridge University Press, 1990. 75. Reason J. Human error: models and management. BMJ 2000; 320(7237): 768–770. 76. Williamson J, Goldschmidt P, Jillson I. Medical Practice Information Demonstration Project-Final Report (Contract #282-77-0068GS). Baltimore, MD: Ofﬁce of the Asst. Secretary of Health, DHEW, 1979. 77. McDonald CJ, Wilson GA, McCabe G Jr. Physician response to computer reminders. JAMA 1980; 244(14):1579–1581. 78. West J. An autopsy method for evaluating trauma care. J Trauma 1981; 21(1):32–34. 79. Browner W, Newman T. The analogy between diagnostic tests and clinical research. JAMA 1987; 257(18):2459–2463. 80. ACo Physicians. Working conditions and supervision for residents in internal medicine programs: recommendations. Ann Intern Med 1989; 110(8): 657–663. 81. Brook R. Using scientiﬁc information to improve quality of health care. In: Warren K, Mosteller F, eds. Doing More Good Than Harm: The Evaluation of Health Care Interventions. New York: The New York Academy of Sciences, 1993:74–85. 82. Chalmers T, Lau J. Randomized controlled trials and meta-anlayses in gastroenterology: major achievements and future potential. In: Warren K, Mosteller F, eds. Doing More Good Than Harm: The Evaluation of Health Care Interventions. New York: The New York Academy of Sciences, 1993:96–106. 83. Chan L, Schonfeld N. How much information is lost during processing: a case study of emergency department records. Comput Biomed Res 1993; 26: 582–591. 84. McNeil B. Use of claims data to monitor patients over time: acute myocardial infarction as a case study. In: Warren K, Mosteller F, eds. Doing More Good

596

85.

86.

87. 88. 89. 90. 91.

92. 93.

94. 95.

96.

97. 98. 99. 100.

101.

Morris Than Harm: The Evaluation of Health Care Interventions. New York: The New York Academy of Sciences, 1993:63–73. O’Connor G, Plume S, Wennberg J. Regional organization for outcomes research. In: Warren K, Mosteller F, eds. Doing More Good Than Harm: The Evaluation of Health Care Interventions. New York: The New York Academy of Sciences, 1993:44–51. Wennberg J, Barry M, Fowler F, Mulley A. Outcomes research, PORTs, and health care reform. In: Warren K, Mosteller F, eds. Doing More Good Than Harm: The Evaluation of Health Care Interventions. New York: The New York Academy of Sciences, 1993:52–62. Tyson NdG. Signal versus noise. Natural History 1996; 105(10):72–76. Schulz E, Barrett J, Price C. Read Code quality assurance: from simple syntax to semantic stability. J Am Med Inform Assoc 1998; 5(4):337–346. Sudeep N, Anuradha L, Obasanjo O, Chaisson R. Errors in the treatment of tuberculosis in Baltimore. Chest 2000; 117(3):734–737. Palevsky PM, Bhagrath R, Greenberg A. Hypernatremia in hospitalized patients [see comments]. Ann Intern Med 1996; 124(2):197–203. Nakhleh RE, Zarbo RJ. Amended reports in surgical pathology and implications for diagnostic error detection and avoidance: a College of American Pathologists Q-probes study of 1,667,547 accessioned cases in 359 laboratories [see comments]. Arch Pathol Lab Med 1998; 122(4):303–309. Krizek TJ. Surgical error: ethical issues of adverse events [in process citation]. Arch Surg 2000; 135(11):1359–1366 [MEDLINE record in process]. Gopher D, Olin M, Badihi Y, et al. The nature and causes of human errors in a medical intensive care unit. In: Human Factors and Ergonomics Society Annual Meeting Proceedings, 1989. 1989:956–960. Ontario-Intensive-Care-Study-Group. Evaluation of right heart catheterization in critically ill patients. Crit Care Med 1992; 20(7):928–933. Guidelines-Committee-Society-of-Critical-Care-Medicine. Guidelines for the care of patients with hemodynamic instability associated with sepsis. Crit Care Med 1992; 20(7):1057–1059. Morris A. Hemodynamic guidelines [see editorial: PC Gay. Of pundits, professors, and pollsters. Crit Care Med 1993; 21(7):959–60]. Crit Care Med 1993; 21(7):1096. Starﬁeld B. Quality-of-care research: Internal elegance and external relevance. JAMA 1998; 280(11):1006–1008. Cooper M. The Dartmouth Atlas of Health Care 1999 (http://www.dartmouth.edu/atlas). Chicago, Illinois: Health Forum (AHA), 1999. Chassin M, Galvin R. Quality at NRoHC. The urgent need to improve health care quality. JAMA 1998; 280(11):1000–1005. Evans RS, Pestotnik SL, Burke JP, Gardner RM, Larsen RA, Classen DC. Reducing the duration of prophylactic antibiotics use through computer monitoring of surgical patients. Ann Pharmacother 1990; 24(4):351–354. Evans RS, Pestotnik SL, Classen DC, et al. A computer-assisted management program for antibiotics and other antiinfective agents. N Engl J Med 1998; 338(4):232–238.

Protocol-Directed Patient Management

597

102. Lesar T, Briceland L, Stein D. Factors related to error in medication prescribing. JAMA 1997; 277(4):312–317. 103. Leape LL, Brennan TA, Laird N, et al. The nature of adverse events in hospitalized patients. Results of the Harvard Medical Practice Study II. N Engl J Med 1991; 324(6):377–384. 104. Montori VM, Wilczynski NL, Morgan D, Haynes RB, for the Hedges Team. Optimal search strategies for retrieving systematic reviews from Medline: analytical survey. BMJ 2004 (bmj. 38336.804167.47). 105. Nelson E, Splaine M, Batalden P, Plume S. Building measurement and data collection into medical practice. Ann Intern Med 1998; 128:460–466. 106. Schacker T, Collier AC, Hughes J, Shea T, Corey L. Clinical and epidemiologic features of primary HIV infection [published erratum appears in Ann Intern Med 1997; 126(2):174] [see comments]. Ann Intern Med 1996; 125(4):257–264. 107. Kiernan M, King AC, Kraemer HC, Stefanick ML, Killen JD. Characteristics of successful and unsuccessful dieters: an application of signal detection methodology. Ann Behav Med 1998; 20(1):1–6. 108. Galuska DA, Will JC, Serdula MK, Ford ES. Are health care professionals advising obese patients to lose weight? [see comments]. JAMA 1999; 282(16):1576–1578. 109. Dickerson JE, Hingorani AD, Ashby MJ, Palmer CR, Brown MJ. Optimisation of antihypertensive treatment by crossover rotation of four major classes [see comments]. Lancet 1999; 353(9169):2008–2013. 110. Redman B. Clinical practice guidelines as tools of public policy: conﬂicts of purpose, issues of autonomy, and justice. J Clin Ethics 1996; 5(4):303–309. 111. Cochrane G. Compliance in asthma. Eur Respir Rev 1998; 8(56):239–242. 112. Jones P. Health status, quality of life and compliance. Eur Respir Rev 1998; 8(56):243–246. 113. Lomas J, Anderson GM, Domnick-Pierre K, Vayda E, Enkin MW, Hannah WJ. Do practice guidelines guide practice? The effect of a consensus statement on the practice of physicians. NEJM 1989; 321(19):1306–1311. 114. Greco P, Eisenberg J. Changing physicians’ practices. New Engl J Med 1993; 329(17):1271–1274. 115. Singer M, Haft R, Barlam T, Aronson M, Shafer A, Sands K. Vancomycin control measures at a tertiary-care hospital: impact of interventions on volume and patterns of use. Infect Control Hosp Epidemiol 1998; 19(4):248–253. 116. Pritchard RS, Fisher ES, Teno JM, et al. Inﬂuence of patient preferences and local health system characteristics on the place of death. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Risks and Outcomes of Treatment [see comments]. J Am Geriatr Soc 1998; 46(10):1242–1250. 117. Teno J, Lynn J, Connors AFJ, et al. The illusion of end-of-life resource savings with advance directives. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment [see comments]. J Am Geriatr Soc 1997; 45(4):513–518. 118. Teno JM, Licks S, Lynn J, et al. Do advance directives provide instructions that direct care? SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment [see comments]. J Am Geriatr Soc 1997; 45(4):508–512.

598

Morris

119. Lo B. Improving care near the end of life: Why is it so hard? JAMA 1995; 274:1634–1636. 120. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for Acute Lung Injury and the Acute Respiratory Distress Syndrome. New Engl J Med 2000; 342(18):1301–1308. 121. Bernard GR, Vincent JL, Laterre PF, et al. Efﬁcacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med 2001; 344(10): 699–709. 122. Morris A, Wallace C, Menlove R, et al. Randomized clinical trial of pressurecontrolled inverse ratio ventilation and extracorporeal CO2 removal for ARDS [erratum 1994; 149(3 Pt 1):838; Letters to the editor 1995; 151(1):255–256, 1995; 151(3):1269–1270, and 1997; 156(3):1016–1017]. Am J Respir Crit Care Med 1994; 149(2):295–305. 123. Eidelman L, Sprung C. Why have new effective therapies for sepsis not been developed? Crit Care Med 1994; 22(8):1330–1334. 124. Lefering R, Neugebauer EA. Steroid controversy in sepsis and septic shock: a meta-analysis [see comments]. Crit Care Med 1995; 23(7):1294–1303. 125. Bernard G. Sepsis trials. Intersection of investigation, regulation, funding, and practice. Am Rev Respir Crit Care Med 1995; 152:4–10. 126. Bone R. Sepsis and controlled clinical trials: the odyssey continues. Crit Care Med 1995; 23(8):1313–1315. 127. Cronin L, Cook DJ, Carlet J, et al. Corticosteroid treatment for sepsis: a critical appraisal and meta-analysis of the literature [see comments]. Crit Care Med 1995; 23(8):1430–1439. 128. Zapol WM, Snider MT, Hill JD, et al. Extracorporeal membrane oxygenation in severe acute respiratory failure. JAMA 1979; 242:2193–2196. 129. Bartlett RH, Morris AH, Fairley HB, Hirsch R, O’Connor N, Pontoppidan H. A prospective study of acute hypoxic respiratory failure. Chest 1986; 89:684–689. 130. Brower R, Lanken P, MacIntyre N, et al. Higher versus lower positive endexpiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351(4):327–336. 131. Morris A. Treatment algorithms and protocolized care. Curr Opin Crit Care 2003; 9(3):236–240. 132. Hamberger E. Transdisciplinarity: A Scientiﬁc Essential. Ann NY Acad Sci 2004; 1028(1):487–496. 133. Marshall JC. Clinical trials of mediator-directed therapy in sepsis: what have we learned? Intensive Care Med 2000; 26(6 suppl 1):S75–S83. 134. Cook DJ, Reeve BK, Guyatt GH, et al. Stress ulcer prophylaxis in critically ill patients. Resolving discordant meta-analyses [see comments]. JAMA 1996; 275(4):308–314. 135. Morris A, Cook D. Clinical trial issues in mechanical ventilation. In: Marini J, Slutsky A, eds. Physiologic Basis of Ventilatory Suport. New York: Marcel Dekker, Inc., 1998:1359–1398.

Protocol-Directed Patient Management

599

136. LeLorier J, Gregoire G, Benhaddad A, Lapierre J, Derderian F. Discrepancies between meta-analyses and subsequent large randomized, controlled trials [see comments]. New Engl J Med 1997; 337(8):536–542. 137. Johnson BE. Review: prophylactic use of vitamin D reduces falls in older persons. Evid Based Med 2004; 9(6):169. 138. NIH. Overview of the NIH Roadmap (http://nihroadmap.nih.gov/overview. asp). In: Zerhouni EA, ed. NIH Roadmap: Accelerating Medical Discovery to Improve Health. Bethesda: NIH, 2003. 139. Giancoli D. Physics. 3rd. Englewood Cliffs, NJ: Prentice Hall, 1995. 140. Rothman K, Greenland S. Modern Epidemiology. 2nd. Philadelphia, Pennsylvania: Lippincott-Raven, 1998. 141. Barrow JD. The Book of Nothing. New York: Pantheon Books, 2000. 142. Babbie E. Observing Ourselves: Essays in Social Research. Belmont, CA: Wadsworth Pub. Co., 1986. 143. Guyatt G, Sackett D, Cook D. User’s guide to the medical literature: II. How to use an article about therapy or prevention; A. Are the results of the study valid? JAMA 1993; 270(21):2598–2601. 144. Guyatt G, Sackett D, Cook D. User’s guide to the medical literature: II. How to use and article about therapy or prevention; B. What were the results and will they help me in caring for my patient? JAMA 1994; 271(1):59–63. 145. Justice AC, Covinsky KE, Berlin JA. Assessing the generalizability of prognostic information. Ann Intern Med 1999; 130(6):515–524. 146. Emanuel EJ, Wendler D, Grady C. What makes clinical research ethical? JAMA 2000; 283(20):2701–2711. 147. Hawe P, Shiell A, Riley T. Complex interventions: how ‘‘out of control’’ can a randomised controlled trial be? BMJ 2004; 328(7455):1561–1563. 148. Babbie E. Survey Research Methods. Belmont, CA: Wadsworth Publishing Co., 1990. 149. Pocock SJ. Clinical Trials: A Practical Approach. New York, NY: John Wiley & Sons, 1983. 150. Bailey RW. Human Performance Engineering. 3rd. Upper Saddle River, NJ: Prentice Hall, 1996. 151. Piantadosi S. Clinical Trials—A Methodologic Perspective. New York: John Wiley & Sons, Inc., 1997. 152. Brook RH, McGlynn EA, Shekelle PG. Deﬁning and measuring quality of care: a perspective from US researchers. Int J Qual Health Care 2000; 12(4):281–295. 153. Hulley S, Cummings S, Warren S, Grady D, Hearst N, Newman T. Designing Clinical Research. 2nd. Philadelphia: Lippincott Williams and Wilkins, 2001. 154. Sackett D, Haynes R, Guyatt G, Tugwell P. Clinical Epidemiology: A Basic Science for Clinical Medicine. 2nd. Boston: Little, Brown, 1991. 155. Gigerenza G, Swijtink Z, Porter T, Daston L, Beatty J, Kru¨ger L. The Empire of Chance. Cambridge: Cambridge University Press, 1989. 156. Matthews J. Quantiﬁcation and the Quest for Medical Certainty. Princeton: Princeton University Press, 1995. 157. McDonald CJ. Medical heuristics: the silent adjudicators of clinical practice [see comments]. Ann Intern Med 1996; 124(1 Pt 1):56–62.

600

Morris

158. Morris AH. Paradigms in management. In: Pinsky M, Dhainaut J, eds. Pathophysiologic Foundations of Critical Care Medicine. Baltimore: Williams and Wilkins, 1993:193–206. 159. Blesser W. A Systems Approach to Biomedicine. New York: McGraw-Hill Book Company, 1969. 160. Hebert PC, Cook DJ, Wells G, Marshall J. The Design of Randomized Clinical Trials in Critically Ill Patients. Chest 2002; 121(4):1290–1300. 161. Friedman C, Gatti G, Elstein A, Franz T, Murphy G, Wolf F. Are clinicians correct when they believe they are correct? Implications for medical decision support. Medinfo 2001; 10(Pt 1):454–458. 162. Morris A, East T, Wallace C, et al. Ethical implications of standardization of ICU care with computerized protocols. In: Ozbolt J, ed. Proceedings of the 18th Annual Symposium on Computer Applications in Medical Care, 1994; Philadelphia: Hanley & Belfus, Inc., 1994:501–505. 163. Morris A. Ethical implications of standardizing clinical decisions with computerized protocols. In: Vincent J-L, ed. Yearbook of Intensive Care and Emergency Medicine 1999. Berlin: Springer, 1999:691–697. 164. Morris A. Evaluating and reﬁning a hemodynamic protocol for use in a multicenter ARDS clincial trial. Am J Respir Crit Care Med (ATS Proceedings Abstracts) Suppl 2000; 161(3):A378. 165. Morris A. Oscillation and damping in human decision-making systems. Am J Respir Crit Care Med (ATS Proceedings Abstracts) Suppl 2000; 161(3):A552. 166. Dawes R, Faust D, Meehl P. Clinical versus actuarial judgement. Science 1989; 243:1668–1674. 167. Leiner G, Abramowitz S, Small M. Pulmonary function testing in laboratories associated with residency training programs in pulmonary diseases. Am Rev Respir Dis 1969; 100:240–244. 168. Hirsh J. Oral anticoagulant drugs. N Engl J Med 1991; 324(26):1865–1875. 169. Classen DC, Evans RS, Pestotnik SL, Horn SD, Menlove RL, Burke JP. The timing of prophylactic administration of antibiotics and the risk of surgicalwound infection. N Engl J Med 1992; 326:281–286. 170. Pestotnik S, Classen D, Evans R, Burke J. Implementing antibiotic practice guidelines through computer-assisted decision support: clinical and ﬁnancial outcomes. Ann Intern Med 1996; 124(10):884–890. 171. Evans R, Classen D, Pestotnik S, Clemmer T, Weaver L, Burke J. A decision support tool for antibiotic therapy. In: Safran C, ed. Proceedings 19th Annual Symposium on Computer Applications in Medical Care (SCAMC), Oct 28Nov 2, 1995. New Orleans: McGraw-Hill, Inc., 1995:651–655. 172. Evans RS, Burke JP, Pestotnik SL, Classen DC, Menlove RL, Gardner RM. Prediction of hospital infections and selection of antibiotics using an automated hospital database. In: Proceedings 14th Annual Symposium on Computer Applications in Medical Care, 1990. Baltimore, MD: IEEE Computer Society Press, 1990:663–667. 173. Tierney W, McDonald C, Martin D, Hui S, Rogers M. Computerized display of past test results. Ann Intern Med 1987; 107(4):569–574. 174. McDonald C, Tierney W. Computer-stored medical records—their future role in medical practice. JAMA 1988; 259:3433–3440.

Protocol-Directed Patient Management

601

175. Tierney W, McDonald C, Hui S, Martin D. Computer predictions of abnormal test results—effects on outpatient testing. JAMA 1988; 259:1194–1198. 176. Tierney W, Miller M, McDonald C. The effect on test ordering of informing physicians of the charges for outpatient diagnostic tests. N Engl J Med 1990; 322:1499–1504. 177. Pestotnik SL, Evans RS, Burke JP, Gardner RM, Classen DC. Development of a computerized infectious disease monitor (CIDM). Comput Biomed Res 1990; 18:103–113. 178. Classen DC, Pestotnik SL, Evans RS, Burke JP. Computerized surveillance of adverse drug events in hospital patients. JAMA 1991; 266(20):2847–2851. 179. Evans RS, Burke JP, Classen DC, et al. Computerized identiﬁcation of patients at high risk of hospital-acquired infection. Am J Infect Control 1992; 20(20):4–10. 180. Classen DC, Pestotnik SL, Evans RS, Burke JP. Intensive surveillance of midazolam use in hospitalized patients and the occurrence of cardiordespiratory arrest. Pharmacotherapy 1992; 12(3):213–216. 181. Evans R, Pestotnik S, Classen D, SB B, Burke J. Prevention of adverse drug events through computerized surveillance. In: Proceedings 16th Annual Symposium on Computer Applications in Medical Care, 1992. Baltimore, MD: IEEE Computer Society Press, 1992:437–441. 182. Tierney WM, Overhage JM, Takesue BY, et al. Computerizing guidelines to improve care and patient outcomes: the example of heart failure. J Am Med Inform Assoc 1995; 2(5):316–322. 183. Iregui M, Ward S, Clinikscale D, Clayton D, Kollef MH. Use of a handheld computer by respiratory care practitioners to improve the efﬁciency of weaning patients from mechanical ventilation. Crit Care Med 2002; 30(9): 2038–2043. 184. Hull RD, Raskob GE, Rosenbloom D. Heparin for 5 days as compared with 10 days in the initial treatment of proximal venous thrombosis. N Engl J Med 1990; 322:1260–1264. 185. Cruikshank MK, Levine MH, Hirsh J, Roberts R, Siguenza M. A standard heparin nomogram for the management of heparin therapy. Arch Intern Med 1991; 151:333–338. 186. Hull RD, Raskob GE, Rosenbloom D, et al. Optimal therapeutic level of heparin therapy in patients with venous thrombosis. Arch Intern Med 1992; 152:1589–1595. 187. Kuhn T. The Structure of Scientiﬁc Revolutions. 2nd. Chicago: The University of Chicago Press, 1970. 188. Mayr E. Speciational evolution or punctuated equilibria. In: Somit A, Peterson S, eds. The Dynamics of Evolution. New York: Cornell University Press, 1992. 189. Schultz SG. William Harvey and the circulation of the blood: the birth of a scientiﬁc revolution and modern physiology. News Physiol Sci 2002; 17(5):175–180. 190. Foss L, Rothenberg K. The Second Medical Revolution. Boston: New Science Library, 1988:Shambhala. 191. Needham J. Biochemistry and Morphogenesis. Cambridge: Cambridge Univ. Press, 1950.

602

Morris

192. Schultz S. Homeostasis, humpty dumpty, and integrative biology. News Physiol Sci 1996; 11:238–246. 193. Hall JE. Integration and regulation of cardiovascular function. Adv Physiol Edu 1999; 277(6):174S–186S. 194. Marks AR. Cardiac intracellular calcium release channels: role in heart failure. Circ Res 2000; 87(1):8–11. 195. Cowan SE, Gilbert E, Liepmann D, Keasling JD. Commensal interactions in a dual-species bioﬁlm exposed to mixed organic compounds. Appl Environ Microbiol 2000; 66(10):4481–4485. 196. Chong L, Ray LB. Whole-istic biology. Science 2002; 295:1661. 197. Davidson EH, Rast JP, Oliveri P, et al. A genomic regulatory network for development. Science 2002; 295(5560):1669–1678. 198. Kitano H. Systems biology: a brief overview. Science 2002; 295(5560):1662–1664. 199. Wood JD. The ﬁrst Nobel prize for integrated systems physiology: Ivan Petrovich Pavlov, 1904. Physiology 2004; 19(6):326–330. 200. Gaucher E, Graddy L, Li T, et al. The planetary biology of cytochrome P450 aromatases. BMC Biol 2004; 2(1):19. 201. Mandelbrot B. The Fractal Geometry of Nature. New York: W.H. Freeman and Company, 1983. 202. Penrose L, Penrose R. Impossible objects: a special kind of visual illusion. Br J Psychol 1958; 49(Pt 1):31–33. 203. Altman I, Rogoff B. World views in psychology: trait, interactional, organismic, and transactional perspectives. In: Stokols D, Altman I, eds. Handbook of Environmental Psychology. New York: John Wiley & Sons, 1987:7–40. 204. Werner C, Brown B, Altman I. Planning and doing transactionally-oriented research: examples and strategies. In: Stokols D, Altman I, eds. Handbook of Environmental Psychology II. New York: John Wiley & Sons, 2002. 205. Morris A. Algorithm-based decision making. In: Tobin M, ed. Principles and Practice of Intensive Care Monitoring. New York: McGraw-Hill, Inc.; 1998: 1355–1381. 206. Li DY, Faury G, Taylor DG, et al. Novel arterial pathology in mice and humans hemizygous for elastin. J Clin Invest 1998; 102(10):1783–1787. 207. Lind J. A Treatise of the Scurvy. Edinborough: Souds Murray and Cochran, 1753. 208. Choi S, Lagakos SW, Schooley RT, Volberding PA. Articles: CD4þ lymphocytes are an incomplete surrogate marker for clinical progression in persons with asymptomatic HIV infection taking zidovudine. Ann Intern Med 1993; 118(9):674–680. 209. Fleming TR, DeMets DL. Surrogate end points in clinical trials: are we being misled? Ann Intern Med 1996; 125(7):605–613. 210. Kirby RR, Downs JB, Civetta JM, et al. High level positive end expiratory pressure (PEEP) in acute respiratory insufﬁciency. Chest 1975; 67(2):156–163. 211. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome [see comments]. N Engl J Med 1998; 338(6):347–354.

Protocol-Directed Patient Management

603

212. Papadakos PJ, Lachmann B. The open lung concept of alveolar recruitment can improve outcome in respiratory failure and ARDS. Mt Sinai J Med 2002; 69(1–2):73–77. 213. Tusman G, Bohm SH, Tempra A, et al. Effects of recruitment maneuver on atelectasis in anesthetized children. Anesthesiology 2003; 98(1):14–22. 214. Oczenski W, Ho¨rmann C, Keller C, et al. Recruitment maneuvers after a positive end-expiratory pressure trial do not induce sustained effects in early adult respiratory distress syndrome. Anesthesiology 2004; 101(3):620–625. 215. Van Speybroeck L, De Waele D, Van de Vijver G. Theories in early embryology: close connections between epigenesis, preformationism, and self-organization. Ann NY Acad Sci 2002; 981(1):7–49. 216. Van de Vijver G, Van Speybroeck L, De Waele D. Epigenetics: a challenge for genetics, evolution, and development?. Ann NY Acad Sci 2002; 981(1):1–6. 217. Griesemer J. What Is ‘‘Epi’’ about epigenetics? Ann NYAcad Sci 1; 981:97–110. 218. Gilbert SF. The genome in its ecological context: philosophical perspectives on interspecies epigenesis. Ann NY Acad Sci 2002; 981(1):202–218. 219. Sutherland JE, Costa M. Epigenetics and the environment. Ann NY Acad Sci 2003; 983(1):151–160. 220. Bucaro MA, Fertala J, Adams CS, et al. Bone cell survival in microgravity: evidence that modeled microgravity increases osteoblast sensitivity to apoptogens. Ann NY Acad Sci 2004; 1027(1):64–73. 221. Mukundakrishnan K, Ayyaswamy PS, Risbud M, Hu HH, Shapiro IM. Modeling of phosphate ion transfer to the surface of osteoblasts under normal gravity and simulated microgravity conditions. Ann NY Acad Sci 2004; 1027(1):85–98. 222. Arun CP. The importance of being asymmetric: the physiology of digesta propulsion on earth and in space. Ann NY Acad Sci 2004; 1027(1):74–84. 223. Hennekens C, Burin J. Epidemiology in Medicine. 1st. Boston: Little, 1987:Brown and Company. 224. Peto R, Collins R, Gray R. Large-scale randomized evidence: large, simple trials and overviews of trials. In: Warren K, Mosteller F, eds. Doing More Good Than Harm: The Evaluation of Health Care Interventions. New York: The New York Academy of Sciences, 1993:314–340. 225. Zielstorff R. Online practce guidelies: issues, obstacles, and future prospects. J Am Med Inform Assoc 1998; 5(3):227–236. 226. Greenes RA, Peleg M, Boxwala A, Tu S, Patel V, Shortliffe EH. Sharable computer-based clinical practice guidelines: rationale, obstacles, approaches, and prospects. Medinfo 2001; 10(Pt 1):201–205. 227. Senn S. Controversies concerning randomization and additivity in clinical trials. Stat Med 2004; 23(24):3729–3753. 228. Friedman LM, Furberg CD, DeMets DL. Fundamentals of Clinical Trials. 3rd. New York: Springer-Verlag, 1998. 229. Chow S-C, Liu J-P. Design and Analysis of Clinical Trials. Hoboken: John Wiley & Sons, Inc., 2004. 230. Wood G. Effect of antihypertensive agents on the arterial partial pressure of oxygen and venous admixture after cardiac surgery. Crit Care Med 1997; 25:1807–1812.

604

Morris

231. Fisher R. Statistical Methods for Research Workers. London: Oliver & Boyd, 1925. 232. McCall W. How to Experiment in Education. New York: Macmillan, 1923. 233. Steiner JF. Talking about treatment: the language of populations and the language of individuals. Ann Intern Med 1999; 130(7):618–622. 234. Petty T, Bone R, Gee M, Hudson L, Hyers T. Contemporary clinical trials in acute respiratory distress syndrome. Chest 1992; 101:550–552. 235. Bernard G, Artigas A, Brigham K, et al. The American-European consensus conference on ARDS: deﬁnitions, mechanisms, relevant outcomes and clinical trial coordination. Am J Respir Crit Care Med 1994; 149(3):818–824. 236. Schultz KF. Subverting randomization in controlled trials. JAMA 1995; 274(18):1456–1458. 237. Atkins H. The three pillars of clinical research. BMJ 1958; 2(27 December): 1547–1553. 238. James BC, Weed M, Lewis SW, Busboom S, Darricades G, Ingram B. Final Analysis of the IHC Transurethral Prostatectomy Utilization Study. In: Intermountain Health Care Department of Medical Affairs Technical Report No.1. Salt Lake City: Intermountain Health Care, Inc., 1987. 239. Baird M et al. Final Analysis of the IHC Coronary Artery Bypass Graft Quality Utilization and Efﬁciency (Q) Study. Salt Lake City: Intermountain Health Care, Dept of Medical Affairs, 1988. 240. Baird M, Busboom S, French T, et al. Final Analysis of the IHC 1987 Permanent Pacemaker Implant Quality, Utilization, and Efﬁciency (QUE) Study. Salt Lake City: Intermountain Health Care, Inc., 1989. 241. Baird M, Busboom S, Ingram B, et al. Final Analysis of the IHC Uncomplicated Total Hip Arthroplasty Quality, Utilization, and Efﬁciency (QUE) Study. Salt Lake City: Intermountain Health Care, Inc., 1988. 242. Baird ML, Busboom SW, French TK, et al. Final Analysis of the IHC 1987 Permanent Pacemaker Implant Quality, Utilization and Efﬁciency (QUE) Study. Salt Lake City: Intermountain Health Care, Inc., 1989. 243. Roos NP, Wennberg JE, Malenka DJ, et al. Mortality and reoperation after open and transurethral resection of the prostate for benign prostatic hyperplasia. NEJM 1989; 320(17):1120–1124. 244. Panageas KS, Schrag D, Riedel E, Bach PB, Begg CB. The effect of clustering of outcomes on the association of procedure volume and surgical outcomes. Ann Intern Med 2003; 139(8):658–665. 245. Meyerhardt JA, Catalano PJ, Schrag D, et al. Association of hospital procedure volume and outcomes in patients with colon cancer at high risk for recurrence. Ann Intern Med 2003; 139(8):649–657. 246. Laine C, Sox HC. Does practice really make perfect? Ann Intern Med 2003; 139(8):696–698. 247. Tuman KJ, Roizen MF. Outcome assessment and pulmonary artery catheterization: why does the debate continue? Anesth Analg 1997; 84(1):1–4. 248. Lobo DN, Bostock KA, Neal KR, Perkins AC, Rowlands BJ, Allison SP. Effect of salt and water balance on recovery of gastrointestinal function after elective colonic resection: a randomised controlled trial. Lancet 2002; 359(9320):1812–1818.

Protocol-Directed Patient Management

605

249. Brandstrup B, Tonnesen H, Beier-Holgersen R, et al. Effects of intravenous ﬂuid restriction on postoperative complications: comparison of two peri operative ﬂuid regimens: a randomized assessor-blinded multicenter trial. Ann Surg 2003; 238(5):641–648. 250. Lang F, Busch GL, Ritter M, et al. Functional signiﬁcance of cell volume regulatory mechanisms. Physiol Rev 1998; 78(1):247–306. 251. Boldt J, Ducke M, Kumle B, Papsdorf M, Zurmeyer EL. Inﬂuence of different volume replacement strategies on inﬂammation and endothelial activation in the elderly undergoing major abdominal surgery. Intensive Care Med 2004; 30(3):416–422. 252. Cochrane-Collaboration. The Cochrane Reviewer’s Handbook Glossary. Version 4.1.4 ed. The Cochrane Collaboration, 2001. 253. Senn S, Harrell F. On wisdom after the event. J Clin Epidemiol 1997; 50(7):749–751. 254. Horwitz RI, Singer BH, Makuch RW, Viscoli CM. On reaching the tunnel at the end of the light. J Clin Epidemiol 1997; 50(7):753–755. 255. Altman DG. Within trial variation—a false trail? J Clin Epidemiol 1998; 51(4):301–303. 256. Senn S, Harrell FE Jr. On subgroups and groping for signiﬁcance. J Clin Epidemiol 1998; 51(12):1367–1368. 257. Day S. Dictonary for Clinical Trials. Chichester: John Wiley & Sons, Ltd, 1999. 258. Polgar S, Thomas SA. Introduction to Research in the Health Sciences. 3rd. Melbourne: Chruchill Livingstone, 1995. 259. Flather M, Aston H, Stables R. Handbook of Clinical Trials. London: ReMEDICA Publishing, Ltd., 2001. 260. Doig GS. Interpreting and using clinical trials. Crit Care Clin 1998; 14(3):513–524. 261. Miller R, Goodman K. Ethical challenges in the use of decision-support software in clinical practice. In: Goodman K, ed. Ethics, Computing, and Medicine: Informatics and the Transformation of Health Care. Cambridge, UK: Cambridge University Press, 1998:102–115. 262. Field M, Lohr K, eds. Clinical Practice Guidelines: Directions for a New Program (Summary). Washington, D.C.: National Academy Press, 1990. 263. Fridsma D, Gennari J, Musen M. Making generic guidelines site-speciﬁc. In: Cimino J, ed. Proceedings 1996 AMIA Annual Fall Symposium (Formerly SCAMC), 1996. Washington, D.C.: Hanley & Belfus, Inc., 1996:597–601. 264. Ely EW, Meade MO, Haponik EF, et al. Mechanical ventilator weaning protocols driven by nonphysician health-care professionals : evidence-based clinical practice guidelines. Chest 2001; 120(90060):454S–463S. 265. MacIntyre NR. Evidence-based guidelines for weaning and discontinuing ventilatory support: a collective task force facilitated by the American College of Chest Physicians; the American Association for Respiratory Care; and the American College of Critical Care Medicine. Chest 2001; 120(90060):375S–396S. 266. Audet A-M, Greenﬁeld S, Field M. Medical practice guidelines: current activities and future directions. Ann Intern Med 1990; 113:709–714.

606

Morris

267. Fletcher R, Fletcher S. Clinical practice guidelines. Ann Intern Med 1990; 113:645–646. 268. Hadorn D, McCormick K, Diokno A. An annotated algorithm approach to clinical guideline development. JAMA 1992; 267(24):3311–3314. 269. Miller P, Frawly S. Trade-offs in producing patient-speciﬁc recommendations from a computer-based clinical guideline: a case study. J Am Med Inform Assoc 1995; 2:238–242. 270. Tierney WM, Overhage JM, McDonald CJ. Computerizing guidelines: factors for success. In: Cimino J, ed. Proceedings 1996 AMIA Annual Fall Symposium (Formerly SCAMC), 1996. Washington, D.C.: Hanley & Belfus, Inc., 1996:459–462. 271. East TD, Bo¨hm SH, Wallace CJ, et al. A successful computerized protocol for clinical management of pressure control inverse ratio ventilation in ARDS patients. Chest 1992; 101(3):697–710. 272. Henderson S, Crapo R, Wallace C, East T, Morris A, Gardner R. Performance of computerized protocols for the management of arterial oxygenation in an intensive care unit. Int J Clin Monit Comput 1992; 8:271–280. 273. Sheppard LC, Kouchoukos NT, Kurtis MA. Automated treatment of critically ill patients following operation. Ann Surg 1968; 168:596–604. 274. Sheppard L, Kirklin J, Kouchoukos N. Chapter 6—Computer-controlled interventions for the Acutely Ill Patient, 1974. 275. Sheppard L. Computer control of the infusion of vasoactive drugs. Ann Biomed Eng 1980; 8(4–6):431–434. 276. The Acute Respiratory Distress Syndrome Network. Mechanical Ventilation Protocol (Complete). World Wide Web: www.ardsnet.org or NAPS Document No. 05542 (Microﬁche Publications, 248 Hempstead Turnpike, West Hempstead, New York 11552), 2000. 277. Armstrong R, Bullen C, Cohen S, Singer M, Webb A. Critical Care Algorithms. New York, NY: Oxford University Press, 1991. 278. Don H, ed. Decision Making in Critical Care. Philadelphia: BC Decker Inc., 1985. 279. Karlinsky J, Lau J, Goldstein R. Decision Making in Pulmonary Medicine. Philadelphia: BC Decker, 1991. 280. Flexner S. The Random House Dictionary of the English Language. 2nd. New York: Random House Inc., 1987. 281. Holcomb BW, Wheeler AP, Ely EW. New ways to reduce unnecessary variation and improve outcomes in the intensive care unit. Curr Opin Crit Care 2001; 7(4):304–311. 282. Grunberg SM, Cefalu WT. The integral role of clinical research in clinical care. N Engl J Med 2003; 348(14):1386–1388. 283. Steinbrook R. How best to ventilate? Trial design and patient safety in studies of the acute respiratory distress syndrome. N Engl J Med 2003; 348(14):1393–1401. 284. Drazen JM. Controlling research trials. N Engl J Med 2003; 348(14): 1377–1380. 285. Reinertsen JL. Zen and the art of physician autonomy maintenance. Ann Intern Med 2003; 138(12):992–995.

Protocol-Directed Patient Management

607

286. Courtney S, Durand D, Asselin J, Hudak M, Aschner J, Shoemaker C. Highfrequency oscillatory ventilation versus conventional mechanical ventilation for very-low-birth-weight infants. N Engl J Med 2002; 347(9):643–652. 287. Johnson AH, Peacock JL, Greenough A, et al. High-frequency oscillatory ventilation for the prevention of chronic lung disease of prematurity. N Engl J Med 2002; 347(9):633–642. 288. Stark A. High-frequency oscillatory ventilation to prevent bronchopulmonary dysplasia—are we there yet? N Engl J Med 2002; 347(9):682–684. 289. Eichacker P, Gerstenberger E, Banks S, Cui X, Natanson C. A metaanalysis of ALI and ARDS trials testing low tidal volumes. Am J Respir Crit Care Med 2002; 166(11):1510–1514. 290. Miller FG, Rosenstein DL. The therapeutic orientation to clinical trials. N Engl J Med 2003; 348(14):1383–1386. 291. Karlawish JHT. Research involving cognitively impaired adults. N Engl J Med 2003; 348(14):1389–1392. 292. Steinbrook R. Trial design and patient safety—the debate continues. N Engl J Med 2003; 349(7):629–630. 293. Tobin MJ. Of principles and protocols and weaning. Am J Respir Crit Care Med 2004; 169(6):661–662. 294. The Cardiac Arrhythmia Suppression Trial II Investigators. Effect of the antiarrhythmic agent moricizine on survival after myocardial infarction. N Engl J Med 1992; 327(4):227–233. 295. Beauchamp T, Childress J. Principles of Biomedical Ethics. 3rd. New York: Oxford University Press, 1989:Inc. 296. Peto R, Baigent C. Trials: the next 50 years. BMJ 1998; 317(7167):1170–1171. 297. Meade MO. An evidence-based approach to pressure- and volume-limited ventilation strategies. Crit Care Clin 1998; 14(3):373–385. 298. Wennberg J, Freeman J, Shelton R, Bubolz T. Hospital use and mortality among medicare beneﬁciaries in Boston and New Haven. New Engl J Med 1989; 321:1168–1167. 299. Gay P. Of pundits, professors, and pollsters. Crit Care Med 1993; 21(7): 959–960. 300. Horn CR, Clark TJ, Cochrane GM. Compliance with inhaled therapy and morbidity from asthma. Respir Med 1990; 84(1):67–70. 301. Cochrane GM. Therapeutic compliance in asthma; its magnitude and implications. Eur Respir J 1992; 5(1):122–124. 302. Cochrane GM. Impact of education on treatment compliance in patients with asthma. Monaldi Arch Chest Dis 1993; 48(4):369–371. 303. Taylor K, Feldstein M, Skeel R, Pandya K, Ng P, Carbone P. Fundamental dilemmas of the randomized clinical trial process: results of a survey of the 1,737 Eastern Cooperative Oncology Group investigators. J Clin Oncol 1994; 12:1796–1805. 304. Yeung M, O’Connor SA, Parry DT, Cochrane GM. Compliance with prescribed drug therapy in asthma. Respir Med 1994; 88(1):31–35. 305. Bosley CM, Fosbury JA, Cochrane GM. The psychological factors associated with poor compliance with treatment in asthma. Eur Respir J 1995; 8(6): 899–904.

608

Morris

306. Cochrane GM. Compliance and outcomes in patients with asthma. Drugs 1996; 52(suppl 6):12–19. 307. Logan P, Schwab R, Salomone J III, Watson W. Patient understanding of emergency department discharge instructions. South Med J 1996; 89(8): 770–774. 308. Bratzler D, Raskob G, Murray C, Bumus L, Piatt D. Underuse of venous thromobembolism prophylaxis for general surgery patients. Arch Intern Med 1998; 158:1910–1912. 309. Hyland M. Types of noncompliance. Eur Respir Rev 1998; 8(review #56): 255–259. 310. Legorreta AP, Christian-Herman J, O’Connor RD, Hasan MM, Evans R, Leung KM. Compliance with national asthma management guidelines and specialty care: a health maintenance organization experience [see comments]. Arch Intern Med 1998; 158(5):457–464. 311. Cabana MD, Rand CS, Powe NR, et al. Why don’t physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999; 282(15): 1458–1465. 312. Weinert CR, Gross CR, Marinelli WA. Impact of randomized trial results on acute lung injury ventilator therapy in teaching hospitals. Am J Respir Crit Care Med 2003; 167(10):1304–1309. 313. McDonald CJ. Use of a computer to detect and respond to clinical events: its effect on clinician behavior. Ann Intern Med 1976; 84:162–167. 314. Lobach D, Hammon W. Computerized decision support based on a clinical practice guideline improves compliance with care standards. Am J Med 1997; 102:89–98. 315. Shojana K, Yokoe D, Platt R, Fiskio J, Ma’luf N, Bates D. Reducing vancomycin use utilizing a computer guideline: results of a randomized controlled trial. J Am Med Inform Assoc 1998; 5:554–562. 316. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157(1):294–323. 317. Fu Z, Costello ML, Tsukimoto K, et al. High lung volume increases stress failure in pulmonary capillaries. J Appl Physiol 1992; 73(1):123–133. 318. Broccard AF, Hotchkiss JR, Kuwayama N, et al. Consequences of vascular ﬂow on lung injury induced by mechanical ventilation. Am J Respir Crit Care Med 1998; 157(6 Pt 1):1935–1942. 319. Broccard AF, Hotchkiss JR, Suzuki S, Olson D, Marini JJ. Effects of mean airway pressure and tidal excursion on lung injury induced by mechanical ventilation in an isolated perfused rabbit lung model. Crit Care Med 1999; 27(8):1533–1541. 320. Mitchell J, Schuller D, Calandrino F, Schuster D. Improved outcome based on ﬂuid management in critically ill patients requiring pulmonary artery catheterization. Am Rev Respir Dis 1992; 145(5):990–998. 321. Martynowicz MA, Minor TA, Walters BJ, Hubmayr RD. Regional expansion of oleic acid- injured lungs. Am J Respir Crit Care Med 1999; 160(1):250–258. 322. Howell J, Permutt S, Proctor D, et al. Effect of inﬂation of the lung on different parts of the pulmonary vascular bed. J Appl Physiol 1961; 16:71–76.

Protocol-Directed Patient Management

609

323. Permutt S, Howell J, Proctor D, et al. Effect of lung inﬂation on static pressurevolume characteristics of pulmonary vessels. J Appl Physiol 1961; 16:64–70. 324. Albert RK, Lakshminarayan S, Kirk W, Butler J. Lung inﬂation can cause pulmonary edema in zone I of in situ dog lungs. J Appl Physiol 1980; 49(5): 815–819. 325. Brower R, Wise RA, Hassapoyannes C, Bromberger-Barnea B, Permutt S. Effect of lung inﬂation on lung blood volume and pulmonary venous ﬂow. J Appl Physiol 1985; 58(3):954–963. 326. Marini JJ, Hotchkiss J. Do variations in cardiac output inﬂuence lung edema? Crit Care Med 2001; 29(5):1079–1080. 327. Feig P, McCurdy D. The hypertonic state. New Engl J Med 1977; 297(26):1444–1454. 328. Windus D. Fluids and electrolyte management. In: Orland M, Saltman R, eds. Manual of Medical Therapeutics. 25th. Boston: Little, Brown and Company, 1986:40–56. 329. Levinsky N. Fluids and electrolytes. In: Wilson J, Braunwald E, Isselbacher K, et al., eds. Harrison’s Principles of Internal Medicine. 12th ed. New York: McGraw-Hill, Inc., 1991:278–283. 330. DeVita M, Michelis M. Perturbations in sodium balance. Clin Lab Med 1993; 13(1):135–148. 331. Rose B. Clinical Physiology of Acid-Base and Electrolyte Disorders. 4th. New York: McGraw-Hill. Inc., 1994. 332. Mange K, Matsuura D, Cizman B, et al. Language guiding therapy: the case of dehydration versus volume depletion. Ann Intern Med 1997; 197(9):848–852. 333. Vanatta J, Fogelman M. Moyer’s Fluid Balance—A Clinical Manual. 4th. Chicago: Year Book Medical Publishers, Inc., 1988. 334. Weaver L, Hopkins R, Churchill S, et al. Outcome of acute carbon monoxide poisoning treated with hyperbaric or normobaric oxygen (double-blind). Am J Respir Crit Care Med (ATS Proceedings Abstracts) 2001; 163(5):A16. 335. Weinberg A, Minaker K. Dehydration. Evaluation and management in older adults. JAMA 1995; 274:1552–1556. 336. Chilton L. Prevention and management of hypernatremic dehydration in breast-fed infants. West J Med 1995; 163(1):74–76. 337. Zornow M, Prough D. Fluid management in patients with traumatic brain injury. New Horizons 1995; 3(3):488–498. 338. Thijs L. Fluid therapy in septic shock. In: Sibbald W, Vincent J-L, eds. Clinical Trials for the Treatment of Sepsis. Berlin: Springer-Verlag, 1995:167–190. 339. Shoemaker W, Appel P, Kram H. Prospective trial of supranormal values of survivors as therapeutic goals in high risk surgical patients. Chest 1988; 94:1176–1186. 340. Tuchschmidt J, Fried J, Astiz M, Rackow E. Supranormal oxygen delivery improves mortality in septic shock patients. Crit Care Med 1991; 19:S66. 341. Shoemaker W, Appel P, Kram H, Bishop M, Abraham E. Temporal hemodynamic and oxygen transport patterns in medical patients: septic shock. Chest 1993; 104:1529–1536. 342. Sailors M, East T, Wallace C, et al. Testing and validation of computerized decision support systems. In: Cimino J, ed. Proceedings of the 1996 AIMIA

610

343. 344.

345.

346.

347.

348.

349. 350. 351.

352.

353.

354.

Morris Annual Fall Symposium, 1996. Philadelphia: Hanley & Belfus, Inc., 1996:234–238. McClave S, Sexton L, Spain D, et al. Enteral tube feeding in the intensive care unit: factors impediing adequate delivery. Crit Care Med 1999; 27(7):1252–1256. The ARDS Network Authors. Ketoconazole for early treatment of acute lung injury and acute respiratory distress syndrome—a randomized controlled trial. JAMA 2000; 283(15):1995–2002. The ARDS Clinical Trials Network. Randomized, placebo-controlled trial of lisofylline for early treatment of acute lung injury and acute respiratory distress syndrome. Crit Care Med 2002; 30(1):1–6. Sailors R, East T, Wallace C, Morris A. A successful protocol for the use of pulse oximetry to classify arterial oxygenation into four fuzzy categories. In: Gardner R, ed. Proceedings 19th Annual Symposium on Computer Applications in Medical Care (SCAMC) Oct 28-Nov 1, 1995. New Orleans, LA: Hanley & Belfus, Inc., 1995:248–252. Morris A, Sorenson D, Orme J, et al. A frame-based tool set for computerized protocols. Am J Respir Crit Care Med (ATS Proceedings Abstracts) 2002; 165(8):A108. Morris A, Sorenson D, Orme J, et al. Version-2 frame- based tools for computerized protocols. Am J Respir Crit Care Med (ATS Proceedings Abstracts) 2003; 167(7):A496. Morris AH, Sorenson D, Warner H, et al. Clinical trial toolbox for constructing study forms. Am J Respir Crit Care Med 2004; 169(7):A329. Morris AH, Sorenson D, Ghosh D, et al. Clinical trial data transfer via the Internet. Am J Respir Crit Care Med 2004; 169(7):A329. Morris AH, Sorenson D, Day J, et al. Electronic tools for conserving clinical coordinator resources in clinical trials. Am J Respir Crit Care Med 2004; 169(7):A329. Brower RG, Lanken PN, MacIntyre N, et al. Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351(4):327–336. Morris AH, Wallace CJ, Beck E, et al. Protocols control respiratory treatment of ARDS. 5th World Congress on Intensive and Critical Medicine, Kyoto, Japan, 3–8 Sept, 1989. 1989:727. East TD, Morris AH, Clemmer TP, et al. Development of computerized critical care protocols—a strategy that really works. In: Proceedings of the Fourteenth Annual Symposium on Computer Applications in Medical Care, Nov 1990, 1990. Washington, DC: IEEE Computer Society Press, 1990:564–568.

23 Crossing the Quality Chasm in Critical Care: Changing Ventilator Management in Patients with ALI

MARGARET J. NEFF and GORDON D. RUBENFELD Division of Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington Seattle, Washington, U.S.A.

I. Introduction For years the concept of ‘‘evidence-based critical care’’ was largely theoretical because of a lack of high-grade clinical trial evidence demonstrating a survival beneﬁt for any speciﬁc therapy for acute lung injury (ALI) and severe sepsis. The last decade has produced a number of clinical trials in critical care demonstrating signiﬁcant effects on important clinical outcomes (lung-protective ventilation for ALI, activated protein C for severe sepsis, and protocolized ventilator weaning) and other trials raising questions about the beneﬁt of treatments thought to be effective (human growth hormone for chronic critical illness, pulmonary artery catheterization, and colloid resuscitation). A natural assumption would be that this evidence would be followed by the rapid integration of these results into clinical practice. Experience tells us otherwise. Whether trying to change practice in business or medicine, change is a slow process. Entire industries are developed to improve systems and processes to incorporate change into the industry. Critical care medicine is new to this phenomenon and also to the situation of having evidence-based medicine with which to guide clinical 611

612

Neff and Rubenfeld

practice. Until recently, the problem was not knowing the answer; now the problem is ﬁguring out how to apply the answers. Critical care medicine is not alone in facing this challenge. Nearly all other ﬁelds of medicine have struggled with and continue to face the challenge of translating the results of clinical trials into practice. Cardiology, a ﬁeld with a long history of large, positive trials in ischemic heart disease, still struggles with this challenge. For example, one-third of survivors of myocardial infarction (MI) are not prescribed aspirin within the ﬁrst two days after the infarct (1), and despite the fact that beta-blockers are known to reduce mortality after an MI, only 21% of eligible elderly patients receive them (2). Failure to employ thrombolytics, beta-blockers, aspirin, and angiotensin-converting enzyme inhibitors in patients with acute MIs may cause as many as 18,000 deaths per year in the United States (3,4). This phenomenon of inadequately providing evidence-based medicine to eligible patients, though, is not isolated to acute, inpatient medicine. Outpatients frequently have their hypertension inadequately managed, preventive services neglected, and diagnoses such as depression missed (5,6). Nor are inadequacies limited to the failure to provide necessary treatments. Antibiotics, hysterectomies, cardiac pacemakers, and coronary angiography have all been shown to be overused in some cases (3). These observations have led to strong responses from the academic, medical, consumer, and health care payer communities. In November 1998, the American Association of Medical Colleges and the American Medical Association convened a Clinical Research Summit devoted to establishing broad, national goals in clinical research. One of the principal recommendations of this commission was to develop a ‘‘broadened agenda of clinical research, related more speciﬁcally to health outcomes and designed to assess the effectiveness of methods for incorporating evidence-based practice into clinical care’’ (7). A recent publication by the Institute of Medicine, Crossing the Quality Chasm: a New Health System for the 21st Century, outlined the case that modern health care frequently fails to deliver optimal medical care even in the absence of barriers such as access to care and limited ﬁnances. This widely cited document charged the U.S. Department of Health and Human Services to ‘‘establish and maintain a comprehensive program aimed at making scientiﬁc evidence more useful and accessible to clinicians and patients’’ (8). To better understand how to translate results into practice, we need to review the available research in the ﬁeld. Research in this area includes studies that evaluate current practice in the community, explore the barriers to changing practice, and evaluate speciﬁc interventions to change the practice of critical care. The better we can understand what the current practice in our ﬁeld is and what the barriers to change are, the better prepared we will be to develop and implement protocols and guidelines to effect change.

Changing Ventilator Management in Patients with ALI

613

II. Understanding Current Practice Because having evidence of beneﬁt is the ﬁrst step in translating research into practice, it is important to ask what interventions in critical care are known to be beneﬁcial. The amount of evidence it takes to convince individual clinicians may vary and will certainly vary with the plausibility, cost, risks, and beneﬁts of the proposed treatment (9). In general, we will need more compelling evidence to convince us to use expensive and possibly harmful interventions like inhaled nitric oxide or prone ventilation for ALI instead of inexpensive interventions like oxygen supplementation to patients with acute hypoxemic respiratory failure. Although the evidence supporting critical care practice is growing, the literature on how this evidence is to be implemented is in its infancy (10–14). There are several techniques for studying the process of care. Clinicians can be surveyed about their attitudes about using different treatments or about their practice in hypothetical case vignettes. Charts can be abstracted retrospectively by research staff using a predetermined protocol. Clinicians can report on their own practice in speciﬁc cases. Databases collected for billing or administrative purposes can be analyzed. As with all research methods, each approach has speciﬁc beneﬁts, limitations, and costs. Surveys of clinicians’ behavior to measure their practice are notoriously unreliable at capturing their actual practice but can still be instructive, especially when self-reporting shows a marked lack of utilization of guidelines or resources. Chart abstraction, while accurate, may not identify all patients with a given critical illness syndrome. Studying only patients recognized with the syndrome may overestimate the use of a particular therapy if recognition of the syndrome is part of the barrier to implementation. Relatively few studies have evaluated current practice in the intensive care unit (ICU), particularly in community settings. Despite evidence for reduced intubation rates, length of stay, and mortality when noninvasive ventilation is used for chronic obstructive pulmonary disease (COPD) exacerbations (15), there are surprisingly few data documenting the use of noninvasive ventilation in current practice. Doherty and Greenstone surveyed 268 hospitals in the United Kingdom and found considerable regional variation in the availability of noninvasive ventilation, with fewer than half of the acute hospitals having noninvasive ventilation available and the rest underutilizing noninvasive approaches in many instances (16). In a single site audit at a teaching hospital, Sinuff et al. found that noninvasive ventilation was used by physicians of different training levels in various settings within the hospital and found important areas for improving the quality of documentation, monitoring, and implementation of noninvasive ventilation (17). In a study that combined patient-level data and survey data from physicians, Heyland et al. identiﬁed noninvasive

614

Neff and Rubenfeld

ventilation, subglottic secretion drainage endotracheal tubes, kinetic bed therapy, small bowel feedings, and elevation of the head of the bed as effective preventive treatments for ventilator-associated pneumonia, which were not being used in a set of Canadian ICUs (18). Several recent studies have explored practice in broad populations of mechanically ventilated patients or patients with ALI (19–22). These studies used a combination of chart abstraction, self-report of practice by clinicians, and survey of attitudes to describe clinical practice. Although these studies were performed before the landmark Acute Respiratory Distress Syndrome (ARDS) Network clinical trial, they were performed during a period when recommendations and clinical experience suggested that mechanical ventilation should be customized for individual diseases. Despite these recommendations, patients received remarkably similar average tidal volumes, positive end-expiratory pressure (PEEP), and fraction of inspired oxygen regardless of whether they were diagnosed with ALI, ARDS, acute respiratory failure (ARF), or COPD (Fig. 1). In one study, 63% of patients managed with a volume-control mode of ventilation received tidal volumes less than 10 mL/kg, but patients diagnosed by their physicians with ALI were no more likely to receive low tidal volumes (e.g., 6 mL/kg) than other patients (19). Additionally, a survey of critical care physician members of the American Thoracic Society reported a wide range of tidal volumes used to treat patients with ARDS (23). Several recent studies have evaluated current practices in the ventilatory management of patients with ALI, including what effect, if any, the publication of the ARDS Network study had on these practices. Young et al. reviewed tidal volume practices both before and after the publication of the ARDS Network low tidal volume paper (reported May 1999, published May 2000), which showed a mortality beneﬁt in patients with

Figure 1 Comparison of tidal volume, PEEP, and FiO2 in large cohorts of mechanically ventilated patients. The values of the Scandinavian, Australian, and international trials are from Refs. 22, 21, and 19, respectively. Abbreviations: ARF, acute respiratory failure; COPD, chronic obstructive pulmonary disease; PEEP, positive end-expiratory pressure; ALI, acute lung injury; ARDS, acute respiratory distress syndrome; FiO2, fraction of inspired oxygen.

Changing Ventilator Management in Patients with ALI

615

ALI managed with tidal volumes of 6 mL/kg of predicted body weight (PBW) (14,24). They reviewed a random sample of charts of mechanically ventilated patients from three tertiary hospitals during those two time frames and identiﬁed patients who met the American–European Consensus Conference deﬁnition for ALI (25). While tidal volumes dropped after the report of the ARDSNet study (from 12.3 to 10.6 mL/kg PBW, p < .001), these were still far larger than 6 mL/kg PBW (24). These results echoed those found in two other studies showing minimal effect of the publication of study results on the practice of ventilatory management of ALI patients (26,27). Finally, a survey conducted by the Canadian Critical Care Trials Group was designed to assess physicians’ opinions regarding preventative and therapeutic interventions for ALI. Understanding of the literature, efﬁcacy of the intervention, and current use were assessed with a wide variation in the management of ALI being identiﬁed (28).

III. Do We Know Why Clinicians Do Not Follow Practice Guidelines? Variation in practice is not necessarily a problem in itself; differences in patients, resources, and expertise may account for ‘‘unexplained’’ variation. However, variation that results in underutilization of proven therapies is a problem. It is this detrimental variation that improved compliance with guidelines and protocols could help eliminate. Many medical specialties have responded to the observations that clinicians fail to incorporate research into their practice by creating a research program directed speciﬁcally at the failure to implement effective practice. With the notable exception of cardiology, which has devoted extensive resources to understanding the care of patients with ischemic heart disease, critical care has not extensively explored these questions (29–31). A recent study by Cabana et al. systematically reviewed the extensive body of literature studying the barriers to implementation of effective treatments and guidelines (32). They developed a conceptual model for categorizing barriers to changing clinical behavior. A wide range of clinical topics were reviewed based on the existing literature, including barriers to appropriate preventive care, obstetric care, pain control, and use of thrombolytic therapy. Similarly, studies evaluating the knowledge, attitudes, and behavior of a broad range of clinicians including general practitioners, cardiologists, radiologists, and surgeons were reviewed. However, this extensive review did not identify a single study of the barriers to implementation of effective therapy in critical care. There was no mention of common critical illness syndromes such as sepsis, ALI, or ARF . Finally, no studies of intensivists, intensive care nurses, or respiratory therapists were reported in this review. In a series of review articles published as a supplement to Chest titled ‘‘Translating

616

Neff and Rubenfeld

Guidelines Into Practice: Implementation and Physician Behavior Change,’’ only two examples from critical care were discussed: prescribing of antimicrobials in the ICU and treatment of MI (33,34). There is a reasonable hypothesis to explain why critical care has not produced a body of research about changing clinical practice in the ICU. Intensivists have grown accustomed to making clinical decisions based on physiologic rationale in the absence of evidence demonstrating an improvement in outcome. Unlike our colleagues in cardiology and oncology, intensivists (until recently) have not enjoyed the luxury of a rich evidence base of positive clinical trials. As typical examples, consider the conclusions of recent reviews on two perennial questions in critical care: ‘‘Is colloid better than crystalloid for ﬂuid resuscitation?’’ and ‘‘Is total parenteral nutrition (TPN) beneﬁcial to critically ill patients?’’ There is no evidence from randomized controlled trials that resuscitation with colloids reduces the risk of death compared to crystalloids in patients with trauma, burns, and following surgery (35). While TPN may have a positive effect on nutritional end points and on even minor complications, the overall results of our meta-analysis fail to support a beneﬁt of TPN on mortality or major complication rates, particularly in critically ill patients (36).

Statements of evidence such as these have allowed intensivists to justify a range of therapeutic decisions. In the absence of compelling evidence of harm or beneﬁt, physicians will tend to base decisions on the basis of biologic rationale, experience, and personal values about cost-effectiveness (9). In a world with few critical care practices of demonstrable beneﬁt, the question of implementing practice is moot. Thankfully, there are now valuable interventions for which improving implementation makes sense. It is important to note that some implementation research occurs without publication in mainstream academic research journals. For example, the Institute for Healthcare Improvement (IHI) is a nonproﬁt organization that sponsors workshops to help clinicians improve the quality of care they provide (37). Many of these quality improvement projects have focused on critical care interventions. The projects are usually single institution, before–after studies, and the results are not peer reviewed. Nevertheless, this is an important source of information about projects designed to change clinical practice at single sites or within collaborative hospitals.

IV. Barriers to Changing Practice in the ICU A considerable body of literature exists evaluating why clinicians do not follow evidence-based clinical practice guidelines (32). The model proposed by Cabana et al. (32) after reviewing the literature in this area identiﬁed seven

Changing Ventilator Management in Patients with ALI

617

categories of barriers: lack of awareness of the guidelines, lack of familiarity with the guidelines, lack of agreement with the guidelines, lack of selfefﬁcacy, lack of outcome expectancy, inability to overcome the inertia of previous practice, and external barriers to performing the recommendations. As an example, focus groups and surveys have been used to explore barriers to implementing semirecumbent positioning in ALI (38). This study showed that nurses and physicians had very different perceptions of the major barriers. Nurses identiﬁed physicians’ failure to order what they wanted for the patient’s position as the major barrier, and physicians identiﬁed nursing preference as the major barrier to using semirecumbent positioning. There was also confusion over real/perceived risks and who was responsible for implementing the intervention, and a lack of enabling/reinforcing strategies. Differences between nurses’ and physicians’ perceptions of barriers were also seen in a study evaluating barriers to implementing evidence-based guidelines for preventing ventilatorassociated pneumonia (39). While a number of barriers were identiﬁed, including lack of resources and disagreement with trial results, the nurses were more likely to identify patient discomfort as a barrier, whereas the physicians were more prone to identify cost as a barrier. These studies highlight the importance of patient-related barriers, as well as including nursing perspectives when planning guideline implementation. Patient-centered factors (e.g., ﬁnances, quality of life, and location of care) were also identiﬁed as signiﬁcant issues for physicians trying to implement guidelines for heart failure management (40). Lack of recognition of the disease and disagreement with the clinical trial’s ﬁndings are two major barriers seen in a number of trials. In a survey of surveillance of patients with Barrett esophagus, adherence to the guidelines was more likely to be associated with agreement with the guidelines than whether or not the physician was aware of the guidelines (41). In the review of Young et al. of whether the ARDS Network publication changed practice, they also found that the physician identiﬁed ALI in a minority of the cases (23% prepublication and 32% postpublication) (24). Physicians in general may also be more likely to advocate interventions with direct/ immediate physiologic consequences (e.g., oxygenation improvement with inhaled nitric oxide or PEEP) even though those surrogate outcomes do not correlate with long-term outcome. Ease of application of the intervention also factors into its use. When experienced ARDS Network nurses and respiratory therapists were questioned about perceived barriers to the initiation and maintenance of low tidal volume ventilation, a number of similar items were identiﬁed as hindering utilization of the study protocol. These included the following: lack of recognition of the disease, physician reluctance to release control of the ventilator, lack of knowledge of the beneﬁts, no protocol in place, and concern over abnormal laboratory values and the perception of patient discomfort (43).

618

Neff and Rubenfeld

Critical care poses other barriers to research on implementation. The ICU itself is essentially an organizational innovation that focuses technology and experienced clinicians into a speciﬁc location in the hospital. Intensivists rely on and work closely with primary care physicians, consulting specialists, ICU nurses, respiratory therapists, pharmacists, nutritionists, and other clinicians in the ICU. The knowledge, attitudes, and behaviors of all of these clinicians must be considered in interventions designed to implement effective ICU care. The multidisciplinary nature of critical care must be considered in designing interventions to change behavior (44). The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT) study, a large multicenter trial designed to change clinical practice, has been criticized because its intervention failed to consider the organizational structure of the ICU and interactions between clinicians (45). Investigators studying the translation of research ﬁndings into practice in the ICU must consider the differences in barriers and facilitators that are likely to be encountered when evaluating interventions targeted at the system level (intensivist coverage, computerized orders, rounding pharmacist, step-down unit, etc.) versus the patient level (lung-protective ventilation, activated protein C, tight glucose control, etc.). For example, identifying the factors affecting the structure of the ICU may require surveying the hospital chief executive ofﬁcer, nonintensivist physicians who admit to the ICU, and hospital ﬁnancial staff. Large capital investments, hospital-wide policy changes, and legal issues may be involved. Patient-level changes in practice may involve system factors, particularly system solutions.

V. Models of Changing Clinical Practice There are a number of conceptual models describing the processes that individuals and organizations go through as they change behavior. Not surprisingly, these models come from ﬁelds that are intimately familiar with trying to change knowledge and behavior: psychology, education, health promotion, and marketing. Models for understanding behavioral change are important because they lead to strategies for changing behavior (Table 1). Although there is some overlap, it is useful to think of these models as falling into broad categories: educational, epidemiological, and marketing strategies (targeting an individual’s internal factors) as well as behavioral, social, organizational, and coercive strategies (targeting factors external to the individual). Educational models are the ones with which physicians are most familiar. Adult learning theory stresses the importance of interactive educational experiences over passive learning in lectures. Examples include Advanced Cardiac Life Support1 or Advanced Trauma Life Support1 courses taught with individual skill stations (55). Epidemiological models focus on

Approach

Theories

Focus on internal processes Educational (46) Adult learning theories Epidemiological Cognitive theories (47) Marketing (48)

Health promotion, innovation and social marketing theories Focus on external inﬂuences Behavioral (49) Learning theory

Social interaction (50,51)

Social learning and innovation theories, social inﬂuence/ power theories

Organizational (52)

Management theories, system theories

Coercive (53)

Focus Intrinsic motivation of professionals Rational information seeking and decision making Attractive product adapted to needs of target audience

Problem-based learning

Controlling performance by external stimuli

Audit and feedback, reminders, economic incentives Peer review in local networks, opinion leaders, academic detailing

Social inﬂuence of signiﬁcant peers/role models

Example Mechanical ventilation course using hands-on demonstrations Consensus conference on mechanical ventilation

Evidence-based guideline development and dissemination Needs assessment, adapting Targeted intervention to increase use of semirecumbent change proposals to local positioning based on focus needs groups of clinicians

Physician prompt that patients have passed a trial of spontaneous breathing Regionally prominent physician, nurse, and respiratory therapist who meet with local clinicians in small groups to convince them to use weaning protocol Re-engineering care process, Development of a weaning team that consults on all patients total quality management, mechanically ventilated for team building, changes to systems more than 72 hr Hospital removes inhaled nitric Regulations, laws, oxide from formulary budgeting, legal procedures

619

Creating structural and organizational conditions to improve care Economic, power, Control and pressure, and learning theories external motivation

Source: From Ref. 54.

Intervention strategies

Changing Ventilator Management in Patients with ALI

Table 1 Evidence Base for Various Behavior Change Strategies

620

Neff and Rubenfeld

synthesizing and presenting the evidence on optimal practice. Examples include published meta-analyses, the Cochrane reviews, and formal guideline–developing activities. Large data warehouses of these resources are available on the Internet (56,57). Marketing strategies rely on research to understand the values, concerns, aspirations, needs, and knowledge of their target audience (58). Marketers realize that selling a product often does not rely on informing their audience about its beneﬁts, but in convincing the target that they will be more popular if they buy it and ‘‘left out’’ if they do not. Similarly, social marketers, trying to ‘‘sell’’ smoking cessation or appropriate antibiotic use, must provide the audience with a reason to act, which may have little to do with the evidence about beneﬁts of the action. A number of models try to inﬂuence behavior by using external factors. Behavioral theory uses feedback and stimulus response, such as automatic reminders or clinician audit and feedback reports, to affect behavior. Social theory takes advantage of information about how individuals behave in groups. Individuals ﬁt into broad categories of innovators, early adopters, early majority, late majority, and laggards based on their willingness to adopt new practices (59). Understanding which group a clinician ﬁts into will allow one to understand the barriers to changing the clinician’s practice. Organizational approaches are adapted from the Total Quality Management and other quality improvement methods used by corporations. The IHI has championed these practices in health care (37). Finally, coercive techniques rely on regulatory, ﬁscal, or legal constraints or incentives to change practice. There have been four recent extensive metareviews (reviews of reviews and meta-analyses) evaluating which techniques are most effective at changing clinical practice (60–63). The authors of these reviews cite common problems with the literature: publication bias, lack of repeat studies to validate methods, and weak study designs. However, the reviews reach remarkably similar conclusions. They rank interventions to change behavior in health professionals into three categories based on the evidence of their effectiveness (Table 2). Table 2 Relative Effectiveness of Strategies to Change Behavior Weak Passive education by distribution of guidelines or continuing medical education lectures or unsolicited written material Source: From Refs. 60–63.

Moderately or variably effective Economic incentives Audit and feedback Local opinion leaders

Relatively strong or consistently effective Multifaceted interventions (combining two or more of feedback, reminders, education, marketing) Academic detailing Reminders or prompting

Changing Ventilator Management in Patients with ALI

621

It is noteworthy that academic clinicians spend a great deal of time engaged in activities known to be only marginally effective at changing clinical practice—lectures and passive dissemination of written materials. VI. Effective Strategies to Change Practice in the ICU It is important to distinguish studies that demonstrate beneﬁt of a particular intervention in critical care from studies that are interested primarily in implementing this approach. Implementation research is not designed to identify effective treatment strategies. The research question is not whether a speciﬁc technique improves outcome—this is presumed to be known. The question is whether this technique can be deployed in a larger community. The success of an intervention in a constrained research protocol environment does not necessarily translate to success of the protocol in the ‘‘real world.’’ The effect of the intervention on patient outcome is important, but it is a secondary research question. There have been no large-scale, multicenter, community-based programs to improve the quality of care for critically ill patients. A computerized decision-support tool to direct mechanical ventilation in patients with ARDS was implemented in a randomized clinical trial at 10 academic sites; however, this study was directed as much at evaluating the efﬁcacy of the ventilator strategy as the feasibility of using a computer to effect practice change (64). In the study of noninvasive ventilation by Sinuff et al., the identiﬁed effective interventions included interactive education coupled with local development and implementation of protocols (17). Berenholtz et al. studied interventions to decrease catheter-related infections in the ICU (65). Compared to preintervention, catheter infections were reduced. The intervention included education, creation of a catheter insertion cart, daily reminders to remove the catheter if possible, development of a checklist, and empowering the nurses to stop procedures that were not being done according to guidelines. These interventions are similar to those found in a systematic review of interventions: interactive education, academic detailing, reminders, audit and feedback, and computerized decision support (61). The evidence from other ﬁelds suggests that effective implementation of critical care interventions will require a multifaceted approach that incorporates local ‘‘buy in’’ of the treatment, local opinion leaders, staff education, incorporation of nursing and respiratory input, audit and feedback, and timely prompts. VII. Conclusions Clinical practice guidelines are increasingly common in all ﬁelds of medicine. In the 1980s, there were but a handful. As of 2002, there were over

622

Neff and Rubenfeld

1000 guidelines approved through the National Guideline Clearinghouse (66). Thanks to recent positive trials in critical care, their use is starting to inﬁltrate the ICU setting. However, few studies have been conducted to assess the implementation of these guidelines, let alone the cost of guideline development. Having a ﬁrm understanding of current practices in the ICU, of the literature that supports clinical practice guidelines, and of the barriers to initiating these changes will provide a solid foundation on which to implement change in the management of ICU patients, including those with ALI. Many of these interventions may involve changes in practice and not necessarily new pharmaceutical agents. The drive for improvement and change will have to come from within the ICU team structure and not necessarily from the industry. An important part of this effort will include identifying the need for education, local development and implementation of easy-to-use protocols, incorporation of all ICU personnel (physicians, nurses, respiratory therapists, etc.) into protocol development, and a system of reminders and feedback. Demonstrating that treatments reduce morbidity and mortality in the ICU has been so challenging that intensivists have not, until recently, focused their full attention on the difﬁculties that lie in ensuring that patients actually receive these treatments. Identifying effective strategies to ‘‘implement’’ practices in critical care will require a research program as rigorous and broad as the research program required to ‘‘develop’’ these interventions. In this regard, intensivists must begin to think like public health professionals—it is not enough for us to develop effective treatments and to use them in our practices. We also have the responsibility to ensure that all patients who can beneﬁt actually receive effective intensive care and, if they do not, to identify solutions to this problem. References 1. Krumholz HM, Radford MJ, Ellerbeck EF, et al. Aspirin in the treatment of acute myocardial infarction in elderly medicare beneﬁciaries: patterns of use and outcomes. Circulation 1995; 92:2841–2847. 2. Soumerai SB, McLaughlin TJ, Spiegehnan D, Hertzmark E, Thibault G, Goldman L. Adverse outcomes of underuse of beta-blockers in elderly survivors of acute myocardial infarction. JAMA 1997; 277:115–121. 3. Chassin MR, Galvin RW. The urgent need to improve health care quality. Institute of Medicine National Roundtable on Health Care Quality. JAMA 1998; 280:1000–1005. 4. Chassin MR. Assessing strategies for quality improvement. Health Aff (Millwood) 1997; 16:151–161. 5. Wells KB, Hays RD, Burnam MA, Rogers W, Greenﬁeld S, Ware JE Jr. Detection of depressive disorder for patients receiving prepaid or fee-for-service care. Results from the Medical Outcomes Study. JAMA 1989; 262:3298–3302.

Changing Ventilator Management in Patients with ALI

623

6. Udvarhelyi IS, Jennison K, Phillips RS, Epstein AM. Comparison of the quality of ambulatory care for fee-for-service and prepaid patients. Ann Intern Med 1991; 115:394–400. 7. Heinig SI, Quon AS, Meyer RE, Korn D. The changing landscape for clinical research. Acad Med 1999; 74:726–745. 8. Institute of Medicine (U.S.). Committee on Quality of Health Care in America. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, D.C.: National Academy Press, 2001:xx, 337. 9. Rubenfeld GD. Understanding why we agree on the evidence but disagree on the medicine. Respir Care 2001; 46:1442–1449. 10. Fagon JY, Chastre J, Wolff M, et al. Invasive and noninvasive strategies for management of suspected ventilator-associated pneumonia. A randomized trial. Ann Intern Med 2000; 132:621–630. 11. van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in the critically ill patients. N Engl J Med 2001; 345:1359–1367. 12. Ely EW, Baker AM, Dunagan DP, et al. Effect on the duration of mechanical ventilation of identifying patients capable of breathing spontaneously. N Engl J Med 1996; 335:1864–1869. 13. Bernard GR, Vincent JL, Laterre PF, et al. Efﬁcacy and safety of recombinant human activated protein C for severe sepsis. N Engl J Med 2001; 344:699–709. 14. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342:1301–1308. 15. Brochard L, Mancebo J, Wysocki M, et al. Noninvasive ventilation for acute exacerbations of chronic obstructive pulmonary disease. New Engl J Med 1995; 333:817–822. 16. Doherty MJ, Greenstone MA. Survey of non-invasive ventilation (NIPPV) in patients with acute exacerbations of chronic obstructive pulmonary disease (COPD) in the UK. Thorax 1998; 53:863–866. 17. Sinuff T, Cook DJ, Randall J, Allen CJ. Evaluation of a practice guideline for noninvasive positive-pressure ventilation for acute respiratory failure. Chest 2003; 123:2062–2073. 18. Heyland DK, Cook DJ, Dodek PM. Prevention of ventilator-associated pneumonia: current practice in Canadian intensive care units. J Crit Care 2002; 17:161–167. 19. Esteban A, Anzueto A, Alia I, et al. How is mechanical ventilation employed in the intensive care unit? An international utilization review. Am J Respir Crit Care Med 2000; 161:1450–1458. 20. Esteban A, Anzueto A, Frutos F, et al. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study. JAMA 2002; 287:345–355. 21. Bersten AD, Edibam C, Hunt T, Moran J, Group TA. Incidence and mortality of acute lung injury and the acute respiratory distress syndrome in three Australian states. Am J Respir Crit Care Med 2002; 165:443–448. 22. Luhr OR, Antonsen K, Karlsson M, et al. Incidence and mortality after acute respiratory failure and acute respiratory distress syndrome in Sweden, Denmark, and Iceland. The ARF Study Group. Am J Respir Crit Care Med 1999; 159:1849–1861.

624

Neff and Rubenfeld

23. Carmichael LC, Dorinsky PM, Higgins SB, et al. Diagnosis and therapy of acute respiratory distress syndrome in adults: an international survey. J Crit Care 1996; 11:9–18. 24. Young MPM, Manning HLM, Wilson DLM, et al. Ventilation of patients with acute lung injury and acute respiratory distress syndrome: has new evidence changed clinical practice? Crit Care Med 2004; 32:1260–1265. 25. Bernard GR, Artigas A, Brigham KL, et al. The American–European Consensus Conference on ARDS. Deﬁnitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149:818–824. 26. Rubenfeld G, Caldwell E, Hudson L. Publication of study results does not increase use of lung protective ventilation in patients with acute lung injury. Am J Respir Crit Care Med 2001; 163:A295. 27. Weinert CR, Gross CR, Marinelli WA. Impact of randomized trial results on acute lung injury ventilator therapy in teaching hospitals. Am J Respir Crit Care Med 2003; 167:1304–1309. 28. Meade MOM, Jacka MJM, Cook DJM, et al. Canadian Critical Care Trials Group. Survey of interventions for the prevention and treatment of acute respiratory distress syndrome. Crit Care Med 2004; 32:946–954. 29. Krumholz HM, Radford MJ, Wang Y, Chen J, Heiat A, Marciniak TA. National use and effectiveness of beta-blockers for the treatment of elderly patients after acute myocardial infarction: National Cooperative Cardiovascular Project. JAMA 1998; 280:623–629. 30. O’Connor GT, Quinton HB, Traven ND, et al. Geographic variation in the treatment of acute myocardial infarction: the Cooperative Cardiovascular Project. JAMA 1999; 281:627–633. 31. Mehta RH, Montoye CK, Gallogly M, et al. Improving quality of care for acute myocardial infarction: the Guidelines Applied in Practice (GAP) Initiative. JAMA 2002; 287:1269–1276. 32. Cabana MD, Rand CS, Powe NR, et al. Why don’t physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999; 282:1458–1465. 33. Payne TH. Computer decision support systems. Chest 2000; 118:47S–52S. 34. Borbas C, Morris N, McLaughlin B, Asinger R, Gobel F. The role of clinical opinion leaders in guideline implementation and quality improvement. Chest 2000; 118:24S–32S. 35. Alderson P, Schierhout G, Roberts I, Bunn F. Colloids Versus Crystalloids for Fluid Resuscitation in Critically Ill Patients (Cochrane Review). Oxford: The Cochrane Library, 2002, update software. 36. Heyland DSC, MacDonald S, Keefe L, Drover JW. Total parenteral nutrition in the critically ill patient: a meta-analysis. JAMA 1998; 280:2013–2019. 37. The Institute for Healthcare Improvement, www.ihi.org/ihi. Ref. type: electronic citation, 2003. 38. Cook DJ, Meade MO, Hand LE, McMullin JP. Toward understanding evidence uptake: semirecumbency for pneumonia prevention. Crit Care Med 2002; 30:1472–1477. 39. Ricart M, Lorente C, Diaz E, Kollef MH, Rello J. Nursing adherence with evidence-based guidelines for preventing ventilator-associated pneumonia. Crit Care Med 2003; 31:2693–2696.

Changing Ventilator Management in Patients with ALI

625

40. James PA, Cowan TM, Graham RP. Patient-centered clinical decisions and their impact on physician adherence to clinical guidelines. J Fam Pract 1998; 46:311–318. 41. Cruz-Correa M, Gross CP, Canto MI, et al. The impact of practice guidelines in the management of Barrett esophagus: a national prospective cohort study of physicians. Arch Intern Med 2001; 161:2588–2595. 42. Morris AHM. Guideline adoption: a slow process. Crit Care Med 2004; 32:1409–1410. 43. Rubenfeld GD, Cooper C, Carter G, Thompson BT, Hudson LD. Barriers to providing lung-protective ventilation to patients with acute lung injury. Crit Care Med 2004; 32:1289–1293. 44. Curry SJ. Organizational interventions to encourage guideline implementation. Chest 2000; 118:40S–46S. 45. Lo B. Improving care near the end of life. Why is it so hard? JAMA 1995; 274:1634–1636. 46. Slotnick HB. Physicians’ learning strategies. Chest 2000; 118:18S–23S. 47. Weingarten S. Translating practice guidelines into patient care: guidelines at the bedside. Chest 2000; 118:4S–7S. 48. Kotler P, Roberto N. Social Marketing: Strategies for Changing Public Behavior. New York, London: Free Press, Collier Macmillan, 1989:xii, 401. 49. Pervin LA. Personality: Theory, Assessment, and Research. New York: Wiley, 1970:xiv, 632. 50. Rogers EM. Diffusion of Innovations. 4th ed. New York: Free Press, 1995: xvii, 519. 51. Soumerai SB, McLaughlin TJ, Gurwitz JH, et al. Effect of local medical opinion leaders on quality of care for acute myocardial infarction—a randomized controlled trial. JAMA 1998; 279:1358–1363. 52. Deming WE. Out of the Crisis. 1st ed. Cambridge, Massachusetts: MIT Press, 2000:xiii, 507. 53. Yamamoto LG, Wiebe RA, Matthews WJ Jr, Sia CC. The Hawaii EMS-C project data: I. Reducing pediatric emergency morbidity and mortality; II. Statewide pediatric emergency registry to monitor morbidity and morality. Pediatr Emerg Care 1992; 8:70–78. 54. Grol R. Personal paper: beliefs and evidence in changing clinical practice. BMJ 1997; 315:418–421. 55. Kaye W. Research on ACLS training—which methods improve skill and knowledge retention?. Respir Care 1995; 50:538–546. 56. Ely EW, Meade MO, Haponik EF, et al. Mechanical ventilator weaning protocols driven by nonphysician health-care professionals: evidence-based clinical practice guidelines. Chest 2001; 120:454S–463S. 57. Meade MO, Ely EW. Protocols to improve the care of critically ill pediatric and adult patients. JAMA 2002; 288:2601–2603. 58. David SP, Greer DS. Social marketing: application to medical education. Ann Intern Med 2001; 134:125–127. 59. Gelijns A, Dawkins HV. Institute of Medicine (U.S.). Committee on Technological Innovation in Medicine. Adopting new medical technology. Washington, D.C.: National Academy Press, 1994:xiv, 224.

626

Neff and Rubenfeld

60. Davis DA, Taylor-Vaisey A. Translating guidelines into practice. A systematic review of theoretic concepts, practical experience and research evidence in the adoption of clinical practice guidelines. CMAJ 1997; 157:408–416. 61. Bero LA, Grilli R, Grimshaw JM, Harvey E, Oxman AD, Thomson MA. Closing the gap between research and practice: an overview of systematic reviews of interventions to promote the implementation of research ﬁndings. The Cochrane Effective Practice and Organization of Care Review Group. BMJ 1998; 317:465–468. 62. Smith WR. Evidence for the effectiveness of techniques to change physician behavior. Chest 2000; 118:8S–17S. 63. Davis D, O’Brien MA, Freemantle N, Wolf FM, Mazmanian P, Taylor-Vaisey A. Impact of formal continuing medical education: Do conferences, workshops, rounds, and other traditional continuing education activities change physician behavior or health care outcomes?. JAMA 1999; 282:867–874. 64. East TD, Heermann LK, Bradshaw RL, et al. Efﬁcacy of computerized decision support for mechanical ventilation: results of a prospective multi-center randomized trial. Proc AMIA Symp 1999:251–255. 65. Berenholtz SMM, Pronovost PJM, Lipsett PAM, et al. Eliminating catheterrelated bloodstream infections in the intensive care unit. Crit Care Med 2004; 32:2014–2020. 66. Larson E. Status of practice guidelines in the United States: CDC guidelines as an example. Prev Med 2003; 36:519–524.

24 How to Design Clinical Studies for Preventing Ventilator-Induced Lung Injury

LAURENT BROCHARD

CHRISTIAN ME´LOT

Medical ICU, Henri Mondor Teaching Hospital, AP-HP, Paris 12 University Cre´teil, France

ICU, Erasme Teaching Hospital, Free University of Brussels Brussels, Belgium

ALAIN MERCAT Medical ICU, Angers Teaching Hospital Angers, France

I. Introduction—Questions to Be Addressed A. General Principles

Choosing the study question is the critical step when initiating a research project. Hulley suggested the following criteria (FINER) for choosing a good study question (1):

Feasible in terms of resources, expertise, etc. Interesting to the investigator Novel: ideally the question should generate new data or conﬁrm or refute earlier ﬁndings Ethical Relevant

The randomized, controlled trial (RCT) is now considered the ideal proving ground for new treatments. RCTs must have sufﬁcient power to detect a clinically important difference in outcome between patients receiving the experimental treatment and controls receiving standard treatment only. Strict methodological rules must be followed, and many of them are 627

628

Brochard et al.

described in this chapter. As an introduction to the conduct of RCTs, however, several major points must be emphasized. 1.

2.

3.

Most RCTs provide clinical proof of a concept. The National Institutes of Health–Acute Respiratory Distress Syndrome (NIHARDS) Network study comparing tidal volumes of 6 mL/kg versus 12 mL/kg of predicted body weight deﬁnitively established that ventilator-induced lung injury (VILI) or ventilator-associated lung injury exists and worsens the clinical outcome (2). It did not determine the optimal tidal volume or the optimal mechanical ventilation strategy for ARDS patients. The heated debate sparked by the release of this important study highlights the fact that other analysis strategies will be needed to answer most of the clinical questions related to ventilator settings (3–6). RCTs are expensive and time consuming, and they expose participants not only to possible beneﬁts but also to possible harm. Therefore, RCTs cannot be used to answer all clinical questions. When designing an RCT, deﬁning what ‘‘usual care’’ is may be difﬁcult, most notably in the ﬁeld of nonpharmacological treatments. In addition, ‘‘usual care’’ and ‘‘standard of care’’ do not have the same meaning. When textbooks, guidelines, or consensus conferences have established a standard of care, that standard should be used in the controls. For instance the standard of care for the duration of antibiotic therapy in patients with ventilatorassociated pneumonia was two weeks, and an RCT could compare a shorter duration to a two-week course (7). However, actual clinical practice (or ‘‘usual care’’) may be difﬁcult to determine and may depart substantially from published recommendations given the many obstacles to implementing protocols or recommendations in the clinical setting (8,9). One can imagine different possibilities for the designing of RCTs that have a ‘‘usual care’’ arm. One is to have several control groups, each receiving one pattern of widely used care, so that most clinicians can recognize a control group that matches best their usual practice. Alternatively, multicenter observational studies can be done before the RCT to collect information on ‘‘usual care.’’ Finally, the RCT can include an arm in which the treatment is left at the discretion of the clinicians. An additional difﬁculty with unblinded RCTs versus usual care at the discretion of the clinicians is that knowledge of the experimental treatment may inﬂuence treatment choices made for patients in the ‘‘usual care’’ arm. It is very unlikely that this difﬁculty can be entirely overcome. Clearly, none of these solutions is ideal, and a major drawback is decreased feasibility related to the need for a much larger number

How to Design Clinical Studies for Preventing VILI

4.

5.

629

of patients. On the other hand, administration of ‘‘usual care’’ to the controls, at least as closely as possible, is crucial to ensure external validity: for instance, if clinicians usually give 8 or 9 mL/kg of tidal volume, they are unlikely to believe they should switch to 6 mL/kg just because this lower volume is better than 12 mL/kg (10). RCTs must represent the ﬁnal stage in an investigative program that may begin with biologic studies, animal experiments, human physiological studies, clinical observations, or all of the above. The clinical and physiological relevance of the study question is very important, and an RCT should not be a ﬁshing expedition, testing the hypothesis that a new approach may be better than another strategy simply because it is new. For all these reasons, RCTs cannot constitute the deﬁnitive or universal tool for solving clinical problems. Obstacles to blinding and informed consent, a low incidence of the disease under study, or an unavoidable inﬂuence of the study setting on ‘‘usual care’’ may make RCTs impractical or infeasible. Other designs, when used rigorously, may provide solutions. It is important to understand that strict methodological rules aimed at minimizing bias can be used in before–after studies, case–control studies, and observational studies and that these designs can make a valuable contribution to answering important clinical questions.

B. Clinical Questions

Many questions remain to be addressed regarding mechanical ventilation in ARDS patients and the prevention of VILI (11,12). Noninvasive Ventilation

Patients with acute lung injury (ALI) are usually considered poor candidates for noninvasive mechanical ventilation (NIMV) because they have other organ failures and high levels of minute ventilation leading to high ventilatory requirements (13). Several studies of carefully selected patients found that NIMV improved physiological criteria or clinical outcomes, whereas other studies raised a number of concerns. Most notably, in most studies, patients who received endotracheal mechanical ventilation after failing NIMV had very high mortality rates (14,15). Conceivably, these patients may experience extremely wide transpulmonary pressure swings, generating an increased risk of VILI; alternatively, failure of NIMV may merely constitute a marker for greater disease severity.

630

Brochard et al. Tidal Volume

One of the key questions today regarding ventilatory settings is the optimal tidal volume for preventing VILI. This question, which has fueled much debate, is discussed in Chapter 20 in this book. Brieﬂy, three views have been expressed, all of which are based on post hoc analyses of available RCTs. The ﬁrst view is that the risk of VILI becomes high at or beyond a threshold of lung distension, as assessed by plateau pressure as a surrogate marker. According to another view, the relationship between lung distension or plateau pressure and VILI is U shaped, with small tidal volumes being harmful via mechanisms that differ from those associated with large tidal volumes (16). The third view is that the risk of VILI decreases with tidal volume. Clearly, further research is needed (17). Designs other than RCTs will undoubtedly be helpful, because clinicians use a variety of strategies. Adjusted comparisons of prospective cohort series may prove particularly valuable. Modes of Mechanical Ventilation

Ventilatory modes that have been tested in ARDS with the main goal of improving oxygenation and thereby outcomes could be tested regarding their potential for preventing VILI. We will discuss a few of these modes. From a purely theoretical standpoint, and provided a number of improvements are made, proportional-assist ventilation could help in preventing VILI (18). The tidal volume is entirely chosen by the patient, and an increase in the level of assistance results in a reduction in the effort of breathing associated with the desired tidal volume, instead of increasing tidal volume. Some patients seem comfortable with extremely small tidal volumes and may conceivably select the optimal tidal volume based on lung-to-brain signals indicating that safe lung distension is maximal. To use proportional-assist ventilation in practice, accurate information on the respiratory system’s resistance and compliance must be available. These variables could be measured automatically and be used to control the ventilator (19,20). Proportional-assist ventilation could be tested during NIMV. The same reasoning may apply to neurally adjusted ventilatory assist, which is based on the patient’s diaphragmatic electromyography signal (21,22). Again, the patient fully controls the ventilator output, although no clinical trials are available yet. High-frequency oscillation (HFO) should, in theory, minimize lung injury by ensuring ventilation of lung regions with sufﬁcient reopening of the edematous tissue while avoiding overdistension at end-expiration (23). One of the main limitations, however, is that the clinician cannot determine either the absolute lung volume that is reached or the regional distension. During standard mechanical ventilation, end-inspiration is the most hazardous phase for the circulation because the alveolar pressure is higher than both the arterial and venous capillary pressures (zone I). Conceivably, if set too high, HFO may produce a continuous zone I in a large part of the lungs,

How to Design Clinical Studies for Preventing VILI

631

thereby producing major adverse effects. We believe this is an example of a research area where much more physiological work is needed before a meaningful RCT can be performed. The optimal combination of lung recruitment, distension risk, and circulatory collapse needs to be addressed. Unfortunately, the circulatory effects of mechanical ventilation often receive little or no research attention. Of note, the importance of right ventricular failure and cor pulmonale was not recognized when designing RCTs to test the efﬁcacy of ventilatory modalities. Acute cor pulmonale results from an increase in right ventricular afterload with pulmonary hypertension (24). Factors involved in acute cor pulmonale include lung disease, hypoxemia, hypercapnia, and ventilatory pressures. Ignoring acute cor pulmonale, which is a problem different from impediments to venous return, can lead to inappropriate ventilator settings that may precipitate circulatory failure. Circulatory failure is a major prognostic factor in ARDS. Therefore, adjusting the ventilator settings to minimize the risk of circulatory failure is a key goal (25). This goal has not yet been incorporated into ventilatory strategies aimed at maximizing recruitment. Nonventilatory treatments such as exogenous administration of surfactant or small doses of partial liquid ventilation are potential avenues for clinical research. New pharmacological treatments targeting the molecular mechanisms of VILI will perhaps be developed in the future and will be worth testing. This is discussed in Chapters 25, 26, and 28 in this book.

II. Inclusion and Exclusion Criteria The primary objective of clinical trials in ALI/ARDS is to provide an accurate and reliable evaluation of the effects of a given intervention in the target population of patients with ALI/ARDS. The choice of inclusion and exclusion criteria heavily inﬂuences both the scientiﬁc and the clinical value of a trial. This is especially true in ALI/ARDS because the well-known heterogeneity of this condition decreases the accuracy, reliability, and generalizability (external validity) of clinical trials. Extremely stringent inclusion and exclusion criteria can increase the internal validity of a trial by minimizing the number of confounding variables (26). Unfortunately, this strategy may lead to the selection of a highly speciﬁc group of patients that may not be representative of the general population of patients with ALI/ ARDS. The result is low external validity and therefore severely limited clinical usefulness. Experimental data suggest that VILI can occur very rapidly after the initiation of mechanical ventilation (27). Therefore, any intervention aimed at reducing VILI should be applied as soon as possible. In clinical trials, the time needed to check inclusion and exclusion criteria and to obtain informed consent may delay initiation of the study intervention.

632

Brochard et al.

The most critical inclusion criterion in studies on ALI/ARDS is the presence of bilateral pulmonary inﬁltrates consistent with edema. Indeed, several studies found poor interobserver agreement for the radiological diagnosis of ARDS (28,29). Formal training of investigators in radiograph interpretation seems necessary to decrease the heterogeneity of patients enrolled in multicenter clinical trials (29).

III. Outcomes Outcomes range from hard end points to soft end points (Fig. 1). Mortality is the hardest end point for studies of VILI. Time-to-death is among the most widely used end points. Nonparametric tests such as Kaplan– Meier curves are best for RCTs. In observational studies, the Kaplan–Meier curves must be completed by a multivariable approach (Cox model) to adjust for between-group differences in confounding variables. When mortality is evaluated over a short and predetermined period (e.g., 28 days) in each patient, logistic regression is used to model the predictors of death. As with the Cox model, multivariable logistic regression can be used to adjust for confounders. Clinical end points based on physiological measurements are appropriate when difﬁculty in detecting a mortality decrease is anticipated. There is general agreement that the most clinically relevant outcomes are those representing a direct link between the treatment and the patient’s health

Figure 1 ‘‘Hard’’ and ‘‘soft’’ end points in clinical studies on preventing VILI in ALI/ARDS. Abbreviations: ALI, acute lung injury; ARDS, acute respiratory distress syndrome; VILI, ventilator-induced lung injury.

How to Design Clinical Studies for Preventing VILI

633

status. With continuous study variables (e.g., arterial PCO2 or PaO2/FiO2), the effect of treatment is usually evaluated by comparing changes versus baseline. Response data are often collected at several time points during the trial. In intensive care units (ICUs), due to the complexity of clinical problems in patients with multiorgan failure, the primary end point may be adjusted on disease severity scores [e.g., simpliﬁed acute physiology score (SAPS II), sequential organ failure assessment (SOFA), or acute physiology and chronic health evaluation (APACHE)]. A. Primary End Point

Regulatory agencies that evaluate new pharmacological interventions recommend that clinical trials have a single primary end point. In conﬁrmatory trials, the primary hypothesis typically deals with the compared efﬁcacy of the study treatment and one or more other treatments. The hypothesis may be superiority or noninferiority of the new treatment. The primary end point is the measure that provides the best direct evidence about the primary hypothesis. In studies of serious or life-threatening diseases, the primary end point (e.g., mortality) can be assessed either as a binary variable or as the time to occurrence of the criterion. It is important that the primary end point be assessed without bias, in a reliable manner, using validated instruments that are sufﬁciently sensitive to detect real changes in a patient’s health status. These assessments should be made prior to initiation of the trial, using experience from previous trials, and not be based on a post hoc analysis. The power and sample size for the trial should be selected based on the primary end point. B. Composite End Points

Composite end points are built from multiple measurements or end points to provide a summary of the patient’s outcome. At the individual level, this involves combining univariate responses in a clinically sensible manner. Composite end points are often used in quality-of-life assessments. They can also be constructed from multiple clinical events. Composite end points may offer increased statistical power due not only to the reduction in dimensionality of the end point combination, but also to the higher incidence of the composite event when incidences of the individual events are low. However, this advantage is offset if the treatment does not affect all individual end points consistently. Also, the clinical interpretation of composite end points can be difﬁcult. C. Surrogate End Points

There are situations in which short-term indicators of a treatment response reliably predict long-term treatment effects. Time on mechanical ventilation

634

Brochard et al.

with a lung-protection strategy (e.g., low tidal volume) may deserve consideration as a surrogate end point. However, because mortality is high in ARDS patients, time on mechanical ventilation is biased by the death of a substantial proportion of patients after a few days. To circumvent this bias, ventilatorfree days (or organ dysfunction–free days) are preferred. Ventilator-free days may be a viable surrogate marker for a better health outcome (30). D. Secondary End Points

Secondary outcome variables are used either to help interpret the primary results or to investigate secondary objectives or hypotheses. Secondary end points may also include explanatory variables that serve for generating hypotheses to be tested in future studies.

IV. Study Designs Studies are traditionally classiﬁed as either observational or experimental (Fig. 2). Evidence from observational studies is considered weaker and less robust. Observational study designs range from purely descriptive studies to studies involving complex statistical testing, such as cohort studies. Experimental studies evaluate an intervention. Their major advantage is their ability to provide strong support for a causal link between the treatment and the outcome. In contrast, it is very difﬁcult to evaluate causality based on observational studies. Also, effective control of confounders can be achieved in experimental studies. The study design directly inﬂuences the level of

Figure 2

Evidence-based medicine: hierarchy of clinical study designs.

How to Design Clinical Studies for Preventing VILI

635

evidence provided by the study and therefore the reliability of recommendations that can be drawn from the study using the evidence-based approach. Experimental studies are generally reserved for research questions about which a large body of data has been obtained. Considerable ground work must be done before a clinical trial is considered. An experimental design is chosen when: (i) the research question cannot be addressed by observational studies; (ii) earlier observational studies have not answered the research question; (iii) existing knowledge is not sufﬁcient to establish a clinical or public health policy; and/or (iv) an experiment is expected to substantially expand existing knowledge. V. The RCT The RCT is widely viewed as the ideal study design—the gold standard against which all other designs are compared. The sequence of the RCT (parallel) design is shown below (Fig. 3). The study population is usually selected from a far larger source population. Only those individuals who meet a set of inclusion and exclusion criteria are eligible for the RCT. In addition, an individual may participate only if the individual provides informed consent to the study. The selected participants are then allocated at random to either the investigational group or the control group (placebo or standard treatment). Randomization achieves three important goals: allocation to treatment groups depends on chance

Figure 3

Validity and biases of randomized, controlled trials.

636

Brochard et al.

alone and is therefore free of selection bias and investigator-related bias; known and unknown confounders are evenly distributed between the two treatment groups; and statistical inferences are valid. Randomization usually produces two groups that are comparable at baseline. When chance introduces a signiﬁcant difference for a potential confounder, the statistical analysis must include adjustment for that confounder. After randomization, the intervention is started. Ideally, the study should be blinded, with neither the investigator nor the patient knowing which treatment is being used. Assessment of the study end points during and/or after the intervention is performed by a blinded investigator to protect against detection bias. The results are then analyzed to look for differences in outcome rates between the two treatment arms. Two approaches can be used: the per protocol approach, which includes only those patients who completed the trial as planned in the study protocol, or the intention-to-treat approach, which includes all patients in the group to which they were allocated by the randomization process. Intention-to-treat testing eliminates the transfer bias that arises when differences in dropout rates result in an imbalance between the two randomized groups and therefore in loss of statistical power. Thus, the core questions in an RCT are the following:

Is Is Is Is Is Is

the trial justiﬁed? the control group appropriate? allocation randomized? blinding effective for the intervention and assessment? outcome assessment blinded? the intention-to-treat approach used for data analysis?

A. Is the Trial Justiﬁed?

The ﬁrst issue in any clinical trial is whether the trial is appropriate. There is universal agreement that a clinical trial is warranted only when there is a state of ‘‘equipoise.’’ Freedman deﬁnes equipoise thus: ‘‘There is no consensus within the expert clinical community about the comparative merits of the alternatives to be tested’’ (31). In other words, if the investigator is sure that the new therapy is better than the earlier one, then a trial is not warranted. Equipoise, in this case, is disturbed by the existence of evidence indicating superiority of the new treatment over the earlier one. Using the earlier treatment would be unethical. Equipoise is both a fascinating and a difﬁcult concept. It is the responsibility of every clinician to prove (to an ethics committee or institutional review board) that equipoise exists before starting a trial. At times, equipoise is disturbed when the trial is under way. For instance, new evidence from other studies may settle the research question. The trial must then be terminated before the planned sample size is reached. Otherwise, loss of chance would occur in patients allocated to the inferior treatment.

How to Design Clinical Studies for Preventing VILI

637

Early trial termination can be a very difﬁcult and painful decision, both for the investigators and for the participants. Nevertheless, the safety and chances for improvement of the participants are more important than the research study. B. Is the Control Group Appropriate?

Many authors argue that using a placebo is unethical when a treatment is available. No patient should be denied treatment, even one with limited effectiveness. Mienert suggested the following requirements for the test and control treatments (32):

They must be distinguishable from each other. They must be medically justiﬁable. There must be an ethical basis for using each treatment. Either treatment must be acceptable to study patients and to physicians. There must be reasonable doubt regarding the efﬁcacy of the test treatment. There should be reason to believe that the beneﬁts will outweigh the risks of the new treatment.

C. Is Allocation Randomized?

Once an eligible patient has agreed to participate in the trial, allocation to the test or control group should be free of selection bias. To eliminate selection bias, the patient must be allocated to a group at random, and both the patient and the physician must be unaware of the group to which the patient is allocated. This is done by double-blind randomization. Randomization also ensures that the baseline characteristics in the test and the control groups are similar, so that any differences noted later on can be ascribed to the difference in treatments. When allocation is not randomized, susceptibility bias may occur: patients with characteristics predicting a good treatment response may be allocated to the treatment group and those with less favorable characteristics to the control group. D. Is Blinding Effective for the Intervention and Assessment?

Blinding ensures that outcome assessment is free of detection bias. Blinding is logistically difﬁcult but essential. Some authors use the word ‘‘masking’’ instead of blinding. In a double-blinded trial, both the patient and the physician are unaware of the treatment received; in a single-blinded trial, the physicians know the treatment. RCTs usually report the effectiveness of blinding. Known adverse effects of drugs may unblind the physician (e.g., bradycardia due to beta-blockers). Such blinding, however, is difﬁcult or

638

Brochard et al.

impossible for many interventions in the ICU, such as mechanical ventilation, hemoﬁltration, early interventions, etc. Ideally, data collection, assessment of end points and side effects, and classiﬁcation procedures on individual patients should be carried out by observers who are unaware of the clinical details or treatment group. For instance, if chest radiographs must be read, they can be sent to another site for evaluation by radiologists who have no knowledge about the patients. The end points should be as objective and clinically relevant as possible. They should be assessable in a blinded fashion. For instance, pain is a subjective outcome that is difﬁcult to measure in a blinded manner. On the other hand, a biochemical parameter is objective and easily blinded. E. Is the Intention-to-Treat Approach Used for Data Analysis?

This is a very important issue when analyzing data from RCTs. All patients allocated to each treatment arm are included in the analysis of results with that treatment, whether or not they received or completed the planned regimen. Failure to use the intention-to-treat approach defeats the main purpose of randomization and can invalidate the study results. For instance, if a patient is randomized to the placebo but subsequently is switched to the study treatment then this patient should be included in the analysis of the placebo group. VI. Ethical Issues in a Clinical Trial

Is the equipoise requirement met? Is informed consent obtained? Is conﬁdentiality protected? Is the choice of the control group justiﬁed? Is there a predetermined set of criteria for premature study termination?

All clinical trials must be cleared by an ethics committee or institutional review board. Equipoise and the choice of a control group have been discussed above. Informed consent is another important issue. The potential participant must be told that the treatment may consist of a placebo and that the study may result in adverse events or even death. Only then can consent to participation in the trial be sought. Conﬁdentiality also needs to be protected. Premature study termination is appropriate in some situations. In some trials, an independent data safety and monitoring board (DSMB) periodically reviews the study data. When a signiﬁcant difference between the treatment groups is found at one of these interim analyses, the DSMB may decide to stop the study. This protects study participants from loss of chance related to the use of an inferior treatment. Early termination may also occur when the adverse event or death rate is unacceptably higher

How to Design Clinical Studies for Preventing VILI

639

in one of the arms or when the difference between the two treatments is so small that continuing the trial would be futile. A. The Moral Obligation to Design a Good Trial

Experimental designs pose many dilemmas. It may be unethical to introduce an untested or inadequately tested drug into widespread use. As stated by Sir Austin Bradford, ‘‘The ethical problem is, indeed, not solely one of human experimentation; it can also be one of refraining from human experimentation’’ (33). Furthermore, a clinical trial should not be undertaken when it is unlikely to provide a conclusive answer because of the absence of randomization, blinding, or a sufﬁcient number of participants (1). The investigator embarking on a clinical trial must make every effort to design the trial well and to consider all core issues. A study that fails to produce information about treatment efﬁcacy because of small sample size is an unethical study: human lives were put at risk and considerable resources expended for a study that was unlikely to answer the research question. Therefore, the number of patients to be included must be calculated before the trial based on realistic baseline rates and expected improvements. The components of an RCT are intended to avoid errors that may arise in the design and conduct of a nonrandomized study, biasing the results. Nevertheless, an RCT shares many design features with a prospective, concurrently controlled, cohort study. These features include the following: (a) every component of the trial is deﬁned before inclusion of the ﬁrst patient, by means of a detailed protocol developed to guide data collection for all study participants; (b) the comparison group is constituted and studied concurrently to the test group; and (c) all participants are followed forward in time. The main difference is this: in a prospective cohort study, the treatment is chosen based on the interaction between the physician and patient, whereas in an RCT the effects of this interaction are eliminated via random treatment allocation. Moreover, randomization validates statistical inferences. B. Study Designs Based on Variations in Informed Consent or Randomization

Numerous authors have approached the issue of clinical trial designs from a philosophical or ethical perspective. They have raised concerns about the manner in which patients are asked for informed consent or randomized to treatment. A number of variations in the conventional timing have been suggested to answer these concerns. Prerandomization Method

Zelen suggested a novel approach for clinical trials comparing a new treatment to standard treatment (34). He suggested that patients prerandomized

640

Brochard et al.

to standard treatment need not be requested to provide informed consent or be told that they were enrolled in a clinical trial. The group prerandomized to the investigational therapy must be asked to provide informed consent; those who refuse consent receive the standard treatment, but their data are analyzed with the investigational group (intention-to-treat analysis). If the number of refusals is substantial, the difference between treatment groups is reduced, and the comparison loses sensitivity. Moreover, this procedure is suitable only for nonblinded RCTs. Ellenberg vigorously criticized this design and stated that it had not been used for any major clinical trial (35). Double Consent Prerandomized Method

This model, also suggested by Zelen, is a variation of the method described above. The main difference is that patients in both treatment groups provide informed consent (36). This design requires a larger number of patients to achieve the same power as conventional methods, but it facilitates patient enrollment, in theory and sometimes in practice. Deferred Consent Process

In emergencies, it is usually impossible for the patient to provide informed consent to participation in a study testing a treatment that requires immediate initiation. Consent can be deferred in this situation until a relative can be contacted. Later, when the patient recovers decision-making capacity, the patient is asked to consent to the trial and to the use of the data collected so far. Cluster Randomization

In a cluster randomization trial, clusters of individuals, rather than individuals themselves, are allocated at random to different intervention groups (Fig. 4). Cluster randomization trials, also known as group randomization trials, are now widely used for evaluating nontherapeutic interventions including lifestyle modiﬁcation, educational programs, and innovation in the provision of health care. Cluster size ranges from fairly small groups (e.g., families) to entire communities. Hospital wards or medical practices are useful as clusters. Cluster randomization can be used for experimental studies such as a trial where all patients in a general practice are allocated to the same intervention so that the general practice constitutes a cluster. An observational study with clustering might, for instance, involve interviewing patients in several hospitals, with the patients in each hospital forming a cluster. Members of a cluster resemble one another more than they resemble members of other clusters. Standard statistical methods that ignore clustering are misleading, because they assume that each participant yields an independent observation. In a cluster design, observations within each cluster are correlated (clustering effect or design effect). The intraclass

How to Design Clinical Studies for Preventing VILI

Figure 4

641

Classiﬁcation of comparative studies.

correlation (q) measures the strength of this correlation, i.e., the similarity of responses within a cluster. Because of this correlation, a cluster randomized trial requires a larger sample size than does an individually randomized trial, and using standard sample-size formulas leads to inadequate statistical power. The design effect [DE ¼ 1 þ (m 1) q] for clusters of size m gives a measure of how many more patients must be added to each group to achieve the statistical power that individual randomization would provide. Failure to make allowances for the design effect may lead to excessively narrow conﬁdence intervals (CI) or excessively small P values and therefore to overestimation of statistical signiﬁcance. Investigators conducting cluster randomized trials must make a multitude of design choices regarding selection of the primary unit of inference, the degree to which clusters should be matched or stratiﬁed on prognostic factors at baseline, and cluster subsampling. Moreover, ethical principles developed for individually randomized trials may also require modiﬁcation. Thus, as stated in recently released guidelines for cluster randomized trials (37), ‘‘ . . . the roles of the guardians of the patients’ interests during the trial, the gatekeepers of access to patient groups, and sponsors of the research are even more important in cluster randomized trials where individuals may not have the opportunity to give informed consent to participation.’’ Although this recommendation is directed primarily toward trials of therapeutic interventions, it may be just as relevant to prevention trials and to evaluations of nontherapeutic interventions. When permission from gatekeepers associated with each cluster is needed for assigning interventions, some indication should be provided as to who these gatekeepers are and how they were identiﬁed. Information about the consent procedure used for individual study participants should

642

Brochard et al.

also be provided. In particular, it is helpful to know what opportunities, if any, could be used by cluster members to avoid the risks inherent in the intervention (38). C. Sequential Design vs. Fixed-Sample Design

Much of the statistical methodology currently used to design and analyze clinical trials grew out of principles developed for agricultural ﬁeld trials in the 1920s. An important structural difference between ﬁeld trials and clinical trials is the time-pattern of data accumulation. Field trials are governed by the natural pattern of the seasons. Data from all plots become available for analysis at the same time. If they are insufﬁcient to allow a deﬁnitive conclusion, a new experiment must be planned. By contrast, data from a clinical trial are accumulated gradually over a period that can cover several months or years. Results from patients recruited early in the study are available for interpretation at a time when additional patients are being recruited and allocated to treatment groups. Evidence from the earliest patients can be used to decide when the study should be stopped. This design ensures the earliest possible identiﬁcation of the less successful treatment and therefore minimizes exposure of individuals to this treatment. Interim analyses of the data as they accumulate and the use of stopping rules based on the results would appear to be natural consequences of gradual data collection. In ﬁxed-sample studies, however, the data are not examined until the end of the study. In this case, the required sample size must be determined before patient recruitment starts. The sample size depends on the desired power (deﬁned as the ability of a negative trial to detect a true difference if it exists, usually 80% or more) and the type I error rate (the highest accepted risk of ﬁnding a signiﬁcant difference when none exists, usually 5% or less). In contrast, in the sequential design the data are examined periodically throughout patient recruitment. In group sequential designs, the data are examined when k responses are recorded (k > 1). Each interim analysis consists of computing two statistics, Z and V. Positive Z values indicate superiority of the experimental treatment, zero indicates equivalence, and negative Z values inferiority. V is approximately proportional to sample size and measures the information in the trial. The values for Z and V are plotted against one another, and the resulting points are compared to the stopping boundaries computed using triangular tests with the desired power and type I error rate. When the results cross the boundaries, the trial is stopped (Fig. 5) (39). D. Data Safety and Monitoring Board

A DSMB is a panel of clinical research experts (statisticians, scientists, bioethicists, and clinicians) independent from the trial promoter and investigators (40,41). The DSMB safeguards the patients enrolled in the trial and

How to Design Clinical Studies for Preventing VILI

Figure 5

643

Graphical analysis of a sequential trial (triangular test).

guarantees the validity and credibility of the data. To these ends, throughout the study, the DSMB conducts regular assessments of the safety and efﬁcacy of the interventions used and monitors the overall conduct of the study. Reports of adverse events or results of planned interim analyses may lead the DSMB to ask that the promoter stop the study, either because one treatment is clearly beneﬁcial or harmful compared to the other or because continuing the study would fail to answer the research question. The role of the DSMB in protecting trial participants is of particular importance in studies of critically ill patients with a high risk of death (42). When the investigational treatment seems associated with an increased mortality rate, early termination of the study based on closely spaced interim analyses may prevent a substantial number of treatment-related deaths. Freeman et al. retrospectively analyzed the results of the study on human growth hormone in critical illness, which had no assessment by the DSMB (42,43). They showed that early stopping rules for harm could have prevented 8 to 44 deaths. The ethical consequences are less momentous when a study is stopped early because superiority of the new treatment is established by an interim analysis. Nevertheless, early study discontinuation decreases the number of patients exposed to the less effective treatment and allows for prompter dissemination of the study results. The main risk is that the new treatment may be erroneously deemed effective (41). This risk can be minimized, but not completely eliminated, by using extremely stringent stopping rules. The much publicized ‘‘6 mL/kg versus 12 mL/kg study’’ conducted by the NIH-ARDS Network was stopped when the fourth interim analysis

644

Brochard et al.

established superiority of the smaller volume (P ¼ 0.005, with a P value for the stopping boundary of 0.023) (2). The main goal of rules for stopping futile trials is not to protect the safety of trial participants but to reduce the time and money spent on a clinical trial that will not show a statistically signiﬁcant difference. The risk here is that the new treatment may be erroneously deemed similar to the comparator. This risk explains the controversy surrounding the futility-stopping rule (44). Another disadvantage of futility-stopping rules is that the results are difﬁcult to interpret. The assessment of low tidal volume and elevated end-expiratory volume to obviate lung injury (ALVEOLI) study comparing higher to lower end-expiratory pressures in patients with ARDS was stopped for futility after the second interim analysis (11). This decision was in accordance with the predetermined futility-stopping rule. However, the early discontinuation contributed to the study’s inability to provide conclusive information on whether higher positive end-respiratory pressure (PEEP) levels inﬂuence mortality in ARDS patients (45).

VII. Understanding the Results of a Clinical Trial A. Absolute and Relative Risk Reduction

Let us consider the ARDS-Network trial comparing two strategies for ventilating ARDS patients (2). The end point was the mortality rate in each arm, that is, a dichotomous criterion (alive or dead). At the end of the trial, the death rate in the group ventilated with a small tidal volume (6 mL/kg) was compared to that in the comparator group (12 mL/kg). A far smaller death rate in the low tidal volume group would argue in favor of the 6 mL/kg strategy. The study showed a death rate of 31.0% with 6 mL/kg and 39.8% with 12 mL/kg. These results can be presented in many ways. The absolute risk reduction (ARR) is the difference between the death rates: 39.8% minus 31.0%, or 8.8%. Ventilation with 6 mL/kg reduced the risk of death by 8.8% as compared to ventilation with 12 mL/kg. The relative risk reduction (RRR) is the percentage of deaths in the control group that would have been prevented had this group received the investigational treatment. The RRR is computed as the difference in outcome rates between the control and investigational groups divided by the outcome rate in the control group. In our example, RRR ¼ 100 [(39.8% 31.0%)/ 39.8%] ¼ 22.1%. Ventilating with 6 mL/kg reduced the risk of death by 22.1% compared to ventilating with 12 mL/kg. The greater the RRR, the more effective the treatment. The number needed to treat (NNT) is the number of patients that must receive the study treatment to prevent one adverse outcome. The NNT was designed to overcome a weakness inherent in the RRR concept. If, for example, the 6 mL/kg strategy had reduced the risk from 3.98% to

How to Design Clinical Studies for Preventing VILI

645

3.10%, the RRR would still be 22.1%. Thus, the RRR does not provide information on the overall death rate. The NNT is the inverse of the ARR times 100. In our example, the ARR was 8.8%, which yields a ratio slightly greater than 11 and an NNT of 12. In other words, 12 patients would have to be treated to prevent one death. If we now go back to the hypothesis of a risk reduction from 3.98% to 3.10% in the investigational arm, we ﬁnd that the NNT is 114, indicating a far smaller therapeutic effect. The lower the NNT, the more effective the new therapy. B. Precision of Rates

All the above rates are only point estimates. They should therefore be reported with their 95% CI. Consider the scenario that yielded a 22% RRR. If the 95% CI for this point estimate were 2% to 46%, for instance, then we would have to conclude that the new treatment can yield worse outcomes than the control treatment (RRR ¼ 2%). This wide 95% CI indicates that the new treatment is no better than the control. The P value in this case would not be signiﬁcant (P 0.05). Now let us consider a point estimate of 22% with a 95% CI of 16% to 28%. The worst possible performance of the investigational treatment is an RRR of 16%. Here the result is statistically signiﬁcant (P value < 0.05) and the new strategy would be considered signiﬁcantly better than the conventional one. It is easy to appreciate that smaller sample sizes yield wider CIs. In very small trials, it is virtually impossible to obtain a statistically signiﬁcant difference in outcomes. In other words, small trials do not have sufﬁcient statistical power to pick up a difference that exists; therefore, absence of a difference in outcomes may reﬂect either absence of a difference between the treatments or inadequate statistical power.

VIII. Nonrandomized Cohort Studies In addition to RCTs, prospective (or retrospective) cohort studies using different designs to compare different clinical approaches deserve greater emphasis. The main issue in nonrandomized trials is to minimize bias and to adjust or control for confounding variables, i.e., variables that are not studied but may explain or greatly inﬂuence the apparent result of a nonrandomized comparison. This is especially true for before–after studies or retrospective cohort studies. Effective statistical methods are available for dealing with confounders, including multivariable analysis and pairedmatched case–control studies (46,47). Interestingly, several reports have shown that carefully designed observational studies conducted using these statistical methods yield evidence that is similar in quality to results of RCTs (48,49).

646

Brochard et al.

Before–after studies are often retrospective studies in which clinicians describe an important and abrupt change in the management of patients and compare the period before to the period after the change. Alternatively, the design may be prospective, in which case three phases are usually described. The ﬁrst phase is an observational period during which outcomes are recorded with usual care. The second phase is marked by implementation of a new strategy or protocol, for instance, with the aim of preventing VILI. The third phase consists of assessing the implementation and effect of the new strategy. A major advantage of these studies is that they provide information on the feasibility of implementing a protocol in real life. The short duration of the study may be a limitation here, however. When huge efforts are expended, implementation of a difﬁcult protocol may be successful for a few weeks or months but less successful later on. It is therefore very important to repeat the study after several months to reassess protocol implementation and efﬁcacy. A second advantage of prospective before–after studies lies in their ability to evaluate the effect of a new strategy on a clinically relevant outcome variable. Here, several problems may arise. The case-mix may change between the two periods, making differences difﬁcult to interpret. It may be difﬁcult to ensure that no management changes occurred apart from the new strategy, and investigators must pay close attention to this problem. However, when the difference is very large and consistent with sound physiological and clinical reasoning, the results may be extremely valuable. For instance, Jardin et al. described the outcome of ARDS patients admitted 15 years apart in the same ICU (50). Between the two periods, high tidal volumes (> 12 mL/kg) and high PEEP levels were changed to low tidal volumes and low PEEP levels, resulting in a large plateau-pressure reduction. Mortality dropped by 50%. Although differences in case-mix may have occurred, most ICUs admit sicker and older patients than in the past. Therefore, this striking mortality difference is probably ascribable in large part to the reduction in plateau pressure. Prospective cohort studies can be used to compare the outcome of patients treated with different ventilatory strategies. Again, the main difﬁculty inherent in the absence of randomization consists of understanding what guides the clinicians’ choices. If the clinicians used the different strategies randomly, the comparison would be easy. However, clinicians choose a speciﬁc strategy based on disease-related factors such as severity or clinical presentation, and they are also inﬂuenced by personal factors. Several statistical methods are available for adjusting factors that inﬂuence treatment choices. These methods are described in statistical textbooks and include multivariable analysis, paired-matched case–control studies, and the use of the propensity score (47). The case–control design has been used in epidemiology to compare outcomes in exposed and unexposed individuals. This approach led to the identiﬁcation of the ‘‘French paradox,’’ i.e., the cardiovascular beneﬁts of drinking moderate amounts of wine. Although

How to Design Clinical Studies for Preventing VILI

647

it has been suggested that an RCT should be done to conﬁrm this unequivocal result (51), we believe this would ﬂy in the face of common sense. A similar statistical approach can be used in large patient cohorts. When two approaches exist to solving a given problem (e.g., endotracheal ventilation versus noninvasive ventilation for acute respiratory failure), patients treated with one or the other of these approaches can be selected to obtain two groups that are closely similar before the treatment decision. This selection of ‘‘matched pairs’’ is made on an individual basis; it should be based on the most clinically relevant criteria assessing severity and on the clinical indication. The limitation of this approach is that such criteria may be unavailable. Noninvasive ventilation studies illustrate the ability of the prospective cohort approach to predict the results of RCTs very early on (52). Prospective cohort studies also helped to clarify the impact of noninvasive ventilation on infection rates in the ICU (52,53). An important advantage of this study design is that the data come from ‘‘real life,’’ whereas patients in RCTs are not only carefully selected but also monitored and treated in an unusually standardized manner. Multivariable analysis is designed to identify all factors that independently inﬂuence the outcome of interest. It is widely used to identify factors independently associated with death in patients with ARDS, with two possible objectives. One is to identify important prognostic factors that need to be taken into account in clinical trials. The other objective is to identify a factor that could be manipulated to inﬂuence the outcome. In a large international multicenter survey on mechanical ventilation, Esteban et al. found that patients having secondary ARDS, i.e., ARDS not present at admission, had a far higher mortality rate than patients with ARDS at admission (54). This is an interesting ﬁnding with two potential implications: ﬁrst, these two types of ARDS should be distinguished in clinical trials, and second, studies are needed to determine whether the development of ‘‘secondary’’ ARDS is related to VILI and could be prevented. Looking at the tidal volumes used at admission and at plateau pressures might be a ﬁrst step toward addressing this question, which could hardly be solved by an RCT (55). There are numerous technical approaches for multivariable analysis, and the choice among them depends on the study questions and on the characteristics of the study variables (47). One important question is whether variables are time dependent or are simply baseline descriptive characteristics identiﬁed at admission, because this inﬂuences the choice between logistic regression and Cox models (56). Also, because these approaches are based on the generation of mathematical models, great care is needed in selecting the variables to be entered into the model. It is generally recommended that the number of variables be limited based on the number of events to be analyzed (e.g., the number of deaths); one variable per 10 events is a commonly used ratio (46). A larger number of variables means a greater

648

Brochard et al.

probability of missing data and therefore a greater number of excluded patients, which may considerably weaken the model. Several tests are used to describe model robustness, such as the goodness-of-ﬁt test. While these analyses are extremely attractive to clinicians, their major limitations must be recognized. An example of these limitations can be found in a study by Vieillard-Baron et al. Following the heated debate on risks associated with the use of the Swan–Ganz catheter, Vieillard-Baron et al. endeavored to analyze the independent factors associated with death in ARDS patients (57). They built two models in which they introduced the use of the Swan–Ganz catheter. In the ﬁrst model, the catheter was signiﬁcantly associated with mortality, in addition to the usual prognostic factors. The authors reasoned, however, that the Swan–Ganz catheter was used chieﬂy in patients with severe hemodynamic instability, and they introduced epinephrine or norepinephrine use as a marker for hemodynamic instability. This marker was strongly associated with mortality in the new model, whereas the Swan– Ganz catheter was not. Interestingly, that the Swan–Ganz catheter had no impact on mortality in this population of patients was subsequently conﬁrmed by a large RCT (58). This illustrates the considerable skill and creative thinking needed to build appropriate statistical models. In part, as a means of circumventing these difﬁculties, a new type of multivariable analysis called the propensity score has been developed (59– 61). The propensity score reﬂects the likelihood of receiving the investigational treatment rather than the control treatment for a patient with speciﬁc prognostic variables (61). It is built as a score describing the likelihood that a treatment (e.g., noninvasive ventilation compared to endotracheal ventilation) or a diagnostic tool (e.g., pulmonary artery catheter) will be used in a given patient. It is based on a multivariable analysis that incorporates the main variables differentiating the two groups (e.g., noninvasive ventilation versus endotracheal ventilation or pulmonary artery catheter versus no pulmonary artery catheter) and gives a score value for each individual patient (62,63). The score is then entered in a multivariable analysis of the outcome of interest (e.g., impact of noninvasive ventilation on infections or impact of using a pulmonary artery catheter on mortality). This minimizes bias regarding selection of the treatment or diagnostic procedure under study.

IX. Evidence-Based Medicine and Hierarchy of Study Designs In 1972, Cochrane suggested that scientiﬁc knowledge, most notably from prospective RCTs, may be superior to personal experience as a basis for making treatment decisions (64). He coined the term ‘‘evidence-based medicine’’ (EBM) to designate this approach. The principles of EBM were then used for clinical recommendations, and scientiﬁc data were classiﬁed into

How to Design Clinical Studies for Preventing VILI

649

ﬁve categories of decreasing reliability, which in turn were associated with different grades of recommendations (65,66). The EBM approach has been widely used to determine the optimal mechanical ventilation strategy for ARDS patients, but has generated much debate (67–69). Clearly, EBM has merit, and managing patients based on the best available research data is hardly criticizable. Regarding VILI, however, the hierarchical ranking of study quality introduced by EBM and used widely in review articles and consensus conferences is open to criticism (70). Large RCTs are described as providing the best possible evidence for guiding treatment decisions. We have seen, however, that RCTs have many limitations, are not suited to all situations, and may fail to reﬂect real-life conditions. On the other hand, case-series including the case–control studies are classiﬁed as having the least scientiﬁc value. As discussed above, we believe this contradicts common sense and clinical experience. In many instances, case–control studies provided answers that were later conﬁrmed by RCTs (24,52,53). Experimental studies are not even classiﬁed in the EBM hierarchy, although the description of VILI, the topic of this book, stems entirely from experimental work. Similarly, simple physiological observations can constitute the basis for a comprehensive patient-management strategy, as in the case of auto-PEEP in patients with obstructive lung disease (71). Therefore, although a universal grid to assess research and to make optimal treatment decisions may apparently hold considerable appeal, this approach may be too simplistic. References 1. Hulley SB, Cummings SR. Designing Clinical Research. Baltimore: Williams and Wilkins, 1998. 2. Network ARDS. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342:1301–1308. 3. Brower RG, Matthay M, Schoenfeld D. Meta-analysis of acute lung injury and acute respiratory distress syndrome trials. Am J Respir Crit Care Med 2002; 166:1515–1517. 4. Amato M, Brochard L, Stewart T, Brower R. Metaanalysis of tidal volume in ARDS. Am J Respir Crit Care Med 2003; 168:612–613. 5. Brower RG, Bernard G, Morris A. Ethics and standard of care in clinical trials. Am J Respir Crit Care Med 2004; 170:198–199. 6. Moran JL, Bersten AD, Solomon PJ. Meta-analysis of controlled trials of ventilator therapy in acute lung injury and acute respiratory distress syndrome: an alternative perspective. Intensive Care Med 2005; 31:227–235. 7. Chastre J, Wolff M, Fagon J-Y, et al. Comparison of 8 vs. 15 days of antibiotic therapy for ventilator-associated pneumonia in adults: a randomized trial. JAMA 2003; 290:2588–2598. 8. Namen AM, Ely EW, Tatter SB, et al. Predictors of successful extubation in neurosurgical patients. Am J Respir Crit Care Med 2001; 163:658–664.

650

Brochard et al.

9. Krishnan JA, Moore D, Robeson C, Rand CS, Fessler HE. A prospective, controlled trial of a protocol-based strategy to discontinue mechanical ventilation. Am J Respir Crit Care Med 2004; 169:673–678. 10. Ferguson ND, Frutos-Vivar F, Esteban A, et al. Airway pressures, tidal volumes, and mortality in patients with acute respiratory distress syndrome. Crit Care Med 2005; 33:21–30. 11. Brower RG, Lanken PN, MacIntyre N, et al. Higher versus lower positive endexpiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351:327–336. 12. Brower R, Brochard L. Mechanical ventilation in Acute Respiratory Distress Syndrome. In: Matthay, ed. Acute Respiratory Distress Syndrome. Marcel Dekker, 2001:29–34. 13. Brochard L, Mancebo J, Elliott M. Noninvasive ventilation for acute respiratory failure. Eur Respir J 2002; 19:712–721. 14. Antonelli M, Conti G, Moro M, et al. Predictors of failure of noninvasive positive pressure ventilation in patients with acute hypoxemic respiratory failure: a multi-center study. Intensive Care Med 2001; 27:1718–1728. 15. Delclaux C, Brochard L, Mancebo J. Continuous positive airway pressure in the hypoxemic patient. In: Brochard L, ed. Mechanical Ventilation and Weaning. Springer Verlag, 2002:336–347. 16. Eichacker P, Gerstenberger E, Banks S, Cui X, Natanson C. Metaanalysis of ALI and ARDS trials testing low tidal volumes. Am J Respir Crit Care Med 2002; 166:1510–1514. 17. Gattinoni L, Carlesso E, Cadringher P, Valenza F, Vagginelli F, Chiumello D. Physical and biological triggers of ventilator-induced lung injury and its prevention. Eur Respir J Suppl 2003; 47:15S–25S. 18. Younes M. Proportional assist ventilation. In: Tobin M, ed. Principles and Practice of Mechanical Ventilation. New York: McGraw Hill, 1994:349–369. 19. Younes MJ, Kun J, Masoiwski B, Webster K, Roberts D. A method for noninvasive determination of inspiratory resistance during proportional assist ventilation. Am J Respir Crit Care Med 2001; 163:829–839. 20. Younes M, Webster K, Kun J, Roberts D, Masoiwski B. A method for measuring passive elastance during proportional assist ventilation. Am J Respir Crit Care Med 2001; 164:50–60. 21. Sinderby C, Navalesi P, Beck J, et al. Neural control of mechanical ventilation in respiratory failure. Nat Med 1999; 5:1433–1436. 22. Sinderby C. Neurally adjusted ventilatory assist (NAVA). Minerva Anestesiol 2002; 68:378–380. 23. Kacmarek RM. High frequency oscillation versus conventional ventilation: Is one superior? Eur Respir J Suppl 1999; 14:733–734. 24. Vieillard-Baron A, Schmitt J-M, Augarde R, et al. Acute cor pulmonale in acute respiratory distress syndrome submitted to protective ventilation: incidence, clinical implications, and prognosis. Crit Care Med 2001; 29:1551–1555. 25. Page B, Vieillard-Baron A, Beauchet A, Aegerter P, Prin S, Jardin F. Low stretch ventilation strategy in acute respiratory distress syndrome: eight years of clinical experience in a single center. Crit Care Med 2003; 31:765–769.

How to Design Clinical Studies for Preventing VILI

651

26. Rothwell P-M. External validity of randomised controlled trials: ‘‘to whom do the results of this trial apply?’’ Lancet 2005; 365:82–93. 27. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 28. Beards SC, Jackson A, Hunt L, et al. Interobserver variation in the chest radiograph component of the lung injury score. Anaesthesia 1995; 50:928–932. 29. Meade MO, Cook RJ, Guyatt GH, et al. Interobserver variation in interpreting chest radiographs for the diagnosis of acute respiratory distress syndrome. Am J Respir Crit Care Med 2000; 161:85–90. 30. Schoenfeld DA, Bernard GR. Statistical evaluation of ventilator-free days as an efﬁcacy measure in clinical trials of treatments for acute respiratory distress syndrome. Crit Care Med 2002; 30:1772–1777. 31. Freeman B. Equipoise and the ethics of clinical research. N Engl J Med 1987; 317:141–145. 32. Mienert C. Clinical Trials. Oxford: Oxford University Press, 1986. 33. Hill AB. Bradford Hill’s Principle of Medical Statistics. 12th ed. New York: Oxford University Press, 1991. 34. Zelen M. A new design for randomized clinical trials. N Engl J Med 1979; 300:1242–1245. 35. Ellenberg SS. Randomization designs in comparative clinical trials. N Engl J Med 1984; 310:1404–1408. 36. Zelen M. Strategy and alternate randomized designs in cancer clinical trials. Cancer Treat Rep 1982; 66:1095–1100. 37. MRC Clinical Trials Series. Cluster Randomized Trials: methodological and ethical considerations. London, MRC Publications, 2002. 38. Donner A, Klar N. Pitfalls of and controversies in cluster randomization trials. Am J Public Health 2004; 94:416–422. 39. Whitehead J. The Design and Analysis of Sequential Clinical Trials. 2nd ed. New York: John Wiley & Sons, 2000. 40. A proposed charter for clinical trial data monitoring committees: helping them to do their job well. Lancet 2005; 365:711–722. 41. Slutsky AS, Lavery JV. Data safety and monitoring boards. N Engl J Med 2004; 350:1143–1147. 42. Freeman BD, Danner RL, Banks SM, Natanson C. Safeguarding patients in clinical trials with high mortality rates. Am J Respir Crit Care Med 2001; 164:190–192. 43. Takala J, Ruokonen E, Webster NR, et al. Increased mortality associated with growth hormone treatment in critically ill adults. N Engl J Med 1999; 341:785–792. 44. Schoenfeld DA, Meade MO. Pro/con clinical debate: it is acceptable to stop large multicentre randomized controlled trials at interim analysis for futility. Crit Care Med 2005; 9:34–36. 45. Levy MM. PEEP in ARDS—how much is enough? N Engl J Med 2004; 351:389–391. 46. Concato J, Feinstein AR, Holdford TR. The risk of determining risk with mutivariable models. Ann Intern Med 1993; 118:201–210. 47. Katz MH. Multivariable analysis: a primer for readers of medical research. Ann Intern Med 2003; 138:644–650.

652

Brochard et al.

48. Benson K, Hartz A. A comparison of observational studies and randomized, controlled trials. N Engl J Med 2000; 342:1878–1886. 49. Concato J, Shah N, Horwitz RI. Randomized controlled trials, observational studies and the hierarchy of research designs. N Engl J Med 2000; 342:1887–1892. 50. Jardin F, Fellahi J, Beauchet A, Vieillard-Baron A, Loubieres Y, Page B. Improved prognosis of acute respiratory distress syndrome 15 years. Intensive Care Med 1999; 25:936–941. 51. Goldberg IJ. To drink or not to drink? N Engl J Med 2003; 348:163–164. 52. Brochard L, Isabey D, Piquet J, et al. Reversal of acute exacerbations of chronic obstructive lung disease by inspiratory assistance with a face mask. N Engl J Med 1990; 323:1523–1530. 53. Girou E, Schortgen F, Delclaux C, et al. Association of noninvasive ventilation with nosocomial infections and survival in critically III patients. JAMA 2000; 284:2361–2367. 54. Esteban A, Anzueto A, Frutos F, et al. Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study. JAMA 2002; 287:345–355. 55. Gajic O, Dara SI, Mendez JL, et al. Ventilator-associated lung injury in patients without acute lung injury at the onset of mechanical ventilation. Crit Care Med 2004; 32:1817–1824. 56. Chevret S. Logistic or Cox model to identify risk factor of nosocomial infection: still a controversial issue. Intensive Care Med 2001; 27:1559–1560. 57. Vieillard-Baron A, Girou E, Valente E, et al. Predictors of mortality in acute respiratory distress syndrome. Focus on the role of right heart catheterization. Am J Respir Crit Care Med 2000; 161:1597–1601. 58. Richard C, Warsawski J, Anguel N, et al. Early use of the pulmonary artery catheter and outcomes in patients with shock and acute respiratory distress syndrome: a randomized controlled trial. JAMA 2003; 290:2713–2720. 59. Rosenbaum PR, Rubin DB. Reducing bias in observational studies using subclassiﬁcation on the propensity score. J Am Stat Assoc 1984; 79:516–524. 60. Drake C, Fisher L. Prognostic models and the propensity score. Int J Epidemiol 1995; 24:183–187. 61. Braitman LE, Rosenbaum PR. Rare outcomes, common treatments: analytic strategies using propensity scores. Ann Intern Med 2002; 137:693–695. 62. Connors AF, Speroff T, Dawson NV, et al. The effectiveness of right heart catheterization in the initial care of critically ill patients. SUPPORT investigators. JAMA 1996; 276:889–897. 63. Girou E, Brun-Buisson C, Taille´ S, Lemaire F, Brochard L. Secular trends in nosocomial infections and mortality associated with noninvasive ventilation in patients with exacerbation of COPD and pulmonary edema. JAMA 2003; 290:2985–2991. 64. Cochrane AL. Effectiveness and Efﬁciency. Random Reﬂections on Health Services. London: Nufﬁeld Provincial Hospitals Trust, 1972. 65. Sackett DL. Rules of evidence and clinical recommendations on the use of antithrombotic agents. Chest 1986; 89:2S–3S.

How to Design Clinical Studies for Preventing VILI

653

66. Guyatt GH, Sackett DL, Cook DJ. Users’ guides to the medical literature. II. How to use an article about therapy or prevention. A. Are the results of the study valid? Evidence-Based Medicine Working Group. JAMA 1993; 270:2598–2601. 67. Kollef MH, Schuster DP. The acute respiratory distress syndrome. N Engl J Med 1995; 332:27–37. 68. Kopp R, Kuhlen R, Max M, Rossaint R. Evidence-based medicine in the therapy of the acute respiratory distress syndrome. Intensive Care Med 2002; 28:244–255. 69. Dreyfuss D, Saumon G. Evidence-based medicine or fuzzy logic: what is best for ARDS management? Intensive Care Med 2002; 28:230–234 (doi: 10.1007/S00134-002-1231-8). 70. Brochard L, Mancebo J, Tobin M. Searching for evidence: don’t forget the foundations. Intensive Care Med 2003; 29:2109–2111. 71. Pepe PE, Marini JJ. Occult positive end-expiratory pressure in mechanically ventilated patients with airﬂow obstruction: the auto-PEEP effect. Am Rev Respir Dis 1982; 216:166–169.

This page intentionally left blank

25 Perﬂuorocarbons and Acute Lung Injury

BRADLEY P. FUHRMAN Division of Pediatric Critical Care, Department of Pediatrics, State University of New York at Buffalo and Women’s and Children’s Hospital of Buffalo Buffalo, New York, U.S.A.

I. Introduction Interest in perﬂuorocarbon liquids, as a means to ventilate the lung, grew out of Leland Clark’s discovery that mice can breathe these oxygen solvents without mechanical assistance (1). But perﬂuorocarbon liquids may prove most valuable to patients with respiratory failure because of their unanticipated anti-inﬂammatory and antioxidant properties. II. Perﬂuorocarbon Liquids as Media for Breathing Mammals can breathe perﬂuorocarbon liquids because they have high solubilities for oxygen, carbon dioxide, and nitrogen. Most perﬂuorocarbons will dissolve about 50 volumes of oxygen per 100 volumes of liquid, when equilibrated to 100% oxygen. They will also dissolve about 200 volumes of carbon dioxide per 100 volumes of liquid at a partial pressure of 760 Torr, a property that greatly facilitates carbon dioxide elimination during tidal liquid ventilation (TLV). Perﬂuorocarbon liquids also have relatively low viscosities, which facilitates their movement back and forth in the airway. 655

656

Fuhrman

This property compensates the mammal constrained to breathe liquid perﬂuorocarbon for the disadvantage it suffers by virtue of its branching airway, which evolved to support the to-and-fro movement of gas from mouth to alveolus. (Fish, in contrast, survive by extracting oxygen from water, which dissolves only 2.1 volumes of oxygen per 100 volumes of liquid when equilibrated to 100% oxygen. Fish propel water over their gills in one direction with almost no resistance to ﬂow.) Perﬂuorocarbon liquids also have remarkably low surface tension—not as low as a compressed ﬁlm of surfactant, but low enough to reduce surface tension at the biologic liquid–liquid alveolar interface within the perﬂuorocarbon-ﬁlled lung. Early medical interest in liquid ventilation had little to do with effects of perﬂuorochemicals on pulmonary inﬂammation. It had more to do with elimination of the air–ﬂuid interfaces in the premature lung where elevated surface tension resists lung inﬂation in neonatal respiratory distress syndrome (2). The observation that perﬂuorocarbon-associated gas exchange (PAGE), also known as partial liquid ventilation (PLV), reduces inﬂammation in the lungs of piglets injured by gastric aspiration (3) drew attention to the effects of perﬂuorocarbons on inﬂammation and oxidative injury.

III. Effects of Perﬂuorocarbons on Inﬂammation and Oxidative Injury Perﬂuorocarbon liquids are not metabolized or degraded in the body. They are metabolically inert. There has, therefore, been a tendency to view their biologic effects as mere extensions of their physical properties. PLV makes the small lung with surfactant deﬁciency more compliant, and this can occur almost immediately after instillation of the liquid. The abrupt nature of this effect suggests that it is mechanical. It is only logical to suppose that the observed quenching of inﬂammation represents a reduction in mechanical trauma to the lung achieved by normalizing the mechanics of breathing, and that it represents a form of mechanical protection from ventilator-induced lung injury. To distinguish those effects of perﬂuorocarbons on the lung that are purely mechanical from those that are more ‘‘biologic,’’ it is instructive to review certain effects of perﬂuorocarbons, which are clearly different from their mechanical effects during liquid ventilation. A. Perﬂuorocarbon Emulsion Blood Substitutes

Perﬂuorocarbons are only sparingly soluble in water. To take advantage of their gas-carrying properties and to fashion them into blood substitutes, they must be emulsiﬁed. Tiny droplets of perﬂuorocarbon are coated with phospholipid to create a stable emulsion, which can then be transfused

Perﬂuorocarbons and Acute Lung Injury

657

into the patient. These particles are rapidly cleared from the blood and sequestered by the reticuloendothelial system (RES). They are later eliminated from the body by gradual diffusion of molecules of perﬂuorocarbon back into the blood and evaporation from the blood into alveolar air. These features of the clearance of perﬂuorocarbon-based artiﬁcial blood products from the body are responsible for several biological effects. Uptake of particles into the RES can stimulate macrophages to release cytokines and other metabolites that cause transient cutaneous ﬂushing and fever (a ﬂu-like syndrome associated with infusion of lipid and perﬂuorocarbon emulsions) (4). Intense bombardment of the RES by large doses of emulsions can produce RES blockade and impairment of RES clearing capacity. Perﬂuorocarbon and lipid emulsions are also known to impede platelet aggregation (5). This has been attributed to interactions involving the surface phospholipid, which coats the perﬂuorocarbon micelles. A third effect of perﬂuorocarbon emulsions, which appears to be a direct consequence of their physical properties, is ‘‘pulmonary gas trapping.’’ High vapor pressure perﬂuorocarbons appear to bubble out of the blood to become transiently lodged under the alveolar surfactant lining. Such perﬂuorocarbon-rich bubbles may then grow by inward diffusion and osmosis of respiratory gases. The result is a lung that will not collapse at necropsy. This phenomenon is species speciﬁc (6). It has been observed in pigs, rabbits, and monkeys, but is virtually nonexistent in mice and dogs. It does not appear to affect humans. It has been clearly related to the vapor pressure of the perﬂuorocarbon and is more pronounced at high vapor pressure (7). When given intravenously, FluosolTM (FC-43), for instance, increased functional residual capacity (FRC) of rabbit lungs fourfold, whereas OxygentTM [perﬂuorooctylbromide (PFOB)] did not alter FRC (8). Other effects of perﬂuorocarbon emulsions are not so readily explained on the basis of known physical properties. Perﬂuorocarbon emulsions can be administered intravenously to limit infarct size. Transient reduction of cerebral hemispheric blood ﬂow followed by reperfusion caused loss of brain electrical activity in dogs. Pretreatment with either mannitol (a free-radical scavenger) or Fluosol allowed incomplete but distinct recovery of electrical activity, and the combination of the two agents caused marked recovery (9). The combination of Fluosol, mannitol, vitamin E, and dexamethasone prevented loss of electrical activity during ischemia and allowed apparent full recovery during reperfusion (10). Some beneﬁt of postinjury perﬂuorocarbon infusion was also seen in a complete ligation stroke model (11,12). It is not reasonable to ascribe these beneﬁcial effects to the limited quantity of oxygen that can be dissolved in the perﬂuorocarbon emulsion, certainly not after complete ligation. After permanent ligation of the left anterior descending coronary artery, Fluosol decreased infarct size and inﬂammatory inﬁltrate (13). Reperfusion

658

Fuhrman

after transient ischemia causes neutrophil activation and inﬁltration with plugging of capillary lumens and endothelial cell disruption. This progressively impedes reperfusion, causing a ‘‘no-reﬂow’’ phenomenon. The combination of perﬂuorocarbon emulsion (Fluosol) and adenosine ameliorates this sequence of events, suppresses chemotaxis and neutrophil inﬁltration, and reduces myocardial infarct size in dogs after 90 minutes of ischemia (14). Free-radical generation during reperfusion contributes to the reperfusion injury of cardiac transplantation. Fluosol administration during reperfusion has been shown to decrease free-radical generation (15). Perﬂuorocarbon emulsion has also been shown to reduce free radical– induced lipid peroxidation during reperfusion of the kidney and to preserve structure and function of reperfused lung. The oxygen-carrying properties of the perﬂuorocarbon emulsion appear to have played no direct role in these reperfusion studies, because reperfusion blood carries more oxygen per mL than the emulsion. Rather, the perﬂuorocarbon emulsion seems to have interfered with the cellular events that trigger inﬂammation, free-radical generation, and oxidative injury. Cardiopulmonary bypass circuits are known to activate leukocytes, a process that is believed to contribute to the acute inﬂammatory response that often follows cardiac operations. Blood circulated through an extracorporeal circuit (in the absence of a patient) undergoes gradual depletion of leukocytes, raising the expression of leukocyte adhesion protein CD11b, and increasing the production of reactive oxygen species. PFOB emulsion appears to quench this inﬂammatory process (16). Burns reduce plasma antioxidant capacity and cause free radical– mediated damage to erythrocytes. Fluosol reduces postburn erythrocyte malonyl dialdehyde concentration and oxidative hemolysis, but not by increasing blood antioxidant levels (17). Endothelial cells also participate in the inﬂammatory response. They may be activated by interleukin (IL)-1, tumor necrosis factor (TNF), or endotoxin [lipopolysaccharide (LPS)] to express leukocyte adhesion molecules, which marginate leukocytes. Perﬂubron emulsion virtually blocks endotoxin-induced activation of cultured umbilical vein endothelial cells, but does not impede IL-1 or TNF activation of endothelial cells (18). This speciﬁcity argues against a ‘‘barrier’’ mechanism. In all of the above-mentioned anti-inﬂammatory effects, perﬂuorocarbon emulsions act by blocking a stimulus to inﬂammation, often resulting in decreased free-radical production or diminished oxidative damage. When administered to healthy volunteers, Oxygent produced the ﬂu-like syndrome described above but did not affect delayed hypersensitivity skin reactions, lymphocyte proliferative responses to mitogenic stimulation, circulating levels of immunoglobulins, complement activation, or plasma levels of inﬂammatory cytokines, TNF, or IL-1 (19). Perﬂuorocarbon emulsions

Perﬂuorocarbons and Acute Lung Injury

659

appear to block certain triggers of inﬂammatory processes, rather than impair global immune function. B. Effects of Neat Perﬂuorocarbons on Inﬂammation

Effects of neat (pure rather than emulsiﬁed) perﬂuorocarbon liquids on inﬂammation have been studied both in vitro and in vivo. A variety of cell lines have been explored, including the alveolar macrophage, mononuclear blood cells, neutrophils, and alveolar epithelial cells. Exposure of both rabbit and piglet alveolar macrophages to PFOB blunts their responsiveness to potent stimuli (20), diminishing their production of free radicals and oxidative species. Both human alveolar macrophages and blood mononuclear cells exposed to perﬂuorohexane show reduced responsiveness to LPS as measured by IL-1, TNF, and tissue-factor production (21). Human neutrophils exposed to perﬂuorocarbons showed decreased activation, less hydrogen peroxide production, and impaired chemotaxis (22). When neutrophils and epithelial cells were incubated together, exposure to perﬂuorocarbon protected the target epithelial cells from neutrophil adhesion and neutrophil-induced injury (23). Different perﬂuorocarbon liquids appear to have highly complex, speciﬁc, and disparate effects on inﬂammatory cell responses (24). There appear to be barrier effects, whereby perﬂuorocarbons interfere with alveolar epithelial cell signaling, blocking IL production (25). The immiscibility of perﬂuorocarbons and water leads to coating of cells. This may play a substantial role in the in vivo modulation of inﬂammation by perﬂuorocarbons. Yet, direct contact with perﬂuorocarbon is not required for modulation of epithelial cell behavior. Diffusion of perﬂubron to nearby endothelial monolayers in vitro reduced neutrophil binding in response to LPS because of reduction in surface levels of E-selectin and intercellular adhesion molecules (ICAM) (26). So these are not entirely barrier effects. Bowell ischemia followed by reperfusion causes oxidant lung injury, which may be mediated by activated neutrophils. Instillation of oxygenated perﬂuorocarbon into the lumen of the bowel during ischemia and reperfusion reduced portal vein concentrations of reactive oxygen species and oxidative lung injury (27). This could involve a barrier effect within the bowel, but peritoneal lavage with oxygenated perﬂuorocarbon during intestinal ischemia-reperfusion injury was also shown to protect the intestinal mucosa and to ameliorate secondary lung injury (28). Other examples of ‘‘remote’’ suppression of inﬂammation by perﬂuorocarbons have been documented. For example, intraperitoneal administration of FC-77 has been shown to reduce neutrophil inﬁltration of the lung after acid aspiration (29). These ‘‘remote’’ effects must be mediated by steps that entail dissolution of sparingly soluble perﬂuorocarbons in water.

660

Fuhrman

Lipid solubility of perﬂuorocarbons is quite variable, and, generally, far exceeds aqueous solubility. The in vitro cellular effects of perﬂuorochemicals correlate with lipid solubility (30). Almost all of the red cell content of perﬂuorocarbon, for instance, is associated with the cell membrane. It seems likely that perﬂuorocarbons accumulate in cell membranes by diffusion and produce many of their remote effects by altering cell membrane structure and function. C. Perﬂuorocarbon Protection from Free-Radical Attack

One outcome of inﬂammation is oxidative damage. Numerous studies have shown that perﬂuorochemicals reduce oxidative injury. This might occur by two distinct mechanisms. Perﬂuorocarbons might reduce inﬂammation and thereby block oxidative injury, which is its natural consequence. Alternatively, perﬂuorocarbons may directly protect lipids and cell membranes from free-radical attack, or both may be the case. Clearly perﬂuorocarbons modulate inﬂammation, thereby reducing oxidative injury. Rotta et al. have shown that neat perﬂuorocarbon protects both cells and nonbiologic lipids in a cell-free, aqueous medium from freeradical attack, a mechanism completely independent of anti-inﬂammatory effects (31). Perﬂuorocarbons do not appear to be free-radical scavengers. Rather, they appear to protect cell membrane and lipid micelles from freeradical oxidation. This protection can be conferred at a distance. That is, the perﬂuorocarbon need not contact the cells or lipid to be protective. Diffusion of trace quantities of perﬂuorocarbon from a reservoir of neat perﬂuorocarbon to an aqueous phase will sufﬁce. There is a time delay in acquisition of this protection, which probably can be attributed to gradual diffusion to and accumulation in target lipids.

IV. In Vitro Effects of Neat Perﬂuorocarbon Liquids Involving Surface Tension Adherent rat alveolar type II pneumocytes incubated with perﬂuorocarbon incorporated it into lamellar bodies within 10 minutes. Rimar 101 and FC-5080 both appear to stimulate secretion of surfactant, though they also decrease surfactant synthesis (32). The net effect on pneumocytes may inﬂuence lung function and susceptibility to ventilator-induced lung injury. Meconium aspiration syndrome is a common cause of respiratory failure in neonates. Intrauterine aspiration of meconium causes airway obstruction. Meconium also inactivates surfactant, thereby altering lung mechanics, creating alveolar heterogeneity and predisposing the lungs to ventilator-induced lung injury. Meconium–saline suspensions do not, however, alter the surface tension properties of PFOB in vitro (33).

Perﬂuorocarbons and Acute Lung Injury

661

V. Effects of Ventilation with Perﬂuorocarbons on Lung Injury Numerous perﬂuorocarbons have been applied to the management of lung disease by many different techniques in various animal models. This complicates any effort to draw simple conclusions about the effects of perﬂuorocarbons on lung injury. A. Effects of Different Perﬂuorocarbons on Lung Injury

In a study of aerosolized perﬂuorocarbons, FC-77, PFOB, and FC-43 were compared using a piglet model of saline lavage–induced surfactant depletion/dysfunction. Some differences between perﬂuorocarbons were noted, but all proved suitable for aerosol delivery, improved gas exchange and lung compliance, and reduced inﬂammation (34). Comparison of ﬁndings between animal studies using different perﬂuorocarbons is inevitably confusing. Especially confusing is the issue of drug dosing. Vapor pressure dramatically inﬂuences evaporation and elimination rates. It is important to ascertain for each study the appropriateness of the dosing scheme. The ‘‘dry’’ lung promptly loses some of the characteristics of the ‘‘wet’’ lung. B. Effects of Different Means of Perﬂuorocarbon Ventilation on Lung Injury in Various Animal Models

Perﬂuorocarbons have been administered during spontaneous breathing by tidal instillation and removal of oxygenated perﬂuorocarbon liquid (TLV), by gas ventilation of the lung after intratracheal ﬁlling with perﬂuorocarbon [PAGE (35) or PLV], by aerosol instillation during gas ventilation, and by vaporization of neat liquid during gas ventilation. The precise means of perﬂuorocarbon ventilation will inevitably have a profound inﬂuence on lung injury. There may be common threads, biologic effects unrelated to the means of ventilation, but there will also inevitably be mechanical effects, just as there are during conventional ventilation. For example, early reports of PAGE used large tidal volumes, which must have inﬂuenced outcomes, histology, and effects on inﬂammation. Reports that compare one means of perﬂuorocarbon ventilation to another are exquisitely sensitive to optimization of the treatments being compared. Such studies inevitably compare not only the means employed but also the skill of the investigators in the application of those means. Anti-inﬂammatory effects of perﬂuorocarbons have been observed in many models of lung injury, in a wide variety of species, and over a broad range of animal weight. The injuries have included oleic acid infusion, surfactant dysfunction/depletion, surfactant deﬁciency, meconium aspiration, gastric acid aspiration, hydrochloric acid aspiration, endotoxin infusion, cardiopulmonary bypass, lung reperfusion injury, and lung overdistension

662

Fuhrman

injury. The species have included lambs, piglets, dogs, cats, rats, and rabbits. Animal size has ranged from less than 1 kg to over 60 kg. Tidal Liquid Ventilation

In a study of sheep injured with oleic acid followed by saline lung lavage, the combination of TLV followed by PLV (both at high tidal volume, 15 mL/kg) improved gas exchange and reduced alveolar hemorrhage, pulmonary edema formation, and inﬂammation compared to high tidal volume conventional ventilation (36). In a piglet surfactant depletion model, TLV reduced peroxide generation and oxidative injury to proteins (37). Partial Liquid Ventilation

Gastric acid aspiration causes a dramatic inﬂammatory response, especially if particulate matter is present. Histologic evidence of inﬂammation in a piglet model was eliminated by PLV (3). Acid aspiration also causes pulmonary inﬂammation. In a rat acid aspiration model, PLV decreased serum (but not tracheal ﬂuid) levels of TNF-a (38). It also suppressed the release of lipid mediators such as leukotriene B4, thromboxane A2, and 6-keto-prostaglandin F1-alpha into the blood (as well as the inﬂammatory rise seen in tissue homogenates of injured controls) (39). These ﬁndings suggest that PLV may modulate systemic sequelae of lung injury and inﬂammation as well as reduce inﬂammation in the lung. Endotoxin infusion causes lung injury in rats. Treatment with PLV using conventional gas ventilation decreased pulmonary neutrophil accumulation, myeloperoxidase activity, alveolar edema, and cell necrosis (40). A recruitment strategy, high-frequency oscillatory ventilation (HFOV), showed results comparable to those of PLV. In rabbits, endotoxin also caused oxidative damage to proteins and lipids. This damage was attenuated by PLV with perﬂubron (41). Premature, surfactant-treated lambs were promptly treated using all combinations of PLV, HFOV, and inhaled nitric oxide (NO). Control gas-ventilated, surfactant-treated lambs showed lung neutrophil accumulation and progressive deterioration in gas exchange. All treatments that included PLV enhanced gas exchange and reduced neutrophil accumulation (42). So did HFOV alone, which has no pharmacologic mechanism. In a similar study of delayed (rescue) application of ‘‘recruitment’’ treatments (HFOV, PLV with conventional gas ventilation, PLV with HFOV), all three means of ventilation improved gas exchange, but none improved histology or reduced leukocyte accumulation (43). It appears that once the inﬂammatory cascade is fully in motion, inﬂammation is more difﬁcult to modify. In a longer (24-hour) trial comparing early institution of PLV, HFOV, and surfactant, the best protection from inﬂammation

Perﬂuorocarbons and Acute Lung Injury

663

was conferred by PLV, which reduced leukocyte inﬁltration and levels of leukotriene B4 and IL-6 (44). Saline-lavaged surfactant-depleted rabbits treated by PLV had less alveolar hemorrhage, less pulmonary edema, fewer hyaline membranes, less neutrophil sequestration in pulmonary capillaries, and less migration of neutrophils into air spaces. Type II pneumocyte architecture was better preserved after PLV than in controls. Electron microscopy showed less alveolar wall damage and less disruption of type II cells in the PLV group (45). When HFOV was combined with PLV in surfactant-depleted swine, lung injury scores and myeloperoxidase activity were reduced, but evaporation of the perﬂuorocarbon correlated with a deterioration in oxygenation, emphasizing the importance of maintaining the perﬂuorocarbon dose by appropriately replacing evaporative losses (46). Reperfusion injury, produced by clamping the hilum of the left lung of rabbits for 90 minutes followed by reperfusion, was treated with PLV with PFOB after reperfusion. Oxygenation was improved, there was marked reduction in the quantity of alveolar hemorrhage and edema accumulation, and there was diminished inﬂammatory inﬁltration (47). Oleic acid injury to dog lungs was treated with PLV using perﬂuorodecalin. There was a signiﬁcant reduction in the bronchoalveolar lavage neutrophil count and improvement in lung histology with less inﬂammation (48). In an infant piglet oleic acid injury model, PLV reduced both lipid and protein oxidation (49). Rabbits with meconium aspiration were treated with PLV during spontaneous breathing with proportional-assist ventilation and compared to gas-ventilated controls. There was less inﬂammation, less atelectasis, and less hemorrhage in PLV animals (50). Respiratory syncytial virus (RSV) is the predominant cause of bronchiolitis in infants. In this infection, the intensity of cellular inﬁltration is determined by concentrations of inﬂammatory chemokines, which are induced in lung tissue by the virus. Elaboration of these cytokines is transcriptionally regulated by nuclear factor-kappa B. Mice infected with RSV were treated with intranasal PFOB six hours later and were allowed to breathe spontaneously. This treatment reduced inﬂammation, though it did not alter virus replication. PLV with PFOB reduced the expression of chemokines in this model by the inhibition of activation of nuclear factor-kappa B (51). Piglets exposed to low ﬂow, hypothermic cardiopulmonary bypass for 90 minutes were treated with PFOB instillation before or during bypass, then supported by PLV. Lung neutrophil inﬁltration was reduced (52). Perﬂuorocarbon Aerosol Therapy

Aerosolized perﬂuorocarbon (FC-77) administered continuously (10 mL/ kg/hr) to surfactant-depleted piglets reduced inﬂammation as effectively

664

Fuhrman

as PLV using full FRC dosing (30 mL/kg bolus instillation), and more effectively than did low-dose PLV (10 mL/kg per hour bolus instillation) (53). In a similar protocol in which lung neutrophil accumulation and mRNA expression of E-selectin, P-selectin, and ICAM-1 were measured, aerosolized FC-77 was again equally as effective as FRC-dosed PLV for reducing inﬂammation (54). Also in this model, laser-assisted microdissection of different lung cell types allowed the determination of mRNA expression of IL-8 and ICAM-1 in individual cell lines. Alveolar macrophages and bronchiolar epithelial cells were found to be most active in the inﬂammatory process, although alveolar septum cells, bronchiolar smooth muscle cells, and vascular smooth muscle cells were also involved. Perﬂuorocarbon aerosol diminished the expression of these markers of inﬂammation in all of these cell lines (55). This means of delivering perﬂuorocarbon may consume no less liquid than PLV (properly performed to replace evaporative losses) and the volume of perﬂuorocarbon resident in the lung may approach that present during PLV, but it appears to distribute perﬂuorocarbon very evenly, may prove effective with less perﬂuorocarbon resident in the lung than during PLV, and eliminates bolus dosing. Although PLV, TLV, and aerosol delivery of perﬂuorocarbon liquids are very different mechanical techniques, all have been shown to reduce pulmonary inﬂammation in injured lungs. Although differences in the mechanics of ventilation may inﬂuence the anti-inﬂammatory efﬁcacy of perﬂuorocarbon administration, much of the beneﬁt appears to be intrinsic to the liquid. C. Effects of Ventilation with Perﬂuorocarbon Liquids on Capillary Leak

Capillary leak and impaired capillary integrity are hallmarks of acute lung injury. PLV has been shown to reduce capillary leak in some but not all models of lung injury. PLV does not by itself promote capillary leak. It attenuates leak in cobra venom–induced lung injury (56), but not in oleic acid–induced injury (57). It also attenuates leak in the isolated, bloodperfused, acid-injured rabbit lung. In that model, it immediately decreased leak even when instilled into the trachea after leak was well established (58). Moreover, the nature of the perﬂuorocarbon inﬂuenced the degree to which it reduced capillary leak (in bowel ischemia/reperfusion-induced lung injury) (59). The capillary leak produced by high tidal volume–induced ventilator-induced lung injury is diminished by PLV (60). In overinﬂation lung injury in rats, capillary leak is mitigated by PLV to varying degrees depending on the speciﬁcs of ventilation (61). Success or failure of the modulation of capillary leak by PLV may be predicted based on the shape of the pressure–volume curve and pressure at the lower inﬂection point (62). These ﬁndings strongly suggest a physical explanation for the effect of PLV on capillary leak in these models of acute lung injury, though some

Perﬂuorocarbons and Acute Lung Injury

665

degree of modulation of inﬂammatory edema by anti-inﬂammatory or antioxidant properties may also be involved. D. Perﬂuorocarbons and Bacterial Growth in the Lung

Bacterial pneumonia is a potent trigger of acute respiratory distress syndrome (ARDS). Nosocomial infection is a serious complication of acute lung injury of other causes. In clinical practice, the effects of perﬂuorocarbons on lung injury will be strongly inﬂuenced by the impact of perﬂuorocarbon therapy on bacterial growth and virulence. The combination of PLV and parenteral antibiotics has been found more effective against pneumococcal pneumonia than antibiotics alone (63). Antibiotics administered into the trachea during either PLV (64,65) or TLV (66) achieve greater lung parenchymal concentration for a given blood concentration. In vitro Escherichia coli adhesion to biologic surfaces was impaired by perﬂubron and by FC-77, but not by Rimar (67). The same study reported that perﬂubron is not bactericidal to Pasteurella multocida, but that PLV does impede proliferation by 90%. In a neonatal rabbit model of group B streptococcal pneumonia, recovery of bacteria ﬁve hours after inoculation was reduced 10-fold. Fewer bacteria were recovered from PLV animals than the dose inoculated, whereas bacteria proliferated in conventionally ventilated animals (68). On the other hand, rats inoculated with Pseudomonas aeruginosa and ventilated for four hours before instituting low-dose PLV with PFOB showed less phagocytosis by neutrophils in alveoli than did inoculated controls (69), suggesting that PLV may actually impair clearance of pseudomonas. VI. Mechanical Protection from Lung Injury by Perﬂuorocarbon Ventilation A thorough discussion of the various forms that ventilation with perﬂuorocarbon can take exceeds the scope of this chapter. It is obvious, however, that how the lung is ventilated inﬂuences the severity of ventilator-induced lung injury (70). There is no reason to suppose this would be different when the lungs are ﬁlled with perﬂuorocarbon than when the lungs are not. What follows is a brief overview of variations on the theme of liquid ventilation, intended merely as a glimpse of the mechanical differences among the methods. A. Tidal Liquid Ventilation

TLV entails movement of tidal volumes of liquid into and out of the lungs. The liquid must be processed to add oxygen and remove carbon dioxide. The liquid ventilator, in its simplest form, comprises a gas exchanger and pumping apparatus.

666

Fuhrman

TLV need not entail wide swings in airway pressure from expiration to inspiration, but enough proximal airway pressure must be applied in inspiration to cope with airway resistance to perﬂuorocarbon ﬂow. Mean airway and mean alveolar pressures are very similar during TLV (71) and vary with depth below the top of the lung (72). TLV has little effect on cardiac output in properly hydrated animals (73), though it does entail greater changes in left ventricular stroke volume over the respiratory cycle than does PLV (74). TLV has been used to support premature lambs (75–77), premature and term monkeys (78), animals with surfactant depletion combined with oleic acid injury (79), and piglets after meconium aspiration (80), to name but a few models. Although most studies have been brief, TLV has been used for 24 hours in full-term newborn lambs (81) and 72 hours in premature lambs (82). In general, gas exchange data, mechanics of ventilation, and results of histology have been impressive. The process of TLV faces challenges in dealing with the large lung, for which high ﬂow rates are essential. Expiratory ﬂow limitation also poses some challenges. Airway collapse during expiration may cause liquid volume trapping and hypercarbia (83) if not properly managed (84). B. Partial Liquid Ventilation

PLV is gas ventilation of the perﬂuorocarbon-ﬁlled lung. A volume of perﬂuorocarbon less than or equal to the normal FRC is instilled into the lungs and gas ventilation is accomplished, which ‘‘bubble oxygenates’’ the liquid in situ (in vivo), with each breath, using a conventional gas ventilator (35,85). This technique allows liquid to be used to recruit atelectatic lung and reduce surface tension at the alveolar lining. In expiration, the liquid FRC represents an incompressible reservoir of oxygen, occupying alveoli that would otherwise collapse and permit intrapulmonary shunting. In inspiration, tidal volumes of gas purge that reservoir of carbon dioxide to be exhaled in expiration. PLV has been described in many forms: with conventional gas ventilation, combined with HFOV (42,86–89), combined with tracheal gas insufﬂation (90), using various levels of positive end-expiratory pressure (91), using various doses of perﬂuorocarbon (92), in the prone position (93), in combination with NO, and before or after surfactant therapy (94,95). It has also been reported using various perﬂuorocarbons (96,97). It has been applied to numerous models of lung injury, including premature lambs with infant respiratory distress syndrome (98), premature lambs pretreated with surfactant (99–101), small animal surfactant dysfunction (saline lavage) (102), large animal surfactant dysfunction in pigs (103–105), intravenous oleic acid injury in small animals (106,107) and in large

Perﬂuorocarbons and Acute Lung Injury

667

ones (108), hydrochloric acid aspiration in sheep (109), meconium aspiration (110,111), lung hypoplasia secondary to congenital diaphragmatic hernia (112–115), smoke inhalation (116), and burn-related ARDS (117). Long-term studies have been performed in neonatal piglets (24 hours old) (118) and in near-term newborn baboons (4–5 days old) (119). There are several uncontrolled reports (120) and series of patients treated with PLV (121–128). Size makes a big difference in PLV. The mechanical effects of the lung heavy with perﬂuorocarbon often seem to offset the other beneﬁcial effects. In general, gas exchange improves during PLV using a wide variety of techniques, but lung mechanics improve little in most large lung models. Large tidal volumes tend to improve lung function during PLV, but may be detrimental in the long term. It has taken many decades to identify tidal volume as a key determinant of lung injury. It is not surprising that we have not yet deﬁned the best ventilator strategy by which to apply perﬂuorocarbon technology to the injured lungs. It appears to be critical to keep the lung ‘‘wet’’ with perﬂuorocarbon during PLV. Mechanical beneﬁts clearly evaporate with the liquid (129). The extent to which the beneﬁts of PLV are mechanical, as opposed to biologically lung protective, remains uncertain. C. Perﬂuorocarbon Aerosol Therapy

Perﬂuorocarbon liquids are readily nebulized to ﬁne particulate aerosols. These deposit in the lung and may confer many of the same beneﬁts as PLV, if delivered in sufﬁcient quantity. The liquid distributes very evenly by this technique. Efﬁcacy of this means of administration appears to rival that of PLV (130). This technique has been successfully applied to the large lung (131). While it may merge into PLV, differing largely in the means by which the liquid is distributed in the lungs, it poses new opportunities. D. Perﬂuorocarbon Vapor Therapy

Perﬂuorocarbons produce many of their in vitro effects by diffusion through water to the targets they affect. The concentrations of some perﬂuorocarbons dissolved in water are lower than the concentrations of their vapor in air. That is, there are more molecules of PFOB per mL of saturated air than there are per mL of water that has been equilibrated to the neat perﬂuorocarbon liquid. It is not unreasonable to anticipate the efﬁcacy of vaporized perﬂuorocarbons, just as inhalation anesthetics have biologic efﬁcacy. Perﬂuorohexane is a high–vapor pressure perﬂuorocarbon (364 Torr). It can, however, be delivered at a partial pressure low enough to maintain acceptable oxygen fraction in the inspired air. Perﬂuorohexane vapor has beneﬁcial effects on gas exchange and lung mechanics in oleic acid–injured sheep (132–134). Its effects on lung inﬂammation have not yet been described.

668

Fuhrman VII. Conclusions

Perﬂuorocarbon ventilation techniques are recruitment strategies. To the extent that they reduce pulmonary surface tension, present an open lung for ventilation, remove injurious debris, and improve both gas exchange and lung compliance, they represent lung-protective strategies. In addition, perﬂuorochemical liquids can be shown to modulate inﬂammatory processes ex vivo. These intrinsic properties of perﬂuorocarbons may act to downregulate inﬂammation and oxidative injury in vivo during applications of perﬂuorocarbon liquids to the injured lung. References 1. Clark LC Jr, Gollan F. Survival of mammals breathing organic liquids equilibrated with oxygen at atmospheric pressure. Science 1966; 152:1755–1756. 2. Shaffer TH. Gaseous exchange and acid base balance in premature lambs during liquid ventilation since birth. Pediatr Res 1976; 10:227. 3. Nesti FD, Fuhrman BP, Steinhorn DM, et al. Perﬂuorocarbon associated gas exchange in gastric aspiration. Crit Care Med 1994; 22(9):1445–1452. 4. Flaim SF. Pharmacokinetics and side effects of perﬂuorocarbon-based blood substitutes. Artif Cells Blood Substit Immobil Biotechnol 1994; 22(4): 1043–1054. 5. Smith DJ, Lane TA. Effect of high concentration perﬂubron emulsion on platelet function. Biomater Artif Cells Immobil Biotechnol 1993; 21(2):173–181. 6. Leakakos T, Schutt EG, Cavin JC, et al. Pulmonary gas trapping differences among animal species in response to intravenous infusion of perﬂuorocarbon emulsions. Artif Cells Blood Substit Immobil Biotechnol 1994; 22(4): 1199–1204. 7. Schutt E, Barber P, Fields T, et al. Artif Cells Blood Substit Immobil Biotechnol 1994; 22(4):1205–1214. 8. Eckmann DM, Swartz MA, Glucksberg MR, et al. Artif Cells Blood Substit Immobil Biotechnol 1998; 26(3):259–271. 9. Mizoi K, Yoshimoto T, Suzuki J. Combined use of mannitol and perﬂuorochemicals in experimental cerebral ischemia. Brain Nerve 1983; 35(7): 669–676. 10. Fujimoto S, Mizoi K, Oba M, et al. Neurol Surg 1984; 12(2):171–180. 11. Peerless SJ, Nakamura R, Rodriguez-Salazar A, et al. Modiﬁcation of cerebral ischemia with Fluosol. Stroke 1985; 16(1):38–43. 12. Kline RA, Negendank W, McCoy L, et al. Beneﬁcial effects of isolvolemic hemodilution using a perﬂuorocarbon emulsion in a stroke model. Am J Surg 1991; 162(2):103–106. 13. Kolodgie FD, Dawson AK, Forman MB, et al. Effect of perﬂuorochemicals (Fluosol-DA) on infarct morphology in dogs. Virchows Archiv B Cell Pathol 1985; 50(2):119–134. 14. Forman MB, Virmani R, Puett DW. Mechanisms and therapy of myocardial reperfusion injury. Circulation 1990; 81(suppl 3):IV69–IV78.

Perﬂuorocarbons and Acute Lung Injury

669

15. Bando K, Teramoto S, Tago M, et al. Oxygenated perﬂuorocarbon, recombinant human superoxide dismutase and catalase ameliorate free radical induced myocardial injury during heart preservation and transplantation. J Thorac Cardiovasc Surg 1988; 96(6):930–938. 16. McDonagh P, Cernney K, Hokama J. Perﬂubron emulsion reduces inﬂammation during extracorporeal circulation. J Surg Res 2001; 99(1):7–16. 17. Bekyarova G, Yankova T, Kozarev I. Suppresive effect of FC-43 perﬂuorocarbon emulsion on enhanced oxidative haemolysis in the early postburn phase. Burns 1997; 23(2):117–121. 18. Lane T, Smith D, Wancewicz E, et al. Inhibition of endotoxin-mediated activation of endothelial cells by a perﬂuorocarbon emulsion. Biomater Artif Cells Immobil Biotechnol 1993; 21(2):163–172. 19. Noveck RJ, Shannon EJ, Leese PT. Randomized safety studies of intravenous perﬂubron emulsion. II. Effects on immune function in healthy volunteers. Anesth Analg 2000; 91(4):812–822. 20. Smith TM, Steinhorn DM, Thusu K, et al. A liquid perﬂuorochemicals decreases the in vitro production of reactive oxygen species by alveolar macrophages. Crit Care Med 1995; 23(9):1533–1539. 21. Koch T, Ragaller M, Haufe D, et al. Perﬂuorohexane attenuates proinﬂammatory and procoagulatory response of activated monocytes and alveolar macrophages. Anesthesiology 2001; 94(1):101–109. 22. Rossman JE, Caty MG, Rich GA, et al. Neutrophil activation and chemotaxis after in vitro treatment with perﬂuorocarbon. J Pediatr Surg 1996; 31(8):1147–1150. 23. Varani J, Hirschl RB, Dame M, et al. Perﬂuorocarbon protects lung epithelial cells from neutrophil-mediated injury in an in vitro model of liquid ventilation therapy. Shock 1996; 6(5):339–344. 24. Nakstad B, Wolfson MR, Shaffer TH, et al. Perﬂuorochemical liquids modulate cell-mediated inﬂammatory responses. Crit Care Med 2001; 29(9):1731–1737. 25. Baba A, Kim YK, Zhang H, et al. Perﬂuorocarbon blocks tumor necrosis factor-alpha-induced interleukin-8 release from alveolar epithelial cells in vitro. Crit Care Med 2000; 28(4):1113–1138. 26. Woods CM, Neslund G, Kornbrust E, et al. Perﬂubron attenuates neutrophil adhesion to activated endothelial cells in vitro. Am J Physiol Lung Cell Mol Physiol 2000; 278(5):L1008–L1017. 27. Rossman JE, Caty MG, Zheng S, et al. Mucosal protection from intestinal ischemia-reperfusion oxidant injury to the lung. J Surg Res 1997; 73(1):41–46. 28. Ohara M, Unno N, Mitsuoka H, et al. Peritoneal lavage with oxygenated perﬂuorochemicals preserves intestinal mucosal barrier function after ischemiareperfusion and ameliorates lung injury. Crit Care Med 2001; 29(4):782–788. 29. Nader ND, Knight PR, Davidson, et al. Systemic perﬂuorocarbons suppress the acute lung inﬂammation after gastric acid aspiration in rats. Anesth Analg 2000; 90(2):356–361. 30. Obraztsov VV, Neslund GG, Kornbrust ES, et al. In vitro cellular effects of perﬂuorochemicals correlate with their lipid solubility. Am J Physiol Lung Cell Mol Physiol 2000; 278:L1018–L1024.

670

Fuhrman

31. Rotta AT, Gunnarsson B, Fuhrman BP, et al. Perﬂuoroctyl bromide (perﬂubron) attenuates oxidative injury to biological and nonbiological systems. Pediatr Crit Care Med 2003; 4(2):233–238. 32. Rudiger M, Wissel H, Ochs M, et al. Perﬂuorocarbons are taken up by isolated type II pneumocytes and inﬂuence its lipid synthesis and secretion. Crit Care Med 2003; 34(4):1190–1196. 33. Fuloria M, Wu Y, Brandt ML, et al. Effect of meconium on the surface properties of perﬂubron. Pediatr Crit Care Med 2004; 5(2):167–171. 34. Von der Hardt K, Kandler MA, Brenn G, et al. Comparison of aerosol therapy with different perﬂuorcarbons in surfactant-depleted animals. Crit Care Med 2004; 32(5):1200–1206. 35. Fuhrman BP, Paczan PR, DeFrancisis M. Perﬂuorocarbon-associated gas exchange. Crit Care Med 1991; 19(5):712–722. 36. Hirschl RB, Tooley R, Parent A, et al. Evaluation of gas exchange, pulmonary compliance, and lung injury during total and partial liquid ventilation in the acute respiratory distress syndrome. Crit Care Med 1996; 24(6):1001–1008. 37. Dani C, Constantino ML, Martelli E, et al. Perﬂuorocarbons attenuate oxidative lung damage. Pediatr Pulmonol 2003; 36(4):322–329. 38. Kawamae K, Pristine G, Chiumello D, et al. Partial liquid ventilation decreases tumor necrosis factor-alpha concentrations in a rat acid aspiration lung injury model. Crit Care Med 2000; 28(2):479–483. 39. Hirayama Y, Hirasawa H, Oda S, et al. Partial liquid ventilation with FC-77 suppresses the release of lipid mediators in rat acute lung injury model. Crit Care Med 2004; 32(10):2085–2089. 40. Rotta AT, Steinhorn DM. Partial liquid ventilation reduces pulmonary neutrophil accumulation in an experimental model of systemic endotoxemia and acute lung injury. Crit Care Med 1998; 26(10):1707–1715. 41. Rotta AT, Gunnarsson B, Hernan LJ, et al. Partial liquid ventilation with perﬂubron attenuates in vivo oxidative damage to proteins and lipids. Crit Care Med 2000; 28(1):202–208. 42. Kinsella JP, Parker TA, Galan H, et al. Independent and combined effects of inhaled nitric oxide, liquid perﬂuorochemicals, and high frequency oscillatory ventilation in premature lambs with respiratory distress syndrome. Am J Respir Crit Care Med 1999; 159(4 Pt 1):1220–1227. 43. Gothberg S, Parker TA, Abman S, et al. High-frequency oscillatory ventilation and partial liquid ventilation after acute lung injury in premature lambs with respiratory distress syndrome. Crit Care Med 2000; 28(7):2450–2456. 44. Merz U, Klosterhalfen LB, Hausler M, et al. Partial liquid ventilation reduces release of leukotriene B4 and interleukin-6 in bronchoalveolar lavage in surfactant-depleted newborn pigs. Pediatr Res 2002; 51(2):183–189. 45. Van Eeden SF, Klut ME, Leal MA, et al. Partial liquid ventilation with perﬂuorocarbon in acute lung injury: light and transmission electron microscopy studies. Am J Respir Cell Mol Biol 2000; 22(4):441–450. 46. Doctor A, Al-Khadra E, Tan P, et al. Extended high-frequency partial liquid ventilation in lung injury: gas exchange, injury quantiﬁcation and vapor loss. J Appl Physiol 2003; 95(3):1248–1258.

Perﬂuorocarbons and Acute Lung Injury

671

47. Momoki Y. Experimental study in partial liquid ventilation for acute respiratory failure after ischemia reperfusion pulmonary injury in a rabbit model. Jpn J Thorac Cardiovas Surg 1998; 46(1):65–70. 48. Suh GY, Chung MP, Park SJ, et al. Partial liquid ventilation with perﬂuorocarbon improves gas exchange and decreases inﬂammatory response in oleic acid-induced lung injury in beagles. J Kor Med Sci 1999; 14(6):613–622. 49. Steinhorn DM, Papo MC, Rotta AT, et al. Liquid ventilation attenuates pulmonary oxidative damage. J Crit Care 1999; 14(1):20–28. 50. Hummler HD, Thome U, Schulze A, et al. Spontaneous breathing during partial liquid ventilation in animals with meconium aspiration. Pediatr Res 2001; 49(4):572–580. 51. Haeberle HA, Nesti F, Dieterich JH, et al. Perﬂubron reduces lung inﬂammation in respiratory syncytial virus infection by inhibiting chemokine expression and nuclear factor-kappa B activation. Am J Respir Crit Care Med 2002; 165(10):1433–1438. 52. Williams EA, Welty SE, Geske RS, et al. Liquid lung ventilation reduces neutrophil sequestration in a neonatal swine model of cardiopulmonary bypass. Crit Care Med 2001; 29(4):789–795. 53. Von der Hardt K, Schoof E, Kandler MA, et al. Aerosolized perﬂuorocarbon suppresses early pulmonary inﬂammatory response in a surfactant-depleted piglet model. Pediatr Res 2002; 51(2):177–182. 54. Schoof E, von der Hardt K, Kandler MA, et al. Aerosolized perﬂuorocarbon reduces adhesion molecule gene expression and neutrophil sequestration in acute respiratory distress. Eur J Pharmacol 2002; 457(2-3):195–200. 55. Von der Hardt K, Kandler MA, Fink L, et al. Laser-assisted microdissection and real-time PCR detect anti-inﬂammatory effects of perﬂuorocarbon. Am J Physiol Lung Cell Mol Physiol 2003; 285(1):L55–L62. 56. Colton DM, Till GO, Johnson KJ. Partial liquid ventilation decreases albumin leak in the setting of acute lung injury. J Crit Care 1998; 13(3):136–139. 57. Lange NR, Kozlowski JK, Gust R, et al. Effect of partial liquid ventilation on pulmonary vascular permeability and edema after experimental acute lung injury. Am J Respir Crit Care Med 2000; 162(1):271–277. 58. Loer SA, Tarnow J. Partial liquid ventilation reduces ﬂuid ﬁltration of isolated rabbit lungs with acute hydrochloric acid-induced edema. Anesthesiology 2001; 94(6):1045–1049. 59. Shashikant MP, Badellino MM, Cooper B, et al. Physicochemical properties of perﬂuorochemicals liquids inﬂuence ventilatory requirements, pulmonary mechanics, and microvascular permeability during partial liquid ventilation following intestinal ischemia/reperfusion injury. Crit Care Med 2002; 30(10): 2300–2305. 60. Lewis DA, Colton D, Johnson K, et al. Prevention of ventilator-induced lung injury with partial liquid ventilation. J Pediatr Surg 2001; 36(9):1333–1336. 61. Ricard JD, Dreyfuss D, Laissy JP, et al. Dose-response effect of perﬂuorocarbon administration on lung microvascular permeability in rats. Am J Respir Crit Care Med 2003; 168(11):1378–1382.

672

Fuhrman

62. Dreyfuss D, Martin-Lefevre L, Saumon G. Hyperinﬂation-induced alveolar ﬂooding in rats: effect of perﬂuorocarbon instillation. Am J Respir Care Crit Care Med 1999; 159(6):1752–1757. 63. Dickson EW, Heard SO, Chu B, et al. Partial liquid ventilation with perﬂuorocarbon in the treatment of rats with lethal pneumococcal pneumonia. Anesthesiology 1998; 88(1):218–223. 64. Cullen AB, Cox CA, Hipp SJ, et al. Intra-tracheal delivery strategy of gentamicin with partial liquid ventilation. Respir Med 1999; 93(11):770–778. 65. Cox CA, Cullen AB, Wolfson, et al. Intratracheal administration of perﬂuorochemical-gentamicin suspension: a comparison to intravenous administration in normal and injured lungs. Pediatr Pulmonol 2001; 32(2):142–151. 66. Fox WW, Weis CM, Cox C. Pulmonary administration of gentamicin during liquid ventilation in a newborn lamb lung injury mode. Pediatrics 1997; 100(5):E5. 67. Sajan I, Scannapieco FA, Fuhrman BP, et al. The risk of nosocomial pneumonia is not increased during partial liquid ventilation. Crit Care Med 1999; 27(12):2741–2747. 68. Rudiger M, Some M, Jarstrand C, et al. Inﬂuence of partial liquid ventilation on bacterial growth and alveolar expansion in newborn rabbis with group B-streptococcal pneumonia. Pediatr Res 2003; 54(6):808–813. 69. Rezaiguia-Delclaux S, Yang K, Stephan F, et al. Effect of partial liquid ventilation on bacterial clearance during pseudomonas aeruginosa-induced lung injury in rats. Intensive Care Med 2003; 29(7):1151–1156. 70. Cox PN, Frndova H, Tan PS, et al. Concealed air leak associated with large tidal volumes in partial liquid ventilation. Am J Respir Crit Care Med 1997; 156(3 Pt 1):992–997. 71. Curtis SE, Fuhrman BP, Howland DF. Airway and alveolar pressures during perﬂuorocarbon breathing in infant lambs. J Appl Physiol 1990; 68(6): 2322–2328. 72. Foley CK, Dowhy MS, Fuhrman BP, et al. Vertical distribution of airway pressure during gas or liquid inﬂation of the liquid-ﬁlled lung. Am J Respir Crit Care Med 1995; 151(4):A445. 73. Curtis SE, Fuhrman BP, Howland DF, et al. Cardiac output during liquid (perﬂuorocarbon) breathing in newborn piglets. Crit Care Med 1991; 19(2): 225–230. 74. Degraeuwe PL, Vos GD, Geskens GG, et al. Effect of perﬂuorochemical liquid ventilation on cardiac output and blood pressure variability in neonatal piglets with respiratory insufﬁciency. Pediatr Pulmonol 2000; 30(2):114–124. 75. Wolfson MR, Greenspan JS, Deoras KS, et al. Comparison of gas and liquid ventilation: clinical, physiological, and histological correlates. J Appl Physiol 1992; 72(3):1024–1031. 76. Valls-i-Soler A, Wolfson MR, Kechner N, et al. Comparison of natural surfactant and brief liquid ventilation rescue treatment in very immature lambs. Clinical and physiologic correlates. Biol Neonate 1996; 69(4):275–283. 77. Degraeuwe PL, Thunnissen FB, Jansen NJ, et al. Conventional gas ventilation, liquid assisted high-frequency oscillatory ventilation and tidal liquid

Perﬂuorocarbons and Acute Lung Injury

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88. 89.

90.

91.

673

ventilation in surfactant-treated preterm lambs. Int J Artif Organs 2000; 23(11):754–764. Jackson JC, Standaert TA, Troug WE, et al. Full-tidal liquid ventilation with perﬂuorocarbon for prevention of lung injury in newborn non-human primates. Artif Cells Blood Substit Immobil Biotechnol 1994; 22(4):1121–1132. Hirschl RB, Parent A, Tooley R, et al. Lung management with perﬂuorocarbon liquid ventilation improves pulmonary function and gas exchange during extracorporeal membrane oxygenation. Artif Cells Blood Substit Immobil Biotechnol 1994; 22(4):1389–1396. Kuo CY, Hsueh C, Wang CR. Liquid ventilation for treatment of meconium aspiration syndrome in a piglet model. J Formos Med Assoc 1998; 97(6): 392–399. Stavis RL, Wolfson MR, Cox C, et al. Physiologic, biochemical, and histologic correlates associated with tidal liquid ventilation. Pediatr Res 1998; 43(1): 132–138. Cox C, Stavis RL, Wolfson MR, et al. Long-term tidal liquid ventilation in premature lambs: physiologic, biochemical and histological correlates. Biol Neonate 2003; 84(3):232–242. Schoenﬁsch WH, Kylstra JA. Maximum expiratory ﬂow and estimated carbon dioxide elimination in liquid-ventilated dogs’ lungs. J Appl Physiol 1973; 35: 117–121. Foley DS, Brah R, Bull JL, et al. Total liquid ventilation: dynamic airway pressure and the development of expiratory ﬂow limitation. Am Soc Artif Intern Organs J 2004; 50(5):485–490. Tutuncu AS, Faithfull NS, Lachmann B. Comparison of ventilatory support with intratracheal perﬂuorocarbon administration and conventional mechanical ventilation in animals with acute respiratory failure. Am Rev Respir Dis 1993; 148(3):785–792. Doctor A, Mazzoni MC, DelBalzo U, et al. High-frequency oscillatory ventilation of the perﬂuorocarbon-ﬁlled lung: preliminary results in an animal model of acute lung injury. Crit Care Med 1999; 27(11):2500–2507. Doctor A, Price B, Bhargava N, et al. High-frequency oscillatory ventilation of the perﬂuorocarbon-ﬁlled lung: dose-response relationships in an animal model of acute lung injury. Crit Care Med 2001; 29(4):847–854. Chen X, Cui N, Geng R. High frequency oscillatory ventilation with perﬂuorocarbon in rabbits with acute lung injury. Chin Med J 2001; 81(2):97–101. Choi CW, Koh SY, Chang YS, et al. Effects of perﬂuorocarbon associated high frequency oscillatory ventilation on hemodynamics and gas exchange in the newborn piglet with respiratory distress. J Kor Med Sci 2003; 18(5): 663–668. Zhu G, Shaffer TH, Wolfson MR. Continuous tracheal gas insufﬂation during partial liquid ventilation in juvenile rabbits with acute lung injury. J Appl Physiol 2004; 96(4):1415–1424. Kaisers U, Kuhlen R, Keske U, et al. Superimposing positive end-expiratory pressure during partial liquid ventilation in experimental lung injury. Eur Respir J 1998; 11(5):1035–1042.

674

Fuhrman

92. Tutuncu AS, Faithfull NS, Lachmann B. Intratracheal perﬂuorocarbon administration combined with mechanical ventilation I experimental respiratory distress syndrome: dose-dependent improvement of gas exchange. Crit Care Med 1993; 21(7):962–969. 93. Max M, Kuhlen R, Lopez F, et al. Combining partial liquid ventilation and prone position in experimental acute lung injury. Anesthesiology 1999; 91(3): 796–803. 94. Mrozek JD, Bing DR, Meyers PA, et al. High-frequency oscillation versus conventional ventilation following surfactant administration and partial liquid ventilation. Pediatr Pulmonol 1998; 26(1):21–29. 95. Baumgart S, Shaffer TH. Liquid perﬂuorochemical priming before surfactant therapy: this time the egg really does precede the chicken. Crit Care Med 1999; 27(9):2053–2054. 96. Su BH, Hu PS, Lin TW, et al. Partial liquid ventilation in normal rabbits: comparison of three kinds of perﬂuorocarbon. Acta Paediatr Taiwan 2000; 41(6):313–317. 97. Al-Rahmani A, Awad K, Miller TF, et al. Effects of partial liquid ventilation with perﬂuorodecalin in the juvenile rabbit lung after saline injury. Crit Care Med 2000; 28(5):1459–1464. 98. Leach CL, Fuhrman BP, Morin FC 3rd, et al. Perﬂuorocarbon-associated gas exchange (partial liquid ventilation) in respiratory distress syndrome: a prospective, randomized, controlled study. Crit Care Med 1993; 21(9):1270–1278. 99. Leach CL, Holm B, Morin FC 3rd, et al. Partial liquid ventilation in premature lambs with respiratory distress syndrome: efﬁcacy and compatibility with exogenous surfactant. J Pediatr 1995; 126(3):412–420. 100. Davidson A, Heckman JL, Donner RM, et al. Cardiopulmonary interaction during partial liquid ventilation in surfactant-treated preterm lambs. Eur J Pediatr 1998; 157(2):138–145. 101. Wolfson MR, Kechner NE, Roache RF, et al. Perﬂuorochemical rescue after surfactant treatment: effect of perﬂubron dose and ventilatory frequency. J Appl Physiol 1998; 84(2):624–640. 102. Tutuncu AS, Akpir K, Mulder P, et al. Intratracheal perﬂuorocarbon administration as an aid in the ventilatory management of respiratory distress syndrome. Anesthesiology 1993; 79(5):1083–1093. 103. Houmes RJ, Verbrugge SJ, Hendrik ER, et al. Hemodynamic effects of partial liquid ventilatin with perﬂuorocarbon in acute lung injury. Intensive Care Med 1995; 21(12):966–972. 104. Kaisers U, Max M, Walter J, et al. Partial liquid ventilation with small volumes of FC 3280 increases survival time in experimental ARDS. Eur Respir J 1997; 10(9):1955–1961. 105. Max M, Kuhlen R, Falter F, et al. Effect of PEEP and inhaled nitric oxide on pulmonary gas exchange during gaseous and partial liquid ventilation with small volumes of perﬂuorocarbon. Acta Anaesthesiol Scand 2000; 44(4): 383–390. 106. Papo MC, Paczan PR, Fuhrman BP, et al. Perﬂuorocarbon-associated gas exchange improves oxygenation, lung mechanics, and survival in a model of adult respiratory distress syndrome. Crit Care Med 1996; 24(3):466–474.

Perﬂuorocarbons and Acute Lung Injury

675

107. Zobel G, Rodl S, Urlesberger B, et al. The effect of positive end-expiratory pressure during partial liquid ventilation in acute lung injury in piglets. Crit Care Med 1999; 27(9):1934–1939. 108. Overbeck MC, Pranikoff T, Yadao CM, et al. Efﬁcacy of perﬂuorocarbon partial liquid ventilation in a large animal model of acute respiratory failure. Crit Care Med 1996; 24(7):1208–1214. 109. Hernan LJ, Fuhrman BP, Kaiser RE, et al. Perﬂuorocarbon-associated gas exchange in normal and acid-injured sheep. Crit Care Med 1996; 24(3): 475–481. 110. Marraro G, Bonati M, Ferrari A, et al. Perﬂuorocarbon broncho-alveolar lavage and liquid ventilation versus saline broncho-alveolar lavage in adult guinea pig experimental model of meconium inhalation. Intensive Care Med 1998; 24(5):501–508. 111. Barrington KJ, Singh AJ, Etches PC, et al. Partial liquid ventilation with and without inhaled nitric oxide in a newborn piglet model of meconium aspiration. Am J Respir Crit Care Med 1999; 160(6):1922–1977. 112. Wilcox DT, Glick PL, Karamanoukian HL, et al. Perﬂuorocarbon-associated gas exchange improves pulmonary mechanics, oxygenation, ventilation and allows nitric oxide delivery in the hypoplastic lung congenital diaphragmatic hernia lamb model. Crit Care Med 1995; 23(11):1858–1863. 113. Major D, Cadenas M, Cloutier R, et al. Combined gas ventilation and perﬂuorochemical tracheal instillation as an alternative treatment for lethal congenital diaphragmatic hernia in lambs. J Pediatr Surg 1995; 30(8): 1178–1182. 114. Wilcox DT, Glick PL, Karamanoukian HL, et al. Partial liquid ventilation and nitric oxide in congenital diaphragmatic hernia. J Pediatr Surg 1997; 32(8):1211–1215. 115. Major D, Cadenas M, Cloutier R, et al. Morphometrics of normal and hypoplastic lungs in preterm lambs with gas and partial liquid ventilation. Pediatr Surg Int 1997; 12(2/3):121–125. 116. Fitzpatrick JC, Jordan BS, Salman N, et al. The use of perﬂuorocarbonassociated gas exchange to improve ventilation and decrease mortality after inhalation injury in a neonatal swine model. J Pediatr Surg 1997; 32(2): 192–196. 117. Lozano JA, Castro JA, Rodrigo I. Partial liquid ventilation with perﬂuorocarbons for treatment of ARDS in burns. Burns 2001; 27(6):635–642. 118. Salman NH, Fuhrman BP, Steinhorn, et al. Prolonged studies of perﬂuorocarbon associated gas exchange and of the resumption of conventional mechanical ventilation. Crit Care Med 1995; 23(5):919–924. 119. Sekins KM, Coalson JJ, deLemos RA, et al. Long-term partial liquid ventilation (PLV) with perﬂubron in the near-term baboon neonate. Artif Cells Blood Substit Immobil Biotechnol 1994; 22(4):1381–1387. 120. Pranikoff T, Gauger PG, Hirschl RB. Partial liquid ventilation in a child on extracorporeal life support. ASAIO J 1996; 42(4):317–320. 121. Toro-Figueroa LO, Meliones JN, Curtis SE, et al. Perﬂubron partial liquid ventilation (PLV) in children with ARDS: a safety and efﬁcacy pilot study. Crit Care Med 1996; 24(supp 1):A150.

676

Fuhrman

122. Leach CL, Greenspan JS, Rubenstein SD, et al. Partial liquid ventilation with perﬂubron in premature infants with severe respiratory distress syndrome. The LiquiVent Study Group. N Engl J Med 1996; 335(11):761–776. 123. Gauger PG, Pranikoff T, Schreiner RJ, et al. Initial experience with partial liquid ventilation in pediatric patients with the acute respiratory distress syndrome. Crit Care Med 1996; 24(1):16–22. 124. Pranikoff T, Gauger PG, Hirschl RB. Partial liquid ventilation in newborn patients with congenital diaphragmatic hernia. J Pediatr Surg 1996; 31(5): 613–618. 125. Greenspan JS, Fox WW, Rubenstein SD, et al. Partial liquid ventilation in critically ill infants receiving extracorporeal life support. Philadelphia Liquid Ventilation Consortium. Pediatrics 1997; 99(1):E2. 126. Gentili M, Pagni R, Gentili S. Partial liquid ventilation (PLV) in the newborn and infants: initial experiences in Italy. Pediatr Med Chir 1998; 20(5): 315–316. 127. Croce MA, Fabian TC, Patton JH Jr, et al. Partial liquid ventilation decreases the inﬂammatory response in the alveolar environment of trauma patients. J Trauma-Injury Infect Crit Care 1998; 45(2):273–280. 128. Fedora M, Nekvasil R, Seda M, et al. Partial liquid ventilation in the therapy of pediatric acute respiratory distress syndrome. Bratisl Lek Listy 1999; 100(9):481–485. 129. Merz U, Klosterhalfen B, Kellinghaus M, et al. Effects of single and multiple doses of perﬂuorocarbon in comparison with continuous partial liquid ventilation on gas exchange and lung pathology in newborn surfactant-depleted pigs. Crit Care Med 2001; 29(3):645–651. 130. Kandler MA, von der Hardt K, Schoof E, et al. Persistent improvement of gas exchange and lung mechanics by aerosolized perﬂuorocarbon. Am J Respir Crit Care Med 2001; 164(1):31–35. 131. Ragaller M, Bleyl J, Tscho U, et al. Effects of inhalation of perﬂuorocarbon aerosol on oxygenation and pulmonary function compared to PGI2 inhalation in a sheep model of oleic acid-induced lung injury. Intensive Care Med 2001; 27(5):889–897. 132. Bleyl JU, Ragaller M, Tscho U, et al. Vaporized perﬂuorocarbon improves oxygenation and pulmonary function in an ovine model of acute respiratory distress syndrome. Anesthesiology 1999; 91(2):461–469. 133. Hubler M, Souders JE, Shade ED. Effects of vaporized perﬂuorocarbon on pulmonary blood ﬂow and ventilation/perfusion distribution in a model of acute respiratory distress syndrome. Anesthesiology 2001; 95(6):1414–1421. 134. Bleyl JU, Ragaller M, Tscho U, et al. Changes in pulmonary function and oxygenation during application of perﬂuorocarbon vapor in healthy and oleic acid-injured animals. Crit Care Med 2002; 30(6):1340–1347.

26 Prospects for Reduction of Ventilator-Induced Lung Injury with Surfactant

CLEMENS RUPPERT, PHILIPP MARKART, WERNER SEEGER, and € NTHER ANDREAS GU Department of Internal Medicine, University of Giessen Lung Center (UGLC), Medical Clinic II, Justus-Liebig University School of Medicine Giessen, Germany

I. Introduction—The Pulmonary Surfactant System Pulmonary surfactant (‘‘surface active agent’’) is a complex mixture of lipids and proteins, which is secreted by type II cells into the alveolar space of all mammalian lungs (1). Its main function is to lower the surface tension at the air–water interface and thereby promote lung expansion during inspiration and prevent lung collapse during end-expiration at the low transpulmonary pressures associated with normal breathing. A. Composition

Composition of the pulmonary surfactant is remarkably similar among different mammalian species; it consists of about 90% lipids and about 10% proteins (1). Apart from a minor amount of neutral lipids (10–20%), of which cholesterol is the most abundant, phospholipids (PLs) represent the predominant class of lipids. Among these, phosphatidylcholine (PC) represents the predominant PL class (80%) and contains an unusually high amount of saturated fatty acids, mainly palmitic acid (16:0) (1). About 50% to 60% of all PC molecules are dipalmitoylated PC (DPPC), thus 677

678

Ruppert et al.

representing the most abundant surfactant component. This high DPPC content is known to be an absolute requirement for the high compressibility of the PL ﬁlm during expiration and for the lowering of the surface tension to values near 0 mN/m (2). The second major surfactant PL is phosphatidylglycerol (PG), which accounts for up to 10% of all PLs (1). Because of its relatively high content of unsaturated fatty acids (40–50% oleic acid), PG alters the ﬂuidity of DPPC and displays favorable adsorption characteristics (3). Other PLs regularly found in low percentages (<2%) are phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylinositol (PI), and sphingomyelin (1). Four surfactant-speciﬁc apoproteins have been discovered so far, called surfactant protein (SP)-A, SP-B, SP-C (4), and SP-D (5). SP-B and SP-C are extremely hydrophobic, low-molecular-weight proteins, whereas SP-A and SP-D are hydrophilic, high-molecular-weight proteins belonging to the family of collectins (C-type lectins).

B. Surfactant Metabolism

Synthesis of pulmonary surfactant takes place in the endoplasmic reticulum of alveolar type II cells. PLs and proteins are then transferred via the Golgi apparatus to multivesicular bodies that fuse with the so-called ‘‘lamellar bodies’’ (LB) (Fig. 1). These LB represent intracellular storage organelles in which surfactant is organized in the form of concentrically arranged and tightly packed PL bilayers (6,7). Upon the inspiratory stretch of the alveolar cell layer, alveolar type II pneumocytes secrete the LB into the alveolar hypophase, which are then reorganized into tubular myelin (1,8) and large, multilamellar vesicles (Fig. 1). Reorganization of PLs into tubular myelin requires at least two of the surfactant proteins, namely SP-A and SP-B (8). It is characterized by a highly ordered lattice-like structure, consisting of long tubules and displaying a high surface activity. Surfactant obtained by bronchoalveolar lavage (BAL) can be separated into different subfractions. Equilibrium buoyant density gradient centrifugation separates ultraheavy, heavy, and light fractions, while differential centrifugation results in large and small aggregates (9,10). Ultraheavy and heavy fractions containing LB, tubular myelin, and large multilamellar vesicles are summarized as ‘‘large surfactant aggregates’’ (LA), whereas the light subtypes consisting of small unilamellar vesicles are referred to as ‘‘small surfactant aggregates’’ (SA). The LA fraction contains the surfactant-associated proteins SP-A, SP-B, and SP-C and displays excellent biophysical activity in vitro and in vivo (11,12) and is assumed to be the precursor of the interfacial surface ﬁlm (9,10), whereas SA display only poor surface activity and most likely represent catabolic products from the interfacial ﬁlm. Surfactant lipids are removed from the alveoli with a turnover time of approximately 5 to 10 hours by recycling into the type II cell, degradation by alveolar

Prospects for Reduction of VILI with Surfactant

679

Figure 1 Metabolic pathways of pulmonary surfactant. Surfactant is synthesized in the ER of type II cells, transported via the Golgi apparatus (Golgi) and MVB, and stored in LB. (1) exocytosis of LB into the alveolar hypophase, (2) conversion of LB into TM, (3) formation of a surface ﬁlm at the air–water interface, (4) reﬁning of the surface ﬁlm with formation of small UV, (5) reuptake of surfactant material by the type II cell (6) uptake of surfactant material by AM. The involvement of SPs is indicated by (þA, B, C). Abbreviations: AM, alveolar macrophages; ER, endoplasmic reticulum; LB, lamellar bodies; MVB, multivesicular bodies; SPs, surfactant proteins; TM, tubular myelin; UV, unilamellar vesicles.

macrophages, ciliary transport up the airways, or movement through epithelium and endothelium into the blood or lymph (Fig. 1). The transition (conversion) of large aggregates to small aggregates can be mimicked in vitro by surface area cycling (10,11,13,14) and is dependent on the magnitude of surface area changes. On the basis of inhibitor studies, the requirement of an enzymatic activity for the conversion process was proposed (15). A diisopropylﬂuorophosphate-binding protein, later named ‘‘convertase,’’ was isolated from BAL ﬂuid (BALF), puriﬁed, and characterized as a member of the carboxylesterase family (16,17). The physiologic substrate of the esterase, however, is not fully settled. Because of the broad substrate speciﬁcity of

680

Ruppert et al.

carboxylesterases, lipids and proteins could both be the target. The assumption that DPPC, the most abundant and biophysically most important PL, is the substrate was recently disproved (18). Other data suggest that SP-B may be a candidate for the convertase attack, thereby promoting LA conversion (18,19). At present, there is no information available as to the regulation of the convertase under physiologic or inﬂammatory conditions. The ﬁnding that this enzyme is constitutively expressed in type II cells and possibly cosecreted with surfactant (20–22) suggests a possible role. An increased convertase expression and contribution to the increased LA-to-SA conversion under inﬂammatory conditions may be anticipated. Recently, this enzyme was isolated from BALFs and characterized as a member of the carboxylesterase family (16,23) displaying 98% homology with mouse liver carboxylesterase (17). C. Surfactant Function

The main function of the surfactant is to reduce the surface tension at the air–water interface. According to the law of Laplace (Dp ¼ 2 c r1; where p ¼ pressure, c ¼ surface tension, and r ¼ radius), surfactant lowers the work of breathing during inspiration and prevents lung collapse at end-expiration. Reduction of surface tension is achieved by the formation of a stable surface ﬁlm that is highly enriched in DPPC and lipid/protein structures closely attached to it, forming a lipid reservoir (24) from which lipids and proteins can be adsorbed during inspiration. In a variety of in vitro experiments, it has been shown that the hydrophobic SPs SP-B and SP-C accelerate the adsorption process, thus reaching equilibrium surface tension values within a few seconds (25). In addition, the hydrophobic surfactant proteins greatly contribute to the stability of the interfacial surface ﬁlm by increasing the packing during dynamic compression and mechanically stabilizing the surfactant ﬁlm at near zero minimum surface tension (26). Additional biophysical functions of surfactant include the prevention of alveolar edema formation by balancing hydrostatic ﬁltration forces (27), stabilization of small airways by maintaining patency (28), improvement of the mucociliary transport (29), and displacement of particles into the aqueous phase toward the epithelium (30). Next to the surface tension–reducing properties, pulmonary surfactant participates in the alveolar host defense system. Especially, the hydrophilic SPs SP-A and SP-D exert distinct functions in the innate immune response to microbial challenge and inﬂuence immunological and inﬂammatory processes in the lung (31). In detail, the collectins SP-A and SP-D directly bind and agglutinate microorganisms; stimulate phagocytosis by alveolar macrophages, monocytes, and neutrophils (acting as an opsonin or activating ligand); inﬂuence the chemotaxis of immune-competent cells; and regulate the release of cytokines and reactive oxygen species by macrophages.

Prospects for Reduction of VILI with Surfactant

681

II. Surfactant Alterations and Replacement Treatment in ALI/ARDS A. Surfactant Alterations in ALI/ARDS

In studies analyzing BALFs from patients with acute lung injury (ALI) and acute respiratory distress syndrome (ARDS), profound alterations of the pulmonary surfactant system have repeatedly been observed, and they may well contribute to the pathophysiologic sequelae of ARDS (32–36). Complex changes in biophysical and biochemical surfactant properties have been noticed, including alterations of the PL and fatty acid proﬁle, decreased levels of SPs, reduced content of LA, inhibition of surfactant function by leaked plasma proteins, and inhibition by inﬂammatory mediators. Changes in the PL, Fatty Acid, and Apoprotein Proﬁles

Lipid changes in ARDS include the following: 1. 2.

3.

4.

A modest (if at all) reduction in the overall PL content. Signiﬁcant changes in the distribution of PL classes, with a marked decrease in PG levels and a compensatory increase in the relative amounts of the minor components (e.g., PI, PE, PS, and sphingomyelin). Interestingly, PC, the most abundant PL class, was found to be only slightly reduced throughout all studies. Signiﬁcantly decreased relative amounts of palmitic acid, the most abundant fatty acid of PC, and increased relative amounts of unsaturated fatty acids in this PL class. Even more, the relative amount of the biophysically important dipalmitoylated species of PC was greatly reduced to 50% of the controls and this has been shown to correlate with the surface activity of these surfactant isolates. A signiﬁcant increase in the relative amount of neutral lipids in the LA fraction and a moderate change in the neutral lipid proﬁle (e.g., increase in cholesterol and triglycerides).

Studies addressing surfactant protein content in samples from ARDS patients (32–34,36,37) revealed an impressive decline of SP-A, but not of SP-D, in original BALF. Concentrations of SP-B and SP-C were also markedly reduced in original BALF and particularly within the LA fraction. The reported changes in the biochemical surfactant composition (PLs, fatty acids, apoproteins, etc.) in ARDS are likely to reﬂect injury of alveolar type II cells, with consequently altered metabolism or secretion of lipids and apoproteins by these cells. Alteration of Surfactant Subtype Distribution

Under physiological conditions, LA and SA are present in a consistent proportion that is balanced through a dynamic process involving secretion of

682

Ruppert et al.

surfactant, formation of LA, conversion of LA into SA, and clearance of SA from the alveolar airspace. In normal lungs, approximately 80% to 90% of the extracellular surfactant material is recovered in the highly active LA fraction. In patients with inﬂammatory lung disease, including ARDS and severe pneumonia, this important surfactant subtype has been shown to be dramatically decreased (34,37,38). In addition, this decrease was paralleled by a loss of SP-B and surface activity within the LA fraction. Inhibition of Surfactant Function by Plasma Protein Leakage

Under inﬂammatory conditions, plasma proteins leak into the alveolar space due to the increased permeability of the lung and may substantially contribute to surfactant alterations in ARDS. Plasma proteins such as albumin, hemoglobin, and in particular ﬁbrinogen and ﬁbrin monomers possess strong surfactant-inhibitory properties. Addition of these proteins to surfactant in vitro resulted in a severe impairment of the surface tension–lowering function (39– 42). One possible mechanism for this inhibition is that these proteins compete with surfactant components and integrate into the surface ﬁlm at the air–water interface. In addition, polymerization of ﬁbrin in the presence of surfactant material results in a loss of surfactant PL from the soluble phase due to binding to or within ﬁbrin strands, paralleled by a virtually complete loss of surface activity (43). These ﬁndings obviously suggest that PLs and hydrophobic apoproteins are incorporated into the growing ﬁbrin matrix, with severe loss of biophysically important surfactant compounds in areas with alveolar ﬁbrin and hyaline membrane formation, which are commonly found in ARDS. Damage of Surfactant Compounds by Inﬂammatory Mediators

A complex network of humoral or cellular effector systems contributes to the inﬂammatory response in ARDS. Proinﬂammatory mediators such as inﬂammatory cytokine or reactive oxygen/nitrogen species may be produced locally in the alveolar compartment by activated neutrophils and macrophages, lung epithelial cells, or ﬁbroblasts. Increased proteolytic activities including elastase, collagenase, and phospholipolytic (44–46) activities have all been encountered in BALF from patients with ARDS. In accordance, proteolytic degradation of SP-A has been directly proven from the BALF of ARDS patients (47). The overall consequence of the severe alterations of the pulmonary surfactant system as described above is a pronounced impairment of the surface tension–lowering properties of surfactant, which may directly contribute to the pathophysiological events encountered in ARDS (48). B. Surfactant Replacement Therapy in ALI/ARDS

As outlined above, there is strong evidence that alterations in the pulmonary surfactant system play an important role in the pathophysiology of

Prospects for Reduction of VILI with Surfactant

683

ALI/ARDS. However, in contrast to infant respiratory distress syndrome (IRDS) in which a lack of surfactant material is the primary cause of the disease and in which surfactant replacement therapy has evolved as the gold standard (49), matters are much more complex in ARDS. In various animal models of ARDS, transbronchial surfactant application improved gas exchange and compliance and reduced the inﬂammatory response. Initial case reports and pilot studies addressing the safety and efﬁcacy of a transbronchial administration of 300 to 800 mg/kg body weight (bw) of natural surfactant preparations mostly showed improvement of gas exchange, alongside a reduction of shunt ﬂow (50–55). Respiratory compliance, however, remained largely unchanged under these conditions. In view of the limitations in natural surfactant material, other clinical studies employing PL mixtures in the absence of surfactant proteins [Exosurf1 (56)], upon reconstitution with a peptide mimicking SP-B [KL4 (57)], or upon reconstitution with recombinant SP-C [Venticute (58–60)] were performed. KL4 has been used only in the frame of smaller Phase II studies; Exosurf and Venticute have undergone extensive testing in the frame of Phase III studies. Unfortunately, the results of the Exosurf trial (725 patients) were disappointing, because no beneﬁt from an aerosol application of Exosurf was found (56). However, the nebulizer technique used did only allow a pulmonary deposition of the total amount of surfactant (only 5 mg/kg bw delivered to the lungs), which was approximately two orders of magnitude below the possibly effective dose under conditions of a high alveolar protein burden in ARDS (34), and still approximately one order of magnitude below that currently used in IRDS. In addition, because of the absence of hydrophobic surfactant proteins, Exosurf is extremely sensitive to inhibition by plasma proteins (61). In view of two randomized, open label, multicenter Phase II studies in North America and Europe/South Africa addressing the feasibility and efﬁcacy of a tracheal application of Venticute in ARDS (58,59), in which a favorable improvement in gas exchange and ventilator-free days was encountered, two large follow-up Phase III trials (prospective, randomized, double-blinded multicenter studies) using this recombinant SP-C–based surfactant were performed in parallel in North America (n ¼ 224) and Europe/South Africa (n ¼ 224) (60). In both studies, tracheal instillation of up to 200 mg/kg bw Venticute resulted in a signiﬁcant improvement of gas exchange as compared to untreated controls. Mortality and ventilatorfree days, however, remained unchanged. Interestingly, post-hoc analysis in this study revealed that in patients with severe ARDS due to a direct injury of the lung parenchyma (pneumonia, aspiration, etc.), outcome was better upon surfactant treatment as compared to controls (60). Very recently, a randomized, blinded, and placebo-controlled trial investigating the effect of endotracheal instillation of a natural calf lung surfactant (Calfactant, Infasurf1) in infants, children, and adolescents with ALI was

684

Ruppert et al.

completed (62). Patients in the verum group showed decreased mortality, more rapid improvement in oxygenation, and better response to conventional mechanical ventilation [no need for the use of high-frequency oscillatory ventilation, nitric oxide, or extracorporeal membrane oxygenation (ECMO)]. However, the primary outcome variable duration of respiratory failure, as measured by ventilator-free days, and secondary outcome measures such as length of stay in the intensive care unit or hospital, did not signiﬁcantly differ from the placebo group (62). In summary, the precise role of surfactant treatment in ARDS still remains to be settled. The current data suggest that improvement in gas exchange may well be achieved by such therapy, but that optimizing gas exchange may not necessarily translate into better outcome. Patients with direct ARDS may represent a more focused population in which a beneﬁcial inﬂuence of surfactant treatment on outcome may be achieved. There is still ongoing discussion in view of the time of treatment (e.g., prophylactic in patients at risk and early or late after onset of ARDS), the type of the surfactant preparation (e.g., natural or synthetic and protein-containing or protein-free), as well as the amount, dosing frequency, and method of delivery (e.g., instillation, aerosolization, and bronchoscopic administration). III. Role of the Pulmonary Surfactant System in VILI A. Alteration of the Pulmonary Surfactant System in Animal Models of VILI

As early as 1959, Mead and Collier addressed the consequences of mechanical ventilation per se in healthy dogs and observed a reduction in pulmonary compliance (63). Greenﬁeld et al. were able to demonstrate that this loss of compliance was related to an impairment of the pulmonary surfactant system (64). Following this, several studies revealed that some ventilation strategies, in particular those involving high tidal volumes and/or high inspiratory peak pressures, were accompanied by a deterioration of the surface tension–lowering properties of the endogenous surfactant pool (65–68) in otherwise healthy lungs. By employing radiolabeled PL precursors, it was found that repeated hyperinﬂation in open-chested cats for three hours promoted the release of surfactant material; however, the released material was inactivated (68). Comparison of different ventilation strategies [low vs. medium vs. high tidal volume, high positive end-expiratory pressure (PEEP) or no PEEP] in otherwise healthy lungs similarly forwarded a signiﬁcant increase in the total PL and LA fraction, alongside a loss of lung compliance and gas exchange properties, when the lung was ventilated for a short-term period (up to 120 minutes) (69,70) using an injurious ventilation mode (high tidal volume, no PEEP). In addition, the antigen levels of SP-A and SP-D in

Prospects for Reduction of VILI with Surfactant

685

BALF were elevated, indicating increased secretion (69). However, mRNA levels of SP-A, SP-B, and SP-C were even downregulated under these conditions, thus documenting the absence of any compensatory upregulation of protein expression. Regarding the release of surfactant material from the type II cell, it had been shown that stretch of the type II cell represents the most effective physiological trigger mechanism (71), and such signaling may initially also take place in lungs undergoing injurious ventilation. Although more surfactant lipid material was shown to be released under these experimental conditions of high tidal volume ventilation, a pronounced impairment of the surface activity was regularly noted (69,70,72). The reasons for this marked impairment in surface activity may be as complex as in ARDS and may include altered composition, synthesis, and secretion pathways; inhibition of extracellular surfactant material; and a direct inﬂuence of the tidal volume and/or PEEP on the extracellular surfactant metabolism. In view of the surfactant composition, no differences were observed with regard to the PL composition, but some slight reduction in the DPPC, the most abundant and biophysically most important PL, was encountered (72). In addition, decreased SP-B and SP-C concentrations may contribute to such ﬁndings, but have not yet been proven under experimental conditions. Surface activity may also be inhibited by a variety of inﬂammatory mediators such as proteases, phospholipases, and plasma proteins, the latter easily entering the alveolar space under conditions of an injurious ventilation mode (69,70). In line with such reasoning, it was found that alveolar edema may occur early and progress rapidly following injurious mechanical ventilation and that the extent of edema formation is mainly determined by the end-inspiratory lung volume (73). Morphometrical analysis of surfactant subtypes by employing electron microscopy conﬁrmed a loss of highly organized lipid structures such as tubular myelin and an increase in small vesicular forms during an injurious ventilation pattern in otherwise healthy lungs (72,74). If the inﬂuence of mechanical ventilation was assessed by comparison between healthy and acutely inﬂamed lungs (by either induction of sepsis by cecal ligation and puncture or N-nitroso-N-methylurethane application), surfactant changes seemed to be much more prominent. A highly increased surfactant subtype conversion was encountered in vivo, with a loss of the LA fraction, paralleled by a loss of surface activity within the LA fraction (75–77). In this regard, ventilation strategies using high tidal volumes in combination with zero PEEP resulted in higher LA-to-SA conversion rates as compared to strategies using lower tidal volumes and/or using higher PEEP. Under these conditions, the LA-to-SA conversion signiﬁcantly correlated with the tidal volume, but not with the respiratory rate (75). Most of these ventilation-induced changes of the surfactant system may be easily explained by either stretch- or ‘‘overstretch’’-induced changes in gene expression and intracellular signaling in the type II cell and by

686

Ruppert et al.

the induction of an inﬂammatory reaction in the alveolar compartment. Besides, a direct inﬂuence of a high tidal volume on the alveolar surfactant ﬁlm may exist, which is detailed in the following text. As already outlined above, 80% to 90% of the extracellular surfactant material is recovered in the LA fraction under normal conditions, and this subfraction represents the biophysically active precursor fraction, which is characterized by a high SP-B content and excellent biophysical activity. Under normal conditions, a slow conversion of this subtype into the SA occurs and this conversion may be promoted by the action of a carboxylesterase titled ‘‘convertase.’’ The overall ratio of LA to SA is high, and it may be maintained by (i ) low conversion rate, (ii) adequate secretion of new surfactant material into the alveolar hypophase, and (iii) rapid uptake of SA by alveolar macrophages and type II cells. When subjecting BALFs from healthy rabbits to extensive surface area changes (approximately ninefold, ‘‘in vitro cycling’’), as they may occur in ARDS lungs upon injurious ventilation, a dramatic increase in the LA-to-SA conversion rate, alongside a rapid loss of surface activity of the LA fraction, was observed (78). Changes in the PL, fatty acid, or neutral lipid pattern were not found to be responsible for this; instead, a rapid decline of the antigen signal for SP-B was found (78). In more recent observations, these ﬁndings could be accomplished by the proof of a decline in the levels of mature dimeric SP-B and an increased recovery of SP-B degradation products in immunoblots upon both the in vitro cycling of BALF and the coincubation of carboxylesterases and isolated SP-B (Ruppert et al., unpublished observation). Taken together, the enzymatically driven degradation of SP-B by a carboxylesterase (convertase), which seems to be possible only under great surface area changes, provides the basis for an increased LA-to-SA conversion, a loss of SP-B within the LA fraction, and a dramatic loss of surface activity of the LA fraction itself. A similar increase in LA-to-SA conversion could also be encountered in animal models of ventilator-induced lung injury (VILI) (70,75–77). Moreover, after transbronchial surfactant application in this model, the magnitude of the LA-to-SA conversion was dependent on the tidal volume: the lower the tidal volume, the lower the LA-to-SA conversion of exogenously administered surfactant (79). In contrast, the PEEP level had no inﬂuence on aggregate conversion, but improved oxygenation during the initial phase after surfactant administration (79). Thus, ventilation of a lung in an injurious ventilation mode may directly affect the homeostasis of the extracellular surfactant ﬁlm and thus represent a fourth way of surfactant inactivation in VILI (Fig. 2). B. Role of Surfactant in the Treatment of VILI

It is emphasized that under clinical conditions, it is currently impossible to determine to which extent the alterations of the surfactant system are due to

Prospects for Reduction of VILI with Surfactant

687

Figure 2 Vicious cycles of surfactant changes in VILI. Mechanical ventilation with high tidal volumes and/or high PIP results in an impaired surfactant function by the mechanisms indicated. The increase in minimum surface tension (cmin ) causes loss of compliance. As a result, even higher ventilation pressures are needed to maintain tidal volumes. Abbreviations: VILI, ventilator-induced lung injury; PIP, peak inspiratory pressure.

the underlying disease (ARDS/ALI) on one hand and the ventilation mode on the other. Therefore, all considerations made in this passage are largely speculative at the current time. On the other hand, however, there seems to be no great difference in view of the demands for an exogenous surfactant preparation being used for the treatment of ALI/ARDS and/or VILI. It must rapidly reduce alveolar surface tension even in the presence of a large number of inhibitors; it should easily integrate into the endogenous surfactant pool and maintain in the LA fraction as long as possible and thus have a long half-time and should be easily uptaken and recycled by type II cells. As outlined above, transbronchial surfactant treatment has been proven beneﬁcial in a large number of ARDS models, and surfactant application also improves gas exchange in ARDS patients (53,60). Drawn against the above-mentioned background of surfactant abnormalities in various VILI models and also in ARDS patients, improvement of alveolar surfactant function thus appears to be a reasonable approach to treat or prevent

688

Ruppert et al.

VILI. Indeed, in animal models of VILI, administration of exogenous surfactant was shown to signiﬁcantly improve oxygenation and lung function, as well as surfactant composition and function (80,81), suggesting that surfactant abnormalities were largely contributing to the deterioration of gas exchange and compliance. Even more, prophylactic treatment of lungs with surfactant material being delivered to the lungs prior to the onset of an injurious ventilation mode was shown to largely preserve lung mechanics and oxygenation under injurious ventilation (82,83), and this observation pretty much resembles the experience in some transplantation models, where surfactant administration prior to the onset of ischemia largely prevented the loss of gas exchange and compliance after transplantation and reperfusion (84–86). It was furthermore observed that the ventilation mode used in the VILI models directly affects surfactant subtype conversion of the exogenous surfactant material. Ventilation with low tidal volumes resulted in less LA-to-SA conversion of exogenously administered surfactant as compared with normal tidal volumes (79). The PEEP level had no inﬂuence on aggregate conversion but improved oxygenation during the initial phase after surfactant administration (79). In contrast to the prevention of barotrauma/ volutrauma, however, proinﬂammatory cytokine release (biotrauma) was not affected by surfactant treatment (81) or even augmented (83). This experience is somehow reﬂected in the recent results by the two ARDS Network studies addressing tidal volume and PEEP regimen in ARDS patients, in which a reduction of the tidal volume from 12 to 6 mL/kg bw alone was found to yield a reduction in mortality by 22% (87), but a further PEEP increment did not substantially inﬂuence survival (88). The current clinical experience with transbronchial surfactant replacement therapy in ARDS patients has been carefully described above. In this regard, surfactant therapy seems to improve gas exchange in ARDS patients in general, but improvement in mortality may only be achieved in patients with direct forms of lung injury such as severe pneumonia necessitating mechanical ventilation. In view of the pathophysiological considerations, the best way, of course, to treat VILI may be to avoid ventilation. In this regard, prophylactic surfactant treatment of ALI patients with surfactant prior to intubation certainly represents an interesting therapeutic approach. However, only inhalative routes of application could be used then, and there is currently no established mode to deliver suitably large amounts of a functionally surfactant preparation in an acceptable time window by a nebulizer device. Next to delivery of exogenous surfactant material into the distal lung, any other pharmacological approach that restores or preserves surfactant function may represent another therapeutic option. A variety of pharmacological agents do increase surfactant synthesis and secretion, among them b-adrenergic agonists (e.g., isoproterenol, epinephrine, and norepinephrine),

Prospects for Reduction of VILI with Surfactant

689

cholinergic agonists (e.g., pilocarpine), purinergic agonists (e.g., ATP, ADP, and adenosine), prostaglandins, leukotrienes, vasopressin, calcium ionophores (e.g., ionomycin), glucocorticoids, thyroid hormones, and estrogens. Some of them have been successfully used in the past in ARDS models. However, to date, no speciﬁc pharmacologic approach for the prevention or treatment of ARDS has been conclusively validated in clinical trials. Evidence that such pharmacological intervention may be suitable to improve or maintain surfactant function under clinical conditions of ALI/ARDS and/ or VILI is currently completely lacking. The ﬁnding that surfactant aggregate conversion is at least partially mediated via an enzymatic activity could have clinical implications. For example, blocking convertase activity during mechanical ventilation might result in a lower rate of aggregate conversion and may keep the vast majority of surfactant in its biophysically active subtype. By this, further deterioration in lung function might be avoided. Recently, highly speciﬁc synthetic low-molecular-weight inhibitors of the conversion process were identiﬁed (89), which, next to an optimized ventilation mode, may serve as therapeutic tools to avoid ventilator-associated lung injury (VALI) and to improve outcome.

IV. Conclusions There is good evidence that severe abnormalities of the pulmonary surfactant system contribute to the pathophysiologic sequelae in VALI/VILI. The surfactant alterations seen in VILI models resemble those seen in patients with ARDS. In most of these models, however, lung injury developed upon exposure to a noxious agent (e.g., lipopolysaccharide, N-nitroso-N-methylurethane, etc.) followed by an injurious ventilation. Thus, a clear differentiation between the impacts of inﬂammation and ventilation on the surfactant alterations is not possible. If ventilation itself was responsible for the development of ALI, as in animal models employing injurious ventilation modes in otherwise healthy organs, the observed changes in the surfactant system still resembled those seen in ARDS. Induction of inﬂammatory changes affecting surfactant composition, metabolism, and function may serve as a possible explanation but deserves further investigation. It is noteworthy that by inducing high surfacearea changes, high tidal volumes may alter the surfactant subtype distribution and biophysical activity in the absence of any additional inﬂammatory or cellular process. In experimental models of VILI, surfactant replacement and/or prophylactic treatment improved lung mechanics and oxygenation, suggesting that surfactant treatment may prevent barotrauma/volutrauma. Especially, the concept of a very early surfactant treatment in ALI/ARDS, with the goal of reducing the invasiveness of mechanical ventilation, clearly represents an attractive therapeutic option that needs to be investigated in future experimental and clinical studies.

690

Ruppert et al. References

1. Creuwels LA, Van Golde LM, Haagsman HP. The pulmonary surfactant system: biochemical and clinical aspects. Lung 1997; 175:1–39. 2. Veldhuizen R, Nag K, Orgeig S, Possmayer F. The role of lipids in pulmonary surfactant. Biochim Biophys Acta 1998; 1408:90–108. 3. Fleming BD, Raynor CM, Keough KMW. Some characteristics of monolayers of l-palmitoyl-2-oleyl-phosphatidylglycerol with and without dipalmitoylphosphatidylcholine during dynamic compression and expansion. Biochim Biophys Acta 1983; 732:243–250. 4. Possmayer F. A proposed nomenclature for pulmonary surfactant-associated proteins. Am Rev Respir Dis 1988; 138:990–998. 5. Persson A, Chang D, Rust K, Moxley M, Longmore W, Crouch E. Puriﬁcation and biochemical characterization of CP4 (SP-D), a collagenous surfactantassociated protein. Biochemistry 1989; 28:6361–6367. 6. Batenburg JJ. Surfactant phospholipids: synthesis and storage. Am J Physiol 1992; 262:L367–L385. 7. Williams MC. Ultrastructure of tubular myelin and lamellar bodies in fastfrozen adult rat lung. Exp Lung Res 1982; 4:37–46. 8. Williams MC, Hawgood S, Hamilton RL. Changes in lipid structure produced by surfactant proteins SP-A, SP-B, and SP-C. Am J Respir Cell Mol Biol 1991; 5:41–50. 9. Magoon MW, Wright JR, Baritussio A, et al. Subfractionation of lung surfactant. Implications for metabolism and surface activity. Biochim Biophys Acta 1983; 750:18–31. 10. Gross NJ, Narine KR. Surfactant subtypes in mice: characterization and quantitation. J Appl Physiol 1989; 66:342–349. 11. Higuchi R, Lewis J, Ikegami M. In vitro conversion of surfactant subtypes is altered in alveolar surfactant isolated from injured lungs. Am Rev Respir Dis 1992; 145:1416–1420. 12. Ikegami M, Jobe A, Yamada T, et al. Surfactant metabolism in surfactanttreated preterm ventilated lambs. J Appl Physiol 1989; 67:429–437. 13. Gross NJ, Narine KR. Surfactant subtypes of mice: metabolic relationships and conversion in vitro. J Appl Physiol 1989; 67:414–421. 14. Ueda T, Ikegami M, Jobe A. Surfactant subtypes. In vitro conversion, in vivo function, and effects of serum proteins. Am J Respir Crit Care Med 1994; 149:1254–1259. 15. Gross NJ, Schultz RM. Serine proteinase requirement for the extra cellular metabolism of pulmonary surfactant. Biochim Biophys Acta 1990; 1044: 222–230. 16. Krishnasamy S, Gross NJ, Teng AL, Schultz RM, Dhand R. Lung ‘‘surfactant convertase’’ is a member of the carboxylesterase family. Biochem Biophys Res Commun 1997; 235:180–184. 17. Krishnasamy S, Teng AL, Dhand R, Schultz RM, Gross NJ. Molecular cloning, characterization, and differential expression pattern of mouse lung surfactant convertase. Am J Physiol 1998; 275:L969–L975.

Prospects for Reduction of VILI with Surfactant

691

18. Dhand R, Young J, Teng A, Krishnasamy S, Gross NJ. Is dipalmitoylphosphatidylcholine a substrate for convertase? Am J Physiol Lung Cell Mol Physiol 2000; 278:L19–L24. 19. Veldhuizen RA, Hearn SA, Lewis JF, Possmayer F. Surface-area cycling of different surfactant preparations: SP-A and SP-B are essential for large-aggregate integrity. Biochem J 1994; 300:519–524. 20. Clark H, Allen L, Collins E, et al. Localization of a candidate surfactant convertase to type II cells, macrophages, and surfactant subtractions. Am J Physiol 1999; 20:L452–L458. 21. Oulton M, Edwards E, Handa K. Convertase activity in alveolar surfactant and lamellar bodies in fetal, newborn, and adult rabbits. J Appl Physiol 1999; 86:71–77. 22. Barr F, Clark H, Hawgood S. Identiﬁcation of a putative surfactant convertase in rat lung as a secreted serine carboxylesterase. Am J Physiol 1998; 274: L404–L410. 23. Gross NJ, Schultz RM. Requirements for extracellular metabolism of pulmonary surfactant: tentative identiﬁcation of serine protease. Am J Physiol 1992; 262:L446–L453. 24. Schurch S, Qanbar R, Bachofen H, Possmayer F. The surface-associated surfactant reservoir in the alveolar lining. Biol Neonate 1995; 67(suppl 1):61–76. 25. Oosterlaken Dijksterhuis MA, Haagsman HP, Van Golde LM, Demel RA. Characterization of lipid insertion into monomolecular layers mediated by lung surfactant proteins SP-B and SP-C. Biochemistry 1991; 30:10965–10971. 26. Qanbar R, Cheng S, Possmayer F, Schurch S. Role of the palmitoylation of surfactant-associated protein C in surfactant ﬁlm formation and stability. Am J Physiol 1996; 271:L572–L580. 27. Nieman GF, Bredenberg CE. High surface tension pulmonary edema induced by detergent aerosol. J Appl Physiol 1985; 58:129–136. 28. Enhorning G. Photography of peripheral pulmonary airway expansion as affected by surfactant. J Appl Physiol 1977; 42:976–979. 29. De Sanctis GT, Tomkiewicz RP, Rubin BK, Schurch S, King M. Exogenous surfactant enhances mucociliary clearance in the anaesthetized dog. Eur Respir J 1994; 7:1616–1621. 30. Schurch S, Gehr P, Im Hof V, Geiser M, Green F. Surfactant displaces particles toward the epithelium in airways and alveoli. Respir Physiol 1990; 80:17–32. 31. Crouch E, Wright JR. Surfactant proteins A and D and pulmonary host defense. Ann Rev Physiol 2001; 63:521–524. 32. Pison U, Seeger W, Buchhorn R, et al. Surfactant abnormalities in patients with respiratory failure after multiple trauma. Am Rev Respir Dis 1989; 140:1033–1039. 33. Gregory TJ, Longmore WJ, Moxley MA, et al. Surfactant chemical composition and biophysical activity in acute respiratory distress syndrome. J Clin Invest 1991; 88:1976–1981. 34. Gunther A, Siebert C, Schmidt R, et al. Surfactant alterations in severe pneumonia, acute respiratory distress syndrome, and cardiogenic lung edema. Am J Respir Crit Care Med 1996; 153:176–184.

692

Ruppert et al.

35. Schmidt R, Yabut-Perez U, Meier M, et al. Alteration of fatty acid proﬁles in different pulmonary surfactant phospholipids in acute respiratory distress syndrome and severe pneumonia. Am J Respir Crit Care Med 2001; 163:95–100. 36. Greene KE, Wright JR, Steinberg KP, et al. Serial changes in surfactantassociated proteins in lung and serum before and after onset of ARDS. Am J Respir Crit Care Med 1999; 160:1843–1850. 37. Gunther A, Schmidt R, Feustel A, et al. Surfactant subtype conversion is related to loss of surfactant apoprotein B and surface activity in large surfactant aggregates. Experimental and clinical studies. Am J Respir Crit Care Med 1999; 159:244–251. 38. Veldhuizen RA, McCaig LA, Akino T, Lewis JF. Pulmonary surfactant subtractions in patients with the acute respiratory distress syndrome. Am J Respir Crit Care Med 1995; 152:1867–1871. 39. Seeger W, Stohr G, Wolf HR, Neuhof H. Alteration of surfactant function due to protein leakage: special interaction with ﬁbrin monomer. J Appl Physiol 1985; 58:326–338. 40. Fuchimukai T, Fujiwara T, Takahashi A, Enhorning G. Artiﬁcial pulmonary surfactant inhibited by proteins. J Appl Physiol 1987; 62:429–437. 41. Holm BA, Notter RH. Effects of hemoglobin and cell membrane lipids on pulmonary surfactant activity. J Appl Physiol 1987; 63:1434–1442. 42. Seeger W, Gunther A, Thede C. Differential sensitivity to ﬁbrinogen inhibition of SP-C- vs. SP-B-based surfactants. Am J Physiol 1992; 262:L286–L291. 43. Seeger W, Elssner A, Gunther A, Kramer HJ, Kalinowski HO. Lung surfactant phospholipids associate with polymerizing ﬁbrin: loss of surface activity. Am J Respir Cell Mol Biol 1993; 9:213–220. 44. Christner P, Fein A, Goldberg S, Lippmann M, Abrams W, Weinbaum G. Collagenase in the lower respiratory tract of patients with adult respiratory distress syndrome. Am Rev Respir Dis 1985; 131:690–695. 45. Lee CT, Fein AM, Lippmann M, Holtzman H, Kimbel P, Weinbaum G. Elastolytic activity in pulmonary lavage ﬂuid from patients with adult respiratorydistress syndrome. N Engl J Med 1981; 304:192–196. 46. Hallman M, Spragg R, Harrell JH, Moser KM, Gluck L. Evidence of lung surfactant abnormality in respiratory failure. Study of bronchoalveolar lavage phospholipids, surface activity, phospholipase activity, and plasma myoinositol. J Clin Invest 1982; 70:673–683. 47. Baker CS, Evans TW, Randle BJ, Haslam PL. Damage to surfactant-speciﬁc protein in acute respiratory distress syndrome. Lancet 1999; 353:1232–1237. 48. Gunther A, Ruppert C, Schmidt R, et al. Surfactant alteration and replacement in acute respiratory distress syndrome. Respir Res 2001; 2:353–364. 49. Gnanaratnem J, Finer NN. Neonatal acute respiratory failure. Curr Opin Pediatr 2000; 12:227–232. 50. Lachmann B. Animal models and clinical pilot studies of surfactant replacement in adult respiratory distress syndrome. Eur Respir J Suppl 1989; 3:98S–103S. 51. Richman PS, Spragg RG, Robertson B, Merritt TA, Curstedt T. The adult respiratory distress syndrome: ﬁrst trials with surfactant replacement. Eur Respir J 1989; 3(suppl 3):109S–111S.

Prospects for Reduction of VILI with Surfactant

693

52. Spragg RG, Gilliard N, Richman P, et al. Acute effects of a single dose of porcine surfactant on patients with the adult respiratory distress syndrome. Chest 1994; 105:195–202. 53. Walmrath D, Gunther A, Ghofrani HA, et al. Bronchoscopic surfactant administration in patients with severe adult respiratory distress syndrome and sepsis. Am J Respir Crit Care Med 1996; 154:57–62. 54. Gregory TJ, Steinberg KP, Spragg R, et al. Bovine surfactant therapy for patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 1997; 155:1309–1315. 55. Willson DF, Zaritsky A, Bauman LA, et al. Instillation of calf lung surfactant extract (calfactant) is beneﬁcial in pediatric acute hypoxemic respiratory failure. Members of the Mid-Atlantic Pediatric Critical Care Network. Crit Care Med 1999; 27:188–195. 56. Anzueto A, Baughman RP, Guntupalli KK, et al. Aerosolized surfactant in adults with sepsis-induced acute respiratory distress syndrome. Exosurf Acute Respiratory Distress Syndrome Sepsis Study Group. N Engl J Med 1996; 334:1417–1421. 57. Wiswell TE, Smith RM, Katz LB, et al. Bronchopulmonary segmental lavage with Surfaxin (KL(4)-surfactant) for acute respiratory distress syndrome. Am J Respir Crit Care Med 1999; 160:1188–1195. 58. Spragg RG, Lewis JF, Wurst W, et al. Treatment of acute respiratory distress syndrome with recombinant surfactant protein C surfactant. Am J Respir Crit Care Med 2003; 167:1562–1566. 59. Walmrath D, De Vaal JB, Bruining HA, et al. Treatment of ARDS with a recombinant SP-C (rSP-C) based synthetic surfactant. Am J Respir Crit Care Med 2000; 161:A379. 60. Spragg RG, Lewis JF, Walmrath HD, et al. Effect of recombinant surfactant protein C-based surfactant on the acute respiratory distress syndrome. N Engl J Med 2004; 351:884–892. 61. Seeger W, Grube C, Gunther A, Schmidt R. Surfactant inhibition by plasma proteins: differential sensitivity of various surfactant preparations. Eur Respir J 1993; 6:971–977. 62. Willson DF, Thomas NJ, Markovitz BP, et al. Pediatric acute lung injury and sepsis investigators. Effect of exogenous surfactant (calfactant) in pediatric acute lung injury: a randomized controlled trial. JAMA 2005; 293:470–476. 63. Mead J, Collier C. Relationship of volume history of lungs to respiratory mechanics in anaesthetized dogs. J Appl Physiol 1959; 14:669–678. 64. Greenﬁeld LJ, Ebert PA, Benson DW. Effect of positive pressure ventilation on surface tension properties of lung extracts. Anesthesiology 1964; 25:312–316. 65. Faridy EE, Permutt S, Riley RL. Effect of ventilation on surface forces in excised dogs’ lungs. J Appl Physiol 1966; 21:1453–1462. 66. McClenahan JB, Urtnowski A. Effect of ventilation on surfactant, and its turnover rate. J Appl Physiol 1967; 23:215–220. 67. Forrest JB. The effect of hyperventilation on pulmonary surface activity. Br J Anaesth 1972; 44:313–320. 68. Wyszogrodski I, Kyei Aboagye K, Taeusch HW Jr, Avery ME. Surfactant inactivation by hyperventilation: conservation by end-expiratory pressure. J Appl Physiol 1975; 38:461–466.

694

Ruppert et al.

69. Veldhuizen RA, Tremblay LN, Govindarajan A, et al. Pulmonary surfactant is altered during mechanical ventilation of isolated rat lung. Crit Care Med 2000; 28:2545–2551. 70. Verbrugge SJ, Bohm SH, Gommers D, Zimmerman LJ, Lachmann B. Surfactant impairment after mechanical ventilation with large alveolar surface area changes and effects of positive end-expiratory pressure. Br J Anaesth 1998; 80:360–364. 71. Wirtz HR, Dobbs LG. Calcium mobilization and exocytosis after one mechanical stretch of lung epithelial cells. Science 1990; 250:1266–1269. 72. Veldhuizen RA, Welk B, Harbottle R, et al. Mechanical ventilation of isolated rat lungs changes the structure and biophysical properties of surfactant. J Appl Physiol 2002; 92:1169–1175. 73. Dreyfuss D, Saumon G. Role of tidal volume, FRC, and end-inspiratory volume in the development of pulmonary edema following mechanical ventilation. Am Rev Respir Dis 1993; 148:1194–1203. 74. Savov J, Silbajoris R, Young SL. Mechanical ventilation of rat lung: effect on surfactant forms. Am J Physiol 1999; 277:L320–L326. 75. Veldhuizen RA, Marcou J, Yao LJ, Mccaig L, Ito Y, Lewis JF. Alveolar surfactant aggregate conversion in ventilated normal and injured rabbits. Am J Physiol 1996; 270:L152–L158. 76. Ito Y, Veldhuizen RA, Yao LJ, McCaig LA, Bartlett AJ, Lewis JF. Ventilation strategies affect surfactant aggregate conversion in acute lung injury. Am J Respir Crit Care Med 1997; 155:493–499. 77. Malloy JL, Veldhuizen RA, Lewis JF. Effects of ventilation on the surfactant system in sepsis-induced lung injury. J Appl Physiol 2000; 88:401–408. 78. Ruppert C, Pucker C, Markart P, et al. Impact of surface tension on the conversion rate of large to small surfactant aggregates. Biophys Chem 2003; 104:229–238. 79. Ito Y, Manwell SE, Kerr CL, et al. Effects of ventilation strategies on the efﬁcacy of exogenous surfactant therapy in a rabbit model of acute lung injury. Am J Respir Crit Care Med 1998; 157:149–155. 80. Vazquez de Anda GF, Lachmann RA, Gommers D, Verbrugge SJ, Haitsma J, Lachmann B. Treatment of ventilation-induced lung injury with exogenous surfactant. Intensive Care Med 2001; 27:559–565. 81. Welk B, Malloy JL, Joseph M, Yao LJ, Veldhuizen AW. Surfactant treatment for ventilation-induced lung injury in rats: effects on lung compliance and cytokines. Exp Lung Res 2001; 27:505–520. 82. Verbrugge SJ, Vazquez de Anda G, Gommers D, et al. Exogenous surfactant preserves lung function and reduces alveolar Evans blue dye inﬂux in a rat model of ventilation-induced lung injury. Anesthesiology 1998; 89:467–474. 83. Stamme C, Brasch F, von Bethmann A, Uhlig S. Effect of surfactant on ventilation-induced mediator release in isolated perfused mouse lungs. Pulmon Pharmacol Ther 2002; 15:455–461. 84. Novick RJ, Gilpin AA, Gehman KE, et al. Mitigation of injury in canine lung grafts by exogenous surfactant therapy. J Thorac Cardiovasc Surg 1997; 113:342–353.

Prospects for Reduction of VILI with Surfactant

695

85. Novick RJ, MacDonald J, Veldhuizen RA, et al. Evaluation of surfactant treatment strategies after prolonged graft storage in lung transplantation. Am J Respir Crit Care Med 1996; 154:98–104. 86. Friedrich I, Borgermann J, Splittgerber FH, et al. Bronchoscopic surfactant administration preserves gas exchange and pulmonary compliance after single lung transplantation in dogs. J Thorac Cardiovasc Surg 2004; 127:335–343. 87. ARDS Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342:1301–1308. 88. Brower RG, Lanken PN, MacIntyre N, et al. Higher versus lower positive endexpiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351:327–336. 89. Ruppert C, Pucker C, Markart P, et al. Selective inhibition of large-to-small surfactant aggregate conversion by serine protease inhibitors of the bis-benzamidine type. Am J Respir Cell Mol Biol 2003; 28:95–102.

This page intentionally left blank

27 Rationale for High-Frequency Oscillatory Ventilation in Acute Lung Injury

JEFFREY M. SINGH, NIALL D. FERGUSON, and THOMAS E. STEWART Interdepartmental Division of Critical Care Medicine and Department of Medicine, Division of Respirology, University Health Network and Mount Sinai Hospital, University of Toronto Toronto, Ontario, Canada

I. Introduction Since its advent in the early 1950s, positive pressure mechanical ventilation has become instrumental in the support of the critically ill patient. In recent years, there has been a greater understanding of the pathogenesis of acute lung injury, and it is now clear that mechanical ventilation can potentiate or cause further lung damage. This evolving understanding has spurred the search for ventilation strategies that mitigate ventilator-induced lung injury (VILI). The principles of lung protection have been applied by using conventional mechanical ventilation (CMV) with demonstrable clinical beneﬁt, but this does not necessarily imply that further reduction of VILI is not attainable using other methods. First described in 1972 (1), high-frequency oscillatory ventilation (HFOV) is a form of high-frequency ventilation that has the potential to accomplish many of the goals of lung-protective ventilation: limiting lung overdistention and preventing cyclic lung collapse by maintaining endexpiratory lung volume. The past decade has seen signiﬁcant advances in both the theory and understanding of HFOV as well as its clinical use in the 697

698

Singh et al.

ventilation of the severely injured lungs of adults. This article sets out to review the rationale for and the clinical experience with HFOV, especially in the context of adults with acute lung injury. II. Background High-frequency ventilation comprises a group of ventilation modes characterized by small subphysiologic tidal volumes delivered at high respiratory rates (typically greater than 100 breaths per minute). HFOV is a mode of high-frequency ventilation in which a reciprocating diaphragm generates pressure waves in a specialized noncompliant ventilator circuit attached to the patient. HFOV is differentiated from other modes of high-frequency ventilation by: (i) active expiration and (ii) delivering the smallest tidal volumes at the highest frequencies (2). A. Mechanisms of Gas Transport in HFOV

In CMV most of the gas exchange occurs through the bulk ﬂow of gas to the alveolus, with intermixing of gas as it ﬂows through the bifurcating bronchial tree. During HFOV, tidal volumes are much smaller than during CMV, and are thought to approach or be smaller than the anatomical dead space. At these small tidal volumes, the contribution of bulk ﬂow to gas exchange is considerably smaller, and the inspiratory gas front may only reach a small number of very proximal alveoli. The ﬁnding that adequate CO2 elimination can be achieved with HFOV implies that gas exchange must occur through a variety of alternative mechanisms. Several alternative mechanisms of gas exchange during HFOV have been proposed, including convective streaming due to asymmetric velocity proﬁles, pendelluft, cardiogenic mixing, and diffusion (3). Convective streaming refers to the convective transport of gas along the edges of airways as a result of differences in gas ﬂow from inspiration to expiration. Taken as a whole, the cross-sectional area of the tracheobronchial tree increases as gas moves into the lung. The resultant conical shape and the branching airways lead to different axial velocity proﬁles (the patterns in which the gas moves in or out of the airway) on inspiration and expiration. When added together in a summation vector they lead to a net movement of fresh gas down the middle of the airway into the lung, with expired gas moving up along the outside airway walls. This asymmetry in velocity proﬁles has the added effect of creating axial concentration gradients that augment the diffusion of gas in the proximal airways (4). The time required for the ﬁlling and emptying of a lung unit at a given pressure gradient depends on its time constant (the product of its resistance and compliance). Pendelluft describes the local movement of gases between parallel lung units due to heterogeneity in their respective time constants,

Rationale for HFOV in Acute Lung Injury

699

whereby gas mixing may be enhanced at rapid breathing rates. At end inspiration, gas may ﬂow locally from fast-ﬁlling units to slow-ﬁlling lung units, whereas slow-emptying units will ﬁll parallel fast-emptying units at end-expiration. Pendelluft is likely exaggerated at the very high respiratory frequencies seen in HFOV, where gas ﬂow occurs at high ﬂow rates. Gas exchange is further augmented by the beating heart, which agitates adjacent lung tissue and enhances local gas mixing. Finally, molecular diffusion also plays a role both close to the alveolar membrane and in the proximal airways (along concentration gradients created by convective streaming). Although these proposed mechanisms remain incompletely understood, and the predominance of one mechanism over another in a given patient at any one time remains unclear, it is indisputable that adequate gas exchange and alveolar ventilation can be achieved using high-frequency, low tidal-volume mechanical ventilation. The fact that gas exchange can be augmented by mechanisms other than bulk ﬂow has important implications for the clinical application of HFOV as part of a lung-protective ventilation strategy. Further discussion of the rationale for HFOV demands an understanding of VILI as well as current principles of lung protection.

III. Rationale for HFOV Current understanding of VILI now centers around two major concepts: lung injury related to high lung volumes and distending pressures (volutrauma and barotrauma), and injury caused by low end-expiratory lung volumes (atelectrauma and lung unit collapse). A. Barotrauma

Lung injury and adverse events due to high airway pressures were some of the ﬁrst recognized complications of positive-pressure mechanical ventilation. Pneumothoraces, pneumomediastinum, and air leaks have long been associated with high inﬂation pressures, though modern pressure- and volume-limited ventilation strategies mitigate these complications (5). B. Atelectrauma and Volutrauma

In the injured lung, heterogeneity of lung injury, alveolar damage, and absolute or qualitative deﬁciencies in alveolar surfactants alter mechanical properties. This leads to alveolar instability and localized lung unit collapse. These unstable alveoli undergo cyclical collapse through the respiratory cycle, with repetitive opening and closing of the collapsed alveoli generating injurious mechanical forces and causing lung injury. This process has been termed atelectrauma. The generation of these forces, particularly in the

700

Singh et al.

injured lung, is strongly inﬂuenced by the way in which mechanical ventilation is employed. Mechanical ventilation with high tidal volumes is relatively safe in healthy lungs but rapidly causes lung injury in lungs with alveolar instability from a lavage-induced surfactant deﬁciency (6). Animal models have also demonstrated that the application of high levels of positive end-expiratory pressure (PEEP), with the intent of limiting tidal recruitment/derecruitment, reduces this lung injury (7). The understanding that it is high tidal volumes and distending pressures (not necessarily high peak airway pressures) that cause signiﬁcant lung injury is important. Elegant animal experiments have demonstrated that the mechanical limitation of lung volume (by reducing chest wall compliance) minimizes lung injury when potentially injurious high-pressure mechanical ventilation is used, whereas when the same pressures are allowed to generate high tidal volumes, signiﬁcant injury does occur (8). Overdistention, or volutrauma, may cause lung injury through alveolar wall stretch, stress failure of the lung ultrastructure, or mechanotransduction of distending forces with the resultant stimulation of inﬂammatory cascades. Prolonged volutrauma may cause demonstrable damage to lung architecture. One CT-imaging study evaluating 21 patients with severe acute respiratory distress syndrome (ARDS) found that the amount of radiographically observed cystic changes and bronchiectasis positively correlated with the duration of mechanical ventilation and the number of days with an end-inspiratory pressure greater than 35 cmH2O (9). This ﬁnding was consistent with the previous work (10), but in addition the authors found that the majority of damage occurred in nondependent anterior lung segments, areas of the lung that have been shown in previous CT-based studies to be well aerated and relatively spared of atelectasis. This has led to the hypothesis that repetitive overdistention of these relatively healthy, compliant lung units may lead to structural damage, and potentially to chronic lung disease. In summary, it is evident that mechanical ventilation with high tidal volumes and limited PEEP subjects alveoli to both alveolar overdistention as well as shear force injury. These injurious forces may also potentiate further lung injury through the activation of inﬂammatory cascades (see below) or by causing structural damage that predisposes to chronic structural lung disease. C. Oxygen Toxicity

Exposure to high fractional inspired oxygen (FiO2) concentrations was one of the ﬁrst mechanisms to be linked to VILI. This phenomenon has been well studied and is widely accepted. In addition to VILI, oxygen toxicity may potentiate lung injury through absorption atelectasis, secondary hypoventilation, and systemic vasoconstriction (1).

Rationale for HFOV in Acute Lung Injury

701

D. Biotrauma and Activation of Inﬂammatory Cascades

Although the most striking physiologic changes in patients with ARDS are related to the lungs, the major cause of death in these patients is multiple organ failure, not hypoxemia. It is now well established that injurious mechanical forces applied to the lung during ventilation leads to the activation of inﬂammatory cascades with the resultant release of systemic inﬂammatory mediators such as tumor necrosis factor-a (TNF-a) and interleukin (IL)-6 (11). As well as directly inciting an inﬂammatory response through mechanical forces, mechanical ventilation itself may modulate the lung’s inﬂammatory response to other pulmonary and extrapulmonary insults. One elegant animal experiment demonstrated a synergistic increase in the pulmonary inﬂammatory response (manifested by TNF-a, IL-8, and lung permeability) to a lipopolysaccharide infusion after injurious mechanical ventilation (12). Alteration of mechanical ventilation settings to protect the lung has been shown to decrease the levels of inﬂammatory mediators. Animal studies comparing various ventilator strategies have demonstrated lower levels of inﬂammatory mediators with strategies designed to limit VILI (pressure and volume limitation with high PEEP). Most importantly, clinical trials in humans have reproduced these ﬁndings, showing changes in both the pulmonary and circulating levels of inﬂammatory mediators (13,14). In summary, despite being an essential tool for the support of the critically ill patient, mechanical ventilation itself can cause or potentiate lung injury. Increasing appreciation of this fact has resulted in the search for ventilation strategies that are more protective. E. Principles and Practice of Lung Protection

Based on our current understanding of the major contributors of VILI, lung-protective ventilation should ideally accomplish the following four goals: (i) Avoid lung overdistention, thus minimizing volutrauma; (ii) open atelectatic lung units; (iii) maintain end-expiratory lung volume to avoid cyclical end-expiratory lung unit collapse and subsequent atelectrauma; and (iv) avoid high inspired oxygen concentrations (15). The clinical application of certain aspects of these principles has been validated by randomized clinical trials employing conventional volume-limited ventilation modes (14,16). Despite these successes, the optimal strategy to minimize VILI remains unknown. Lung-protective strategies using CMV have primarily been studied to date, although there are compelling reasons to consider HFOV as a lung-protective ventilation strategy. As stated, lung-protective ventilation strategies employing conventional ventilation modes have been shown to decrease mortality in clinical trials. The ﬁrst of these trials compared a protective ventilation strategy (tidal volume, 6 mL/kg; PEEP targeted above the lower inﬂection point

702

Singh et al.

on the quasistatic pressure–volume curve; frequent recruitment maneuvers) with a strategy employing higher tidal volumes and lower PEEP (16). Patients receiving protective mechanical ventilation had improved 28-day survival and a lower incidence of barotrauma. Subsequently, a landmark study conducted by the National Institutes of Health ARDS Network randomized 861 patients with ARDS to receive a targeted tidal volume of 6 mL/kg of predicted body weight, keeping the plateau pressure below 30 cmH2O, or to receive a targeted tidal volume of 12 mL/kg (14). The group receiving the lower tidal volume ventilation had a mortality rate that was 9% lower than that seen in the higher tidal volume group. F. Limits to CMV-Based Lung Protection

Despite the success with CMV-based strategies, there are several considerations that suggest they may not be the optimal solution to VILI. The relatively noncompliant lungs of patients with ARDS frequently require high driving pressures to achieve a given tidal volume. This low compliance means that with pressure-limited approaches (generally with the goal of avoiding plateau pressures in the range of 30–35 cmH2O), there are frequently challenges in achieving adequate CO2 elimination as alveolar ventilation decreases. The clinician is thus faced with either allowing airway pressures/volumes to rise to potentially injurious levels, or accepting higher CO2 levels (permissive hypercapnia), which has several potential downsides (17). There may be circumstances in which high plateau pressures are not dangerous—when there is extrapulmonary restriction (e.g., from a distended abdomen, stiff chest wall, or large pleural effusions)—resulting in high plateau pressures but lower transpulmonary (or distending) pressures. In these situations, clinicians may be tempted to let the airway pressure rise, feeling that the risk of overdistention is lower. The problem with this assumption is that there is signiﬁcant heterogeneity of the lung in ARDS, and as a result some areas of the lung may still be overdistended. For example, CT imaging studies in animal models and humans have demonstrated a nonhomogeneous distribution of lung disease with hyperinﬂated, normally aerated, poorly aerated, and atelectatic lung zones distributed grossly along a ventral–dorsal axis, respectively (10,18). This heterogeneity means that at a given applied plateau pressure, there may be regional volutraumaand atelectrauma-type injuries depending on the local lung condition and compliance. Thus, despite the successes of recent CMV strategies, it may be possible to further minimize VILI by maximizing the tenets of lung protection. G. Lung Protection with HFOV

The unique mechanical properties of HFOV have the potential to accomplish all the goals of lung protection. The relative uncoupling of ventilation

Rationale for HFOV in Acute Lung Injury

703

from swings in airway pressure allows the application of high mean airway pressures, leading to improved recruitment and oxygenation while avoiding overdistention and atelectrauma. Early animal studies demonstrated improved oxygenation with HFOV compared to CMV in oleic-acid injury models (19–21). McCulloch et al. compared two HFOV-based ventilation strategies (low- and high-mean lung volume HFOV) with CMV that was applied with a low mean lung volume in surfactant-deﬁcient rabbits (19). After seven hours of mechanical ventilation, the low lung volume groups (both HFOV and CMV) had lower respiratory system compliance, decreased hysteresis on expiration, and more histologic evidence of diffuse alveolar damage. If HFOV results in a reduction in mechanical injury, this might translate into a reduction of biotrauma, with a resultant decrease in multiple organ dysfunctions and mortality. Three animal studies have attempted to address this hypothesis by comparing lung-protective CMV with HFOV (20–22). Rotta et al. studied adult rabbits that underwent saline lung lavages and were then stabilized using HFOV combined with lung volume recruitment maneuvers (20). After stabilization, all rabbits were randomized to receive CMV with (i) low PEEP (2 cmH2O) and moderate tidal volume, (ii) high PEEP (10 cmH2O) and moderate tidal volume, (iii) low tidal volume (6 mL/kg) with PEEP targeted above the lower inﬂection point, or (iv) HFOV. The animals were ventilated for four hours, after which they were sacriﬁced and markers of lung injury and inﬂammation were measured. The groups ventilated with HFOV and open-lung CMV had similar and improved lung mechanics, compliance, and oxygenation than the groups receiving nonprotective CMV. However, HFOV was found to attenuate the increase in markers of lung injury that was observed in all CMV groups, including the open-lung CMV group. In another animal study, Imai et al. evaluated the effects of various lung-protective CMV and HFOV ventilation strategies on inﬂammatory markers. A rabbit lung lavage model was used to compare a CMV ventilation strategy with moderate tidal volume (10–12 mL/kg) and low PEEP, lung-protective ventilation using CMV with low tidal volume (5–6 mL/kg) and high PEEP (8–10 cmH2O), and HFOV. After mechanical ventilation with the assigned experimental protocol for four hours, gas exchange, respiratory system compliance, inﬂammatory mediators, and lung histology were compared across all groups. Although all of the lung-protective strategies improved the respiratory system compliance and oxygenation compared with the nonprotective strategy, only rabbits ventilated with HFOV had decreased pulmonary inﬂammation. Another recently published study comparing HFOV and CMV in a surfactant-depleted pig model found that ventilation with HFOV resulted in a signiﬁcant reduction in mRNA expression of IL-1b, IL-6, IL-8, and IL-10, TGF-a, and adhesion molecules (22).

704

Singh et al.

Together, these studies suggest that HFOV may be superior (with respect to reduction of biotrauma and inﬂammation) to a lung-protective approach with CMV, even when similar oxygenation is achieved. IV. Clinical Experience with HFOV A. Experience in Neonates and Children

The longest clinical experience as well as the most rigorous evaluation of HFOV has occurred in neonates (23–32). Although initial clinical trials evaluating HFOV in these populations were disappointing, it later became evident that the safe and effective application of HFOV requires utilizing an open-lung approach (23,33,34). Subsequent studies of HFOV with an openlung approach have demonstrated that HFOV is safe, improves oxygenation, and may reduce the risk of air leak and barotrauma (24–28). A current meta-analysis of the existing trials evaluating neonatal HFOV found no mortality difference between CMV and HFOV, but HFOV may have been associated with a modest reduction in chronic lung disease (31,35). Although the effect of HFOV at reducing infant mortality and morbidity is contentious, a recent study suggests that in carefully controlled clinical settings HFOV is beneﬁcial (27). Interpreting the neonatal HFOV literature is made challenging by the differences in study populations (preterm vs. term), interventions (degree of lung recruitment targeted), and the timing of HFOV (immediately after birth vs. later), and by the introduction of exogenous surfactants as a standard therapy in the 1990s. In addition, it is important to realize that the baseline mortality rate in infant respiratory distress syndrome is almost an order of magnitude lower than that of adults with ARDS. All of these factors mean that it is difﬁcult to extrapolate these studies to adult populations. Important lessons that can be learned from the neonatal study of HFOV, however, include the importance of thoroughly understanding the appropriate physiological targets of the therapy, and the recognition that a learning curve may exist in the initiation of an HFOV program (33). B. Experience in Adults

Clinical experience with HFOV in adults is limited. While neonatal and pediatric oscillatory ventilators have been commercially available since the 1980s, it is only in the last 10 years that the adult equivalent has been available, and for most of that time only in a research setting. There are multiple prospective (36–39) and retrospective (40–43) case series reporting the safety and efﬁcacy of HFOV as a rescue therapy in adults with severe ARDS who were failing conventional ventilation. Three reports describe HFOV used in adults with traumatic injuries (37) and burns (40,41). All series reported improvements in oxygenation with the safe application of HFOV at mean

Rationale for HFOV in Acute Lung Injury

705

airway pressures signiﬁcantly higher than was possible with CMV. Unfortunately, direct comparison of mortality rates in these series to those using CMV is confounded by the high severity of illness and the severe hypoxemia of patients enrolled in the HFOV studies. For example, in the largest series to date published by Mehta et al., the mean PaO2/FiO2 ratio was 91 with an acute physiology and chronic health evaluation (APACHE) II score of 23.8 (43). In addition, many of these series included patients with severe burns, hematological malignancies, and allogenic bone-marrow transplants, all factors known to increase mortality. From the data available in all of these case series, it seems reasonable to conclude that HFOV should be considered a viable rescue therapy in patients with severe ARDS who are failing conventional ventilation. In this setting HFOV is usually effective in improving oxygenation and appears to be reasonably safe. To date we are aware of only one published randomized trial comparing HFOV to CMV (44). In this study 148 adults with ARDS (mean PaO2/ FiO2 ¼ 112.5 mmHg) were randomized to either HFOV or pressure-control conventional ventilation. This study was powered to detect large differences in the ‘‘incidence of key adverse outcomes (e.g., new air leak, intractable hypotension, etc.)’’ between the HFOV and CMV arms (44). The investigators observed a more rapid improvement in oxygenation using HFOV, and did not detect any differences in adverse events. There was an absolute difference in the 30-day mortality of 15% favoring the HFOV group, but this did not reach statistical signiﬁcance given the relatively small number of patients included in the study. While this mortality difference is certainly enticing, it should be pointed out that this study was designed and initiated prior to the results of the ﬁrst ARDS Network trial becoming available (14). As such, the protocol for conventional ventilation arguably did not conform to what would now be considered the standard of care. However, the results of this trial do allow us to conclude that HFOV is safe when used outside the rescue setting, and that it appears to have some viable promise as a therapy to further reduce the mortality effects of VILI.

V. Future Directions in the Application of HFOV Evidence to date suggests that HFOV is safe, effective as rescue therapy in patients failing CMV, and shows promise as a primary therapy to minimize VILI in ARDS patients. It is this latter indication that generates both the most excitement and the most uncertainty among clinicians and researchers in ARDS. There is general agreement among experts that HFOV in adults should be employed in the context of an open-lung strategy (34,45,46). There are also increasing calls for the initiation of HFOV early in the course of severe ARDS. A number of studies have found an association between a longer duration of CMV prior to HFOV initiation and death (36,38,39

706

Singh et al.

43,44). Although intriguing, and congruent with the hypothesis that HFOV mitigates VILI and subsequent multiple organ dysfunction, this observation needs be considered with the realization that it comes from uncontrolled studies. However, given what we know about the rapidity of the onset of VILI, it makes sense to study the early application of this therapy, before the biotrauma pathway is irrevocably activated. Despite the published literature, the optimal ventilator setting for HFOV in adults remains unclear. Most studies in adults have used similar initial ventilator settings; frequency 4–5 Hz, bias ﬂow 30–40 L/min, delta P 60–90 cmH2O, and mean airway pressure (mPaw) set at 5 cmH2O above the mPaw on CMV, then titrated upward until improvements in oxygenation were observed. These settings, however, are fairly arbitrary and do not necessarily represent optimal HFOV. In future studies, close attention should be paid to HFOV strategies that attempt to minimize delivered tidal volumes, potentially maximizing both frequency and power settings to take advantage of the alternative mechanisms of gas exchange (46,47). Other important considerations include the development of an explicit strategy for implementing HFOV across centers (because it cannot be simply targeted to a speciﬁc tidal volume), a plan for switching patients back from HFOV to CMV (because adults in general cannot spontaneously breathe well enough on the current HFOV circuit), and using recruitment maneuvers or other ancillary strategies along with HFOV. Our group recently published the results of a single-arm pilot study in which patients with early ARDS were switched to HFOV using an initial series of sustained inﬂation lung-volume recruitment maneuvers (48). Recruitment maneuvers and ventilator setting changes were governed by an explicit protocol, utilizing initial high mean airway pressures with a rapid decremental titration algorithm. An explicit protocol for transitioning patients who had achieved a low FiO2 and mPaw to CMV was also used, whereby patients were monitored for intolerance to ‘‘gentle’’ CMV settings and switched back to HFOV if necessary. This trial found the combination of HFOV and recruitments safe and effective in achieving lung recruitment, as judged by a marked and rapid improvement in oxygenation. The protocol appeared feasible in terms of the excellent compliance and the reasonable rates of success in the conversion back to conventional ventilation (48). Finally, a comparative randomized trial of HFOV and the ‘‘best possible’’ CMV is needed. The low tidal volume strategy of the ARDS Network study may be considered as the standard against which other ventilation strategies should be compared (14,49). However, in the testing of open-lung HFOV, it may be better to compare this to a conventional ventilation strategy that also utilizes an open-lung approach. Already one RCT has not found large differences in outcome between the original ARDS Network approach and one that utilized low tidal volumes with higher PEEP (50).

Rationale for HFOV in Acute Lung Injury

707

If the results of another ongoing study comparing a lung-open ventilation strategy (LOVS) with the ARDS Network protocol show similar (or improved) outcomes in the LOVS group, we feel that this protocol would make a sensible mechanistic comparison group for an HFOV trial (51). Such a trial comparing ‘‘best practices’’ of HFOV and CMV would signiﬁcantly clarify the role of HFOV in the support of the critically ill patient with severe hypoxemic respiratory failure.

VI. Conclusion Our goals for mechanical ventilation have changed signiﬁcantly over the last 20 years as our understanding of VILI has become more complete. We have shifted from ventilating to ‘‘normal’’ blood gases to instead attempting to minimize VILI while still achieving acceptable gas exchange. Signiﬁcant strides have been made in the identiﬁcation of lung-protective conventional ventilation strategies, but there may be still further gains to be made. HFOV is theoretically ideal for preventing VILI—providing adequate lung recruitment without tidal overdistention—something not always possible with conventional ventilation. In adults, HFOV should currently be considered as a rescue therapy for patients with severe oxygenation failure unresponsive to conventional ventilation. Further studies are needed to rigorously test the efﬁcacy of HFOV in reducing VILI and lowering mortality in patients with severe ARDS—we await their results with interest.

References 1. Bryan CL, Jenkinson SG. Oxygen toxicity. Clin Chest Med 1988; 9:141–152. 2. Kacmarek RM, Malhotra A. High-frequency oscillatory ventilation: What large-animal studies have taught us! Crit Care Med 2005; 33:S148–S154. 3. Chang HK. Mechanisms of gas transport during ventilation by high-frequency oscillation. J Appl Physiol: Respir, Environ Exer Physiol 1984; 56:553–563. 4. Taylor GI. Diffusion and mass transport in tubes. Proc Roy Soc A 1953; 216:186. 5. Antonio A, Fernando F, Andres E, et al. Incidence, risk factors and outcome of barotrauma in mechanically ventilated patients. Intensive Care Med 2004; 30:612–619. 6. Steinberg J, Schiller HJ, Halter JM, et al. Tidal volume increases do not affect alveolar mechanics in normal lung but cause alveolar overdistension and exacerbate alveolar instability after surfactant deactivation. Crit Care Med 2002; 30:2675–2683. 7. Steinberg JM, Schiller HJ, Halter JM, et al. Alveolar instability causes early ventilator-induced lung injury independent of neutrophils. Am J Respir Crit Care Med 2004; 169:57–63.

708

Singh et al.

8. Dreyfuss D, Soler P, Basset G, Saumon G. High inﬂation pressure pulmonary edema Respective effects of high airway pressure, high tidal volume, and positive end-expiratory pressure. Am Rev Respir Dis 1988; 137:1159–1164. 9. Treggiari MM, Romand JA, Martin JB, Suter PM. Air cysts and bronchiectasis prevail in nondependent areas in severe acute respiratory distress syndrome: a computed tomographic study of ventilator-associated changes. Crit Care Med 2002; 30:1147–1752. 10. Gattinoni L, Caironi P, Pelosi P, Goodmanm LR. What has computed tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001; 164:1701–1711. 11. Tremblay LN, Slutsky AS. Ventilator-induced injury: from barotrauma to biotrauma. Proc Assoc Am Physicians 1998; 110:482–488. 12. Altemeier WA, Matute-Bello G, Frevert CW, et al. Mechanical ventilation with moderate tidal volumes synergistically increases lung cytokine response to systemic endotoxin. Am J Physiol Lung Cell Mol Physiol 2004. 13. Ranieri VM, Suter PM, Tortorella C, et al. Effect of mechanical ventilation on inﬂammatory mediators in patients with acute respiratory distress syndrome. JAMA 1999; 282(1):54–61. 14. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med 2000; 342:1301–1308. 15. Lachmann B. Open up the lung and keep the lung open. Intensive Care Med 1992; 18:319–321. 16. Amato MB, Barbas CS, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998; 338:347–354. 17. Laffey G, Donall O, Paul M, Kavanagh P. Permissive hypercapnia—role in protective lung ventilatory strategies. Intensive Care Med 2004; 30:347–356. 18. Malbouisson LM, Muller JC, Constantin JM, Lu Q, Puybasset L, Rouby JJ. Computed tomography assessment of positive end-expiratory pressure-induced alveolar recruitment in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163:1444–1450. 19. McCulloch PR, Forkert PG, Froese AB. Lung volume maintenance prevents lung injury during high frequency oscillatory ventilation in surfactant-deﬁcient rabbits. Am Rev Respir Dis 1988; 137:1185–1192. 20. Rotta AT, Gunnarsson B, Fuhrman BP, Hernan LJ, Steinhorn DM. Comparison of lung protective ventilation strategies in a rabbit model of acute lung injury. Crit Care Med 2001; 29:2176–2184. 21. Imai Y, Nakagawa S, Ito Y, Kawano T, Slutsky AS, Miyasaka K. Comparison of lung protection strategies using conventional and high-frequency oscillatory ventilation. J Appl Physiol 2001. 22. von der HK, Kandler MA, Fink L, et al. High frequency oscillatory ventilation suppresses inﬂammatory response in lung tissue and microdissected alveolar macrophages in surfactant depleted piglets. Pediatr Res 2004; 55:339–346. 23. HIFI Study Group. High-frequency oscillatory ventilation compared with conventional mechanical ventilation in the treatment of respiratory failure in preterm infants. N Engl J Med 1989; 320:88–93.

Rationale for HFOV in Acute Lung Injury

709

24. Ogawa Y, Miyasaka K, Kawano T, et al. A multicenter randomized trial of high frequency oscillatory ventilation as compared with conventional mechanical ventilation in preterm infants with respiratory failure. Early Hum Dev 1993; 32:1–10. 25. HiFO Study Group. Randomized study of high-frequency oscillatory ventilation in infants with severe respiratory distress syndrome. J Pediatr 1993; 122: 609–619. 26. Arnold JH, Hanson JH, Toro-Figuero LO, Gutierrez J, Berens RJ, Anglin DL. Prospective, randomized comparison of high-frequency oscillatory ventilation and conventional mechanical ventilation in pediatric respiratory failure. Crit Care Med 1994; 22:1530–1539. 27. Gerstmann DR, Minton SD, Stoddard RA, et al. The Provo multicenter early high-frequency oscillatory ventilation trial: improved pulmonary and clinical outcome in respiratory distress syndrome. Pediatrics 1996; 98:1044–1057. 28. Rettwitz-Volk W, Veldman A, Roth B, et al. A prospective, randomized, multicenter trial of high-frequency oscillatory ventilation compared with conventional ventilation in preterm infants with respiratory distress syndrome receiving surfactant. J Pediatr 1998; 132:249–254. 29. Thome U, Kossel H, Lipowsky G, et al. Randomized comparison of highfrequency ventilation with high-rate intermittent positive pressure ventilation in preterm infants with respiratory failure. J Pediatr 1999; 135:39–46. 30. Moriette G, Paris-Llado J, Walti H, et al. Prospective randomized multicenter comparison of high-frequency oscillatory ventilation and conventional ventilation in preterm infants of less than 30 weeks with respiratory distress syndrome. Pediatrics 2001; 107:363–372. 31. Courtney SE, Durand DJ, Asselin JM, Hudak ML, Aschner JL, Shoemaker CT, and the Neonatal Ventilation Study Group. High-frequency oscillatory ventilation versus conventional mechanical ventilation for very-low- birthweight infants. N Engl J Med 2002; 347:643–652. 32. Johnson AH, Peacock JL, Greenough A, et al., and the United Kingdom Oscillation Study Group. High-frequency oscillatory ventilation for the prevention of chronic lung disease of prematurity. N Engl J Med 2002; 347:633–642. 33. Bryan AC, Froese AB. Reﬂections on the HIFI trial. Pediatrics 1991; 87: 565–567. 34. Froese AB, Kinsella JP. High-frequency oscillatory ventilation: lessons from the neonatal/pediatric experience. Crit Care Med 2005; 33:S115–S121. 35. Henderson-Smart DJ, Bhuta T, Cools F, Offringa M. Elective high frequency oscillatory ventilation versus conventional ventilation for acute pulmonary dysfunction in preterm infants. Cochrane.Database.Syst.Rev. CD000104, 2003. 36. Fort P, Farmer C, Westerman J, et al. High-frequency oscillatory ventilation for adult respiratory distress syndrome-a pilot study. Crit Care Med 1997; 25: 937–947. 37. Claridge JA, Hostetter RG, Lowson SM, Young JS. High-frequency oscillatory ventilation can be effective as rescue therapy for refractory acute lung dysfunction. Am Surg 1999; 65:1092–1096. 38. Mehta S, Lapinsky SE, Hallett DC, et al. Prospective trial of high-frequency oscillation in adults with acute respiratory distress syndrome. Crit Care Med 2001; 29:1360–1369.

710

Singh et al.

39. David M, Weiler N, Heinrichs W, et al. High-frequency oscillatory ventilation in adult acute respiratory distress syndrome. Intensive Care Med 2003. 40. Cartotto R, Cooper AB, Esmond JR, Gomez M, Fish JS, Smith T. Early clinical experience with high-frequency oscillatory ventilation for ARDS in adult burn patients. J Burn Care Rehabil 2001; 22:325–333. 41. Cartotto R, Ellis S, Gomez M, Cooper A, Smith T. High frequency oscillatory ventilation in bum patients with the acute respiratory distress syndrome. Burns 2004; 30:453–463. 42. Andersen FA, Guttormsen AB, Flaatten HK. High frequency oscillatory ventilation in adult patients with acute respiratory distress syndrome—a retrospective study. Acta Anaesthesiol Scand 2002; 46:1082–1088. 43. Mehta S, Granton J, MacDonald RJ, et al. High-frequency oscillatory ventilation in adults: the Toronto experience. Chest 2004; 126:518–527. 44. Derdak S, Mehta S, Stewart TE, et al. High-frequency oscillatory ventilation for acute respiratory distress syndrome in adults: a randomized, controlled trial. Am J Respir Crit Care Med 2002; 166:801–808. 45. Derdak S. High-frequency oscillatory ventilation for acute respiratory distress syndrome in adult patients. Crit Care Med 2003; 31:S317–S323. 46. Fessler HE, Brower RG. Protocols for lung protective ventilation. Crit Care Med 2005; 33:S223–S227. 47. Froese AB. The incremental application of lung-protective high-frequency oscillatory ventilation. Am J Respir Crit Care Med 2002; 166:786–787. 48. Ferguson ND, Chiche JD, Kacmarek RM, et al. Combining high-frequency oscillatory ventilation and recruitment maneuvers in adults with early acute respiratory distress syndrome: The Treatment with Oscillation and an Open Lung Strategy (TOOLS) Trial pilot study. Crit Care Med 2005; 33:479–486. 49. Stewart TE. Controversies around lung protective mechanical ventilation. Am J Respir Crit Care Med 2002; 166:1421–1422. 50. Brower RG, Lanken PN, MacIntyre N, et al. Higher versus lower positive endexpiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351:327–336. 51. Meade MO, Davies A, Arabi Y, et al. An RCT of two lung-protective ventilation strategies in ALI/ARDS—preliminary data. Proc Am Thorac Soc 2005; 2:A436.

28 Gene Therapy for Ventilator-Induced Lung Injury

¨ KHAN M. MUTLU GO

YOCHAI ADIR

Pulmonary and Critical Care Medicine, Feinberg School of Medicine, Northwestern University Chicago, Illinois, U.S.A.

Division of Pulmonary Medicine, Carmel Medical Center, Technion, Institute of Technology Haifa, Israel

PHILLIP FACTOR Pulmonary, Allergy, and Critical Care Medicine, Columbia University College of Physicians and Surgeons New York, New York, U.S.A.

I. Introduction The revolution of gene transfer as an experimental tool and therapy has created a new potential for the delivery of genetic pharmaceuticals that alter cell physiology in previously unimaginable ways. The accessibility of the airspace compartment and the availability of vehicles that efﬁciently transduce epithelial cells offer the possibility of gene therapy to modulate the pathophysiology of acute lung injury (ALI). It is reasonable to speculate that some of the genetic therapies being proposed for ALI might be applicable to ventilator-induced lung injury (VILI). This review includes the summaries of potential genetic therapies for ALI, which may be applicable for VILI. II. Gene Therapy for ALI ALI is characterized by increased alveolar endothelial and epithelial permeability allowing exudation of proteinaceous material and ﬂux of activated leukocytes into the alveolar airspace. Subsequent alterations include surfactant 711

712

Mutlu et al.

dysfunction, impaired alveolar epithelial solute transport, disrupted oxidant/ antioxidant balance, and activation of metalloproteinases. During ALI, alveolar epithelial cells are lost and are replaced with ﬁbrin-rich deposits (hyaline membrane) that likely signal the recruitment of inﬂammatory cells and ﬁbroblasts, fostering progression to lung ﬁbrosis. Genetic approaches that address each of these pathophysiologic processes have been tested in experimental systems (Table 1). As a form of ALI, VILI is associated with atelectasis and ﬂuid-ﬁlled alveolar airspaces (1). Epithelial cells are lost and replaced with hyaline membrane, providing few targets for gene transfer. Recently, a small body of knowledge has appeared showing that lung injury may not be a limitation as once thought. Stammberger et al. have shown that lipid-facilitated gene transfer of plasmid DNA is possible when instilled into the lung made atelectatic by central airway obstruction (2). Factor et al. have recently reported that rat lungs with severe injury and edema caused by exposure to 100% oxygen (O2) for 64 hours are readily transducible with ﬁrst-generation adenovectors (3). Similarly, rat lungs with sepsis-induced injury are more transducible with adenovectors than the controls, possibly due to increased expression of adenoviral receptors on epithelial cells (4). Together, these data suggest that gene transfer to the injured lung is feasible. To date, ﬁve types of gene therapy applicable to ALI have been reported: improvement of surfactant synthesis, enhancement of the ability of the epithelium to clear excess ﬂuid, upregulation of antioxidants, restoration of hypoxic pulmonary vasoreactivity, and modulation of alveolar barrier

Table 1 Potential Genetic Therapies for VILI Direction of therapy Improving alveolar epithelial function

Improving endothelial function Recovery of surfactant production Reversal of V/Q mismatch Augmentation of antioxidants

Genes (model) Na,K-ATPase (hyperoxia, thiourea, VILI) b2-Adrenergic receptor (hyperoxia) CFTR 8-Oxoguanine DNA glycosylase (hyperoxia) VEGF receptor Surfactant protein A and B Phosphocholine cytidylyltransferase Kþ channel Superoxide dismutase (hyperoxia, bleomycin, radiation) Heme oxygenase-1 (inﬂuenza) TGF-b receptor (radiation)

Abbreviations: CFTR, cystic ﬁbrosis transmembrane conductance regulator; VEGF, vascular endothelial growth factor; TGF-b, transforming growth factor-b; VILI, ventilator-induced lung injury; ATPase, adenosine triphosphatases.

Gene Therapy for Ventilator-Induced Lung Injury

713

function. Most studies employed recombinant adenovectors. Although other viral and nonviral transfer systems (5) are promising for ALI, the relative tropism of these vectors for respiratory epithelium and the rapidity of onset (6) and short duration of transgene expression (7) make recombinant adenoviruses particularly well suited for ALI. It is important to appreciate that data published thus far employed gene transfer prior to the onset of ALI. This is a reﬂection of the lack of experimental models that mimic the temporal patterns of VILI in humans. A unique feature of VILI that is distinct from other forms of ALI is that it may be possible to identify the patients at risk of VILI, thereby allowing early delivery of a gene transfer vehicle. A. Improving the Clearance of Pulmonary Edema Fluid from the Alveolar Airspace

It is now apparent that the clearance of excess ﬂuid from the alveolar airspace is dependent on active Naþ transport by alveolar epithelial cells. It is widely accepted that the coordinated function of apical Naþ and Clchannels and basolateral Na,K-adenosine triphosphatases (ATPases) creates a transepithelial osmotic gradient that causes ﬂuid to exit the alveolus. It has been observed in animal models that in some types of lung injury, active Naþ transport is impaired. Thus, pulmonary edema in the injured lung is due to increased ﬂux of ﬂuid into the airspace and impairment of the mechanisms that remove it. Clinical relevance of this paradigm has been provided by Ware and Matthay, who noted that mechanisms of alveolar ﬂuid clearance are impaired in most patients with acute respiratory failure (8). Gene transfer to improve the functions of key active Naþ transport proteins in the alveolar epithelium has been proposed for the purpose of improving the lung’s ability to clear excess ﬂuid and maintain a ‘‘dry’’ airspace capable of gas exchange. Adenoviral-mediated overexpression of rat Na,K-ATPase b1 subunit has been shown to increase Na,K-ATPase function and active Naþ transport in rat alveolar type 2 (AT2) epithelial cells, and pulmonary edema clearance, in normal rat lungs (9). Similar data have been recently reported following cystic ﬁbrosis transmembrane conductance regulator (CFTR) gene transfer in rats and mice (10). Rats overexpressing a Na,K-ATPase b1 subunit gene tolerated better the lung injury caused by exposure to 100% O2 for 64 hours (hyperoxia) had no pleural effusions and had alveolar ﬂuid clearance rates that were 300% greater than in the hyperoxic controls (11). Notably, b1 subunit overexpression was associated with 100% survival through 14 days of hyperoxia (vs. approximately 30% in controls). Similarly, intranasal delivery of a plasmid that expresses both a and b Na,K-ATPase subunits increases whole lung ATPase activity and reduces lung water accumulation in a murine, thiourea-induced model of high-permeability pulmonary edema (12). These studies indicate that Na,K-ATPase upregulation enhances active Naþ transport in alveolar

714

Mutlu et al.

epithelial cells in vitro and speeds the clearance of edema ﬂuid from the alveolar airspace of normal and injured rat lungs. b-adrenergic agonists increase active Naþ transport by increasing epithelial Naþ channel, CFTR, and Na,K-ATPase expression and function in the alveolar epithelium. These effects are mediated by alveolar epithelial b2-adrenergic receptors (b2AR) (13). Overexpression of a human b2AR in the alveolar epithelium of normal rats and mice increases alveolar ﬂuid clearance by 100% and 40%, respectively (13,14), by increasing the expression and function of both the amiloride-sensitive epithelial Naþ channel and Na,K-ATPase. Alveolar b2AR overexpression conferred supranormal survival to C57bl6 mice exposed to 100% normobaric O2 for 64 hours (13). Overexpression of the b2AR in these systems increases b2-receptor sensitivity to endogenous catecholamines. B. Modulation of Alveolar Barrier Function

Vascular endothelial growth factor (VEGF) diminishes alveolar wall barrier function and promotes the ﬂux of ﬂuid into the alveolus (15,16). Adenoviralmediated overexpression of a VEGF gene in the lung causes increased permeability pulmonary edema in mice within 24 hours of infection, suggesting that this growth factor may play a role in the impairment of barrier function during ALI (17). Pretreatment with an adenovector that overexpresses a truncated, soluble extracellular domain of a VEGF receptor (ﬂt-1) three days prior to infection with the VEGF adenovector protected the mice from VEGF-associated pulmonary edema (17). Elevated plasma VEGF levels have been noted in patients with acute respiratory distress syndrome (18). Conversely, variably increased and decreased lung and bronchoalveolar lavage ﬂuid VEGF levels at different time points of ALI have been reported in autopsy specimens from patients with sepsis and in experimental models of sepsis and ALI (19–23). This form of gene therapy could be useful by preserving alveolar barrier function and attenuating injury-induced edema formation, if an important role for VEGF in ALI can be established. C. Improvement of Surfactant Production

Surfactant has promising physiologic effects that include reduction of alveolar surface tension and anti-inﬂammatory and antibacterial properties in experimental models of lung injury (24,25). Both the composition and the function of surfactant can be severely impaired during V1LI (26). Moreover, exogenously administered surfactant may decrease the severity of VILI (27,28). Korst et al. have shown that adenoviral-mediated gene transfer can be used to increase lung expression of surfactant proteins A and B, but did not test for changes in surfactant function (29). A more challenging task is to increase the synthesis of the phospholipid components of

Gene Therapy for Ventilator-Induced Lung Injury

715

surfactants. Phosphocholine cytidylyltransferase (CTP) is the rate-limiting step in the synthesis of disaturated phosphatidylcholine (DSPC) in rat AT2 cells. In vitro, infection of rat AT2 cells with a CTP-expressing adenovirus increased CTP protein expression and caused a 25-fold increase in phosphocholine incorporation into DSPC (30). Although not yet studied in ALI, these approaches demonstrate that gene transfer can be used to augment the expression of lipid and protein components of the surfactant and offer the possibility of surfactant gene therapy for ALI and VILI. D. Reversal of Ventilation–Perfusion Mismatch

Loss of protective, hypoxic pulmonary vasoconstriction due to impairment of O2-sensitive voltage-gated potassium channels is, in part, responsible for altered ventilation–perfusion relationships and hypoxemia during ALI. Adenoviral-mediated expression of a cytomegalovirus promoter–driven human Kþ channel (Kv1.5) in chronically hypoxic rats restores O2-sensitive Kþ channel activity; normalizes hypoxic pulmonary vasoconstriction, reduces pulmonary vascular resistance, and attenuates pulmonary arterial hypertrophy (31). The adenovirus used in this study was delivered intratracheally by aerosol and increased Kv1.5 expression in the airway epithelium as well as the pulmonary arteries. This approach, if combined with an endothelial targeting strategy, represents an attractive supportive therapy that could ameliorate some of the pathophysiologic changes during ALI. Recent work from Reynolds using adenoviruses linked to an anti-angiotensin converting enzyme antibody and endothelial cell speciﬁc promoter [VEGF type-1 (ﬂt-1)] produces a 300,000-fold improvement in the selectivity of transgenes in the lung endothelium as opposed to the liver (32,33). This combined transductional and transcriptional targeting approach offers the possibility of lung-speciﬁc endothelial cell gene therapies. E. Augmentation of Lung Antioxidant Function

Reactive oxygen species generated by inﬂammatory or lung parenchymal cells contribute to the pathogenesis of ALI as well as VILI (34). Gene transfer to augment intracellular antioxidant enzymes has been suggested to limit oxidant injury in the lung. Intratracheally administered adenoviruses encoding human superoxide dismutase and catalase-complementary DNA (cDNA) attenuated hyperoxic ALI in rats (34). Heme oxygenase-1 (HO-1) is an inducible heat shock protein that plays an important role in the cessation of inﬂammation, probably by modulating apoptosis. HO-1 protects rats from endotoxin-induced ALI (35,36). Transduction of the bronchial epithelium of rats with an adenovirus expressing a human HO-1 protected rats from hyperoxia and improved their survival (66% vs. 0% survival to 96 hours) (37). Likewise, adenoviral-mediated transfer of an HO-1 cDNA

716

Mutlu et al.

attenuates epithelial apoptosis and is associated with reduced lung proinﬂammatory cytokine levels and improved survival in a murine model of inﬂuenza-induced ALI (38).

III. Gene Therapy for VILI Ventilation of rat lungs with tidal volumes that produce peak airway opening pressures of 35 cmH2O for 20 to 60 minutes increases alveolar epithelial permeability and decreases alveolar ﬂuid clearance by up to 50% (39). AT2 epithelial cells isolated from these rats have decreased Na,K-ATPase a1 subunit message levels and decreased Na,K-ATPase activity (39). Thus, high, and not low or moderate, tidal volume ventilation of rat lungs downregulates alveolar active Naþ transport. Adir et al. tested the hypothesis that adenoviral-mediated overexpression of a Na,K-ATPase b1 subunit in the alveolar epithelium could upregulate alveolar active Naþ transport and protect the lung from VILI-induced pulmonary edema (40). Gene transfer seven days prior to VILI increased Na,K-ATPase expression and activity in basolateral cell membranes of the distal lung (40). Most importantly, lungs overexpressing the Na,K-ATPase b1 subunit had no pulmonary edema, and the alveolar ﬂuid clearance rates were up to 300% greater than the controls following 40 minutes of high tidal volume ventilation. These experiments highlight the inter-relationship between VILI and alveolar active Naþ transport and lend credence to therapies that protect the lung by improving its ability to maintain a dry alveolar airspace. Recent data from a rat model of VILI indicates that a high tidal volume ventilation impairs cAMP-dependent active Naþ transport (41). This new data suggests that b2AR gene transfer (13,14) might be useful for the preservation of alveolar active Naþ transport in VILI.

IV. Conclusions Recent advances in the understanding of the pathophysiology of ALI illuminate multiple pathways that might be amenable to therapeutic modulation via exogenously delivered transgenes. Only one study has employed gene transfer to alter the pathophysiology of VILI; however, it is probable that forthcoming studies will employ the approaches outlined above, in the experimental models of VILI. The potential of gene transfer to effect the highly compartmentalized expression of genetic tools that are not otherwise deliverable offers the possibility of therapies that interrupt or redirect the maladaptive processes that contribute to VILI. Human gene therapy for VILI will require better gene transfer vectors and clinically applicable methods of tracking transgene expression and function (7).

Gene Therapy for Ventilator-Induced Lung Injury

717

Acknowledgments This work was supported by HL-66211, HL-70241, HL-71081, and the American Heart Association.

References 1. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998; 157:294–323. 2. Stammberger U, Uduehi AN, Kubisa B, Roth T, Schmid RA. Non-viral gene delivery to atelectatic and ventilated lungs. Ann Thorac Surg 2002; 73: 432–436; discussion 436–437. 3. Factor P, Mendez M, Mutlu GM, Dumasius V. Acute hyperoxic lung injury does not impede adenoviral-mediated alveolar gene transfer. Am J Respir Crit Care Med 2002; 165:521–526. 4. Weiss YG, Tazelaar J, Gehan BA, et al. Adenoviral vector transfection into the pulmonary epithelium after cecal ligation and puncture in rats. Anesthesiology 2001; 95:974–982. 5. Qiao R, Zhou B, Harboe-Schmidt E, et al. Subunit-speciﬁc coordinate upregulation of sodium pump activity in alveolar epithelial cells by lentivirus-mediated gene transfer. Hum Gene Ther 2004; 15:457–468. 6. Dumasius V, Jameel M, Burhop J, et al. In vivo timing of onset of transgene expression following adenoviral-mediated gene transfer. Virology 2003; 308:243–249. 7. Richard JC, Factor P, Ferkol T, Ponde DE, Zhou Z, Schuster DP. Repetitive imaging of reporter gene expression in the lung. Mol Imaging 2003; 2:342–349. 8. Ware LB, Matthay MA. Alveolar ﬂuid clearance is impaired in the majority of patients with acute lung injury and the acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163:1376–1383. 9. Factor P, Senne C, Dumasius V, et al. Overexpression of the Na,K-ATPase a1 subunit increases Na,K-ATPase function in A549 cells. Am J Respir Cell Mol Biol 1998; 18:741–749. 10. Mutlu GM, Koenig CS, Burhop J, Welch L, Factor P. CFTR increases alveolar ﬂuid clearance via a beta-receptor dependent mechanism. Am J Respir Crit Care Med 2003; 167:A57. 11. Factor P, Dumasius V, Saldias F, Sznajder J. Adenoviral-mediated overexpression of the NA,K-ATPase beta 1 subunit gene increases lung edema clearance and improves survival during acute hyperoxic lung injury in rats. Chest 1999; 116:S24–S26. 12. Stern M, Ulrich K, Robinson C, et al. Pretreatment with cationic lipidmediated transfer of the NaþKþ-ATPase pump in a mouse model in vivo augments resolution of high permeability pulmonary oedema. Gene Ther 2000; 7:960–966. 13. Mutlu GM, Dumasius V, Burhop J, et al. Upregulation of alveolar epithelial active Naþ transport is dependent on beta2-adrenergic receptor signaling. Circ Res 2004; 94:1091–1100.

718

Mutlu et al.

14. Dumasius V, Sznajder JI, Azzam ZS, et al. Beta(2)-adrenergic receptor overexpression increases alveolar ﬂuid clearance and responsiveness to endogenous catecholamines in rats. Circ Res 2001; 89:907–914. 15. Roberts WG, Palade GE. Increased microvascular permeability and endothelial fenestration induced by vascular endothelial growth factor. J Cell Sci 1995; 108:2369–2379. 16. Esser S, Lampugnani MG, Corada M, Dejana E, Risau W. Vascular endothelial growth factor induces VE-cadherin tyrosine phosphorylation in endothelial cells. J Cell Sci 1998; 111:1853–1865. 17. Kaner RJ, Ladetto JV, Singh R, Fukuda N, Matthay MA, Crystal RG. Lung overexpression of the vascular endothelial growth factor gene induces pulmonary edema. Am J Respir Cell Mol Biol 2000; 22:657–664. 18. Thickett DR, Armstrong L, Christie SJ, Millar AB. Vascular endothelial growth factor may contribute to increased vascular permeability in acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 164:1601–1605. 19. Tsokos M, Pufe T, Paulsen F, Anders S, Mentlein R. Pulmonary expression of vascular endothelial growth factor in sepsis. Arch Pathol Lab Med 2003; 127:331–335. 20. Becker PM, Alcasabas A, Yu AY, Semenza GL, Bunton TE. Oxygen-independent upregulation of vascular endothelial growth factor and vascular barrier dysfunction during ventilated pulmonary ischemia in isolated ferret lungs. Am J Respir Cell Mol Biol 2000; 22:272–279. 21. Maitre B, Boussat S, Jean D, et al. Vascular endothelial growth factor synthesis in the acute phase of experimental and clinical lung injury. Eur Respir J 2001; 18:100–106. 22. Karmpaliotis D, Kosmidou I, Ingenito EP, et al. Angiogenic growth factors in the pathophysiology of a murine model of acute lung injury. Am J Physiol Lung Cell Mol Physiol 2002; 283:L585–L595. 23. Choi WI, Quinn DA, Park KM, et al. Systemic microvascular leak in an in vivo rat model of ventilator-induced lung injury. Am J Respir Crit Care Med 2003; 167:1627–1632. 24. Anzueto A, Baughman RP, Guntupalli KK, et al. Aerosolized surfactant in adults with sepsis-induced acute respiratory distress syndrome. Exosurf Acute Respiratory Distress Syndrome Sepsis Study Group. N Engl J Med 1996; 334:1417–1421. 25. Gregory TJ, Steinberg KP, Spragg R, et al. Bovine surfactant therapy for patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 1997; 155:1309–1315. 26. Verbrugge SJ, Bohm SH, Gommers D, Zimmerman LJ, Lachmann B. Surfactant impairment after mechanical ventilation with large alveolar surface area changes and effects of positive end-expiratory pressure. Br J Anaesth 1998; 80:360–364. 27. Verbrugge SJ, Vazquez De Anda G, Gommers D, et al. Exogenous surfactant preserves lung function and reduces alveolar Evans blue dye inﬂux in a rat model of ventilation-induced lung injury. Anesthesiology 1998; 89:467–474. 28. Haitsma JJ, Uhlig S, Lachmann U, Verbrugge SJ, Poelma DL, Lachmann B. Exogenous surfactant reduces ventilator-induced decompartmentalization of

Gene Therapy for Ventilator-Induced Lung Injury

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

719

tumor necrosis factor alpha in absence of positive end-expiratory pressure. Intensive Care Med 2002; 28:1131–1137. Korst RJ, Bewig B, Crystal RG. In vitro and in vivo transfer and expression of human surfactant SP-A and SP-B associated protein cDNAs mediated by replication deﬁcient, recombinant adenoviral vector. Hum Gene Ther 1995; 6: 277–287. Li J, Marsh JJ, Spragg RG. Effect of CTP: phosphocholine cytidylyltransferase overexpression on the mouse lung surfactant system. Am J Respir Cell Mol Biol 2002; 26:709–715. Pozeg ZI, Michelakis ED, Mcmurtry MS, et al. In vivo gene transfer of the O2-sensitive potassium channel Kv1.5 reduces pulmonary hypertension and restores hypoxic pulmonary vasoconstriction in chronically hypoxic rats. Circulation 2003; 107:2037–2044. Reynolds PN, Nicklin SA, Kaliberova L, et al. Combined transductional and transcriptional targeting improves the speciﬁcity of transgene expression in vivo. Nat Biotechnol 2001; 19:838–842. Reynolds PN, Zinn KR, Gavrilyuk VD, et al. A targetable, injectable adenoviral vector for selective gene delivery to pulmonary endothelium in vivo. Mol Ther 2000; 2:562–578. Danel C, Erzurum SC, Prayssac P, et al. Gene therapy for oxidant injuryrelated diseases: adenovirus-mediated transfer of superoxide dismutase and catalase cDNAs protects against hyperoxia but not against ischemia-reperfusion lung injury. Hum Gene Ther 1998; 9:1487–1496. Otterbein L, Sylvester SL, Choi AM. Hemoglobin provides protection against lethal endotoxemia in rats: the role of heme oxygenase-1. Am J Respir Cell Mol 1995; 13:595–601. Inoue S, Suzuki M, Nagashima Y, et al. Transfer of heme oxygenase 1 cDNA by a replication-deﬁcient adenovirus enhances interleukin 10 production from alveolar macrophages that attenuates lipopolysaccharide-induced acute lung injury in mice. Hum Gene Ther 2001; 12:967–979. Otterbein LE, Kolls JK, Mantell LL, Cook JL, Alam J, Choi AM. Exogenous administration of heme oxygenase-1 by gene transfer provides protection against hyperoxia-induced lung injury. J Clin Invest 1999; 103:1047–1054. Hashiba T, Suzuki M, Nagashima Y, et al. Adenovirus-mediated transfer of heme oxygenase-1 cDNA attenuates severe lung injury induced by the inﬂuenza virus in mice. Gene Ther 2001; 8:1499–1507. Lecuona E, Saldias F, Cornelias A, Ridge K, Guerrero C, Sznajder JI. Ventilator-associated lung injury decreases lung ability to clear edema in rats. Am J Respir Crit Care Med 1999; 159:603–609. Adir Y, Factor P, Dumasius V, Ridge KM, Sznajder JI. Na,K-ATPase gene transfer increases liquid clearance during ventilation-induced lung injury. Am J Respir Crit Care Med 2003; 168:1445–1448. Frank JA, Pittet JF, Lee H, Godzich M, Matthay MA. High tidal volume ventilation induces NOS2 and impairs cAMP-dependent air space ﬂuid clearance. Am J Physiol Lung Cell Mol Physiol 2003; 284:L791–L798.

This page intentionally left blank

Index

Absolute risk reduction (ARR), 644 Absorption atelectasis, 699 N-Acetylcysteine, 540 Acidosis, 520 management of, 505 Acidotic blood, 354 Acinar micromechanics, 27, 28 Acinar stress/strain distributions, 26 Actin cytoskeleton, 73, 210 Actin depolymerization, 75 Activatable ﬂuorophores, 463 Activated protein C, 618 Actomyosin, 270 Acute lung injury (ALI), 25, 97, 158, 350, 611, 656, 681 genomic DNA repository, 421 nickel-induced, 410 Acute Physiology and Chronic Health Evaluation II (APACHE II), 520 Acute respiratory distress (ARD). See ARDS. Acute respiratory distress syndrome (ARDS), 25, 47, 103, 157, 267, 432 Acute respiratory failure (ARF), 614 Adenosine, 658 monophosphate (AMP), 82 triphosphate (ATP), 385 Adenylyl cyclases, 81 Adsorption/squeeze-out cycle, 36

Advanced Cardiac Life Support1, 618 Advanced Trauma Life Support1, 618 Aerobic cellular respiration, universal waste product of, 342, 358 Aerosolization, 684 Aerosolized perﬂuorocarbon, 661, 663 Affymetrix GeneChip Microarray System, 412 Air leaks, 699 Air–liquid interfacial stresses, 157 Airspace nitrite, 386 Airway, occlusion of, 35 Airway and vascular pressures, behavior of, 105 role on genesis of VILI, 106 Airway closure, mechanisms of, 164 compliant collapse, 165, 166 meniscus formation, 165, 166 Airway collapse, 666 Airway pressure–time curve, 135 Airway reopening, 167 Akt activation, 11 Alanine aminotransferase (ALT), 484 Albumin, 457, 682 Albumin–globulin ratio, 138 ALI. See Acute lung injury. ALI/ARDS lung, 224 a-Actinin, 74 Alveolar airspace, pulmonary edema ﬂuid clearance from, 713 Alveolar barrier damage, 103, 714

721

722 Alveolar–capillary barrier, 71, 72, 121, 273, 391 alterations of, 121 properties of, 35 ultrastructural correlates of, 122 Alveolar–capillary impression, 98 Alveolar–capillary membrane, 158 Alveolar–capillary permeability, 277 Alveolar collapse, 315, 393 Alveolar distension, 85, 212 Alveolar duct, 27 Alveolar epithelial cells, 390 Alveolar epithelial death, 322 Alveolar epithelial–endothelial barrier, 483 Alveolar epithelial function, 377 Alveolar epithelial ion, 385 Alveolar epithelium, 104, 158, 332, 377 functions of, 378 stresses encountered by, 378 Alveolar ﬂooding, 500 Alveolar ﬂuids, 318 clearance, 387, 713 Alveolar hemorrhage, 663 Alveolar hypophase, 678 Alveolar lining ﬂuid, 158 Alveolar macrophages, 85, 390, 664 Alveolar mechanical ventilation effects, 379 Alveolar micromechanics, in injury states, 29 Alveolar openings, 27 Alveolar space, 123, 682 Alveolar stabilization, 488 Alveolar vessels, 98 Alveoli, 28 consolidated/atelectatic, 225 normal, 225 study, 644 types of, 225 Amiloride-sensitive epithelial Na,K channel, 714 Amniotic embolism, 433 Amniotic infections, 418 Anaphylatoxins, 477

Index Anatomic imaging, 456 chest radiography, 449 computed tomography, 449 mechanical resonance imaging, 451 positron emission tomography, 453 Angiogenic activity, 244 Angiotensin-converting enzyme (ACE), 406 inhibitors, 612 Annihilation radiation, 454 Antiapoptotic pathways, 11 Anticytokines, 540 Antiendotoxins, 540 Antihypertensive drugs, 573 therapy, 574 Apoptogens, 555 Apoptosis, 268 Arachidonic acid metabolism, 76 ARDS. See Acute respiratory distress syndrome. ARDS Network, 508 Arterial capillary pressure, 630 Arterial hypercapnia, 347 Arterial hypertension, 4 Arterial oxygenation, 499, 552, 574 Arterial pressure, 3 Arterial wall remodeling, 6 Arteriovenous ﬁstula model, 5 Artiﬁcial plastic lung, 245 Ascites, 34 Aspartate aminotransferase (AST), 484 Atelectasis, 35, 136, 346, 500 Atelectasis-prone lung, 140 Atelectatic region, 126, 169 Atelectatic units, 101 Atelectrauma, 230, 268, 478, 699 microscale, 173 Atherosclerotic plaque, 185 ATI phenotype, 55 Auto-PEEP, lung-protective effects of, 511

Baby lung, 29, 33, 224 syndrome, 130 Bacterial pneumonia, 665 BAL. See Bronchoalveolar lavage.

Index Balloon ﬂotation catheter, 562 Barotrauma, 34, 128, 499, 527, 699 Basolateral cell membranes, 380, 387, 716 B-cell precursors, 418 Bernoulli equation, 459 b-Adrenergic agonist, 382, 387, 688, 714 Beta-adrenergic pathway, 386 b2-Adrenergic receptors, 714 Beta-blocking drugs, 563 Bioactive surfactant, 36 Bioluminescence, 463 imaging, 463 Biotrauma, 29, 120, 169, 206, 268, 431 Bird1 ventilator, 574 Bleb formation, 123 plasma membrane, 52, 54 Blinding, 637 Blood–gas barrier, 46, 103 disrupting mechanisms, 103 Bowell ischemia, 659 Bradykinin antagonist (CP-0127), 540 Breslow-Day test, 527 Bronchial distortion, 438 Bronchial epithelial cells, 212 Bronchiolectasis, 433, 437 Bronchiolitis obliterans syndrome (BOS), 236 Bronchioloalveolar overinﬂation. See Overinﬂation. Bronchoalveolar lavage (BAL), 69, 228, 272, 678 cytokines, 489 Bronchoalveolar lavage ﬂuid (BALF). See Bronchoalveolar lavage. Bronchopulmonary dysplasia (BPD), 234, 433 Bronchoscopy, 318 Buffered hypercapnia, 350 Bullae, 438 Bullous emphysema, 34

Cadherins, 73 Calcium channel blockers, 8

723 Calcium-dependent signaling cascades, 9 Calfactant, 683 Calmodulin (CaM), 74, 382 CaM. See Calmodulin. Candidate gene approach, 406, 407, 412 Capillary endothelium, 104 ﬁltration coefﬁcient, 80 leak, 664 number, 179 stress failure, 104, 126 Carboxylesterases, 680 Cardiac Arrhythmia Suppression Trial (CAST), 539 Cardiac pacemakers, 612 Cardiogenic edema, 525 Cardiopulmonary bypass (CPB), 490, 661 circuits, 658 Caspases, 322 Catalase, 10 Catalase-complementary DNA (cDNA), 715 Catheters types of, 318 esophageal balloon, 34 Swan-Ganz, 273, 648 Cecal ligation, 685 Cell arterial endothelial cells, 209 chondrocytes, 209 death, 384 programmed, 58 deformation-associated PM remodeling, 51 endothelial, 1 active, 72 epithelial, 712 apoptosis, 319 tracheal, 390 eukaryotic, 54 human alveolar epithelium-derived, 53 injury, 384 mesangial cells, 209 motility, 413 myocardial, 318

724 [Cell] myocytes, 209 occludin content, 385 osteoblasts, 209 remodeling, 46 repair mechanisms, 49 retractile tension, 72 stretch devices, 268 types, 208, 309 vascular smooth muscle cells, 209 Cell–cell adhesions, 385 Cell–cell tethering proteins, 73 Cell-signaling pathways, 412 Cerebral oxygenation, 347 C-fos, 8, 271 mRNA, 272 Chemoattractants, 206, 242, 245 Chemokines, 206, 242 CC, 246 CXC, 58 ELR CXC, 242 inﬂammatory, 663 superfamily, 242 subfamilies, 242 Chemotaxis, 413 Chemotherapeutic targeting, 58 Cholinergic agonists, 689 Chromosomal regions, 409 Chronic obstructive pulmonary disease (COPD), 613 Cingulin, 73 Circumferential tension, 2 cJun, 388 C-jun N-terminal kinases (JNK), 12, 13, 14, 389 Clara cell secretory protein (CCSP), 69 Clinical scanners, 452 Clinical trials, 532, 534 attributes of, 564 blinding, 564 design, 642 ﬁxed-sample, 642 nonrandomized cohort studies for, 645 principles of, 562 sequential, 642

Index [Clinical trials] human experiments, sources of error, 558 multivariable analysis for, 647 randomization, 564 Clostridium botulinum C3 exoenzyme, 12 Cluster analysis, 332 design effect for, 642 randomization, 641 CO2-induced protection, mechanisms of, 356 Coagulation cascades, 322 Coagulation factor III, 411 Colchicine, 387 Collagenase, 682 Collapse phase, 189 Collectins, 678 Colloid resuscitation, 611 Color coding system, 436 Computational ﬂuid dynamics, 195 simulations, 195 Computed tomography (CT), 24, 143 numbers, 449 Computerized protocol experience, 575 Concavalin A-coated silicone bead, 51 Convertase, 686 Corner vessels, 99 Coronary angiography, 612 Coronary insufﬁciency, 521 Corticosteroids, 540 Cox model, 632, 647 Creatinine, 484 Critical surfactant concentration, 192 C-type lectins, 678 Cyclic adenosine monophosphate (cAMP), 72, 382 Cyclic guanosine monophosphate (cGMP), 78 stimulant, 540 Cyclic pressure-stretching, 245 Cyclic stretch, 209 of lung epithelial cells, 211 Cyclic stretch-induced cell activation, 211

Index Cyclohexamide, 212 Cyclo-oxygenase, 477 inhibitors, 540 Cystic ﬁbrosis transmembrane conductance regulator (CFTR), 713 Cytochalasin d, actin polymerization inhibitor, 210 Cytokines, 223, 226 balance, 321 intra-alveolar, 481 network, modulation of, 475 pleiotropic type, 2, 240 proinﬂammatory, 233, 320 signaling, 357 spillover of, 477 Cytomegalovirus promoter-driven human Kþ channel (Kv1.5), 715 Cytoplasmic proteins, 73 Cytoskeleton, 6, 385 alterations, 80 motility, 81 network, 6, 51 rearrangement, 209 remodeling, 82 stiffness, 53 tension, 74

Data safety and monitoring board (DSMB), 638 Death receptor, 322 Decision-support tools, 540, 565 clinical applications, 577 guidelines, 565 protocols, 565 Deep venous thrombosis (DVT) prophylaxis, 539 Deferred consent process, 641 Deformation-induced lipid trafﬁcking (DILT), 51 Dexamethasone, 657 Dextran, 176 Diabetic ketoacidosis, 364 Differential in gel electrophoresis (DIGE), 330

725 Dipalmitoylphospatidylcholine (DPPC), 189, 684 Disaturated phosphatidylcholine (DSPC), 715 Diuretic, 539 D-luciferin, 463 DNA sequencing, 421 Dobutamine administration, 580 Doppler ﬂow velocity, 459 Dorsal paraspinal regions, 35 Drug dosing, 661 Drug trials double-blind, 564

E-cadherin, 73, 79 proteolysis, 84 Echocardiography, 459 Edema ﬂuid, 318 interstitial, 381 model, 31 Edematous lung, 29 Elastase, 682 Elastic modulus, 33 Elastic recoil pressure, 33 Elastic tissue deformation, 28 Elastin deﬁciency, 550 lamellae, 550 Electrolyte therapy, 562, 574 Electronic coincidence detection, 452 Electronic medical records (EMRs), 545 Emergent properties, conceptual paradigm of, 548 Emphysema, 159, 436 Emphysema-like lesion, 433, 441 Encainide, 550 End-expiration, 630, 684 hold, 33 lung stress, 146 lung volume, 685 pressure, 143, 500 transpulmonary pressure, 34 Endocytic pathway, 54 Endocytosis, 51 Endogenous ﬂuorochromes, 462

726 End-organ dysfunction, 268 Endothelial blebbing, 141 Endothelial junctional permeability, 81 Endothelial NO synthase (eNOS), 11 expression, 101 Endothelial oxidative response, 11 Endothelial permeability, 72 Endothelium, interaction of, 85 Endotoxin infusion, 661, 662 Endotracheal tubes, 614 Epinephrine, 688 Epithelial barrier function, 383 Epithelial fragmentation, 103 Epithelial inﬂammatory mediators, 387 Epithelial sodium channel (ENaC), 385 Escherichia coli, 665 bronchopneumonia, 435 E-selectin, 659 Estrogens, 689 Eukaryotic initiation factor (EIF), 15 Evidence-based medicine (EBM), 648 Exocytosis, 51 Exosurf1, 683 Experimental group equation, 564 Experimental replicability, 541 Experimental scientiﬁc principles aims of, 541 conceptual paradigms of, 547 scale of investigation of, 549 Expression arrays, 213 Extracellular matrix, 7 Extracellular signal-regulated kinase (ERK), 12, 13, 271 Extracorporeal gas exchange, 540 Extracorporeal membrane oxygenation study, 432, 543 Extraphysiological inﬂation. See Overinﬂation. Extravascular albumin space, 71 Extravascular lung water (EVLW), 456

[18F]FDG-6-monophosphate, 462 [18F]ﬂuorodeoxyglucose, 462

Index F-actin, 385 ﬁlaments, 9 Fas-dependent apoptosis pathway, 323 Fas expression, 323 FDG-PET imaging, 462 Fibrin membrane formation, 682 monomers, 682 Fibrinogen, 411, 682 Fibroblast growth factor, 83 Fibrolytic activity, 236 Fibronectin, 7 Fibroplasias, 227 Fibroproliferative diseases, 229, 324 Field gradients, 452 Filtering approach, 413 Filtration coefﬁcient, 72 Filtration constant (KF), 105 Flash MX, 588 Flecainide, 550 Flippases, 53 Flow cytometry, 49 Fluid and Catheter Treatment Trial (FACTT), 577 Fluid leakage, 121 resuscitation, 616 transport, 385 transudation, 108 Fluid–structure interactions, micromechanical damage to, 175 Fluorescence imaging, 463 Fluorescent marker technique, 384 Fluorescent tracer, 385 Fluosol, 657 Focal adhesion, 12 plagues, 210 Focal adhesion kinase (FAK), 7, 74, 210 Fractalkine/CX3CL1, 242 Fraction of inspired oxygen (FiOa), 478, 524, 699 Free-radical, 658, 660 scavenger, 657 French paradox, 646 Functional polymorphisms, 422 Functional residual capacity (FRC), 99, 121, 456, 627

Index G allele, 421 Gadolinium-sensitive Ca2þ channels, 10 Gadolinium-sensitive ion channels, 211 Gas transport, mechanisms of, 698 Gastric acid aspiration, 656, 661 Gelatinase, 321, 389 Gene array expression data, 414 Gene clusters, 388 Gene expression endogenous, 465 imaging, 411, 462, 463 microdissection techniques, 237 Gene transcription, 210 Gene transfer, 711 lipid-facilitated, 712 therapeutic, 448 Genetic pharmaceuticals, 711 Genetic susceptibility, 412 Genistein, 80 Genome medicine, 403 Genomic technologies, 422 Glutathione, 390 Glutathione peroxidase, 10 Glycogen synthase kinase-3, 73 Glycoproteins, 7 Glycosphingolipid, 54 Glycosylphosphatidylinositol-anchored proteins, 54 Goodpasture’s syndrome, 47 G-protein–coupled receptors (G-PCRs), 10, 100, 242 G-protein receptors, 6 Gradient-echo experiments, 452 Green ﬂuorescent protein, 463 Growth arrest and DNA-damageinducible gene-45 (GADD-45), 214 Growth factor receptors, 75 Guanosine diphosphate (GDP), 242 Guanosine triphosphate (GTP), 242 Gut circulation, 270

Haplotype frequency, 421 protective, 421

727 Hardy–Weinberg equilibrium, 421 Harvard ventilator, 232 Heat shock protein (HSP), 70, 214, 237, 388, 409 Heat shock protein-70, 388 Hematocrits, 454 Hematoxylin and eosin (H&E), 48 Heme oxygenase-1 (HO-1), 715 Hemodynamic alterations, 270 Hemodynamic instability, 363 Hemodynamic support, 574 Hemoptysis, 104 Hemosiderin-laden macrophages, 104 Heparin-binding epidermal growth factor (HB-EGF), 211 Hepatic ﬁbrosis, 229 Hepatocellular enzymes, 318 Hepatocyte growth factor (HGF), 83, 211 Heterodimer, 56 G proteins, 9 High-frequency oscillation (HFO), 630 High-frequency oscillatory ventilation (HFOV), 233, 697 High-frequency ventilation active expiration, 698 delivering smallest tidal volumes, 698 modes, 698 Homophilic occludins, 73 Hoop stress, 27 Houndsﬁeld units (HU), 34, 436 Human alveolar macrophages, 332 Human experiments, holistic, 552 Human genome project, 422 Human superoxide dismutase, 715 Hyaline membrane, 244, 663, 712 formation, 682 plasma protein–rich, 158 Hydrochloric acid aspiration, 661, 667 Hydrophobic interactions, 55 Hydrostatic pressure gradient, 22 Hypercapnia, 270, 342, 520, 631 clinical studies of, 365 deﬁnitions and terminology of, 342 permissive, 342, 344, 358

728 [Hypercapnia] physiologic effects of, 345 therapeutic, 341, 365 Hypercapnia-induced tissue injury molecular mechanisms of, 359 sepsis, 360 tissue nitration, 359 Hypercapnia-induced vasodilation, 354 Hypercapnic acidosis, 49, 342 anti-inﬂammatory effects of, 357 effects on free-radical generation and activity, 358 protective effects of, 357 Hyperinﬂation, See Overinﬂation. Hyperoxia, 49, 713 Hyperoxic environment. See Hyperoxia. Hyperplasia, 4 Hyperpolarized (HP) gases, 458 Hypotension, drug therapy for, 563 Hypoventilation, 344, 347 secondary, 699 Hypoxemia, 158, 499, 631, 715 Hypoxic pulmonary vasoconstriction, 715 Hypoxic pulmonary vasoreactivity, 712 Hypoxic-ischemic brain injury, 354

Ibuprofen, 540 Ideal body weight (IBW), 524 IEL fenestrations, 6 IL-1 receptor antagonist (IL-1RA), 321, 476 IL-6 genotyping, 422 Ileus, 34 Imaging, types of, 448 anatomic, 448 functional, 456 pulmonary hemodynamics, 459 radioaerosol inhalation, 456 ventilation and ventilationperfusion imaging, 458 molecular, 456 types of, 461. See also Molecular imaging methods. Immunoglobulins, 332

Index Immunohistochemical techniques, 237 Immunosuppressive therapy, 525 In situ hybridization, 464 In vitro cycling, 686 In vivo hybridization, 464 Inducible nitric oxide synthase (INOS), 386 Infant respiratory distress syndrome (IRDS), 683 Infasurf1, 190, 683 Inﬂammation, acute, 320, 324 Inﬂammatory cascades, activation of, 699 Inﬂammatory cell trafﬁcking, 448 Inﬂammatory mediators, 57, 268, 270, 483, 681 In-hospital mortality, 524 Injurious ventilation, 386 mode, 685 strategies, 270 Inotrope, 583 Inotropic agents, 269 Insertion/deletion polymorphism, 408 Inspiratory plateau pressures, 503, 509 Institute for Healthcare Improvement (IHI), 616 Insulin IV infusion values, 584 Integrin-extracellular matrix interactions, 7 Integrin-mediated signaling, 10 Integrin-rich focal adhesion sites, 9 Integrins, 6, 7 clusters, 12 Intercellular adhesion molecule-1 (ICAM-1), 58, 359, 659 Interdependence, 379 Interleukin (IL)-1b, 206, 235 Internal elastic lamina (IEL), 6 Interstitial pressure, 98 Intracellular lipid vesicles, 209 Intracellular signal transduction, 11 focal adhesion kinase, 12 mitogen-activated protein kinase cascade, 13 nitric oxide (NO) and Akt, 11 pathways, 477

Index Intracellular vesicular organelles, 51 Intracranial hypertension, 521 Intracranial pressure, 520 Intraparenchymal pseudocysts, 432 Intrapulmonary prostaglandins, 344 Intratracheal endotoxin instillation, 141 Intravital microscopy, 28 video, 393 Ion channels, 6 mechanogated, 75 Ionic surfactant, 192 Ionomycin, 689 Ischemia-reperfusion–induced ALI, 350 Ischemic heart disease, 615 Ischemic optic neuropathy decompression trial, 539 Ischemic-reperfusion injury, 354 Isolated, ventilated, and perfused (IVP) system, 105 Isoprostane, 351 Isoproterenol, 688 Isotope coded afﬁnity tag (ICAT), 331

Junctional catenins, 73

Keratinocyte growth factor (KGF), 83, 385 Kerley’s lines, 449 6-Keto-prostaglandin f1-alpha, 662 Kinetic bed therapy, 614 Kl-6 protein, 320 Klebsiella pneumoniae, 275 Knockout and transgenic technologies, 407, 410

Lactate dehydrogenase (LDH), 484 Lactosyl ceramide, 54 Lamellar bodies (LB), 678 Lamellar units, 4 Laminin, 7 Laplace’s law, 2, 680 L-arginine/NO pathway, 5 Laser confocal microscopy, 29 Law of Laplace. See Laplace’s law.

729 Leukocyte adhesion protein, 658 Leukocyte sequestration, 139, 225, 234 Leukotrienes, 689 L-glucuronolactone oxidase, 549 Ligation stroke model, 657 Linkage mapping, 405 Lipid hydroperoxides, 10 microdomains, 54 proﬁle, neutral, 681 solubility, 660 trafﬁcking, 124, 385 nonsecretory, 54 Lipid–protein interactions, 51 Lipid–protein mixture, 36 Lipopolysaccharide (LPS), 207, 231 recognition molecule, 270 Liposome physics, 51 Liquid lining viscosity, 166 Liquid ventilator, 667 Liqui-Vent1, 143 Lobectomy, 526 Logistic regression, 647 Longitudinal relaxation, 452 Low peak pressure/stretch protocol, 245 Low tidal volume ventilation, 520, 530, 534, 617 Low volume injury cytopathologic sequelae of, 46 Lubrication ﬁlm, 180 Luciferase reporter gene assay, 421 Lung antioxidant function, augmentation of, 715 architecture, 27 cell types, 209 alveolar and bronchial epithelial cells, 209 bronchial and vascular smooth muscle cells, 209 ﬁbroblasts, 209 macrophages and monocytic cells, 209 microvascular endothelial cells, 209

730 [Lung] distention, 71 passive effects of, 71 expansion, heterogenous, 29, 30 ﬁbroblasts, 390 ﬁbrosis, 438 hypoplasia, 667 inﬂammation, 207, 238 inﬂation, 656 inhomogeneity, inﬂuence of, 126 injury, 46 effect on Crs and position of UIP, 130 modulation of, 341 overdistension, 661 unilateral saline lavage–induced, 412 instability, 136 ischemia, 433 mechanical properties, 143 mechanics during VILI, 119 model, nonperfused, 392 morphology, 28 neutrophil inﬁltration of, 663 normal, micromechanics of, 26 parenchyma, 436 parenchymal strain, 26 protein leak, 76 recruitment technique, 552 reperfusion injury, 661 tissue density, 101 ultrastructure, 699 Lung-protective ventilation, 498, 500, 520, 527, 530 strategy, 324, 480 mechanical, 475 Lymphotactin-a/XCL1, 242 Lymphotactin-b/XCL2, 242

Macrophage inﬂammatory protein (MIP)-2, 206 Macrophages, 476 Mannitol, 657 MAP kinase cascade, 14, 15 MAP kinase kinase kinase (MEKK), 12 Marangoni ﬂow, 187 Marangoni stress, 187

Index Marker protein, 54 Markers, 320 biochemical, 485 biological, examples of, 317 of VILI, 315 recent progress in, 318 of apoptosis, 322 of bacterial products, 332 of inﬂammation, 320, 406 of repair, 323 physiological, 316 structural, 318 Mass spectroscopy, 330 Matrix-assisted laser desorption, 331 Matrix-assisted laser ionization, 331 Matrix metalloproteinases (MMPs), 6, 84, 209 Mead equation, 105 Mean plateau airway pressures, 529 Mean plateau pressure, 345 Mechanical injury, cyclic nucleotide modulation of, 81 Mechanical resonance–active nuclides, 452 Mechanical stress cellular plasma membrane response to, 45 deﬁnitions of, 170 types of from airway opening, 172 normal, 170 tangential (or shear), 174 Mechanical stress–related genes, 413 Mechanical ventilation (MV), 205, 207, 267, 540, 545, 621, 630 computerized protocol instructions, 545 conventional, 229, 699 impact of, 546 modes of, 630 physiological effects of, 269 strategy, 563 systemic effects of, 267 Mechanical ventilation–associated injury, 21

Index Mechanosensing, 209, 210 Mechanosensitive channels, types of shear-activated potassium, 8 stretch-activated, 8 Mechanotransduction, 1, 212, 225, 379, 699 pathways, 499 Meconium aspiration, 661, 667 syndrome, 660 Mediastinum, 34, 436 Membrane depolarization, 8 disruption, 384 nontethered components, 174 oxidase NADH/NADPH, 10 signal transduction, 6 Membrane type 1 MMP (MT1-MMP), 84 Meniscus curvature, 167 Meniscus-frame streamlines, 192 Mesenteric ischemia-reperfusion, 351 Messenger RNA (mRNA), 464 transcripts, 212 Meta-analysis, 527, 530, 531 Metabolic acidotic reperfusates, 354 Metalloproteinase activation, 11 mechanical strain by, 83 Metallothionein, 411 Micropet scanners, 454 Microtubule assembly, 387 Microvascular barrier, 121 Microvascular environment, cyclic effect on, 110 Microvascular permeability, 69 defect of, 122 endothelial, 122 Microvascular pressures, 124 Mitogen-activated protein (MAP), 80 Mitogen-activated protein kinases (MAPK), 12, 210, 389 Mitogenic stimulation, 658 Mitral stenosis, 47 Molecular imaging methods, 448 inﬂammation imaging, 461 transgene expression imaging, 461

731 [Molecular imaging methods] platforms, 462 strategies, 464 Monocyte chemotactic protein (MCP)-1, 207, 214 Monolayer dynamics, 190 Morbid obesity, 34 Moricizine, 550 MRI experiments, types of, 452 gradient-echo, 452 spin-echo, 452 Mucus glycoprotein (MUC-1), 320 Multifactorial injury, 86 Multilayer dynamics, 190 Multimodality imaging sessions, 448 Multiorgan dysfunction syndrome (MODS), 158, 224, 267, 278 Multiorgan failure syndrome (MOSF), 158, 391 Multiple inert gas elimination technique (MIGET), 346 Multiple trauma, 525 MV. See Mechanical ventilation. Myocardial contractility, 347, 521 Myocardial infarction, 408, 550, 563, 612 Myosin heavy chain, 8 Myosin light chain (MLC), 74 phosphorylation, 80 Myosin light chain kinase (MLCK), 74, 409 gene, 405

Na,K-adenosine triphosphatases, 713 Naloxone, 540 Near-infrared ﬂuorophores, 463 Necrosis, 322 Negative airway pressure, 447 Negative pleural pressure, 24 Neonatal respiratory distress syndrome, 656 Neurohumoral abnormalities, 4 Neurokinins, 85 Neutrophil elastase, 83

732 [Neutrophil] sequestration, 245, 663 transendothelial diapedesis of, 47 Neutrophilic alveolitis, 418 N-nitroso-N-methylurethane, 392 application, 685 Non-cell contact–dependent pathways, 58 Non-clathrin–dependent endocytosis, 54 Nonequilibrium normal stress, 188 Noninvasive mechanical ventilation (NIMV), 629 Noninvasive ventilation, 613, 629, 647 No-reﬂow phenomenon, 658 Normocapnia, 342, 358 Normocapnic acidosis, 354 Nosocomial infection, 224, 665 Nuclear factor kB (NFkB), 58

O-15-labeled water, 455 Oleic-acid injury, 663 model, 455, 663 Open lung concept, 230 Organ dysfunction, nonpulmonary, 504 Organ injury, acute brain, 347, 354 myocardial, 354 Ortholog database, 413 Osmolar loads, 364 Osteoblasts, 555 Overdistension. See Overinﬂation. Overinﬂation, 47, 71, 120, 433, 699 Ovid1, 565 Oxygen toxicity, 699 Oxygenation support, 500 Oxygen-free radicals, 10 OxygentTM, 657 Oxyluciferin, 463

Pancreatitis, acute, 246 PaO2, arterial oxygen pressure, 524, 528 PaO2/FiO2 ratio, 277, 524, 569 Paracellular permeability, 72, 385

Index [Paracellular permeability] modulators of, 81 Parenchyma, 28 Parenchymal tethering, 168 Parietal tension, 2 Partial liquid ventilation (PLV), 143, 656, 662 Pasteurella multocida, 665 Paxillin, 74, 210 PBEF. See Pre-B-cell colony-enhancing factor. PC. See Phosphatidylcholine. Peak airway pressures, 521 Peak inﬂation pressures (PIP), 69, 223 ventilation, 76 Peak inspiratory pressure, 520 Perﬂubron, 143 Perﬂuorocarbon-associated gas exchange (PAGE), 656 Perﬂuorocarbons, 655 aerosol therapy, 663, 667 effects on inﬂammation and oxidative injury, 656 emulsion blood substitutes, 656 emulsions, 657 liquids, 656 micelles, 657 partial liquid ventilation, 666 tidal liquid ventilation (TLV), 665 vapor therapy, 667 ventilation techniques, 665 Perﬂuorohexane, 667 Perﬂuorooctylbromide (PFOB), 657 Peritonitis, 437 Perivascular hemorrhage, 47 PET. See Positron emission tomography. Phagocytosis, 680 Phosphatidylcholine (PC), 54, 684 dipalmitoylated species of, 681 Phosphatidylethanolamine (PE), 678 Phosphatidylglycerol (PG), 678 Phosphatidylinositol (PI), 678 Phosphatidylinositol-3-kinase, 11 Phosphatidylserine (PS), 678

Index Phosphocholine, 715 Phosphocholine cytidylyltransferase (CTP), 715 Phosphodiesterase inhibitor, 382 Phosphodiesterase IV inhibitor, 82 Phosphoinositide 3-kinase (PI3K)mediated pathway, 389 Phosphoinositol-3 kinase (PI3K) target, 73 Phospholipase, 477 Phospholipase A2 (PLA2), 75 Phospholipase C (PLC), 75 Phospholipase-dependent signaling pathways, 51 Phospholipid, 656, 684 content, 189 Phosphorylation signaling, 79 Photomultiplier tubes, 454 Physico-chemical hydrodynamics, 162 Pilocarpine, 689 Plasma membrane (PM) lesions, 58 lipid bilayer, 50 phosphatidylinositol 4,5-bisphosphate (PIP2), 51 repair, 55 self-sealing, 55 tension, determinants of, 50 trafﬁcking, 384 wounding effects, on gene expression and cell survival, 57 Plasma protein leakage, 682 Plasminogen activator, 411 Plasminogen activator inhibitor type 1 (PAI-1), 411 Plateau airway pressure, 277 Platelet activating factor, 477 Platelet aggregation, 657 Platelet-activating factor antagonist (BN 52021), 540 Platelet-derived growth factor (PDGF), 9 receptor-a, 10 Pleural pressure, 22, 167 gradient, 101 PM. See Plasma membrane. Pneumatoceles, 441 Pneumomediastinum, 699

733 Pneumonectomy, 526 Pneumothorax, 143, 441, 525, 699 Polymorphic variants, 407 Polymorphonuclear neutrophil (PMN), 126, 206 Positional cloning, 405 Positive end-expiratory pressure (PEEP), 30, 46, 119, 159, 229, 380, 497, 684 effect of, 141, 142 Positive pressure ventilation, 113 techniques, 498 Positron emission tomography (PET), 453 Positron-emitting isotope, 452 Positron-emitting isotope oxygen-15 (H215O), 452 Positron-emitting radionuclide-labeled tracer, 467 Postalveolar vascular pressure, effect on development of VILI, 111 Postnatal hemodynamic adaptation, 351 Post-traumatic pulmonary insufﬁciency, 341, 365 Pre-B-cell colony-enhancing factor (PBEF), 418, 420 genotyping, 420 Premature ventricular contractions (PVCs), 550 Prerandomization method, 639 Pressure gradients, 175 Pressure–basement membrane area, 28 Pressure–volume (PV) curve explanations of, 131 lower inﬂection point (LIP) of, 128, 160, 521 upper inﬂection point (UIP) of, 128, 160 Preterm newborn, 234 Proapoptotic cascades, 11 Proapoptotic mediators, 393 Proapoptotic stimuli, 58 Procollagen peptides, 323 Proinﬂammatory mediators, 207, 390

734 Prone positioning, 144 Propidium iodide (PI), 48 Prostaglandins, 689 Protective ventilation, 478 Protective ventilator strategy, 230, 270, 277 Protein channel, 51 ﬁlaments, types of intermediate ﬁlaments, 6 map, 330 microﬁlaments, 6 microtubules, 6 phosphorylation, 75, 389 reﬂection coefﬁcient, 122 spots, 331 tracer, systemic, 136 Protein- and energy-dependent transport mechanism, 54 Protein kinase C (PKC)-dependent mechanism, 9 Proteolysis, 331 Proteomics, 330 Protocol decision support, 556 Protocol-directed patient management, 537 Pseudocysts, 433, 438 Pseudomonas aeruginosa, 665 Pulmonary artery balloon occlusion pressure (PAOP), 539 Pulmonary artery catheter, 648 catheterization, 611 pressures, 99 resistance, 351 Pulmonary aspirate chemokine levels, 484 Pulmonary capillary endothelium, 158 Pulmonary capillary wedge pressure (PCWP), 521 Pulmonary edema, 119, 229, 358, 663 acute, 120 noncardiogenic, 34, 223 Pulmonary ﬂuid–structure interactions organ level (macroscale), 162 tissue level (microscale), 162, 170

Index Pulmonary gas trapping, 657 Pulmonary hypertension, 346, 520, 631 Pulmonary imaging, 458 sliding, 458 Pulmonary ischemia-reperfusion, 351 Pulmonary lesions, 432 Pulmonary mechanics, 163 Pulmonary micromechanics, 21 Pulmonary microvascular ﬁltration coefﬁcient, 350 Pulmonary surfactant, 158, 677 composition of, 677 function of, 680 metabolism of, 678 properties of, 158 protective effect of, 185 role in VILI, 684 Pulmonary vascular resistance (PVR), 98, 344 Pulmonary vascular tree, 98 pulmonary expansion effect on, 98 Pulmonary venous pressures, 99 Pulsating bubble surfactometer (PBS), 189 Purinergic agonists, 75, 689 Pyruvate ratio, 269

Quality chasm, in critical care, 611 Quantitative trait loci (QTLs), 406, 409

Radiolabeled antisense oligodeoxynucleotide (RASON), 464 Radiolabeled protein, 457 Radiotracer, 455 imaging, 464 Randomization, 563 Randomized controlled trials (RCTs) core questions in, 636 meta-analyses of, 527 of tidal volume reduction, 521, 628, 635 RCTs. See Randomized controlled trials. Recruitment/derecruitment damage, 190

Index Regional air content, 25 Relative risk reduction (RRR), 644 Renal cell apoptosis, 392 Reperfusion, 351 Reperfusion-induced lung injury, 663, 664 Reporter transgene, 467 RES. See Reticuloendothelial system. Rescue therapy, 542 Resonant frequency, 451 Respiratory acidosis, 347 beneﬁcial and adverse effects of, 512 Respiratory distress syndrome (RDS), 157, 356 Respiratory gating, 452 Respiratory rate (RR), 233 Respiratory syncytial virus (RSV), 663 Respiratory system compliance (CRS), 128 Respiratory therapists, 617 Reticuloendothelial system (RES), 319, 657 blockade, 657 clearing capacity, 657 Reverse transcriptase-polymerase chain reaction (RT-PCR), 412 RGD peptide-covered ferromagnetic beads, 9 RGD peptides, 7 RhoA kinase, 15 Rigid reopening model, 184 RNA-binding factor, 15

Saline lavage, 667 Saline lavage-induced surfactant depletion/dysfunction, 661 Scintillation materials, 454 Scurvy, 549 Secondary end points, 634 Sensitive factor attachment protein receptors (SNAREs), 56 Sepsis syndrome, 316 Shc, adapter protein, 8 Shear force, endothelial response to, 100 Shear stress gradients, 175

735 [Shear stress] responding element, 212 Shear stress-activated cytoskeletal remodeling, 100 Shear stress-dependent remodeling, 5 Shrunken lung, 130 Shunt fraction, 316 Siemens1 ventilator, 574 Signal-to-noise (S/N) ratio, 550, 561 Single nucleotide polymorphism (SNP), 407 screening, 422 Single-photon-emission detection systems, 452 Small guanosine triphosphatase (GTPase), 79 Smooth muscle cell (SMC), 3 hypertrophy of, 4 polyploidy of, 4 Sodium dodecyl (lauryl) sulfatepolyacrylamide gel electrophoresis (SDS-PAGE), 419 Sodium dodecyl sulfate (SDS), 192 Sol–gel transition, 167 Spectrin, 73 Sphingolipids, 54 Sphingomyelin (SM), 53, 678 Splanchnic circulation, 347 Splanchnic organs, 269 Squeezeout theory, 189 Src kinase, 210 Stable-peeling reopening, 168 Static compliance, 316 Strain-survey approach, 410 Stress exposure duration, 181 ﬁbers, 210 Stress-activated protein kinases, 14 Stress-bearing elements, 46 Stretch responding element, 212, 215 Stretch-activated cation channels (SACC), 75 Stretch-induced endocytosis, 53

736 Stretch-induced lipid trafﬁcking, 53 Stretched pore phenomenon, 124 Subcortical cytoskeleton, 50 Subglottic secretion drainage, 614 Subpleural alveoli, 29, 385 Subpleural lung regions, 432 Substernal diaphragm-apposed regions, 34 Suction pressure, 24 Superoxide anion (O2), 10 Superoxide dismutase, 10 Supravalvular aortic stenosis (SVAS), 550, 551 Surface active agent, 684 Surfactant aggregates, 678 alterations, in ALI/ARDS, 681 apoprotein, 678 compounds, damage of, 682 conversion, 32 deﬁciency, 661 depletion, 353 inactivation, 35, 686 large, 678 production, improvement of, 714 proteins, 318, 406 replacement therapy, in ALI/ARDS, 682 small, 678 transbronchial treatment, 687 Surfactant-associated proteins, 189 Surfactant-deactivated lungs, 131 Sympathetic stimulation, 270 Synaptotagmins, 57 Synchronized intermittent mandatory ventilation (SIMV), 520 Systemic inﬂammatory response syndrome (SIRS), 224

Talin, 74, 210 T-allele frequency, 421 Teleinspiratory volume, 142 Tensin, 210 Thrombin, 477 Thromboembolic disease, 547

Index Thrombolytic therapy, 615 Thrombolytics, 612 Thromboxane A2, 662 Tidal inﬂation, 488 Tidal liquid ventilation, 656, 662 Tidal ventilation, 131, 206, 501 cycle, 166 Tidal volume, 498, 527, 630 local, 36 role in VILI reduction, 497 ventilation, 387, 525 Time points, types of, 633 composite end point, 633 primary end point, 633 surrogate end point, 633 Tissue buckling, 27 factor, 411 hypoxia, 269 remodeling, 172 Tissue-type plasminogen activator (t-PA), 270 TNF soluble receptor-1 (TNF-r), 391 TNF-a-converting enzyme (TACE), 228 Toll-like receptor-4 (TLR-4), 273 Toll-like receptors, 332 Total lung capacity (TLC), 28 Total parenteral nutrition (TPN), 616 Tracer speciﬁc activity, 454 Tracheal aspirate cytokines, 489 Tracheal gas insufﬂation, 667 Transactional unit, 553 Transcellular pathways, 72 Transcription inhibitor, 212 Transepithelial osmotic gradient, 713 Transgenes, 464 expression, 716 Transient cutaneous ﬂushing, 657 Transient receptor potential (TRP), 76 Transmural pressure, 3, 25, 125 Transpulmonary pressure, 125, 380 Transvascular permeability, 79 Traveling-wave region, 181 Tri-amino acid motif, 243 Tricuspid regurgitation jet, 459 Triple collagen helix, normal, 550

Index Tris-hydroxymethyl aminomethane, 347 Tromethamine buffering, 346 Troponin, 318 Tumor necrosis factor (TNF)-a, 206, 227–229 Two-dimensional electrophoresis, 330 Type II alveolar epithelial cells, 46 Type II pneumocyte architecture, 663 Type IV collagen ﬁbers, 47 Tyrode’s solution, 99 Tyrosine kinase inhibition, 80 receptors, 6 Urea nitrogen, 484 Urokinase receptor (UPAR), 411 Utah Clinical Trial Toolbox, 545 bedside protocols, 588 blood glucose/IV insulin protocol, 584 education and training tools, 585 electronic tools, 575 hemodynamic support protocol, 578 Web-based tools, 584 VALI. See Ventilator-associated lung injury. Vascular derecruitment, 111 Vascular endothelial growth factor (VEGF), 83, 101, 270, 714 Vascular endothelial-cadherin, 73, 79 Vascular hemorrhage, 80 Vascular pathway, segments of, 98 arterial, 98 intermediate, 98 venous, 98 Vascular permeability, 72 active endothelial control of, 72 Vascular recruitment, 111 Vascular remodeling, 3 Vascular resistance systemic, 346 Vascular smooth muscle cell (VSMC), 5 Vascular tracer, 381 Vasoactive agonists, 1

737 Vasoconstriction, systemic, 699 Vasodilator, 11 Vasodilatory effect, 270 Vasopressin, 689 Vasopressor requirements, 526 Venous capillary pressure, 630 Venticute, 683 Ventilation efﬁcacy, 192 Ventilation-induced lung injury (VILI), 45, 69, 205, 404, 660 cellular and molecular basis for, 205 clinical considerations of, 145 gene expression in animal models, 409 gene therapy for, 711 genetics, 406 histology of, 46 mechanisms, 32, 499 ortholog gene database in, 412 prevention of, 627 reduction of, 677 Ventilation-perfusion ratios, 499 Ventilator management, 447 Ventilator-associated lung injury (VALI), 145, 412 gene expression in animal models, 409 genetics and the candidate gene approach, 406 genetics of, 404 ortholog gene database in, 412 preliminary IL-6 genotyping in, 423 Ventilator-associated systemic inﬂammation, 476 Ventilator-induced alveolar epithelial permeability, 379 Ventilator-induced lung inﬂammation, 206 Ventilatory cycle, 131 Ventilatory modes, 630 Ventilatory patterns, 71 Ventilatory stress, 103 Ventricular ﬁbrillation, 550 Ventricular tachycardia, 550

738 Verapamil, 8 Vesicles, patch of, 55 Vesicular organelles, 56 Vesicular trafﬁcking response, 51 Vessel elongation, 98 VILI. See Ventilation-induced lung injury. Vinculin, 74 Visible light ﬂuorophores, 463 Vitronectin, 7 Volutrauma, 143, 169, 268, 344, 699 von Willebrand factor antigen, 319 Wall compliance, 166 Weaning rate, 521

Index Weibel–Palade bodies, 320 West zone I, 99 Wet lung, 661 Wilson–Bachofen model, 30

Xa factor, 477 Xanthine oxidase, 350

Yield pressure, 167 Young–Laplace equation, 186

Zero end-expiratory pressure (ZEEP), 141, 477

Figure 18-2 Effect of state of lung inflation on the measured density of lung parenchyma in each volume element (voxel) of an image, at TLC, FRC during atelectasis (collapse), or in the presence of alveolar edema. (See page 450.)

Figure 18-4 Lung water images from a normal rat and a rat treated with i.v. oleic acid obtained using a Concorde R4 microPET scanner after tail vein injection of approximately 10 mCi of O-15–labeled water. (See page 455.)

Figure 18-7 Example of multimodality inflammation imaging in a mouse with experimental right lung pneumonia. (See page 462.)

Lung Biology in Health & Disease Volume 215 Ventilator-Induced Lung Injury

Proteoglycans in Lung Disease (Lung Biology in Health and Disease)

Lung Biology in Health & Disease Volume 186 Pleural Disease

Lung Biology in Health & Disease Volume 195 Asthma Prevention

Lung Biology in Health & Disease Volume 176 Non-Neoplastic Advanced Lung Disease

Lung Biology in Health & Disease Volume 180 Venous Thromboembolism

Lung Development and Regeneration (Lung Biology in Health and Disease)

Lung Biology in Health & Disease Volume 151 Fetal Origins of Cardiovascular and Lung Disease

Lung Biology in Health & Disease Volume 211 Tropical Lung Disease 2nd Edition

Lung Biology in Health & Disease Volume 210 Sarcoidosis

Lung Biology in Health & Disease Volume 176 Non-Neoplastic Advanced Lung Disease

Lung Biology in Health & Disease Volume 184 Lung Volume Reduction Surgery for Emphysema

Lung Biology in Health & Disease Volume 157 Respiratory-Circulatory Interactions in Health & Disease

Lung Biology in Health & Disease Volume 201 Lung Surfactant Function and Disorder

Lung Biology in Health & Disease Volume 200 Functional Lung Imaging

Lung Biology in Health & Disease Volume 160 Gravity and the Lung: Lessons from Microgravity

Lung Biology in Health & Disease Volume 162 Drug Delivery to the Lung

Lung Biology in Health & Disease Volume 162 Drug Delivery to the Lung

Monitoring Asthma (Lung Biology in Health and Disease, Volume 207)

Severe Pneumonia (Lung Biology in Health and Disease, Volume 206)

Lung Biology in Health and Disease Volume 153 Environmental Asthma

Bronchopulmonary Dysplasia, Volume 240 (Lung Biology in Health and Disease)

Lung Biology in Health & Disease Volume 135 Control of Breathing in Health and Disease

Pleural Disease, Second Edition (Lung Biology in Health and Disease)

Lung Biology in Health & Disease Volume 146 Sleep Apnea: Implications in Cardiovascular and Cerebrovascular Disease

Lung Biology in Health & Disease Volume 191 Long-Term Intervention in Chronic Obstructive Pulmonary Disease

Lung Biology in Health & Disease Volume 182 Pharmacotherapy in Chronic Obstructive Pulmonary Disease

Lung Biology in Health & Disease Volume 170 Disease Markers in Exhaled Breath

Lung Biology in Health & Disease Volume 156 Genetic Models in Cardiorespiratory Biology

Lung Biology in Health & Disease Volume 181 Upper and Lower Respiratory Disease

Lung Biology in Health & Disease Volume 183 Acute Exacerbations of Chronic Obstructive Pulmonary Disease

Lung Biology in Health & Disease Volume 215 Ventilator-Induced Lung Injury

Proteoglycans in Lung Disease (Lung Biology in Health and Disease)

Lung Biology in Health & Disease Volume 186 Pleural Disease

Lung Biology in Health & Disease Volume 195 Asthma Prevention

Lung Biology in Health & Disease Volume 176 Non-Neoplastic Advanced Lung Disease

Lung Biology in Health & Disease Volume 180 Venous Thromboembolism

Lung Development and Regeneration (Lung Biology in Health and Disease)

Lung Biology in Health & Disease Volume 151 Fetal Origins of Cardiovascular and Lung Disease

Lung Biology in Health & Disease Volume 211 Tropical Lung Disease 2nd Edition

Lung Biology in Health & Disease Volume 210 Sarcoidosis

Lung Biology in Health & Disease Volume 176 Non-Neoplastic Advanced Lung Disease

Lung Biology in Health & Disease Volume 184 Lung Volume Reduction Surgery for Emphysema

Lung Biology in Health & Disease Volume 157 Respiratory-Circulatory Interactions in Health & Disease

Lung Biology in Health & Disease Volume 201 Lung Surfactant Function and Disorder

Lung Biology in Health & Disease Volume 200 Functional Lung Imaging

Lung Biology in Health & Disease Volume 160 Gravity and the Lung: Lessons from Microgravity

Lung Biology in Health & Disease Volume 162 Drug Delivery to the Lung

Lung Biology in Health & Disease Volume 162 Drug Delivery to the Lung

Monitoring Asthma (Lung Biology in Health and Disease, Volume 207)

Severe Pneumonia (Lung Biology in Health and Disease, Volume 206)

Lung Biology in Health and Disease Volume 153 Environmental Asthma

Bronchopulmonary Dysplasia, Volume 240 (Lung Biology in Health and Disease)

Lung Biology in Health & Disease Volume 135 Control of Breathing in Health and Disease

Pleural Disease, Second Edition (Lung Biology in Health and Disease)

Lung Biology in Health & Disease Volume 146 Sleep Apnea: Implications in Cardiovascular and Cerebrovascular Disease

Lung Biology in Health & Disease Volume 191 Long-Term Intervention in Chronic Obstructive Pulmonary Disease

Lung Biology in Health & Disease Volume 182 Pharmacotherapy in Chronic Obstructive Pulmonary Disease

Lung Biology in Health & Disease Volume 170 Disease Markers in Exhaled Breath

Lung Biology in Health & Disease Volume 156 Genetic Models in Cardiorespiratory Biology

Lung Biology in Health & Disease Volume 181 Upper and Lower Respiratory Disease

Lung Biology in Health & Disease Volume 183 Acute Exacerbations of Chronic Obstructive Pulmonary Disease

Recommend Documents