The Artificial Lung

TISSUE ENGINEERING INTELLIGENCE UNIT 7

Steven N. Vaslef and Robert W. Anderson

The Artificial Lung

TISSUE ENGINEERING INTELLIGENCE UNIT 7

The Artificial Lung Steven N. Vaslef, M.D., Ph.D. Associate Professor of Surgery and Biomedical Engineering Director, Trauma Services Chief, Section of Trauma and Critical Care Division of General Surgery Duke University Medical Center Durham, North Carolina, U.S.A.

Robert W. Anderson, M.D. The David Sabiston, Jr. Professor of Surgery Professor of Biomedical Engineering Chairman, Department of Surgery Duke University Medical Center Durham, North Carolina, U.S.A. Landes Bioscience GEORGETOWN, TEXAS U.S.A.

Eurekah.com AUSTIN, TEXAS U.S.A.

THE ARTIFICIAL LUNG Tissue Engineering Intelligence Unit Eurekah.com Landes Bioscience Designed by Jesse Kelly-Landes Georgetown, Texas, U.S.A. Copyright ©2002 Eurekah.com All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Printed in the U.S.A. Please address all inquiries to the Publishers: Eurekah.com / Landes Bioscience, 810 South Church Street, Georgetown, Texas, U.S.A. 78626 Phone: 512/ 863 7762; FAX: 512/ 863 0081 ISBN: 1-58706-019-1 hard cover ISBN: 1-58706-167-8 soft cover While the authors, editors and publisher believe that drug selection and dosage and the specifications and usage of equipment and devices, as set forth in this book, are in accord with current recommendations and practice at the time of publication, they make no warranty, expressed or implied, with respect to material described in this book. In view of the ongoing research, equipment development, changes in governmental regulations and the rapid accumulation of information relating to the biomedical sciences, the reader is urged to carefully review and evaluate the information provided herein.

Library of Congress Cataloging-in-Publication Data The artificial lung / [edited by] Steven N. Vaslef, Robert W. Anderson. p. ; cm. -- (Tissue engineering intelligence unit ; 7) Includes bibiliographical references. ISBN 1-58706-019-1 (alk. paper) 1. Artificial organs. 2. Oxygenators. 3. Biomedical Engineering. WF 600 A791 2002] I. Vaslef, Steven N. II. Anderson, Robert W., MD III. Series. RD130 .A76 2002 617.5´420592--dc21 2001002121

CONTENTS Preface ................................................................................................. vii 1. Development of an Implantable Artificial Lung ..................................... 1 Steven N. Vaslef and Robert W. Anderson Introduction .......................................................................................... 1 Historical Aspects of Implantable Artificial Lung Development ............ 2 Design of an Implantable Artificial Lung ............................................... 5 In Vitro Evaluation ............................................................................. 14 In Vivo Evaluation .............................................................................. 19 Summary ............................................................................................. 22 2.

Rationale for an Implantable Artificial Lung ........................................ 24 William R. Lynch and Robert H. Bartlett Introduction ........................................................................................ 24 History of Implantable Artificial Lung Development .......................... 25 Design and Application of an Implantable Artificial Lung ................... 27

3.

Engineering Design of Thoracic Artificial Lungs .................................. 33 Lyle F. Mockros and Keith E. Cook Introduction ........................................................................................ 33 General Design Considerations ........................................................... 34 Attachment Modes .............................................................................. 37 Specific Design Considerations ............................................................ 39 Effects of Pulsatility ............................................................................. 56 Discussion ........................................................................................... 59 Acknowledgement ............................................................................... 63 4. Biocompatibility of Artificial Lungs ..................................................... 65 Keith E. Cook and Lyle F. Mockros Introduction ........................................................................................ 65 Initiation of Blood-Artificial Surface Reaction: Protein Adsorption .......................................................................... 66 Platelet Activation ............................................................................... 71 Leukocyte Activation and the Whole Body Inflammatory Response .... 72 Shear Stress Activation ........................................................................ 74 Organ and System Level Effects ........................................................... 75 Blood Activation During Artificial Lung Use ...................................... 81 Combatting Blood Activation .............................................................. 82 Summary ............................................................................................. 87 5. Testing and Performance Evaluation of Artificial Lungs ...................... 98 Akhil Bidani, Weike Tao and Joseph B. Zwischenberger Introduction ....................................................................................... 98 Gas Exchange in “Artificial Lungs” Versus Human Lung .................... 99 Mass Transfer in Artificial Lungs ....................................................... 100 Measurement of Gas Transfer in Artificial Lung Devices ................... 102 Intracorporeal Gas Exchange ............................................................. 105

Extracorporeal Gas Exchange ............................................................ 119 Arteriovenous extracorporeal CO2 removal ........................................ 122 Conclusion ........................................................................................ 129 Acknowledgments ............................................................................. 130 6. Gas Exchange in the Venous System: Support for the Failing Lung ... 133 Brack G. Hattler and William J. Federspiel Introduction ...................................................................................... 133 Background ....................................................................................... 135 Clinical Trial of an Intravenous Respiratory Support Device ............. 141 The Hattler Respiratory Support Catheter ......................................... 142 Hattler Catheter Development .......................................................... 161 Animal Tests of the Hattler Catheter ................................................ 167 Acknowledgements ............................................................................ 171 Index .................................................................................................. 175

EDITORS Steven N. Vaslef, M.D., Ph.D. Associate Professor of Surgery and Biomedical Engineering Director, Trauma Services Chief, Section of Trauma and Critical Care Division of General Surgery Duke University Medical Center Durham, North Carolina, U.S.A. Chapter 1

Robert W. Anderson, M.D. The David Sabiston, Jr. Professor of Surgery Professor of Biomedical Engineering Chairman, Department of Surgery Duke University Medical Center Durham, North Carolina, U.S.A. Chapter 1

CONTRIBUTORS Robert H. Bartlett General and Thoracic Surgery The University of Michigan Medical Center Ann Arbor, Michigan, U.S.A. [email protected]

William J. Federspiel McGowan Center for Artificial Organ Development University of Pittsburgh Pittsburgh, Pennsylvania, U.S.A. [email protected]

Chapter 2

Chapter 6

Akhil Bidani The University of Texas Medical Branch at Galveston Galveston, Texas, U.S.A. [email protected]

Brack G. Hattler McGowan Center for Artificial Organ Development University of Pittsburgh Pittsburgh, Pennsylvania, U.S.A. [email protected]

Chapter 5

Chapter 6

Keith E. Cook Department of Surgery Northwestern University Chicago, Illinois, U.S.A. [email protected] Chapters 3 and 4

William R. Lynch Milton S. Hershey Medical Center Pennsylvania State University Hershey, Pennsylvania, U.S.A. [email protected] Chapter 2

Lyle F. Mockros Biomedical and Chemical Engineering Northwestern University Evanston, Illinois, U.S.A. [email protected] Chapters 3 and 4

Joseph B. Zwischenberger Division of Cardiothoracic Surgery The University of Texas Medical Branch at Galveston Galveston, Texas, U.S.A. [email protected] Chapter 5

Weike Tao The University of Texas Medical Branch at Galveston Galveston, Texas, U.S.A. [email protected] Chapter 5

PREFACE Although artificial lungs have been used in clinical medicine since the late 1960s, the past decade has seen significant advances relative to artificial lung design, clinical applications and our understanding of blood-surface interactions. Artificial lung technology is now to the point that support of the failing lungs is possible, at least for a short period of time, for acute or chronic respiratory failure. Several approaches are being taken for the application of artificial lungs to the treatment of severe respiratory disease: extracorporeal circuitry, as in extracorporeal membrane oxygenation (ECMO) or arteriovenous carbon dioxide removal (AVCO2R); intravascular gas exchange, whereby a gas exchange device is placed in the vena cavae to affect oxygen and carbon dioxide transfer; and fully implantable artificial lung for total or partial respiratory support. This book attempts to synthesize the current developments in the field of artificial lungs. Contributions by experts in this area have resulted in the final product, the first book of its kind dedicated to artificial lung development. This book is divided into six chapters. Chapter one discusses the current status of the development of an implantable artificial lung, reviews the features of current designs and summarizes the results of recent in vitro and in vivo investigations. Chapter two discusses the rationale for an implantable artificial lung, provides a brief history of artificial lung development and outlines the current and potential applications for artificial lungs. Chapter three presents the engineering aspects of artificial lung design. A detailed discussion of the gas exchange requirements, the hemodynamic requirements and biocompatibility requirements is given, with emphasis on the design of a pumpless, implantable artificial lung. Chapter four is a thorough review of blood-artificial surface interactions and discusses current methods for making artificial lung surfaces more biocompatible. Chapter five describes the in vitro and in vivo testing and performance of artificial lungs and details the extensive experience of the authors in the laboratory and the clinical functional assessment of both intravascular and extracorporeal artificial lung devices. Chapter six is dedicated solely to intravascular lung assist devices. It focuses on the theoretical aspects, design and laboratory tests of a novel intravenous respiratory support catheter to provide partial support for acutely failing lungs. It is the intent of the editors that this book appeals to engineers and clinicians alike. Anyone interested in artificial lung development may find this book useful. Likewise, critical care practitioners keen on cutting edge technology may find their curiosity piqued by the innovations presented herein. Steven N. Vaslef Robert W. Anderson

CHAPTER 1

Development of an Implantable Artificial Lung Steven N. Vaslef and Robert W. Anderson

Introduction

A

n implantable artificial lung is under development for the treatment of patients with advanced respiratory failure. The device is intended primarily as a bridge to lung transplantation for patients with end-stage chronic pulmonary disease, but it may also be considered a treatment option for patients with severe acute respiratory failure. The ultimate goal of implantable artificial lung development is to provide total or near-total respiratory support for a period of several weeks or several months. This chapter reviews the impetus to develop an artificial lung, the historical aspects of implantable lung development, engineering design considerations, and results of preclinical studies. Why develop an artificial lung? The engineering challenges at first glance are arguably more complex than those associated with the development of other artificial organs, but these challenges, as will become clear in this chapter, are not insurmountable. Conceptually, an artificial lung, designed for several weeks to several months of near-total respiratory support, could be used either in advanced chronic or acute respiratory failure. In the case of chronic pulmonary failure, the artificial lung would be used as a bridge to lung transplantation. In the case of acute pulmonary failure, the artificial lung would be a modality providing pulmonary support until the lungs heal when more conventional approaches have failed. The only option at the present time for thousands of patients with chronic, end-stage lung disease is allogeneic lung transplantation. Table 1.1 lists the diagnoses most commonly leading to lung transplantation, the majority falling into the categories of emphysema/chronic obstructive pulmonary disease, cystic fibrosis, and idiopathic pulmonary fibrosis. However, the shortage of organ donors has not kept up with the demand for transplantable organs.1 As a result, as Figure 1.1 illustrates, the annual number of deaths of patients on the lung transplant waiting list is increasing. For instance, in 1998 there were 4557 patients on the lung transplant waiting list. Only 862 patients were transplant recipients, while 485 patients died while on the waiting list. This situation is not likely to improve in the near future, as the number of patients listed for lung transplantation continues to grow and the success of lung transplantation continues to improve. Although xenotransplantation is being actively investigated,2 the notion of transplanting animal

The Artificial Lung, edited by Steven N. Vaslef and Robert W. Anderson. ©2002 Eurekah.com.

The Artificial Lung

2

Table 1.1. Most common diagnoses leading to lung transplantation -Emphysema/COPD -Cystic Fibrosis -Idiopathic Pulmonary Fibrosis -Alpha-1-antitrypsin Deficiency -Primary Pulmonary Hypertension -Re-transplant/Graft Failure -Congenital Disease

organs into humans is still unproven and very controversial. The hurdles associated with cross-species immunologic barriers have, thus far, been difficult to overcome and it is unlikely that clinical xenotransplantation will become a reality in the next decade. Unless the pool of allogeneic organ donors increases, the demand for allogeneic lungs will continue to exceed the supply. An artificial lung, serving as a bridge to lung transplantation, may reduce the death rate among patients awaiting lung transplantation. Acute lung injury leading to the acute respiratory distress syndrome (ARDS) has a mortality rate of approximately 40 to 50 percent. A number of techniques of respiratory support (Table 1.2) have been tried, with variable success rates, to maintain adequate oxygenation and ventilation in patients with severe ARDS. These methods of respiratory support are generally used to either prevent further injury to the lungs from excessive airway pressures or volumes, or to maximize the recruitment of alveoli. There have been conflicting results regarding the efficacy of these treatments in ARDS, but a recent report suggests that low tidal volume ventilation may be associated with improved survival.3 Extracorporeal techniques, as well as intravenous membrane oxygenation, have failed to gain widespread popularity in the treatment of ARDS in adults for a number of reasons, including technical considerations, significant complication rates associated with the methodologies, and failure to demonstrate improved efficacy over conventional techniques.4,5 An implantable artificial lung may be able to harness the attractive features from some of these other modalities and become an alternative therapy for ARDS when other less invasive modalities have failed.

Historical Aspects of Implantable Artificial Lung Development The idea of an implantable artificial lung has been around for several decades. Bodell et al., in 1965,6 described an implantable booster lung that consisted of a teflon graft sutured to the side of the pulmonary artery and to the left atrium (Fig. 1.2). The inside of the graft contained silicone rubber capillary tubing (0.025 inch outer diameter) through which the oxygenating gas flowed. Gas exchange occurred as blood flowed through the graft, contacting the gas-permeable silicone rubber capillary tubes. Inlet and outlet gas lines exited through the chest wall. This early artificial lung, which was tested in sheep and dogs, demonstrated that measurable effects on systemic arterial blood gases could be achieved. However, the device was severely limited by its small gas-exchange surface area (approximately 0.06 m2), kinking of the graft, and thrombosis of the graft. This same group of investigators realized the limitations of their prototype device and redesigned an implantable artificial lung that had greater gas exchange surface area.7,8 The lung eliminated external connections and was tested under conditions of full anticoagulation. This device, shown in Figure 1.3, consisted of a number of silicone rubber

Development of an Implantable Artificial Lung

3

Table 1.2. Oxygenation and ventilation strategies in ARDS -Pressure control ventilation -Low tidal volume ventilation -High frequency oscillation -Jet ventilation -Inverse ratio ventilation -Liquid lung ventilation -Prone positioning -Inhaled nitric oxide therapy -Permissive hypercapnia -Extracorporeal membrane oxygenation (ECMO) -Extracorporeal carbon dioxide removal (ECCCO2R) -Arteriovenous carbon dioxide removal (AVCO2R) -Intravenous membrane oxygenation

capillary tube bundles all potted together in a manifold through which the blood flowed. A compliant “ventilating envelope” moved air in and out of an artificial bronchus with the normal motion of the chest wall and diaphragm. Gas exchange was effected as the air contacted the surface of the 0.025 inch outer diameter capillary tubes containing the flowing blood. As in their previous design, this artificial lung had the inflow and outflow blood connections to the pulmonary artery and left atrium. The device was implanted in the left chest following pneumonectomy. Oxygen transfer rates amounted to less than 15 ml/min. Although some of their devices had substantial membrane surface area, up to 0.95 m2, gas exchange was limited by the low blood flow rate, which was less than 200 ml/ min, low gas diffusion rates in the relatively large blood channels of the capillary tubes, and passive gas flow. Morin et al., in 1977,9 made a significant contribution to the design of artificial lungs by proposing that the blood flow be perpendicular to the oxygenating surface of the capillary tubes. Their device consisted of a set of screens composed of woven silicone rubber capillary tubes through which flowed oxygenating gas. The cross-flow configuration enhances convective mixing of the blood, thereby leading to higher gas transfer rates. These investigators recognized that the tightness of the weave, the number of silicone rubber screens, and the frontal area presented to the blood path were key features that influenced the gas transfer rates and the pressure loss across the device. They performed in vitro tests of devices containing up to 40 grids, but found that the blood-side pressure losses were excessive and overall gas transfer rates low. At a blood flow rate of 1,060 ml/min, a maximum oxygen transfer rate of 48 ml/min•m2 was achieved, but the blood-side pressure drop exceeded 50 mm Hg. They also outlined the specifications for an implantable artificial lung, many of which are still applicable today. Limitations of their design included small membrane surface area and relatively large capillary tubes (0.06 cm outer diameter). The narrow constraints imposed upon artificial lung developers by the limited types and sizes of gas-permeable capillary tubes were also evident in the work of Galletti et al.10 These investigators designed several implantable “booster lung” prototypes using parallel arrangements of coiled capillary tubes made of expanded microfibrillar polytetrafluoroethylene (PTFE). It was apparent that the pulsatile nature of pulmonary blood flow, as well as the serpentine blood flow path through the tubes, contributed to enhanced gas transfer rates. However, the very large inner diameter of the capillary tubes, ranging from

4

The Artificial Lung

Fig. 1.1. UNOS Registry data for lung transplantation showing the mismatch in the number of patients on the waiting list relative to the number of transplant recipients. The rise in patient deaths while awaiting transplant is also evident.

Fig. 1.2. Artificial booster lung described by Bodell6 in 1965, showing Teflon graft anastomosed to pulmonary artery and left atrium. Gas exchange occurred through ten foot lengths of silicone rubber capillary tubing coiled up inside the graft. Adapted from Bodell BR, Head JM, Head LR et al. An implantable artificial lung. JAMA 1965; 191:125-127.

Development of an Implantable Artificial Lung

5

Table 1.3. Design specifications for an implantable artificial lung -Capable of transferring > 200 ml/min of oxygen and carbon dioxide -Blood flow pattern resulting in efficient gas transfer and minimal shunting -Blood-side pressure loss < 15 mm Hg at blood flow rates (cardiac output) of 4-6 l/min -Gas-side pressures less than blood-side pressures to avoid gas embolism -Compliant housing chamber or connecting tubing to minimize impedance to pulsatile blood flow -Reliability and durability to function at unaltered performance for at least 2-3 weeks -Size and configuration to fit in the hemithorax without impingement of surrounding structures -Thromboresistant and otherwise biocompatible

0.20 to 0.29 cm, severely limited membrane surface area and adequate convective mixing at the blood-PTFE interface. The new era of artificial lung design was ushered in by the development of microporous polypropylene hollow fibers. This material, with a thin wall and a smaller outer diameter than the silicone rubber or PTFE tubes of the time, was manufactured with micropores in the wall of the fiber, each micropore measuring just a fraction of a micron in diameter and allowing for excellent gas transfer capability. The polypropylene hollow fibers had an outer diameter approximately half that of the silicone rubber tubes, thereby permitting the design and development of compact hollow fiber membrane oxygenators with large surface area and gas exchange capability.11 Higher rates of oxygen and carbon dioxide transfer, therefore, were achieved with smaller, more efficient devices than previously realized. Microporous polypropylene hollow fibers, over 15 years after their introduction in membrane oxygenators, remain the industry standard for the production of commercial oxygenators for cardiopulmonary bypass circuits. The ability to design a compact, efficient, implantable artificial lung now became more attainable than ever before.

Design of an Implantable Artificial Lung Basic Features The design specifications of an implantable artificial lung are outlined in Table 1.3. Utilizing currently available materials as the gas exchange surface, the implantable artificial lung that is compact, yet efficient, is composed of a bank of hollow fibers (e.g., microporous polypropylene) across which flows the blood and through which flows the oxygenating gas (Fig. 1.4). This cross-flow configuration, as long as the blood channels are not too wide and the blood path is tortuous, ensures good convective mixing at the boundary layer between blood and the fiber surface. The fiber bundle is encased in a housing material that optimally is nonreactive and will not erode into surrounding structures in the thoracic cavity. Blood enters the housed bundle, oxygenation and carbon dioxide removal occur as the blood passes through the fiber bank, and the oxygenated blood exits the fiber bundle on the opposite side. The fibers are potted at either end with a polyurethane-based compound that serves to keep the fibers together and to separate the gas and blood phases. The oxygenating gas enters a manifold at one end, distributes and flows through the lumens of the hollow fibers, enters a manifold at the other end, and exits. There are, therefore, four attachments to the artificial lung: two blood lines for the inlet (deoxygenated blood) and outlet (oxygenated blood), and two gas lines. The inlet and outlet blood

6

The Artificial Lung

Fig. 1.3. Early artificial lung design by Palmer et al.8 Inlet and outlet blood connections were made to the left pulmonary artery and the left atrium, respectively. Blood flowed on the inside of silicone rubber tubing assembled into multiple tubing modules. Gas flow, resulting from the normal motion of the chest and diaphragm, was passive through the silicone rubber “ventilating envelope”; the artificial bronchus exited through the chest wall and was open to the atmosphere. See text for details. Adapted from Palmer AS, Collins J, Head LR. Development of an implantable artificial lung. J Thorac Cardiovasc Surg 1973; 66:521-525.

Fig. 1.4. Fiber bank with overall height, H, width, W, and blood path length, L. The frontal area, Af, is the product of H and W. Blood flow rate is indicated by Qb and gas flow rate by Qg. Individual fibers have outer diameter, d. The cross-flow configuration in which the blood flow is perpendicular to the fibers enhances convective mixing to optimize gas transfer.

Development of an Implantable Artificial Lung

7

lines may be entirely intrathoracic, anastomosed, for example, to the proximal and distal pulmonary artery. The gas lines, on the other hand, are extracorporeal, exiting the thoracic cavity through the chest wall. The inlet gas is generally 100% O2 to achieve maximum oxygen transfer through the device. The exit gas line, up to now, has also been extracorporeal, but conceivably could be intrathoracic if a connection between a bronchus or the trachea was surgically established.

Gas Exchange The absolute transfer rate of oxygen and carbon dioxide through the artificial lung is dependent on a number of factors: 1. the degree of convective mixing achieved in the blood phase, which is dependent on the particular configuration and orientation of the fiber bundle; 2. gas exchange surface area of the fiber bundle; 3. blood flow rate through the device; 4. characteristics of the blood itself, such as hemoglobin concentration, viscosity, oxyhemoglobin saturation, and carbon dioxide content; 5. fiber length or gas flow rate through the fiber bundle; and 6. composition of the oxygenating and ventilating gas.

In order to maximize gas exchange, it is critical to achieve excellent convective mixing of the blood as it passes through the fiber bundle of the artificial lung. The designer of artificial lungs has several parameters that may be manipulated to achieve excellent gas transfer and maintain low blood side pressure losses. Consider blood flowing perpendicular to a bank of uniform fibers of outer diameter d, as illustrated in Figure 1.4. The frontal area, Af, is the product of the overall height, H, and the width, W. The overall length of the fiber bank in the blood flow direction is L. The void fraction, or porosity of the fiber bundle, is defined as: Vf = void volume/total volume. The tighter packed the fibers, the lower the porosity. In general, to achieve low pressure losses and high gas transfer, the porosity of the fiber bundles in artificial lungs should be in the range from about 0.5 to 0.75. A cross-flow configuration, in which the blood path is perpendicular to the fiber bundle, is important to achieve good convective mixing. The diameter of the fibers may also vary, but in the case of microporous polypropylene hollow fibers, there are a limited number of sizes typically available. The outer diameter usually falls within the range of 0.02 to 0.04 cm. Smaller diameter fibers allow the designer to increase the effective gas exchange surface area for a given artificial lung configuration. This can be seen by the relation:

As =

4(1 ! V f ) A f L d

,

(Eqn. 1.1)

in which As, the fiber surface area in this example, also increases with frontal area, blood path length or lower void fraction. Using microporous polypropylene hollow fibers, the amount of surface area required to achieve clinically significant amounts of gas transfer is between 1.5 to 2.0 m2, but may be less if small diameter fibers are used. By specifying the void fraction and fiber diameter, the design engineer may vary the frontal area and path length to meet certain size constraints and surface area requirements. However, the blood

8

The Artificial Lung

Fig. 1.5. Computer-generated family of artificial lung design curves showing the relation of pressure drop, blood path length and frontal area to oxygen transfer rate, mo2, for various blood flow rates. In this example, hemoglobin concentration is 12 g/dl, inlet and outlet oxyhemoglobin saturations are 65% and 95%, respectively, fiber outer diameter is 0.038 cm, and fiber bundle void fraction is 0.53. Similar curves may be generated for various conditions. The highlighted area indicates the area which meets the design specifications for an implantable artificial lung.

path length must not be inordinately long, otherwise the blood-side pressure losses will be excessive. Computer modeling, based on a semi-empirical theory of oxygen transport in artificial lungs, is an effective way to arrive at a prototype design with specified gas transfer rates and blood-side pressure losses. It minimizes the amount of trial-and-error testing and allows the engineer to manipulate the various design parameters described above. The details of the theory are too complex to summarize in this chapter, but may be found elsewhere.11-13 Briefly, in the case of oxygen transport, the designer specifies certain operating conditions, such as hemoglobin concentration, blood flow rate, inlet oxyhemoglobin saturation, and outlet (or desired) oxyhemoglobin saturation. By numerically solving a design equation that takes into account the particular geometric configuration, pressure losses, porosity, frontal area, blood path length, etc., a series of design curves may be constructed. To illustrate, Figure 1.5 depicts a family of design curves that relate blood side pressure drop and blood path length to frontal area at several blood flow rates. In this example, the hemoglobin concentration is assumed to be 12 g/dl, the fiber outer diameter is 0.038 cm, the void fraction of the fiber bundle is 0.53, the inlet blood oxyhemoglobin saturation is 65%, and the outlet blood oxyhemoglobin saturation is 95%. In order to achieve oxygen transfer rates of at least 200 ml/min at the stated conditions, it is evident from the graphs that a blood flow rate of at least 4 l/min is required. The frontal area in

Development of an Implantable Artificial Lung

9

Table 1.4. Material requirements for the artificial lung -Reproducible manufacturing process -Small diameter hollow fibers (200 to 400 micron outer diameter) -High diffusivity for oxygen and carbon dioxide -Relatively inert -Non-leaching components -Fibers that effectively separate the blood and gas phases -No deterioration in performance over time -Elastomeric housing for fiber bundle

this example would be about 150 cm2 and the blood path length between 2.5 to 3.0 cm. Figure 1.6, for the same specified conditions, shows that a fiber surface area of at least 1.5 m2 would be required to achieve an oxygen transfer rate of at least 200 ml/min. It is clear that by manipulating various parameters and blood conditions one can arrive at numerous permutations and graphs to optimize the design of the artificial lung. It is critical to keep the gas side pressures well below the blood side pressures in order to avoid the consequences of gas embolism. The gas side pressure drop (∀Pg) will be dependent on the gas flow rate (Qg), viscosity of the flowing gas (∝), the length of the individual fibers (l), radius of each fiber (r), and total number of fibers (n), according to Poiseuille’s Law:

∀Pg =

8∝Q g l n#r 4

.

(Eqn. 1.2)

For fibers having an inner diameter of 0.024 cm and utilizing 100% oxygen as the flowing gas, ∀Pg = 3.3 mm Hg x (Qg l) x (100/n) fibers, in which Qg is expressed in liters/ min and l is in cm. The entire fiber bundle will be composed of thousands of individual fibers, thereby minimizing the gas pressure build-up within any individual fiber at the gas flow rates utilized. If the device contains 10,000 fibers having a length of 20 cm each, a gas flow of 3 liters/min results in a gas-side pressure loss of about 2 mm Hg. Generally, to keep the gas side pressures low and to have a device that fits into the thoracic cavity, the fiber length should be in the range of 10 to 20 cm. Furthermore, it is important to keep the fiber length relatively short in order to maximize carbon dioxide transfer. Assuming that 100% oxygen is used as the ventilating gas, the gradient for O2 transfer (approximately 650 mm Hg) is much greater than that for CO2 transfer (approximately 40-50 mm Hg). With long fiber lengths, CO2 will build up in the ventilating gas, decreasing the gradient further. Thus, the gas flow rate will affect CO2 transfer more than O2 transfer. Adjusting the gas flow rate in an artificial lung is much like adjusting the minute ventilation on a mechanically ventilated patient. A practical rule of thumb is to have Qg /Qb < 1. Although the gradient for CO2 transfer is lower than that for O2 transfer, overall CO2 exchange usually equals or exceeds overall O2 exchange because the diffusivity of CO2 in blood is greater than that of O2.

10

The Artificial Lung

Fig. 1.6. Computer-generated family of artificial lung design curves showing the relation of fiber surface area and pressure drop for various blood flow rates. Specified conditions are the same as in Figure 1.5. The highlighted area shows that a surface area of at least 1.5 m2, for the set conditions, is required to meet design specifications for the implantable artificial lung.

Hemodynamic Compatibility In addition to adequate gas exchange capability, the artificial lung must be hemodynamically compatible with the cardiovascular system. Two approaches may be followed: the artificial lung derives its blood flow directly from the heart, thereby avoiding the need for an external, mechanical blood pump, and a composite artificial heart-lung system may be devised, incorporating a mechanical pump to provide blood flow through the lung. The latter approach has been taken by Japanese investigators, who have used a pneumatically driven blood pump to generate the pressure to force blood through the hollow fiber artificial lung. The advantages of having an incorporated blood pump are that the blood-side pressure losses across the lung become less of a design issue and adequate blood flows are easily achieved. On the other hand, the addition of a blood pump to the lung introduces another level of complexity to the device, makes the device less compact, increases the potential for mechanical device failure, and is associated with higher rates of thromboembolism and direct trauma to the blood elements. Readers interested in the composite heart-lung system are referred to other sources.14,15 Investigators in the United States have, to date, adopted the simpler approach of designing a pumpless artificial lung, blood flow through which comes from the right ventricle (RV). Systemic gas exchange requirements will not be met if arterial blood passes through the artificial lung, even at high flow rates. Therefore, the design relies on blood originating from the RV in order to meet blood flow and gas exchange requirements. The basic hemodynamic requirements of an artificial lung utilizing this approach are to minimize right ventricular afterload and to provide adequate compliance in the circuit. To minimize right ventricular afterload, the artificial lung must be designed so that there is a very low blood side pressure loss across the fiber bundle. High blood side pressure losses will result in increased RV afterload, decreased blood flow through the device,

Development of an Implantable Artificial Lung

11

and RV strain or failure. The maximum pressure losses must account for a wide range of blood flows, hematocrits, and blood viscosities. Pressure losses across the artificial lung should not exceed 15-20 mm Hg, and RV function may be optimized if the pressure losses are closer to 10-15 mm Hg. In addition to blood flow rate and the physical properties of blood, other factors that influence the pressure drop across the artificial lung include the frontal area of the fiber bundle, the blood path length or depth of the fiber bundle, the void fraction of the fiber bundle, and the diameter of the individual fibers comprising the fiber bundle. These variables are all design-dependent and may be manipulated using computer modeling, as discussed previously, to design the artificial lung and to optimize the pressure losses. Referring again to Figure 1.5, it is evident that the blood path length affects the pressure drop more than the frontal area does. As the blood path length increases, the pressure drop increases. As the frontal area decreases, the pressure drop increases, but to a lesser extent. To design a compact artificial lung with low pressure losses, a balance has to be struck to achieve a reasonable frontal area and blood path length. The void fraction of the fiber bundle, or how tightly packed the fibers are, has already been addressed. The lower the porosity, the higher the pressure drop across the fiber bundle. The porosity in a particular artificial lung configuration should be optimized so that pressure losses are kept to a minimum and the blood channels are narrow enough to achieve good convective mixing. Compliance of the artificial lung is another important design feature. The pulmonary circulation is characterized by low impedance and high compliance. The compliance of the natural pulmonary circulation acts as a windkessel to dampen the pulmonary arterial pulse and to allow blood to flow through diastole. Lack of compliance would lead to greatly diminished flow during diastole and the development of RV strain. To increase the compliance of the artificial lung, a compliance chamber could be added in-line with the fiber bundle, or the housing surrounding the fiber bundle could be made of an elastomeric material. Boschetti et al. have shown in a computational model of the hemodynamic effects of an artificial lung that increased compliance at the inlet of the artificial lung results in reduced right heart power and reduced impedance modulus.16 Furthermore, increased compliance of the fiber bundle housing results in decreased pulsatility of the fiber bundle which leads to more efficient gas exchange. The compliance of the system, as well as the RV afterload, are also affected by the way the artificial lung is connected to the natural circulation. The proximal pulmonary artery (PA) is recognized as the most suitable inlet to the artificial lung. The outlet, however, may be to the distal pulmonary artery, the left atrium (LA), or to both the pulmonary artery and left atrium. The first configuration (PA-PA) is in series with the natural lungs, whereas the second configuration (PA-LA) is in parallel with the natural lungs. The third configuration is a hybrid approach in which a portion of the blood flow is in series and a portion is in parallel with the natural lungs. All anastomoses are constructed with vascular grafts. The PA-PA circuit yields the worst hemodynamic performance in computer models because the impedances of the artificial lung and of the natural lungs have an additive effect. The relative advantages of the PA-PA circuit are: 1. any thromboembolic debris or gas emboli are filtered out of the circulation by the natural lungs; 2. the entire cardiac output potentially flows through the artificial lung, thereby maximizing the gas exchange rate; and 3. flow to the natural lungs is preserved, which may be important for the lungs to continue their nonrespiratory metabolic functions.

12

The Artificial Lung

The in-parallel PA-LA configuration may have the best hemodynamic consequences, resulting in less RV strain than with the in-series configuration. Flow to the native lungs may be preserved and regulated by leaving the pulmonary artery intact and using a flow occluder on the outside of the distal pulmonary artery. Possible disadvantages of the PA-LA circuit include the induction of arrhythmias, increased risk of left atrial and systemic thromboembolism, and reduced flow and gas exchange through the artificial lung if a portion of the cardiac output continues to go through the pulmonary artery to the native lungs. The hybrid configuration would be expected to have intermediate hemodynamic effects, with the relative advantages and disadvantages of the in-series and in-parallel configurations.

Biocompatibility The biocompatibility of any artificial organ must be considered in its design. Both material-specific and device-specific factors should be accounted for. Material requirements for the artificial lung of the hollow fiber type are listed in Table 1.4. Silicone rubber and microporous polypropylene have proven to be the best materials for artificial lungs because of their high diffusivities for CO2 and O2. Hollow fibers made of polypropylene are currently the industry standard for manufacturing membrane oxygenators because of the small diameter and compactness that can be achieved, thus this material has emerged as the choice for artificial lung development, as well. Moreover, a higher gas transfer rate per unit of surface area may be achieved with microporous polypropylene than with silicone rubber. A disadvantage of the microporous polypropylene is that plasma leakage into the gas phase may occur over time. This results in impaired gas exchange and device failure. In order to prevent plasma leakage, chemical modification or surface coating of the fibers is recommended. Surface coating with a thin silicone-based membrane was used for the intravascular oxygenator (IVOX), and is being actively investigated for application to the implantable artificial lung. Development of alternative fiber materials that have high diffusivities for O2 and CO2, but without the micropores, may be an area for future study. Although polypropylene is relatively inert, because of the large surface area of the material exposed to blood, there is enormous potential for blood-surface interactions. Cook and Mockros, in a separate chapter, have detailed the host response to these blood-surface interactions. Others have recently reviewed host-device interactions based on the study of patients undergoing cardiopulmonary bypass or ECMO procedures.17-19 Heparin bonding of the fibers is one tactic that is being taken in order to ameliorate blood-surface interactions and possibly avoid systemic anticoagulation with the use of the artificial lung.

Current Designs Current designs of artificial lungs have taken two general approaches. The first, being pursued by Tatsumi et al. in Japan, is essentially a variant of extracorporeal membrane oxygenation (ECMO). 14,15 An integrated artificial heart-lung implanted in the paracorporeal position, Tatsumi’s device utilizes a polyolefin hollow fiber membrane lung that has been surface-treated to reduce plasma leakage and heparin-bonded to reduce anticoagulation requirements. The artificial lung is integrated with inlet and outlet pusher-plate type pneumatic blood pumps. Venoarterial connections to the blood pumps complete the circuit. Because this device is not intended for intracorporeal implantation and represents a variant of ECMO, it will not be discussed further in this chapter.

Development of an Implantable Artificial Lung

13

Fig. 1.7. Side and top views of the artificial lung being developed by Cook et al.21 The fiber bundle is enclosed in a compliant, elastomeric housing. Blood enters on one side of the fiber bundle, passes through the bundle to become oxygenated and purged of carbon dioxide, and exits on the opposite side of the fiber bundle. Oxygenating gas flows through the fibers via tubing attached to the connectors at the potted ends of the fiber bundle. (Photographs courtesy of Rachid Idriss and Keith Cook).

14

The Artificial Lung

The second approach, adopted by investigators in the United States, is that of a fully implantable, pumpless artificial lung that derives its blood flow from right ventricular output. There are currently two different designs of pumpless artificial lungs that are being developed, both of which employ the cross-flow configuration to enhance convective mixing of the blood. The first design, originally described by Vaslef et al.,20 is being pursued by Cook,21 Mockros, and other researchers at Northwestern University. Their device, shown in side and top views in Figure 1.7 and in a schematic cross-sectional view in Figure 1.8, consists of a microporous hollow fiber polypropylene fiber bundle that is potted at each end and enclosed in a compliant, elastomeric housing. The fiber bundle is potted at each end with a polyurethane compound that keeps the bundle intact and prevents gas and blood side mixing. Blood enters on one side of the fiber bundle, passes cross-flow to the fibers to become oxygenated and purged of carbon dioxide, and exits on the opposite side of the fiber bundle. The oxygenating gas flows through the fibers via connecting tubing attached to manifolds enclosing the potted ends of the fiber bundle. Design improvements over earlier prototypes have included the use of a compliant housing for the fiber bundle, reduction of fiber bundle resistance, and reduction of inlet and outlet connection resistances.21 The device utilizes 209 ∝m outer diameter microporous polypropylene hollow fibers that are woven into mats and wrapped around a flat, polycarbonate frame. The void fraction is approximately 74% and the total fiber surface area is approximately 1.83 m2. The total device volume is only 800 cm3 with a priming volume of 350 ml. The connections to the blood inlet and outlet are via 18 mm inner diameter Dacron grafts. The second implantable lung design is being developed by Montoya et al. at Michigan Critical Care Consultants, Inc., in collaboration with Bartlett and his colleagues at the University of Michigan.22 Their device, shown in Figure 1.9 and in schematic cross-section in Figure 1.10, consists of 300 ∝m outer diameter matted microporous polypropylene hollow fibers that are wrapped around a mandrel and housed in a cylindrical shell with two circular endcaps. Blood flow enters axially through the center of the core and proceeds in a radial direction, in a cross-flow fashion, across the fiber bundle to become oxygenated and purged of carbon dioxide. Once the blood reaches the periphery of the fiber bundle, it is directed by a gutter to exit the device. The housing enclosing the fiber bundle currently is made of a rigid, noncompliant plastic. Total fiber surface area is 2.25 m2, and the priming volume is 240 ml. Inlet and outlet gas lines are attached to ports in the endcaps.

In Vitro Evaluation In vitro evaluation of prototype artificial lung devices allows for the tight control of operating conditions and precise measurement of pressure losses and oxygen and carbon dioxide transfer rates. In vitro tests have been conducted in both water and in animal blood. Testing the artificial lungs in water permits multiple evaluations to be performed on a single device and is much easier to accomplish than testing the devices in blood. In order to correlate artificial lung performance in water with that in blood, we developed a semi-empirical mathematical model of oxygen transfer that allows us to predict the oxygen transfer rate to blood at any set of operating conditions, based on the results of in vitro water tests.12 Figure 1.11 depicts the in vitro circuit for evaluation of artificial lungs. For the water tests, water is warmed to 37°C, and the PCO2 and PO2 are adjusted to the desired inlet conditions using an adjustable gas mixture of O2, CO2, and N2. Using a single pass technique, the water is then pumped through the artificial lung at different flow rates into an

Development of an Implantable Artificial Lung

15

Fig.1.8. Schematic cross-sectional view of the artificial lung being developed by Cook et al.21 Adapted from Cook KE, Makarewicz AJ, Backer CL et al. Testing of an intrathoracic artificial lung in a pig model. ASAIO J 1996; 42:M604-M609.

outlet reservoir. Inlet and outlet water samples are taken at each flow rate for gas analysis on a standard blood gas analyzer. The experimental circuit for in vitro blood tests is the same as for the water tests. Fresh bovine blood, anticoagulated with heparin (10,000 U/l) and EDTA (1g/l), is filtered into the inlet reservoir. The blood is recirculated through the artificial lung until the desired inlet conditions are achieved. Separate rotameters are used to adjust the concentrations of O2, CO2, and N2 in the gas mixture passing through the artificial lung. Blood pH is adjusted, as necessary, by the addition of 1 M sodium bicarbonate. Hemoglobin and percent oxyhemoglobin saturation are measured using cooximetry, and hematocrit is measured using the microcapillary centrifugation technique. Gas flow rate through the device is measured using a mass flow meter, and the CO2 concentration in the exit gas is measured using mass spectroscopy on collected gas samples. Oxygen and carbon dioxide transfer rates are calculated according to:

m˙ o = 1.34[Hgb ]Qb [(S a o 2 )out ! (S vo 2 ) in ] ⋅ 10 + kQb [( Po 2 )out ! ( Po 2 ) in ] ⋅ 10, (Eqn. 1.3) 2

and m˙ co 2 = Q g ⋅ [ Fco 2 ]out ,

(Eqn. 1.4)

where mO2 and mCO2 are the oxygen and carbon dioxide transfer rates, respectively, in ml/ min, [Hgb] is the hemoglobin concentration in g/dl, Qb is the blood (or water) flow rate in l/min, (SaO2)out and (SvO2)in are the outlet and inlet oxyhemoglobin saturations, k is the solubility of oxygen in blood (or water) in ml/ml•mm Hg, (PO2)out and (PO2)in are the

16

The Artificial Lung

Fig. 1.9. End and side views of the artificial lung being developed by Montoya et al. (Michigan Critical Care Consultants, Inc., Ann Arbor, MI). The inlet and outlet blood flow lines are connected to vascular grafts at one end and to the device inlet and outlet. Blood enters into the core of the device, passes radially outward through the concentric fiber bundle, and exits the device on the outside of the fiber bundle. The fiber bundle is potted at either end to prevent admixture of the blood and gas streams and to provide connecting attachments for the inlet and outlet gas lines. (Photograph courtesy of Pat Montoya).

Development of an Implantable Artificial Lung

17

Fig. 1.10. Schematic cross-sectional view of the Michigan artificial lung, showing the blood and gas paths. Adapted from Lynch WR, Montoya JP, Brant DO, et al. Hemodynamic effect of a low-resistance artificial lung in series with the native lungs of sheep. Ann Thorac Surg 2000; 69:351-356.

Fig. 1.11. Experimental circuit for in vitro evaluation of artificial lung. A single pass technique is utilized at various blood flow rates. A mixture of O2, CO2, and N2 is used to adjust fresh anticoagulated bovine blood to venous conditions. Once the pH, PCO2 and PO2 are at the desired venous levels, the gas mixture through the artificial lung is changed to 100% O2 and the single pass of blood through the lung is begun.

18

The Artificial Lung

Fig. 1.12. Predicted vs. measured oxygen transfer rates compared with the line of identity. Measured transfer rates are based on in vitro bovine blood experiments through the prototype device of Vaslef et al. 20 Predicted rates are based on a semi-empirical mathematical model of oxygen transfer through hollow fiber membrane oxygenators using data obtained from water experiments. There is a strong correlation between predicted and measured values (r = 0.995).

Fig. 1.13. O2 and CO2 transfer rates through the artificial lung prototype of Vaslef et al.20 Data are shown as mean ± sd for in vitro bovine blood experiments. See text for full description.

Development of an Implantable Artificial Lung

19

Fig. 1.14. Time course of arterial PO2 and PCO2 in sheep implanted with the Michigan radial flow artificial lung utilizing a PA-PA configuration.22 Gas requirements were met through the duration of the study.

outlet and inlet partial pressures of oxygen, and [FCO2]out is the fractional concentration of CO2 in the exit gas stream. Applying the previously described mathematical model12 to the water results for oxygen transfer in order to predict mO2 in blood, it can be shown (Fig. 1.12) that the predicted and measured O2 transfer rates show a strong correlation (r = 0.995, p < 0.000001).20 The overall measured gas transfer rates for a series of in vitro bovine blood experiments are shown in Figure 1.13. At blood flow rates ranging from 1 l/min to 5 l/min, the mean mO2 ranged from 52 to 225 ml/min. The mean mCO2 at these blood flow rates ranged from 150 to 286 ml/min. Gas flow rate was 5 l/min. The blood conditions included [Hgb] = 9.5 ± 1.6 g/dl, (SvO2)in = 65.4 ± 10.0%, and (PCO2)in = 46.8 ± 2.8 mm Hg. The blood-side pressure losses at these blood flow rates ranged from 4.0 to 29.0 mm Hg, which have been further reduced to less than 10 mm Hg in later prototypes by enlarging the blood-side connectors and tubing. It is readily apparent that the in vitro evaluations have confirmed the validity of computer models and that an artificial lung with O2 and CO2 transfer rates in excess of 200 ml/min and low pressure losses can be achieved without the need for a prosthetic blood pump.

In Vivo Evaluation Pilot studies of paracorporeal implantation of pumpless artificial lungs in large animals were reported in 1994 by Vaslef et al.20 and by Fazzalari et al.22 The prototype devices were similar to those depicted in Figures 1.7 and 1.9. Paracorporeal implantation, rather than intrathoracic implantation, has been the strategy used thus far so that the device can be closely monitored. In these initial short-term experiments, artificial lungs were implanted for up to eight hours. Vaslef ’s group utilized a PA-LA configuration in adult swine,

20

The Artificial Lung

Fig. 1.15. Mean O2 and CO2 gas transfer rates over a 24-hour implantation period in two swine. The artificial lung of Cook et al. was implanted in a paracorporeal position utilizing the PA-LA configuration. Gas exchange, as well as flow through the device, remained stable during the study period.21

whereas Fazzalari’s group employed the PA-PA configuration in adult sheep. Investigators from both groups were able to achieve significant oxygen and carbon dioxide transfer rates. The 2.2 m2 artificial lung of Vaslef et al. transferred approximately 100 ml/min of O2 and from 86 to 144 ml/min of CO2 at a maximum blood flow rate of 1200 ml/min through the device. A pressure drop of 10 mm Hg across the fiber bundle was observed. Blood flow rate through the artificial lung was thought to be limited because of the parallel (PA-LA) attachment mode, whereby some of the cardiac output was directed to the native lungs, the noncompliant housing chamber for the fiber bundle, and the small diameter tubing and connectors used to attach the lung to the native circulation. The 1.95 m2 radial flow device evaluated by Fazzalari et al. was implanted in six anticoagulated sheep via a left lateral thoracotomy. A flow occluder placed between the two anastomoses to the pulmonary artery helped to regulate blood flow through the device. There were no significant changes in the mean arterial pressure or cardiac index over time. The gas exchange requirements for the six animals tested were completely met for over seven hours, when the experiments were terminated. Figure 1.14 shows that the arterial Po2 and Pco2 were maintained within a normal range over the time period of the implantation. Pressure losses were less than 12 mm Hg at a blood flow rate of 5 l/min. Cook improved upon the earlier prototype of Vaslef et al. by making several design changes, including the use of smaller diameter (209 ∝m outer diameter) polypropylene hollow fibers, a larger void fraction (0.74), a smaller fiber surface area (1.83 m2), larger diameter vascular grafts and connecting tubing, and the use of an in-line compliance chamber.21 These modifications resulted in the implantation for up to 24 hours in three swine. One pig died from bleeding complications after 20 hours of support. The mean gas trans-

Development of an Implantable Artificial Lung

21

Fig. 1.16. Awake and active sheep with a paracorporeal implantation of the Michigan artificial lung. (Photograph courtesy of Bill Lynch).

fer rates for the 24 hour period in the other two pigs are shown in Figure 1.15. Overall, the mean blood flow rate through the device was 3.0 l/min. The mean gas transfer rates were 179 ml/min for O2 and 206 ml/min for CO2. The calculated right ventricular power output remained constant, indicating that there was no significant right ventricular strain. Although these experiments were not without complications, they provided ample evidence that the design requirements of an implantable artificial lung could be met and that short-term respiratory support with an artificial lung is feasible. Most recently, Lynch has expanded the work of Fazzalari et al. utilizing the radial flow artificial lung.23 Five of seven sheep survived 24 hours with at least 75% of the cardiac output diverted through the device. The artificial lung was implanted in series with the native lungs (PA-PA) by anastomosing the inlet and outlet of the device to the proximal and distal pulmonary artery. A flow occluder was positioned between the anastomoses to regulate flow through the artificial lung. Two of the seven sheep did not tolerate occlusion of the pulmonary artery and succumbed, presumably to right ventricular failure. The remaining five animals were awake and active (Fig. 1.16). The mean blood flow rate through the device for these animals was 3.2 l/min, and the mean oxygen transfer rate was 194 ml/ min. Carbon dioxide transfer rate was not specifically reported; however, the PCO2 was kept within a normal range by adjusting the sweep gas flow rate. To maximize gas flow rate through the device, i.e., to maximize CO2 transfer, as well as to prevent gas embolism, the outlet gas line was connected to a vacuum of –20 cm H2O. Blood-side pressure losses were minimal, amounting to less than 2 mm Hg, and no untoward cardiovascular effects were observed in the five sheep that survived 24 hours. Current efforts are aimed at determining the optimal attachment mode and extending the period of artificial lung support to several days or weeks.24

The Artificial Lung

22

Summary Both in vitro and in vivo studies have demonstrated the feasibility of an artificial lung for providing near-total respiratory support. The gas exchange requirements can clearly be met with a compact device using materials that are readily available. The hemodynamic consequences of an artificial lung on the cardiovascular system have not yet been fully defined, but it is apparent that an artificial lung perfused by the right ventricle without the need for a prosthetic blood pump is possible. The optimal mode of attachment of the artificial lung to the native circulation has to be further delineated, but it may be desirable to have at least a portion of the cardiac output go to the native lungs to support nonrespiratory metabolic functions. The biocompatibility of the device with the host, as well as device function over extended periods of time (days to weeks), must also be studied further. Future refinements of artificial lung technology will likely include the development of novel surface coatings to minimize blood-surface interactions and to prolong device function. The application of tissue engineering to artificial lung design to create a hybrid bioartificial organ is a possibility for the future, but in the near term it is more likely that investigative efforts will concentrate on the continued progress towards the development of an artificial lung constructed from synthetic materials. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

United Network for Organ Sharing. 1999 annual report of the U.S. scientific registry of transplant recipients and the organ procurement and transplantation network. U.S. Department of Health and Human Services, 1999. Platt JL, Nagayasu T. Current status of xenotransplantation. Clin Exp Pharmacol Physiol 1999; 26:1026-1032. The Acute Respiratory 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. Morris AH, Wallace CJ, Menlove RL et al. Randomized clinical trial of pressure-controlled inverse ratio ventilation and extracorporeal CO2 removal for adult respiratory distress syndrome. Am J Respir Crit Care Med 1994; 149:295-305. Conrad SA, Eggerstedt JM, Grier LR. Intravenacaval membrane oxygenation and carbon dioxide removal in severe acute respiratory failure. Chest 1995; 107:1689-1697. Bodell BR, Head JM, Head LR et al. An implantable artificial lung. JAMA 1965; 191:125-127. Shah-Mirany J, Head LR, Ghetzler R et al. An implantable artificial lung. Ann Thorac Surg 1972; 13:381-387. Palmer AS, Collins J, Head LR. Development of an implantable artificial lung. J Thorac Cardiovasc Surg 1973; 66:521-525. Morin PJ, Gosselin C, Picard R et al. Implantable artificial lung. J Thorac Cardiovasc Surg 1977; 74:130-136. Galletti PM, Richardson PD, Trudell LA et al. Development of an implantable booster lung. Trans Am Soc Artif Intern Org 1980; 26:573-577. Mockros LF, Leonard R. Compact cross-flow tubular oxygenators. Trans Am Soc Artif Intern Org 1985; 31:628-633. Vaslef SN, Mockros LF, Anderson RW et al. Use of a mathematical model to predict oxygen transfer rates in hollow fiber membrane oxygenators. ASAIO J 1994; 40:990-996. Vaslef SN, Mockros LF, Cook KE et al. Computer-assisted design of an implantable, intrathoracic artificial lung. Artif Organs 1994; 18:813-817. Tatsumi E, Eya K, Taenaka Y et al. Long-term cardiopulmonary support with a composite artificial heart-lung system. ASAIO J 1995; 41:M557-M560. Tatsumi E, Takewa Y, Akagi H et al. Development of an integrated artificial heart-lung device for long-term cardiopulmonary support. ASAIO J 1996; 42:M827-M832. Boschetti F, Perlman CE, Cook KE et al. Hemodynamic effects of attachment modes and device design of a thoracic artificial lung. ASAIO J 2000; 46:42-48.

Development of an Implantable Artificial Lung

23

17. Janvier G, Baquey C, Roth C et al. Extracorporeal circulation, hemocompatibility, and biomaterials. Ann Thorac Surg 1996; 62:1926-1934. 18. Peek GJ, Firmin RK. The inflammatory and coagulative response to prolonged extracorporeal membrane oxygenation. ASAIO J 1999; 45:250-263. 19. Salzman EW, Merrill EW, Kent KC. Interactions of blood with artificial surfaces. In: Colman RW, Hirsh J, Marder VJ, Salzman EW, eds. Hemostasis and Thrombosis: Basic Principles and Clinical Practice. 3rd ed. Philadelphia: J.B. Lippincott Company, 1994:1469-1485. 20. Vaslef SN, Cook KE, Leonard RJ et al. Design and evaluation of a new, low pressure loss, implantable artificial lung. ASAIO J 1994; 40:M522-M526. 21. Cook KE, Makarewicz AJ, Backer CL et al. Testing of an intrathoracic artificial lung in a pig model. ASAIO J 1996; 42:M604-M609. 22. Fazzalari FL, Montoya JP, Bonnell MR et al. The development of an implantable artificial lung. ASAIO J 1994; 40:M728-M731. 23. Lynch WR, Montoya JP, Brant DO et al. Hemodynamic effect of a low-resistance artificial lung in series with the native lungs of sheep. Ann Thorac Surg 2000; 69:351-356. 24. Lynch WR, Haft JW, Montoya JP et al. Partial respiratory support with an artificial lung perfused by the right ventricle: Chronic studies in an active animal model. (Abstract). ASAIO J 2000; 46:202.

CHAPTER 2

Rationale for an Implantable Artificial Lung William R. Lynch and Robert H. Bartlett

Introduction

T

he concept of devices for pulmonary assistance is easy to imagine. The device would be continuously perfused by blood at flow rates high enough to exchange most or all of the oxygen and carbon dioxide requirements. Design considerations would include vascular access, thrombogenicity, durability, gas supply, and infection. Candidates for this technology would include patients with acute or chronic respiratory failure, from which tens of thousands die each year. Membrane gas exchange devices have been used in extracorporeal support systems for over 30 years. The devices made cardiopulmonary bypass possible and as the technology was refined, support of longer duration outside of the operating rooms became practical. The concept of applying membrane gas exchange as a pulmonary assist device has been suggested in the literature during the same 30 years, but is only now realizing laboratory success making clinical application realistic. Why is this happening now? The past decade has seen the maturation of temporary extracorporeal life support and lung transplantation into clinically successful technologies. Extracorporeal membrane oxygenation (ECMO) has become a reliable technology making prolonged support a reality. The application of ECMO allows support of patients in respiratory failure that, prior to this technology, had little chance for survival. ECMO provides a window of opportunity for recovery from respiratory insult; however, the practical application for adults is only 30 days. Once past this time frame, complications limit further support. Lung transplantation has also become a clinical reality through advances in surgical technique and immunosuppression. This treatment can extend the lives of those suffering from end-stage irreversible respiratory failure. The therapy is limited because of donor supply and many patients die waiting for a suitable organ. A device that would simplify ECMO technology and lengthen the therapeutic window would offer relief to patients dying from irreversible respiratory failure and provide a bridge to transplantation. An artificial lung could fulfill this role. The device would need to be compact, efficient in gas exchange and durable. Meeting these goals would make the device potentially implantable, minimizing the risk of infection. An artificial lung, like the artificial heart and the ventricular assist device, is unlikely to be a permanent organ replacement in the foreseeable future. Reliable performance for 1 to 6 months is feasible and such a device

The Artificial Lung, edited by Steven N. Vaslef and Robert W. Anderson. ©2002 Eurekah.com.

Rationale for an Implantable Artificial Lung

25

could support prolonged reversible acute respiratory failure or bridge those in irreversible failure to transplantation. The stage is set for design, development and testing of an implantable artificial lung.

History of Implantable Artificial Lung Development The technological advances making the development of an artificial lung possible provide an interesting review. Most significant has been the science and engineering of oxygenator technology but the story is not complete without an appreciation for the evolution of cardiopulmonary bypass and ECMO.

Cardiopulmonary Bypass and ECMO The modern advances towards extracorporeal support began with the story of John Gibbon’s heart-lung machine. On October 3, 1930, a woman in her thirties suffered a pulmonary embolism two weeks after cholecystectomy. John Gibbon was the young resident assigned to care for the patient as she struggled to survive. After 18 hours, the woman’s condition suddenly deteriorated and she was rushed to the operating room. A pulmonary embolectomy, the procedure of Trendelenberg, was performed but she did not survive. This experience started Dr. Gibbon’s lifelong pursuit to develop a machine to replace heart and lung function. His vision was to create a means of temporary support, making cardiac and pulmonary surgery possible. His work included design and development of oxygenators, pumps, and cannulae along with management of anticoagulation. The result was cardiopulmonary bypass.1,2 Cardiopulmonary bypass began to have an impact in the 1950’s, making possible operations that could not previously have been tried. As the technique was refined in the operating room, investigators explored ways to prolong support and utilize the technology outside of the operating room. Early on, it became apparent that extracorporeal circulation became life threatening when prolonged even for a few hours. Thrombocytopenia, coagulopathy, hemolysis, generalized edema and deterioration of organ function occurred in proportion to the duration of bypass.3 Experiments by Lee,4 Dobell5 and others revealed the culprit to be direct exposure of blood to oxygen, thus emphasizing the need for improved gas exchange techniques. The membrane interface proved to be a successful solution and a number of lung designs based on this technology was the result. Bartlett, Drinker, and others6-9 demonstrated that lower doses of heparin were practical for prolonged support and reduced bleeding complications. As the newer lungs and lower anticoagulation regimens became incorporated into the application of extracorporeal support, successful support began to be measured in days instead of hours. The physiology of prolonged support began to be characterized10 and extending the therapy from days to weeks seemed possible. The hemodynamics were easy to regulate and prolonged support was hematologically well tolerated. Acidosis, capillary permeability and organ deterioration no longer limited prolonged support. Bleeding was minimal, hemolysis was negligible and the inevitable thrombocytopenia was manageable. Extracorporeal support was ready for intensive care application, where it came to be known as ECMO.11

Oxygenator Development Oxygenator development progressed as cardiopulmonary bypass evolved. Gibbon’s first oxygenators were revolving cylinders over which the patient’s blood was poured. This created a thin film of blood over the surface of the cylinder for direct exposure to oxygen.

26

The Artificial Lung

Later techniques, such as bubble oxygenators and rotating disk oxygenators, also directly exposed blood to oxygen. Direct exposure proved to be a very good way of achieving gas exchange, but was damaging to the proteins and formed elements of the blood. Kolff, during his work with membranes for hemodialysis, noted that oxygen transfer occurred across these membranes.12 This observation, made during the early days of cardiac surgery, led researchers to investigate membranes as a means of eliminating the blood-gas interface.

Solid Membranes

The first membrane oxygenator was designed and built by Clowes, Kolobow et al.13,14 The membrane was a sheet of solid polyethylene, which had very low gas transfer capabilities. In order to exchange sufficient quantities, a very large surface area was required. Flat sheets of this solid membrane were used and assembly of the device was very complicated. Nonetheless, gas exchange occurred and the device was used in the laboratory and in some clinical cases. Teflon (tetrafluoroethylene) membranes were used and then replaced by silicone. Thomas used a silicone rubber film deposited over a nylon screen, which improved strength and stability.15 E. Converse Peirce, working with General Electric Company, used a copolymer of silicone rubber and polycarbonate.16,17 Silicone was very efficient in gas exchange but lacked structural integrity. The polycarbonate skeleton gave strength to the silicone rubber, yielding an efficient and durable membrane. As membrane technology improved, how to best arrange the membrane became a consideration. The membrane required supporting structures and together they created the blood and gas compartments. Design strategies to optimize gas exchange, reduce thrombogenicity and simplify manufacturing prompted the development of different compartment configurations and a variety of devices. The GE Peirce lung, which used long flat sheets of the reinforced silicone membrane in a sandwich arrangement, was used successfully for cardiac surgery and prolonged support. Thrombogenicity inherent to this design limited further development. A flat sheet sandwich design of Clowes was modified by Landé18 and proved to be quite efficient for cardiac surgery and longer support. Another sandwich type of flat sheet device was designed by Bramson19 and used by Hill for cardiac surgery. These two used this device in 1971 for the first successful prolonged extracorporeal support of a patient with respiratory failure.20 The device was assembled by hand for each use and, although functionally quite good, the cumbersome nature of the device made it impractical for continued use. The spiral coil membrane lung designed by Kolobow21 has been the only solid membrane lung to withstand the test of time. The device consisted of a long envelope of silicone rubber reinforced with fabric and coiled around a mandrel, producing the “spiral coil”. Gas flowed inside the envelope as blood coursed outside. Originally manufactured by SciMed Corporation, the device is still available from Avecor, Inc. using a design that has been virtually unchanged in 30 years. The device is currently the only oxygenator approved for prolonged use.

Hollow Fibers The solid membrane was the basis for many design platforms, yet manufacturing challenges and thrombogenicity led others to explore different approaches. One solution was the use of small hollow tubes of silicone referred to as fibers. These hollow fibers could be drawn into strands and then woven into sheets. Gas was blown through the core of the hollow fibers while blood circulated over and through the weave of the “fabric”. This flow pattern enhanced mixing of the blood, which in turn improved gas exchange and

Rationale for an Implantable artificial lung

27

reduced thrombogenicity. A mat of woven silicone fibers was an early precursor to the current Kuraray membrane lung. High efficiency, low priming volume and minimal thrombogenicity were demonstrated but the problems of manufacturing proved to be major. The Silox Company in Japan manufactured a solid fiber membrane lung using larger caliber silicone tubes, assembled in a fashion similar to a tube-type dialyzer.22

Microporous Membranes Newer materials continued to be evaluated as membranes. The advent of hydrophobic microporous fabrics (Gortex, etc.) offered an attractive alternative to solid polymers. The microporous material was easily manufactured in sheets and had excellent gas exchange properties. Based on an oxygenator reported by Douglas, Wildevuur and others,23,24 Leonard at the Travenol Company manufactured a sandwich type membrane lung using this material. Tests of this lung demonstrated excellent gas exchange, and the hematologic response did not exhibit the adverse effects of direct blood-gas interface. Even though the micropores allowed direct contact of oxygen with the blood, the serum proteins and formed elements were minimally affected. There was, however, a shunting of blood between the envelope sheets, making overall performance less than satisfactory. This led to the development of hollow fiber lungs made from microporous material. This membrane proved to be most successful for routine manufacture of artificial lungs for cardiac surgery. The small hollow fibers were easy and inexpensive to manufacture, had a high efficiency of gas exchange, and worked very well for a period of hours. The original hollow fiber microporous device was manufactured under the auspices of the NIH by Dow Corning Company. Johnson & Johnson, and Medtronics eventually adopted the design. The device, presently marketed as the Maxima membrane oxygenator, is a gas inside, blood outside device that provides excellent gas exchange. The other major manufactures of perfusion devices (Bentley, Cobe, Terumo, Avecor, and others) all make gas exchange devices made of hollow fibers of microporous materials.

Coated Microporous Membranes The hollow fiber microporous devices fail when water and plasma eventually leak through the micropores. This causes the gas side of the membrane to become wet, reducing gas exchange efficiency. The cause is absorption of lipid from the blood, resulting in wetting of the pores and plasma leakage when the surface tension is not sufficient to sustain the hydrostatic pressure.25 To avoid this failure mode, investigators tried coating the microporous material with thin layers of polymers. Using a thin layer of silicone, the plasma leak was eliminated. The microporous fibers served as a skeleton for this very thin layer of silicone and the result was a very efficient and durable membrane. Flat sheet devices utilizing this fabrication technique were reported by Galletti et al. and are currently manufactured by Jostra Company. Mortensen26 employed this same membrane technology in the original IVOX devices.

Design and Application of an Implantable Artificial Lung Oxygenators and their membranes have been designed with cardiac surgery in mind. Some are used for prolonged support in ECMO. These devices are intended for short periods of use and with extensive heparinization. Most have very high resistance to blood flow and areas of stagnation, and little attention has been paid to mixing of blood. Condensation, which accumulates on the gas side of the membrane in a matter of hours, has

The Artificial Lung

28

drawn little attention. These characteristics make the existing oxygenators designed for cardiac surgery unsuitable as implantable devices. How is the transition to an implantable artificial lung to be made?

Design of an Artificial Lung The first step is to define the qualities of an artificial lung. The device should have minimal resistance to blood flow while having high efficiency in gas exchange. Minimizing the thrombogenic potential will be necessary to extend membrane life, and stagnant areas should be avoided. Taking advantage of secondary flows and blood mixing will help reduce thrombogenesis while improving gas exchange efficiency. These efforts should be made in concert with reducing or eliminating dependence on anticoagulation. Protecting the gas side of the membrane from plasma and water will also pose a design challenge necessary to extend membrane life. As each of these goals is accomplished, efficiency will increase, size will decrease, and durability will improve. There are four feasible design strategies to an implantable artificial lung: 1. A gas exchange device perfused by a systemic arteriovenous shunt to provide partial respiratory support. 2. An intravascular gas exchange device placed in a large vessel for gas exchange with venous blood. 3. A gas exchange device, perfused by the right ventricle, to replace, or augment, native lung function. 4. A gas exchange device and a pump replacing the right ventricle and native lungs (or both ventricles and lungs).

Each of these strategies has been, or is presently being investigated by a variety of laboratories, and that research is the subject of this book.

Applications for an Artificial Lung Acute Respiratory Failure Currently, ECMO is the only mechanical support system for patients with acute respiratory failure who are failing on mechanical ventilation. Eighty-five percent of neonates managed in this fashion recover and survive with normal lung function. With the exception of lung hypoplasia, neonatal respiratory failure is successfully treated by ECMO and there would be no application for an implantable device. However, older children and adults with severe acute respiratory failure have only a 50-70% survival with temporary ECMO support. The patients who do not survive usually die from multiple organ failure or from the effects of progressive irreversible pulmonary fibrosis. The ECMO survivors typically require 1 to 4 weeks of ECMO support followed by 1 to 3 months in the ICU. The application for an implantable device in this large patient group would be two fold. First, early implantation of an artificial lung for those requiring prolonged ECMO support could simplify nursing care, reduce costs and allow management outside of an ICU. Second, for the patients who progress to end-stage pulmonary fibrosis, return of pulmonary function is not expected. These patients usually have infections and many are suffering other organ system dysfunction. Because of the acute nature of their illness, these patients are not candidates for lung transplantation and go on to die. An implantable artificial lung could support these patients through the acute illness, allowing recovery of other organ function and making transplantation a reasonable goal.

Rationale for an Implantable artificial lung

29

Fig. 2.1. Current treatment strategies for the management of acute or chronic lung disease.

Graft failure in a recent lung transplant patient leaves few options. Whether graft failure is secondary to acute respiratory deterioration, rejection or bronchiolitis, the acute nature of the process precludes retransplantation. Temporary support with an artificial lung would allow the transplant lung time to recover function or provide time for a second transplant. Use of the artificial lung in this situation could be used to rescue the graft or support patients until a replacement organ is found. The artificial lung, in this application, would be similar to reverting to hemodialysis in patients with failed transplanted kidneys.

Chronic Respiratory Failure The largest potential group of patients to be candidates for an implantable artificial lung will be sufferers of chronic lung diseases of a destructive nature such as emphysema, cystic fibrosis, idiopathic fibrosis or chronic rejection of a transplanted lung. Currently, these diseases are treated with supplemental oxygen and, for a select few, lung transplantation. The average wait for a donor lung is almost two years and many of these patients die before receiving a transplant. Some have worsening respiratory failure, requiring mechanical ventilation. For many centers, this is reason to remove a patient from the transplant list. Many of these patients die on the ventilator. The use of an implantable artificial lung

30

The Artificial Lung

Fig. 2.2. Potential treatment strategies, incorporating the use of artificial lungs, for the future management of acute or chronic lung disease.

could rescue these patients from the ventilator and provide a bridge to transplantation. When supported by the device, there would be an opportunity to improve nutrition, exercise tolerance, and clear infections while waiting for an appropriate donor. The time could also be used to facilitate treatment of the underlying disease, such as lung volume reduction surgery for bullous emphysema, prolonged treatment of transplant rejection, or aggressive lavage of the infected lungs of a cystic fibrosis patient. Patients with chronic respiratory failure might also benefit from application of partial, instead of total support. The device could be either implanted or placed in a paracorporeal position. Perhaps the device could be perfused by an arteriovenous shunt designed to provide only enough flow to remove metabolically produced carbon dioxide while providing a supplemental amount of oxygen. The inability to clear CO2 causes increased work of breathing that is debilitating to many of these patients. A strategy of

Rationale for an Implantable artificial lung

31

partial support directed at relieving this aspect of the disease might be a simple and effective means of stabilizing the patient’s condition, requiring no further treatment.

Chronic Right Ventricular Failure There is a small group of patients who die from high pulmonary vascular resistance that is unresponsive to medical treatment. The etiology might be primary pulmonary hypertension or chronic obstructive pulmonary disease, but death is from progressive right ventricular failure. Lung transplantation is the only treatment when right ventricular function is preserved. If the right ventricle is not salvageable, both heart and lung must be transplanted. The use of an implantable artificial lung perfused by the right ventricle would provide sufficient respiratory support while also reducing right ventricular work. The device would serve as a bridge to lung transplantation while rescuing the native right ventricle.

Theoretical and Speculative Applications Treatment of patients with minimal or no native lung function might one day be possible. Consider infants with bilateral pulmonary hypoplasia, often associated with diaphragmatic hernias. Presently, these infants can be supported with ECMO but with minimal chance for recovery. Promising methods to induce lung growth exist, but these treatments require a prolonged period of time. An implantable or paracorporeal device could sustain life, providing time for lung maturation to occur. Even more speculative would be using an artificial lung in conjunction with lung cancer therapy. Patients with primary or metastatic pulmonary disease could be supported with an artificial lung and their native lungs removed. Perhaps one or both lungs could be removed for treatment of the neoplasm while an implantable artificial lung supports the patient’s respiratory function. The excised native lungs could be perfused with a chemotherapeutic regimen that is too toxic for systemic exposure. Once the lungs were verified to be tumor free, they would be reimplanted into the patient for a cure. Innovative applications would evolve as the device and technique improved with experience. Figure 2.1 represents treatment strategies in use today while Figure 2.2 suggests ways an artificial lung could impact treatment in the future. References 1. 2. 3. 4. 5. 6. 7. 8. 9.

Gibbon JH Jr. Artificial maintenance of circulation during experimental occlusion of the pulmonary artery. Arch Surg 1937; 34:1105-1131. Romaine-Davis A. John Gibbon and his Heart-Lung Machine. Philadelphia: University of Pennsylvania Press, 1991:19-61. Bartlett RH. Extracorporeal life support for cardiopulmonary failure. Curr Probl Surg 1990; 27:623-705. Lee WH Jr, Krumhaar D, Fonkalsrud EW. Denaturation of plasma proteins as a cause of morbidity and death after intracardiac operations. Surgery 1961; 50:29-39. Dobell ARC, Mitri M, Galva R. Biologic evaluation of blood after prolonged recirculation through film and membrane oxygenators. Ann Surg 1965; 161:617-622. Drinker PA, Bartlett RH, Bialer RM et al. Augmentation of membrane gas transfer by induced secondary flows. Surgery 1969; 66:775-781. Bartlett RH, Kittredge D, Noyes BS Jr et al. Development of a membrane oxygenator: Overcoming blood diffusion limitation. J Thorac Cardiovasc Surg 1969; 58:795-800. Bull BS, Korpman RA, Huse WM. Heparin therapy during extracorporeal circulation: Problems inherent in existing heparin protocols. J Thorac Cardiovasc Surg 1975; 69:674-684. Whittlesey GC, Kunda SY, Salley SO. Is heparin necessary for extracorporeal circulation? ASAIO Trans 1988; 34:823-826.

32

The Artificial Lung

10. Bartlett RH, Cilley RE. Physiology of extracorporeal life support. In: Arensman RM, Cornish JD, eds. Extracorporeal Life Support. Boston: Blackwell, 1993:89-104. 11. Kanto WP Jr, Shapiro MB. The development of prolonged extracorporeal circulation. In: Zwischenberger JB, Bartlett RH, eds. ECMO: Extracorporeal Cardiopulmonary Support in Critical Care. Ann Arbor: Extracorporeal Life Support Organization, 1995:15-25. 12. Kolff WJ, Berk HT Jr. Artificial kidney: A dialyser with a great area. Acta Med Scand 1944; 117:121-134. 13. Clowes GHA Jr, Hopkins AL, Neville WE. An artificial lung dependent upon diffusion of oxygen and carbon dioxide through plastic membranes. J Thorac Surg 1956; 32:630-637. 14. Clowes GHA Jr, Hopkins AL, Kolobow T. Oxygen diffusion through plastic films. Trans Am Soc Artif Intern Organs 1955; 1:23-24. 15. Thomas JA. Über eine herz-lungen maschine mit künslicher alveolar-membran. Langenbeck Arch Klin Chir 1958; 289:286-290. 16. Peirce EC II. A new concept in membrane support for artificial lungs. Trans Am Soc Artif Intern Organs 1966; 12:334-339. 17. Peirce EC II. Modifications of the Clowes membrane lung. J Thorac Cardiovasc Surg 1960; 39:438-448. 18. Landé AJ, Dos SJ, Carlson RG. A new membrane oxygenator-dialyzer. Surg Clin North Am 1967; 47:1461-1470. 19. Bramson ML, Osborn JJ, Main FB. A new disposable membrane oxygenator with integral heat exchanger. J Thorac Cardiovasc Surg 1965; 50:391-400. 20. Hill JD, de Leval MR, Fallat RJ et al. Acute respiratory insufficiency: Treatment with prolonged extracorporeal oxygenation. J Thorac Cardiovasc Surg 1972; 64:551-562. 21. Kolobow T, Bowman RL. Construction and evaluation of an alveolar membrane artificial heart lung. Trans Am Soc Artif Intern Organs 1963; 9:238-243. 22. Hirschl RB. Devices. In: Zwischenberger JB, Bartlett RH, eds. ECMO: Extracorporeal Cardiopulmonary Support in Critical Care. Ann Arbor: Extracorporeal Life Support Organization, 1995:159-190. 23. Douglas M, Birnbaum D, Eiseman B. Biological evaluation of a disposable membrane oxygenator. Arch Surg 1971; 103:89-92. 24. Wildevuur CR, Kuipers JR, Spaan JA et al. The use of a membrane oxygenator for treatment of respiratory insufficiency. Trans Am Soc Artif Intern Organs 1971; 17:362-368. 25. Montoya JP, Shanley CJ, Merz SI et al. Plasma leakage through microporous membranes. Role of phospholipids. ASAIO J 1992; 38:M399-M405. 26. Mortensen JD. An intracaval blood gas exchange (IVCBGE) device. A preliminary report. ASAIO Trans 1987; 33:570-573.

CHAPTER 3

Engineering Design of Thoracic Artificial Lungs Lyle F. Mockros and Keith E. Cook

Introduction

A

cute and chronic respiratory insufficiency continue with high mortality rates.1 Ventilators, the common mode of therapy for acute episodes, and oxygen supplies, the common mode of therapy for chronic conditions, are effective in the majority of cases but are ineffective and even damaging in many cases. Other current experimental approaches include liquid ventilation, extracorporeal membrane oxygenation, extracorporeal carbon dioxide removal, intravascular lung assist devices, and thoracic artificial lungs (TALs). The latter, the subject of this chapter, receive blood from the right ventricle and are attached in-series, in-parallel, or some combination of in-series and in-parallel with the natural pulmonary circulation. These devices may be intra-thoracic artificial lungs (ITALs) or extra-thoracic artificial lungs (ETALs). An essential function of natural lungs is to exchange respiratory gases, oxygen and carbon dioxide. Natural lungs, however, have several other functions, including metabolism of vasoactive and coagulation-modulating molecules, secretion of immunoglobulins, and filtration of emboli. The artificial lungs discussed in this chapter are intended to replace or supplement only the gas exchange function of the natural lungs. The degree to which the other functions of natural lungs are essential is largely unknown. Some experiments,2,3 however, suggest that the non gas exchange functions of lung tissue may be essential for homeostasis, in which case total removal of the natural lungs and replacement with these artificial lungs may not be feasible. Some portion of the cardiac output, thus, may have to be diverted through natural lung tissue, and TALs may have to be designed to operate in conjunction with some residual natural pulmonary circulation. Veno-arterial extracorporeal membrane oxygenation, on the other hand, essentially bypasses the pulmonary circulation for extended periods of time. Details of the device design may depend on the specific clinical condition in which it is to be used: acute, as in cases of acute respiratory distress syndrome, or chronic, as a bridge to transplant. The severity of pathologically increased pulmonary vascular resistance may also influence device design. If used for acute respiratory failure, the device possibly could be attached using minimally invasive surgery and be paracorporeal or in the extra-thoracic configuration, as an ETAL, where its function could be easily monitored. The expectation would be that the natural lungs would heal and return to near normal The Artificial Lung, edited by Steven N. Vaslef and Robert W. Anderson. ©2002 Eurekah.com.

The Artificial Lung

34

function. The device, thus, must be designed to insure that the natural lung tissue receives a blood supply that induces healing. Being paracorporeal means there are minimal constraints on the physical size and shape. The device, on the other hand, may have to be implanted intra-thoracically, as an ITAL, if it is to be used for extended chronic treatment in order to minimize the risk of infection. Such a device must be designed to be compact enough to fit within the thoracic or abdominal cavity. It would be less constrained by concerns for preserving and healing the natural lungs, but some blood flow through the natural lung, nevertheless, may be required for the non gas exchange functions of natural lungs.

General Design Considerations Four general design aspects must be considered for any artificial organ: function, hemodynamic compatibility, hematologic compatibility, and size and shape. The following is a discussion of an engineering methodology for designing thoracic artificial lungs. The general strategy, however, should be applicable to designing all blood-bearing artificial organs. The function of most importance for an artificial lung is the ability of the device to transfer oxygen and carbon dioxide to and from blood, respectively, at rates that are dictated by the medical condition. If the artificial lung is inadequate in gas transfer, other design aspects are moot. At basal conditions, the average human has a blood flow rate of about 5 l/min and requires oxygen to be supplied at a rate of about 260 ml/min and carbon dioxide to be removed at a rate of about 200 ml/min. The artificial lung must transfer these gases at these rates, as a minimum, if the natural lung is not functioning at all. If the device is to be used as an assist to a partially functioning lung, however, lesser rates may be effective. Unless the need is for the device to supply at least half of the basal rates, its use probably could not be justified. If, on the other hand, the artificial lung is to be utilized in a chronic condition for extended periods, basal rates may not be adequate. Mild exercise, such as walking around, would elevate cardiac output and place additional gas exchange requirements on the artificial lung. Cotes,4 relates cardiac output, CO, to ˙ , with the relation: oxygen consumption, Vo 2 ˙ + 3400, CO = 6.1Vo 2

(Eqn. 3.1)

˙ are in ml/min. Walking around, thus, may increase oxygen in which both CO and Vo 2 consumption to 750 ml/min and cardiac output to 8 l/min. Artificial lungs should be designed, therefore, for applications requiring, at the low end, the transfer of 130 ml/min of oxygen to 2.5 l/min of blood flow for a bedridden patient in acute respiratory insufficiency with partially functioning natural lungs, and, at the high end, the transfer of, say, 750 ml/min of oxygen to 8 l/min of blood flow for a mobile patient with chronic respiratory insufficiency with essentially total loss of natural lung function. Natural lungs have blood-side passageways at the gas exchange surfaces that are as small as erythrocytes and, as a result, are very efficient in transferring oxygen to the hemoglobin within the red cells, the oxygen carrier in blood. Currently available technology, however, does not permit the manufacture of artificial lungs with such small passages. Since red blood cells are such powerful oxygen sinks and since the blood-side passages in these devices are large relative to red blood cells, efficient oxygen transfer can occur in artificial lungs only with transverse mixing of the cells to and from the exchange surfaces. One of the most efficient such designs has the blood flowing cross-wise over the outside of

Engineering Design of Thoracic Artificial Lungs

35

hollow fibers arranged in a bundle, with the gases flowing inside the fibers. (Most currently available blood oxygenators for heart-lung machines are of this design.) The present design considerations assume such a configuration and are based on currently available technology and materials. Future designs can be expected to be even more efficient. Thoracic artificial lungs would be attached to the natural pulmonary circulation (see discussion below), and, as such, must be designed to be hemodynamically compatible. The right ventricle would supply the blood flow to the artificial lung. Since the right ventricle can fail with a fairly modest overload, especially when weakened by cor pulmonale, any increase in mechanical impedance caused by the addition of an artificial lung to the pulmonary circulation must be quite small. One measure of the mechanical load on the right ventricle is the power required to supply the necessary cardiac output. The time-average and peak power output of the right ventricle with a normally functioning natural lung is about 0.26 and 1.3 watts, respectively, under basal conditions and 0.35 and 1.7 watts, respectively, under walking around conditions. The required power output of the right ventricle depends on the attachment mode, the condition of the natural lung, and the design of the device. Because the right ventricle operates in a pulsatile mode, the artificial lung, like the natural lung, must be designed to have both a low resistance and large compliance. A low resistance lowers pressure drop and, therefore, lowers pulmonary arterial pressures, resulting in lower right ventricular work during the entire cardiac cycle. A large compliance smoothes the blood flow pulse, delivering a greater portion of the flow to the resistance during lower pressure, i.e., during diastole, and thereby reduces the required average and peak work of the right ventricle. In addition, with current technology, a smoother flow pulse results in an increased rate of gas transfer and exposes the blood to neither too high nor too low instantaneous shear rates (see discussion below). An artificial lung must be hemodynamically designed, thus, with due consideration to its resistance, Ra, and compliance, Ca. Blood flow through any artificial organ is well-known to stimulate and even lyse cells, and to activate the coagulation, clot lysis, and inflammatory systems. Stimulation of thrombocytes and leukocytes and activation of the intrinsic coagulation and complement systems are of particular interest. This blood trauma is known to be due to two mechanisms: (i) abnormally high shear stresses, shear-induced trauma, and (ii) reactions with non-biologic synthetic materials, material-induced trauma. The effects of high shear stress are known to depend not only on the magnitude of the shear but also on how long the blood is exposed to the shear. Very high stress is well tolerated if it exists for very short times, e.g., for a few milliseconds. Low shear stress, on the other hand, can cause blood trauma if sustained for long times, e.g., for a few seconds. The blood flow path through most current artificial lungs is tortuous, with complicated, time-varying shear stress. An exact calculation of the time history of this shear stress on any blood element is not practical. The trauma effect of a time-varying shear stress, furthermore, has not been determined. The basic studies that have investigated blood trauma induced by shear have almost always exposed the blood to steady-state shear stress. The present design strategy for minimizing or eliminating possible shear-induced trauma is based, therefore, on this steady-state data and considers the space-time average shear stress and average residence time in the resistive element of the device. The space-time average shear stress is calculated by noting that shear stress in an incompressible fluid is the mechanism by which mechanical energy is dissipated, and the average rate of mechanical energy dissipation on a blood element can be determined from the time-average pressure drop and flow rate (see discussion below). Artificial lungs, thus, can be designed to avoid the flow regimes where shear-induced blood

36

The Artificial Lung

trauma should occur, i.e., by designing the device such that the shear stress is below the critical levels for the required residence time. Current technology requires artificial lungs to be made of synthetic materials. Unfortunately, no synthetic material can match the hematologically compatible environment provided by the blood-endothelial cell environment. Future developments will allow, perhaps, the development of artificial lungs lined with endothelial cells. The approach at this time, however, is to understand the trauma-producing reactions at these surfaces, select minimally stimulating synthetic materials, and chemically treat the blood-contacting materials to further reduce adverse effects. Details of these approaches are discussed in another chapter in this book. The adverse effects of synthetic materials on the blood is dependent, however, not only on the chemical reactions between the material surface and the multitude of blood cells and molecules, but also on the flow-induced transport processes as the blood flows over the exchange surfaces. Stagnating the blood on the surfaces is probably the worst condition for material-stimulated blood trauma. Very slow flow or separated regions near the synthetic surfaces would be locales for the accumulation of activated procoagulant and proinflammatory factors. These activated factors may reach critical concentrations in such regions, whereas if they were carried away from the surfaces they may become diluted and/or inactivated. Material-induced blood trauma can be minimized, therefore, by minimizing the amount of surface area of synthetic material, consistent with having sufficient gas transfer area, and by designing the device such that shear rates are always maintained above certain levels. Too slow a flow, thus, enhances material-induced adverse reactions, whereas too fast a flow enhances shear-induced adverse reactions. Optimal design requires some intermediate flow regime. Size and shape considerations, within reason, are not restrictive in a paracorporeal application. For an implant application, however, the size and shape are constrained by the available space. How much space is available depends upon where the device is to be placed. Although the attachments of the device to the natural circulation would be in the thorax, possible locations of the device itself include the thoracic cavity or the abdominal cavity. If it is to be placed in the thorax, the size restraints would be determined by how much, if any, of the natural lung is to be removed. In either case, the artificial lung should have a gross volume of no more than about 2000 cm3, and preferably 1000 cm3 or less. The device, in addition to not requiring too much volume, needs to have a shape that would be suitable for implant. The shape of the bundle should not be too cube-like nor too flat, for ease of fit with the anatomic structures. The overall shape can be assumed to be a rectangular box, having a frontal area, Af, and blood path length, L, (i.e., bundle thickness). See Figure 3.1. The blood enters the bundle perpendicular to the frontal area and flows through the bundle thickness. The blood path length should be relatively short so the resistance to blood flow is small. If the frontal area is assumed to be more-or-less square, the length of one side of the square (or, more generally, the square root of the frontal area) relative to the bundle thickness, the aspect ratio = A f / L should be somewhere in the range of 2:1 to 5:1. For a given set of specified constraints, the goal is to optimize the fiber bundle with respect to size, shape, surface area, and flow characteristics.

Engineering Design of Thoracic Artificial Lungs

37

Fig. 3.1. Sketch of cross-flow fiber bundle with frontal area, Af, and blood flow path length, L.

Attachment Modes The proposed artificial lungs could be attached to the natural pulmonary circulation in several different modes, with the optimal mode depending, perhaps, on the specific pathology. All modes would route all or some of the right heart output through the artificial lung. Some or all of the cardiac output also may have to pass through at least a portion of the natural lung circulation, however. Exactly how the device should be attached depends on the: 1. 2. 3. 4. 5.

pathology, required rates of gas transfer, work load of the right ventricle with the device attached, need for non gas transfer hematological functions of the natural lung, and blood supply requirements of the natural lung tissue.

Figure 3.2 illustrates possible attachment modes. In all cases, the blood supply to the device is to be delivered by the right ventricle, and the inlet blood line to the device is to be attached to the natural pulmonary circulation via an anastomosis to the main pulmonary artery, PA. The outlet blood line from the device could be attached to the left atrium, LA, to the distal main PA, or split with some blood directed to the LA and some to the PA. A PA to LA attachment would place the device in-parallel with the natural pulmonary circulation; whereas the PA to PA attachment, with a constricting band between anastomoses, would place the device in-series with the natural pulmonary circulation. Splitting the TAL output between the PA and LA is a hybrid series/parallel attachment. The fraction of the right heart output that is routed through the artificial lung is controlled by constricting bands on the PA, the pulmonary veins, or on the outlet line of the device, along with the relative mechanical characteristics of the natural and artificial lungs. Although surgically

38

The Artificial Lung

Fig. 3.2. Possible modes for attaching a thoracic artificial lung to the pulmonary circulation. Depending on whether the various bands are open, closed, or partially open, the TAL may be attached in-parallel, in-series, or hybrid parallel/series with the natural lungs.

more difficult, constricting flow through the pulmonary veins has the advantage of having the compliance of the natural pulmonary arterial system available to the right heart. Routing all blood flow around the natural pulmonary circulation via the parallel path to the LA may not be feasible because of the need for non gas transfer physiological functions of the natural lung and blood supply requirements of natural lung tissue that is being preserved and/or treated. Routing all blood flow through the device and through the natural circulation in a PA to PA attachment also may not be advisable. This arrangement, which would require complete constriction of the PA, increases the work load of the right ventricle and may lead to right heart hypertrophy and/or failure. More feasible attachments would be to divide the right ventricular output between the device and the natural circulation and/or divide the output from the device to the PA and LA. Perhaps the most desirable attachment would be to route the total cardiac output from the right ventricle to the artificial lung with a portion of the outflow from the device diverted to the PA and a portion to the LA. The ability of any artificial lung to transfer oxygen depends on the rate of blood flow supplied to it. The higher the rate of blood flow through the device, the easier it is transfer oxygen. A portion of the outflow from the artificial lung, on the other hand, may have to be routed back to the PA for nonrespiratory functions of the natural lung and/or to supply oxygen to that tissue. The fraction of the right heart output that is routed through the artificial lung directly affects the total amount of oxygen that can be transferred to the blood by the device. The limit on the ml/min of oxygen that can be transferred is mostly determined by the hemoglobin concentration of the blood. Once the hemoglobin is saturated, additional oxygen can only be added in dissolved form, a relatively inefficient form. The artificial lung, thus, is most effective in transferring oxygen if all of the right heart output, i.e., all the hemoglobin, is routed to it. There are trade-offs, however. High flow rates through the artificial lung continually expose more blood to both shear-induced and material-induced blood trauma. Too low flow rates, on the other hand, may cause stagnant regions with their propensity for clot formation and cell adhesion. The rate of

Engineering Design of Thoracic Artificial Lungs

39

blood flow through the device should be enough to handle the gas transfer requirements, but no more. If some blood flow is required for the natural lung circulation, the hybrid attachment mode, although surgically difficult, is probably the most effective. The attachment mode, along with the mechanical characteristics of the device and the natural lungs and the flow rate directed to it, also directly affect the input impedance to the right-ventricle.

Specific Design Considerations Thoracic artificial lungs must be designed to meet requirements for 1. 2. 3. 4.

oxygen and carbon dioxide transfer, resistance and compliance, blood trauma, and size and shape.

They are constrained by the 1. patient’s blood characteristics and 2. available technology and materials.

These devices will have two basic mechanical elements: the resistive element or fiber bundle, where the gas exchange takes place, and the compliant element. The design calculations in this and the next section are based on assuming a steady flow through the device and, thus, are a discussion of the fiber bundle or resistive element. The effects of pulsatility and the compliance part of the device are discussed in a later section. This chapter deals with all aspects of the fiber bundle design, except material selection. A discussion of the effect of material selection and material surface treatments appears in another chapter of this book.

Design: Gas Transfer and Fluid Mechanics The gas transfer and fluid mechanical aspects of the typical design problem would be to specify, depending on the specific pathology, the desired rates of oxygen and carbon dioxide transfer, blood flow rate through the device and required resistance and compliance of the device. Another specification for a particular design would be the outlet PO2. The blood flow rate to be diverted through the device could range from the total expected maximum cardiac output to some fraction thereof. The ability of the device to transfer oxygen depends on having access to a sufficient amount of blood. Transferring oxygen to blood that already has saturated hemoglobin, although possible, is very inefficient, thus the design should be such that exiting hemoglobin is saturated but with little excess oxygen in dissolved form, i.e., the outlet PO2 should be about 150 mm Hg and the hemoglobin 99% saturated. The specifications, thus, are that the blood flow rate through the device should be such that the partial pressure of oxygen in the blood at the outlet is 150 mm Hg, while transferring oxygen at the specified rate, which, in turn depends on metabolic requirements. If, according to Equation 3.1, the cardiac output is 4, 5, 6, 7, or 8 l/ min, the required metabolic rate of oxygen consumption is about 100, 260, 420, 590, or 750 ml/min, respectively. For example, if one assumes the hemoglobin concentration is 0.12 g/ml and the hemoglobin is saturated when the outlet oxygen partial pressure is 150 mm Hg, the venous partial pressure would be 51.6, 35.9, 30.0, 26.3, or 24.1 mm Hg, respectively. These designs are based on the assumption of a normally functioning heart, i.e., the cardiac output is not limiting the ability of the blood to carry the required respiratory oxygen. A specification for the rate of blood flow through the device is essentially dictated

The Artificial Lung

40

by the hemoglobin concentration in the blood, the venous inlet saturation, and the desired rate of oxygen transfer by the artificial lung. The oxygen transfer rate is related to these terms by: (Eqn. 3.2)

in which Qb = blood flow rate, [Hgb] = hemoglobin concentration, Pi = partial pressure of oxygen in the inlet venous blood, Po = partial pressure of oxygen in the outlet blood, n and P50 are the parameters in the Hill equation for the oxyhemoglobin dissociation relation, and k is the oxygen solubility in the blood. Blood has a limited capacity to carry oxygen once the hemoglobin is saturated (the last term in the above equation), so hemoglobin must be diverted through the device at a sufficient rate to absorb oxygen at the desired rate. Assuming, as indicated above, the exchange surfaces are a bundle of fibers, the only choices open to the designer are the (a) fiber size, within the limits of what is available, and (b) fiber spacing, which determines the bundle void fraction, Vf , i.e., blood space volume as a fraction of gross fiber bundle volume. The overall size of the bundle is mostly dictated by the specifications for gas transfer. The shape, on the other hand, is determined by the blood flow rate and the allowable bundle resistance, i.e., the allowable pressure drop across the bundle. A bundle with a large frontal area and short thickness would have a relatively small pressure drop, whereas a bundle with a small frontal area and a long blood path would have a relatively large pressure drop for the same blood flow rate and overall bundle volume. The pressure drop for a given bundle shape will depend, of course, also on the tightness of the fiber spacing, i.e., the bundle void fraction. The rate of gas transfer by the fiber bundle is often expressed as:

Sh = ∃Re % Sc 1 / 3 ,

(Eqn. 3.3)

in which Sh = Sherwood number = KL/ D, K = mass transfer coefficient, L = characteristic length = dh = hydraulic diameter of the bundle = 4(void volume)/(surface area) = Vf d / (1-Vf), d = outside diameter of fibers, D = molecular diffusivity, Re = Reynolds number = dh v&/∋ = Qb d&/(1-Vf)Af ∋, v = average velocity of blood within the bundle, & = blood density, ∋ = blood viscosity, Sc = Schmidt number = ∋/(&D), and ∃ and % are constants that depend on bundle void fraction and fiber arrangement and can be determined from experiments. For a blood flow rate Qb and a bundle with a frontal area Af, v = Qb / Vf Af. Details of the gas transfer theory are presented in Mockros and Leonard,5 Vaslef et al.,6 and Vaslef et al.,7 and will not be repeated here. The governing equation for oxygen transfer is, however: (Eqn. 3.4) in which P = partial pressure of oxygen in the blood, Pb = partial pressure of oxygen at the fiber surface, and (= slope of the oxyhemoglobin dissociation relation. The flow through such a bundle of fibers is akin to flow through a packed bed, and the pressure drop for flow through a packed bed is typically written as:

Engineering Design of Thoracic Artificial Lungs

41

L (Eqn. 3.5) &v 2 , 2d h in which ∀p = pressure drop across the bundle, fD = Darcy friction factor, and L = blood path length, (i.e., bundle thickness). The friction factor will depend on the Reynolds number, Re, of the flow through the bundle. For low Reynolds numbers, as would be the case here, ∀p = f D

fD =

∀p =

B Re

(Eqn. 3.6)

BL(1 ! V f )2 ∋Qb 2d 2V f3 A f

(Eqn. 3.7)

in which B = a constant depending on the bundle void fraction and fiber arrangement and also must be determined from experiments. If one specifies (i) fibers of diameter d, (ii) bundle Vf (which determines B, ∃, and %), (iii) blood flow rate Qb, (iv) bundle resistance or allowable pressure drop, ∀p, for the blood flow rate, (v) blood characteristics (i.e., hematocrit, which affects density, &, viscosity, ∋, and the molecular diffusivity of oxygen and carbon dioxide, and the oxyhemoglobin relation), and (vi) required rate of gas transfer, these equations can be used to calculate the necessary bundle thickness, L, bundle frontal area, Af (and, therefore, gross bundle volume, Af L), and aspect ratio, = A f / L , fiber surface area, As, and the Reynolds number, Re. By considering various fiber diameters and bundle void fractions, the designer can see the tradeoffs for various combinations of diameters and void fractions for the required rate of gas transfer, allowable pressure drop, and specified blood flow rate.

Design: Right Ventricular Power Requirements The work load of the right heart may be estimated by considering the mechanical impedance faced by the right ventricle. Both the natural circulation and the artificial lung may be modeled as simple windkessels, each having compliance, C, and resistance, R. The adjustable bands on the PA, pulmonary veins, and the device outflow track also provide additional resistances. In all cases, however, the mechanical impedance to right heart output may be reduced to an equivalent compliance, Ce, and equivalent resistance, Re, which, in turn, depend on the compliance, Cn, and resistance, Rn, of the natural lung, the compliance, Ca, and resistance, Ra, of the artificial lung, the constriction-resistances, Rcis, and the specific attachment mode. The value of Cn and Rn are set by the physiology and, in many cases of interest here, the pathology. For a healthy natural lung, Cn ) 2.7 ml/mm Hg and Rn ) 1.25 mm Hg/(l/min). With severe pulmonary hypertension, however, Rn may be 6 mm Hg/(l/min) or more. The artificial lung designer, on the other hand, can control the work load of the right heart by properly selecting Ca, Ra, Rcis, the attachment mode, and the position of Ca relative to Ra and the Rcis. Artificial lung compliance, Ca, for instance, could be incorporated in the inlet graft and/or as the fiber bundle housing. Right heart load may be quantified in terms of time-averaged right ventricular power: PRV =

1 TS

TS

∗ p PAQ 0

RV

dt

(Eqn. 3.8)

42

The Artificial Lung

Fig. 3.3. Blood trauma chart. The thresholds for shear-induced hemolysis and platelet release are based on numerous, previously reported studies. ■ Data collected by Wurzinger et al.12 showing the stress-time conditions necessary to shorten the Stypven clot time by 20%. • Data collected by Chow et al.11 showing the stress-time conditions necessary to induce 20% platelet aggregation. The threshold for shear-induced trauma is an arbitrary extrapolation from the above data to mark the speculative limit that one wants to stay below to avoid shear-induced blood activation. The equation of this line is +2 = 900/t + 225. The line for the minimum condition recommended for material-induced blood activation is very speculative and based on anecdotal information; one wants to stay above that line to minimize material-induced trauma. The equation of this line is +2 = -2/t + 64.

in which PRV = time-average right ventricular power requirements, TS = systolic duration, pPA = pulmonary artery pressure, and QRV = right heart output. Both pPA and QRV are functions of time. In addition to the time-average power, the instantaneous peak power requirement for the right ventricle, if excessive, could induce right heart failure.

Design: Blood Trauma Blood flow through artificial organs has the potential to activate cellular and molecular components of the coagulation, inflammation, and plasminogen systems due to the presence of excessively high shear stress, shear-induced trauma, and the synthetic materials, material-induced trauma. The designer, therefore, must address how each combination of fiber diameter and bundle void fraction affects potential blood trauma in the device. Numerous studies have been conducted relating controlled shear stress to various forms of blood trauma: e.g., hemolysis, sub-lethal red cell alterations, platelet lysis, platelet release, and leukocyte activation. These all involve subjecting blood or platelet rich plasma to various levels of shear stress for various exposure times. The trauma depends both on the level of shear stress in the blood and the time of exposure to the shear stress. The many published studies of red cell lysis and platelet release have been summarized by several

Engineering Design of Thoracic Artificial Lungs

43

authors.8-10 Figure 3.3 includes a composite from those summaries, in which relations for the red cell lysis and platelet release thresholds are shown as dashed lines. These thresholds for red cell lysis and platelet release may be functionally expressed as: +2=k1/t+k2, in which + = shear stress in dynes/cm2, t = time in seconds at that shear level, and k1 and k2 are constants. The threshold for hemolysis may be approximated with k1 = 104 (dynes2•sec)/ cm4 and k2 = 2.2 x 106 dynes2/ cm4. The threshold for platelet release may be approximated with k1 = 3 x 105 (dynes2•sec)/cm4 and k2 = 1.2 x 104 dynes2/ cm4. For thoracic artificial lungs, however, the concern is for forms of blood trauma more subtle than cell lysis or platelet granular release, such as up-regulation of surface active molecules and release of pro-coagulant and pro-inflammatory molecules by thrombocytes and leukocytes. Relatively few studies are available that quantitatively relate shear stress to these forms of trauma, but generally these effects would be expected at lower levels of shear for a given exposure time. Chow et al.,11 for instance, using an optical technique with a specially-made cone-plate viscometer, could track intracellular platelet calcium and platelet aggregation as a function of time while subjecting platelet suspensions to uniform shear fields of 15, 30, 60, 90, and 120 dynes/cm2. They found essentially no response at 15 dynes/cm2, but progressively more response with increases in shear stress. Figure 3.3 shows data, based on their measurements at 30, 60, 90, and 120 dynes/cm2, at the times after initiation of the shear stress when the platelets were 20% aggregated. Longer exposures showed progressively more aggregation at each stress level. Wurzinger et al.,12 using a flow-through Couette viscometer in which they could expose platelets to uniform and well controlled shear stress for short exposure times of 7 to 700 ms, found platelet procoagulant activity (PF3), expressed as a percentage shortening of Stypven time, as the parameter most sensitive to shear stress. Figure 3.3 also shows the stress, based on their measurements, at which the Stypven time was shortened by 20% for exposures of 7, 27, and 113 ms. Higher stress levels produced a greater shortening. Insufficient data are available to know exactly the threshold for these more subtle forms of trauma, but an extrapolation may be conjectured from the platelet release threshold summarized above. One such extrapolation, here referred to as the threshold for shear-induced trauma, is given by the relation +2 = 900/t+225, which is shown as a continuous line on Figure 3.3. Virtually all available data relating blood trauma to shear stress, furthermore, are for steady-state exposures to the shear. The blood passing through the fiber bundle of these artificial lungs will be pulsatile on two time scales, one because the flow rate, driven by the cyclically beating right ventricle, will be pulsatile and another because of the complex tortuous path around the fibers in the bundle. Since little data are available for blood trauma due to time-varying shear stresses, the exposure time for the present work is taken as the average transit time, t t , through the bundle, and the shear stress is taken as the time-space average shear stress, + . The average transit time through the bundle is t t = V / Qb in which V = Vf Af L is the blood prime volume of the fiber bundle and Qb is the time-average blood flow rate. The average stress, + , can be calculated by noting that shear stress is the mechanism by which mechanical energy is dissipated in incompressible flows, and the time-space average rate of mechanical energy dissipation is easily calculated and/or measured. (Rather than relate blood trauma to shear stress, one can relate it, alternatively, to the rate of mechanical energy dissipation.13) The local rate of mechanical energy dissipated per unit volume is related to the shear stress by , = (1/ ∋)(+:+), in which , is the dissipation function, ∋ is the fluid viscosity, and + is the stress tensor. The average rate of mechanical energy dissipation, , ,is reflected in the time-average

The Artificial Lung

44

pressure drop across the bundle, ( ∀p) , neglecting gravity effects, and the average transit time, i.e., , = (1 / ∋)(+ : + ) = ( ∀p) Qb /V . The root-mean-square shear stress, + , thus, may be calculated as:

+ − [(+ : + )]1 / 2 = [∋( ∀p)(Qb )/V ]1 / 2 = (B / 2)1 / 2[∋(1 ! V f )/(V f d )]v ,

(Eqn. 3.9)

in which all the terms are easily calculated or measured. For a given bundle configuration and blood viscosity, the space-time average shear stress is proportional to the space-time average velocity, v = Qb /Vf Af, through the bundle. Shear-induced blood trauma decreases as the average velocity decreases. To minimize or eliminate shear-induced blood trauma, therefore, the design should be such that the average shear stress and residence time satisfy the relation:

+ 2 ! 900 / t t ! 225 < 0,

(Eqn. 3.10)

in which the constants, 900 and 225, are from the shear-induced trauma threshold extrapolation mentioned above. The quantity + 2 ! 900 / t t ! 225 may be referred to as the shear-induced trauma factor, Tf,s, i.e., the design should be such that Tf,s < 0. Using the condition that Tf,s < 0 to avoid shear-induced blood trauma, to repeat, is a conjecture based on an extrapolation from the data for more severe forms of trauma. Although some data are available that suggest the functional dependence of shear-induced trauma on shear stress and exposure time, little, if any, quantitative experimental data are available that suggest the functional dependence of material-induced trauma on flow conditions. Furthermore, whereas shear-induced trauma often manifests itself by cell stimulation, molecular processes, particularly the intrinsic clotting cascade, is the major concern with material-induced trauma. The process may be quite complicated, involving the surface properties of the specific synthetic material, the rate of transport of reactants to and from the surfaces, the reaction kinetics, and the maintenance of sufficient concentrations of activated factors at or near the surfaces. The transport rates to and from the synthetic materials depend linearly on the shear rates near the surfaces, which, in turn, depend on the blood velocity past the surfaces. High velocities produce high transport rates to and from the surfaces and provide a mechanism for bringing fresh reactants to the surfaces and/or pushing reactant products away from them. Low velocities, on the other hand, may be conducive to building up the concentration of activated factors at the surfaces to levels sufficient for cascading reactions. It has been generally observed both in vivo and in vitro,14 indeed, that clots tend to develop in regions of slow or stagnant flow. One can speculate that if the velocity is very slow that activated contact factors tend to accumulate near prosthetic surfaces and that these available layers of high concentrations of activated contact factors stimulate the myriad of coagulant and inflammatory reactions in the bulk flow. As Spaeth et al.15 summarized, stagnant blood flow may not initiate thrombogenesis, but it certainly promotes thrombogenesis. High velocities may tend to wash away the surface-activated contact factors, diluting them in the bulk blood and, perhaps in some cases, routing them to sites of physiological clearance. A common observation is that contact factor XII attaches to most “biocompatible” synthetic materials and converts to FXIIa as soon as the blood comes in contact with the material. This leads to conversion of XI to XIa and so on through the common pathway to the production of thrombin and subsequent fibrin clots. Basmadjian et al.16 have devel-

Engineering Design of Thoracic Artificial Lungs

45

oped theoretical models of the kinetics of these convective-diffusive processes and predict several interesting regimes, including cases with (i) little FXIIa, but substantial amounts of thrombin and (ii) substantial levels of initiating coagulants, but vanishing amounts of thrombin. The regimes depend, of course, on the surface reactivity of the synthetic material, the reaction-kinetics, and the convective processes. Their models are for steady state flow within a circular tube, but the general conclusions are probably applicable to the complicated flows through these fiber bundles. In one case of the flow within a 10 cm long tube, they conclude that very low thrombin production occurs if the shear rate is greater than 100 s-1. If one assumes a blood viscosity of about 0.03 poise, this shear rate corresponds to a shear stress of about 3 dynes/cm2. Spaeth et al.,15 on the other hand, have shown, in some carefully designed experiments, that shear stress of as little as one dyne/ cm2 reduces the rate of FXa activation of prothrombin by 50 to 70% compared to the rates in stagnant pools. Conservatively extrapolating from these values, one might want to design the device such that the shear stress exceeds, say, 4 dynes/cm2 for the longer exposure times. Whereas shear-induced trauma is known to depend on both the level of shear and the time of exposure to shear at that level, neither the form nor the magnitude of the time dependence has been quantified for material-induced blood trauma. One might expect, however, that, due to reaction kinetics, very low shear rates may not produce material-induced trauma if they occur for only a very short time, e.g., milliseconds, whereas long exposures, e.g., seconds or tens of seconds, too low shear rates may be conducive to sequential reactions at the surfaces. Also shown on Figure 3.3 with a solid line is a speculative minimum recommended level of shear stress. This lower bound is asymptotic to 4 dynes/cm2 for long exposure times and decreases for exposure times less than one second. The shear stress should exceed this level at all times and locations to avoid, as much as possible, conditions conducive to material-induced blood trauma. Because only anecdotal experimental information is available, this material-induced limit is even more speculative than that for shear-induced trauma. This minimum will depend on the specific synthetic material used as well as any surface treatment of that material, but little is known quantitatively about how the transport processes affect reactions with any specific blood-material interface. An average velocity and, therefore, an average shear stress/shear rate at some intermediate value is expected to produce the least blood trauma. From the standpoint of shear-induced trauma, the lower the shear rate the better. From the standpoint of material-induced trauma, on the other hand, theory and observations suggest that too low a shear rate is disadvantageous. The region between the threshold for shear-induced trauma and the minimum recommended for material-induced trauma may be considered the “safe” zone for a design. The present study, thus, speculates that an upper bound in the stress-time domain, to minimize blood trauma, is provided by Tf,s = 0. The lower bound is provided by a material-induced trauma factor, Tf,m, which for this case is speculated to be Tf,m = + 2 + 2 / t t ! 16 = 0. and the design should be such that: .

Tf,m = + 2 + 2 / t t ! 16 > 0.

(Eqn. 3.11)

Too high an average velocity through the bundle produces shear-induced trauma, whereas too low an average velocity produces material-induced trauma. The “safe” zone for the designer, therefore, is some intermediate range of average velocity, i.e., an intermediate

46

The Artificial Lung

range of the shear stress-time domain. The shear stress and residence time in the fiber bundle, thus, should be such that Tf,s < 0 < Tf,m, i.e., within the “safe” zone, as shown on Figure 3.3. Lastly, it is also desirable to minimize fiber bundle surface area to reduce material-induced blood activation. Even if size and space requirements are minimal, the designer must be careful not to over design the fiber bundle for gas transfer purposes with an excessive fiber surface area. Material-induced blood trauma can only be minimized by both controlling the local shear fields near the material surfaces and by minimizing the amount of that surface area.

Example Steady Flow Designs with Current Technology A specific design of an artificial lung is determined by the specifications (the required O2 and CO2 transfer rates, the blood characteristics, and the compliance and resistance of the natural lungs of the patient) and the choices made by the designer. Based on the required oxygen transfer rate, the designer will have to specify the blood flow rate that must be diverted through the artificial lung. He/she can select the fiber size and the bundle void fraction (i.e., fiber spacing) to meet the required patient specifications. For any set of specifications and designer choices, the above theoretical relations can be used to calculate the necessary fiber surface area, bundle gross volume, bundle aspect ratio, blood flow Reynolds number, and blood trauma factors. The specific design will depend, as mentioned above, on the specific application. In all cases, assume the blood of the patient is that of a normal adult, with the oxygen-dissociation relation given by the Hill equation: S = (P/P50)n/[1+(P/P50)]n, in which S = fractional oxyhemoglobin saturation, P = oxygen partial pressure, and P50 and n are Hill parameters and are taken, for humans at 37°C, as 26.6 mm Hg and 2.7, respectively.17 Assume oxygen is supplied to the gas side at a rate that is adjusted to control carbon dioxide removal so as to maintain a normal respiratory quotient and that the oxygen partial pressure on the gas side is maintained at 700 mm Hg. Assume the body temperature is 37°C, the hematocrit is 36% and the hemoglobin concentration is 0.12 g/ml. The blood density, &, viscosity, ∋, and oxygen solubility, k, are calculated according to: & = 1.09 (Hct) + 1.035 (1-Hct) g/ml, in which Hct = fractional hematocrit, ∋ = 0.0124 exp (2.31(Hct)) poise, and k = 1.803x10-5 (Hct) + 2.855x10-5 ml/(ml•mm Hg), respectively. The effective oxygen diffusivity is calculated using the relations of Fricke.18 The required rate of oxygen transfer by the device is dictated by the specific pathology, i.e., the requirement of a total lung function replacement or a lung assist device, and patient state, e.g., basal conditions, walking around, etc. Since the ability of blood to carry oxygen is essentially dictated by its hemoglobin concentration, the required blood flow rate through the device is dictated by the required oxygen transfer rate, the hemoglobin concentration of the blood, and the venous oxyhemoglobin saturation. According to the relation given in equation 3.1, as the level of exercise increases, the rate of metabolic oxygen consumption goes up more rapidly than the increase of cardiac output, implying the venous oxyhemoglobin saturation, or venous oxygen partial pressure, decreases with increased activity. Since the arterial oxygen partial pressure in normal physiology is about 150 mm Hg and the cardiac output and oxygen transfer rate are related through the Cotes relation, the venous oxygen partial pressure can be calculated using the Hill oxyhemoglobin dissociation relation. The calculations for devices that supply total oxygen requirements, thus, are based on specifying a required rate of oxygen transfer by the device, using the Cotes relation to calculate the required blood flow rate through the device, specifying

Engineering Design of Thoracic Artificial Lungs

47

that the oxygen partial pressure at the device outlet should be 150 mm Hg, and back calculating the venous oxygen partial pressure. For devices that are to act as partial support artificial lungs, the calculations are based, again, on specifying a required rate of oxygen transfer, inputting the expected venous oxygen partial pressure, specifying that the oxygen partial pressure at the device outlet should be 150 mm Hg, and back calculating the required blood flow rate through the artificial lung, which may be a fraction or all of the cardiac output. Two specific cases, as examples, are considered here: (1) Suppose one wants a partial support artificial lung that will supply 208 ml/min of oxygen to a subject that has a venous oxygen partial pressure of 36 mm Hg. To deliver 208 ml/min of oxygen and have an outlet oxygen partial pressure of 150 mm Hg, 4 l/min of blood flow would have to be diverted through the device. Assume the pressure drop across the fiber bundle is to be 5 mm Hg at the blood rate of 4 l/min. (2) Suppose one wants an artificial lung that will supply all the required oxygen for a typical adult that is walking around, an adult with a cardiac output of, say, 8 l/min and a requirement of about 750 ml/min of oxygen. The device is designed such that the outlet oxygen partial pressure is 150 mm Hg. At this blood flow rate, oxygen consumption rate, and hemoglobin concentration, the venous oxygen partial pressure normally would be 24 mm Hg. The calculations are based on the assumption that the device bundle can have an average pressure drop of 10 mm Hg with the blood flowing at a time-average rate of 8 l/min. Both cases consider the effects of making the bundles with microporous fibers that have outer diameters of either 200, 300, or 400 ∝m, reasonable sizes with currently available technology. The fiber walls, being microporous with, perhaps, a one micron thick continuous coating to prevent plasma leakage and/or water accumulation in the micropores, are assumed to offer no resistance to the gas transfer. Both cases also consider the effects of arranging the fibers such that the fiber bundles have void fractions ranging from 0.2 to 0.8. Figures 3.4a – 3.4f show the resulting characteristics of case (1), the artificial lung for partial support. Figure 3.4a indicates that the total surface area can be minimized by choosing the smallest diameter fiber available and utilizing large bundle void fractions. With 0.70 . Vf . 0.80, 200 ∝m diameter fiber bundles would require between 1.0 to 1.5 square meters of surface area. Figure 3.4b indicates the bundle volume increases with increases of fiber diameter or bundle void fraction, but all gross volumes are reasonable for this case, especially if one chooses a 200 ∝m fiber. Figure 3.4c indicates that reasonable bundle aspect ratios could be achieved with any of the fiber diameters considered, but that with fiber diameters of 200 ∝m the fiber bundle should have a bundle void fraction ranging from 0.74 to 0.80 to stay within the reasonable bounds of 2:1 to 5:1. Figure 3.4d indicates that the shear stress increases with decreased fiber diameters, but the root mean square shear stress will be quite low, between 4 and 9 dynes/cm2, for all fiber diameters at the higher void fractions. The issue is not so much the potential for shear-induced trauma, but rather the potential for material-induced trauma brought on by too low a shear stress/shear rate. Figure 3.4e shows the mean transit time through the fiber bundles. It is 2 1/2 to 3 seconds for bundles made with 200 ∝m fibers and 0.70 . Vf . 0.80. Figure 3.4f, showing the trauma factors, indicates the shear-induced trauma factors are strongly negative, as desired. The material-induced trauma factors, although positive as desired, push the limit, especially at large void fractions. Since shear stress is directly proportional to flow rate, this figure also suggests that the potential for material-induced trauma will be significant at lower blood flow rates. Weaning on and off the device must be done with caution. Figure 3.5 shows a section of Figure 3.3 with the loci of stress vs. transit time positions for differ-

48

The Artificial Lung

Fig. 3.4a. The results of design calculations for a fiber bundle based on specifying: (i) a blood flow rate of 4 l/min, (ii) an oxygen transfer rate of 208 ml/min, and (iii) a pressure drop across the bundle of 5 mm Hg for a blood with a hemoglobin concentration of 0.12 g/ml. Bundle characteristics are shown as a function of bundle void fraction for three different fiber diameters. Required bundle surface area.

Fig. 3.4b. Gross volume of bundle.

Engineering Design of Thoracic Artificial Lungs

Fig. 3.4c. Bundle shape.

Fig. 3.4d. Root-mean-square shear stress on blood.

49

50

Fig. 3.4e. Blood transit time through bundle.

Fig. 3.4f. Material-induced and shear-induced trauma factors.

The Artificial Lung

Engineering Design of Thoracic Artificial Lungs

51

Fig. 3.5. The loci, on the shear-stress/time diagram, for different blood flow rates for the fiber bundle used by Cook et al.19 Weaning on or off the device at too low a flow could potentiate material-induced activation.

ent blood flow rates in the ITAL described by Cook et al.19 If one wishes to use large bundle void fractions to minimize material surface area, the best choice is to use fibers with small diameters that increase root-mean-square shear stress. If one chooses fibers with diameters of 200 ∝m and bundle void fractions of 0.70 to 0.80, the shear stress is still only about 8 dynes/cm2 at a blood flow rate of 4 l/min. If one wanted a device such that average shear stress was maintained above, say, 10 dynes/cm2, one would have to use a fiber bundle void fraction of 0.55 or less. This, however, would result in a total bundle surface area of over two square meters, nearly double that required with, say, a void fraction of 0.75. Exactly the trade-off between the amount of surface area and the transport mechanisms to that surface area and the potential for material-induced trauma is unknown. A better idea, however, may be to alter the design specifications and increase the allowable pressure drop across the bundle. The Reynolds number, not shown, is less than 10 for all cases considered and is within a reasonable range. Within this range, the higher the Reynolds number, the higher the mass transfer coefficient. The Reynolds number increases with increased bundle void fraction and increased fiber diameter. It rises sharply as bundle void fraction increases above 0.70, the range desirable for many other reasons. The reason the required fiber surface area decreases with increasing bundle void fraction, for the larger void fractions, undoubtedly is due to the high mass transfer coefficients with these relatively high Reynolds number flows. Figures 3.6a – 3.6f show the resulting characteristics of the artificial lung that would deliver 750 ml/min of oxygen to blood flowing at a rate of 8 l/min, the second case. This,

52

The Artificial Lung

Fig. 3.6a. The results of design calculations for a fiber bundle based on specifying (i) a blood flow rate of 8 l/min, (ii) an oxygen transfer rate of 750 ml/min, and (iii) a pressure drop across the bundle of 10 mm Hg for a blood with a hemoglobin concentration of 0.12 g/ml. Bundle characteristics are shown as a function of bundle void fraction for three different fiber diameters. Required bundle surface area.

Fig. 3.6b. Gross volume of bundle.

Engineering Design of Thoracic Artificial Lungs

Fig. 3.6c. Bundle shape.

Fig. 3.6d. Root-mean-square shear stress on blood.

53

54

Fig. 3.6e. Blood transit time through bundle.

Fig. 3.6f. Material-induced and shear-induced trauma factors.

The Artificial Lung

Engineering Design of Thoracic Artificial Lungs

55

clearly, is a most demanding case for current technology. Figure 3.6a shows the required surface area of the bundle for various choices of fiber diameter and bundle void fraction. To keep the fiber surface area as small as possible, the designer would choose as small a fiber diameter as is available and a fairly large bundle void fraction, say 0.7 . Vf . 0.8. With a fiber diameter of 200 ∝m and a bundle void fraction of 0.75, say, the device would still require a surface area of about 3.5 square meters. Figure 3.6b shows that gross fiber bundle volume is very sensitive to fiber diameter, and that if one wants a gross fiber bundle volume of 1000 cm3 or less and a fiber bundle void fraction of, say, 0.75, one would have to utilize fibers with outer diameters of no more than about 250 ∝m. In all cases, total bundle volume is reduced by reducing bundle void fraction. For most other purposes, however, bundle void fraction should be large. Since the gross bundle volume is quite reasonable in all cases with 200 ∝m fibers, choosing a large void fraction is acceptable. Figure 3.6c shows that the fiber bundle aspect ratio is very sensitive to bundle void fraction. If one wants an aspect ratio of between 2:1 and 5:1, high void fractions are required. If, to keep the surface area as small as possible, one chooses 0.70 . Vf . 0.80, again only small diameter fibers are suitable. With a fiber outer diameter of 200 ∝m and a bundle void fraction of 0.75, the bundle aspect ratio would be about 3.2, a reasonable shape. Figure 3.6d shows the effect of fiber diameter and bundle void fraction on the root-mean-square shear stress that the blood is subjected to as it flows through the bundle. With large void fractions and large diameter fibers, the shear stress is close to pushing the limit for material-induced trauma. Again, 200 ∝m fibers would be the best choice to avoid material-induced trauma. Figures 3.6e and 3.6f show the average blood transit time in the bundle and the trauma factors, respectively. The average transit time is 3 1/2 to 4 seconds, using 200 ∝m fibers and 0.70 . Vf . 0.80. In all cases the shear trauma factor is significantly negative, again indicating there should be little or no shear-induced trauma to the blood components. Figure 3.6f, however, indicates that the shear stresses can be quite low with large bundle void fractions, suggesting a possible source for material-induced trauma. Again, bundles with 200 ∝m fibers produce the most positive material trauma factors with the desirable larger void fractions. Not shown, but also calculated is the Reynolds number of the blood flow for these three fiber diameters and bundle void fractions ranging from 0.20 to 0.80. The above theory for both gas transfer and pressure drop are valid only if the Reynolds number is less that about 10. In all cases considered here, the Reynolds number is less than 10. With a fiber diameter of 200 ∝m and bundle void fractions ranging from 0.70 to 0.80, the Reynolds number would range from 1.5 to 3.9. In summary, a very feasible device is possible for this application if one utilizes fibers with outer diameters of about 200 ∝m and arranges them in bundles with void fractions of between 0.70 and 0.80. The major negative to such an artificial lung would be the relatively large surface area, about 3.6 to 4.0 square meters. This large surface area, relative to that of currently available heart-lung machine oxygenators, is a result of requiring such a low pressure drop, i.e., 10 mm Hg, and the required high rate of oxygen transfer, i.e., 750 ml/min. If, for example, the allowable pressure drop were 100 mm Hg, typical of current heart-lung machine oxygenators, a device required to transfer 750 ml/min of oxygen to 8 l/min of blood would require only 2.1 square meters of surface area if the bundle void fraction were 0.75. Suppose, as a summary for this second example, one sets specifications requiring the material surface area to be less than 4 m2, the gross volume of the fiber bundle to be less than 1000 cm3, the bundle aspect ratio to be between 2 and 5, Tf,s < 0, and Tf,m > 0, these analyses indicate the bounds on the choice of fiber diameter and bundle void fraction. For any combination of fiber diameters between 200 and 400 ∝m and void fractions between

56

The Artificial Lung

Fig. 3.7. A composite of the results of the design calculations for the fiber bundle based on specifying (i) a blood flow rate of 8 l/min, (ii) an oxygen transfer rate of 750 ml/min, and (iii) a pressure drop across the bundle of 10 mm Hg for a blood with a hemoglobin concentration of 0.12 g/ml. The design criteria are that surface area < 4 m2, fiber bundle gross volume < 1000 cm3, aspect ratio between 2 and 5, shear-induced trauma factor is negative, material-induced trauma factor is positive. The trauma factor criteria are met for all combinations of fiber diameter and bundle void fraction. The shaded area indicates the only region where the other three criteria are met.

0.2 and 0.8, the trauma factor criteria are satisfied (although the material trauma may be borderline, see Figure 3.6f ). Figure 3.7, as a composite result, shows the restrictions placed on the design by the specifications for surface area, gross volume, and aspect ratio. The shaded area in the lower right corner is the only combination of fiber diameter and void fraction that satisfies all three restrictions. The design choice should be fiber diameters between 200 and 230 ∝m and bundle void fractions between 0.70 and 0.80. Figure 3.8 shows the loci of design results in a shear stress vs. transit time figure for the two cases considered. These results assume 200 ∝m fibers are utilized and show the design position on the stress vs. time graph for void fractions ranging from 0.20 to 0.80. In all cases, the designs fall within the “safe” zone for blood trauma. One method for moving these designs more into the safe zone for the large void fractions would be to increase the allowable pressure drop. Too high a bundle resistance, however, might place too high a work load on the right ventricle, but this could be manageable by providing adequate device compliance, see below.

Effects of Pulsatility Blood flow through artificial lungs attached to the pulmonary circulation, being generated by the cyclic contraction of the right ventricle, will be pulsatile. The right ventricular pulse is damped before entering the device by the effective time constant, ReCe, of the

Engineering Design of Thoracic Artificial Lungs

57

Fig. 3.8. Loci of design results in the stress-time plane for the range of fiber bundle void fractions considered for the two examples considered above, assuming fiber outer diameters of 200 ∝m.

combined natural and artificial lungs. The pulse can be adequately damped by a large resistance and/or a large compliance. A large resistance, however, would cause a large pulmonary artery pressure and excessive afterload for the right ventricle. An adequate compliance, therefore, must be built into the device. The larger this compliance, the lower the pulse height as it flows through the fiber bundle. Since they are functions of blood flow rate, the instantaneous gas transfer rate, shear stress, and rate of right ventricular work will be time dependent with maximum and minimum values higher and lower than their time-average values. Since the rates of gas transfer and right ventricular work are nonlinear functions of flow rate, furthermore, their time-average values will be a function of pulse height. Relative to their values at the mean flow rate, these functions will deteriorate as the pulse height increases, i.e., relative to their values at the mean flow rate, the rate of gas transfer will be less, whereas the required right ventricular power will be more. Shear stress, being a linear function of flow rate, will show little change in its time-average value with degree of pulsatility. The instantaneous shear stress, however, can be quite high or low if the pulse height is high through the device. The larger the effective compliance, Ce, of the combined natural and artificial lungs, the better the performance of the device. The thoracic artificial lungs, thus, consist of two major components, the fiber bundle, which is where the gas transfer takes place and is essentially a resistance element, Ra, and the compliance, which damps the ventricular pulse and is essentially a capacitance element, Ca. The quantitative importance of the compliance is considered in this section. The effect of compliance on device function can be illustrated, for example, by considering a TAL that has the fiber bundle, referred to here as ITAL Model 1, that was used in the experiments reported in Cook et al.19 The fibers of this bundle have outside diameters of 209 ∝m and are arranged such that the bundle void fraction is 0.74. The bundle

58

The Artificial Lung

Fig. 3.9. The flow pulse wave shape coming out of the right ventricle and that through the fiber bundle with some system compliance.

has a frontal area of 109 cm2 and a thickness of 3.5 cm. Figure 3.9 shows the effect of two different compliances on the shape of the flow pulse through the device. The larger the compliance the lower the flow pulse height and the more the flow through the bundle approaches a steady rate. Reducing the flow pulse height is essential for assuring that the 1. right ventricle is not overloaded, 2. desired rate of oxygen transfer is achieved, and 3. shear stresses are maintained within reasonable bounds throughout the heart cycle.

The pressure pulse height in the proximal main pulmonary artery determines the instantaneous and time-average right heart power requirements, whereas the flow pulse height through the fiber bundle determines the effects on gas transfer rates and shear stress. These pulse heights will depend on the locations and magnitudes of the various resistances and compliances in the system. Figures 3.10, 3.11, and 3.12 show the effects of system relative pulse heights on heart power requirements, oxygen transfer rates and shear stress, respectively. For an average normal human at rest the right heart power output is about 0.27 watts. The TAL system, as seen in Figure 3.10, should have enough compliance such that the relative pressure pulse in the pulmonary artery is about 1.0. Too little compliance could induce right heart failure by causing an excessive afterload for the ventricle. Figure 3.11 shows that the time-average rate of oxygen transfer can be severely impaired (because of the nonlinear dependence of oxygen transfer on Reynolds number*) if the compliance is too small. With a compliance that reduces the relative flow pulse to 2.0 or less, the device would transfer oxygen at essentially the same rate as it would under steady flow conditions. Figure 3.12 shows that the time-average shear stress through the device is insensitive to compliance, a result of the fact that shear stress is proportional to flow rate. The instantaneous shear stress can be both quite high and quite low, however, if the com-

Engineering Design of Thoracic Artificial Lungs

59

Fig. 3.10. Right ventricular power requirements as a function of relative flow pulse height.

pliance is too small. Without adequate compliance, the periods of very low shear would be particularly prone to material induced activation. The greater the compliance, the better from all standpoints.

Discussion A thoracic artificial lung is quite feasible with currently available materials and technology. The device would consist of two major parts: (i) the bundle of fibers where the gas exchange takes place and which acts as a hemodynamic resistance element and (ii) elastic chambers where the pulse from the right heart is smoothed before flowing through the fiber bundle and which acts as a hemodynamic compliance element. The objectives of a design are to: 1. supply the required rates of O2 and CO2 transfer, 2. not overload the right heart, 3. create little or no thrombogenic and inflammatory reactions, and 4. be of a size and shape that can fit into the thoracic and/or abdominal cavities.

A most efficient gas transfer design with current technology is to have the blood flow cross-wise over the outside of hollow fibers packed into a bundle with oxygen flowing inside the fibers. The bundle designs discussed above are based on specifying the required rate of O2 transfer to blood with a specified venous O2 saturations, the blood flow rate (which is somewhat set by the required rate of O2 transfer), and the allowable bundle

* These calculations are based on assuming (i) quasi-steady rate of gas transfer, which should be applicable at these low Reynolds numbers and (ii) the mass transfer coefficient is a power function of the Reynolds number, the convective rate of mass transfer, and reflects any pure diffusional transfer during stopped phases of a sharp pulse.

60

The Artificial Lung

Fig. 3.11. Relative oxygen transfer rate as a function of relative flow pulse height.

resistance. A desirable fiber bundle design is one that has as little surface area as possible, a reasonable gross volume and shape, and flow characteristics within the bundle that are not conducive to either shear-induced or material-induced thrombogenesis and/or inflammatory stimulation. The designer has two aspects of the fiber bundle to select to optimize the design: fiber diameter and bundle void fraction and two aspects of the device compliance: compliance placement and compliance properties. The above calculations suggest that the required surface area is minimized by using fibers with as small a diameter as is practical, probably about 200 ∝m today, and arranging the fibers such that bundle void fraction is 0.70 to 0.80. The required gross volume, although generally not a problem, is also minimized by using small diameter fibers. It is further minimized by selecting small fiber bundle void fractions, but gross volume is relatively insensitive to void fraction and quite reasonable at the larger void fractions as long as one uses small diameter fibers. Reasonable bundle shapes, on the other hand, are only feasible if one selects large void fractions, especially with small diameter fibers. The potential for shear-induced blood trauma is minimized by choosing as small a fiber diameter as feasible and a small bundle void fraction. None of the cases considered here, however, indicates that shear-induced trauma should be a problem, as long as the blood flow is essentially steady through the bundle, and selecting a small fiber diameter and a large bundle void fraction is acceptable. Material-induced blood trauma, on the other hand, is likely to be a problem, especially if the blood flow rate through the bundle is too slow. Material-induced trauma is affected by the specific material used, any surface treatment of that material, the total amount of prosthetic material surface area, and the flow-induced transport to and from

Engineering Design of Thoracic Artificial Lungs

61

Fig. 3.12. Instantaneous maximum and minimum average shear stress in the fiber bundle as a function of relative flow pulse height.

the surfaces. The above calculations only address how the design choices, for a given set of specifications, affect the total required surface area and the transport aspects of material-induced trauma. The specific material of choice would be strongly influenced by those materials that have evolved as the materials of choice for the oxygenators in heart-lung machines, although in the latter case the cost of materials is a much more important consideration than in the case of intrathoracic artificial lungs. Virtually all manufacturers of heart-lung machine oxygenators use, today, microporous polypropylene fibers that are potted using polyurethane. These materials are reasonably blood compatible and could be the best choice of materials for TALs, but other material choices need to be explored. If hollow fibers made of microporous polypropylene, or any other microporous material, are used, however, some means needs to be utilized that will prevent potential filling of the pores with plasma and plasma leakage through the micropores. Such potential plasma leakage is generally not a problem in the relatively short-term use of oxygenators for heart-lung bypass surgery, but is a well-known problem with ECMO and would be a problem with TALs. One promising technique would be to coat the fibers with a one-micron thick layer of silicone rubber as was done by the developers of the IVOX.20 If such a technique is used for TALs, the blood would interface with silicone rubber rather than the microporous material itself. The microporous fibers, in that case, would serve as structural support for the very thin silicone rubber and offer very little, if any, resistance to the gas transport. Silicone rubber is at least as blood compatible as polypropylene. Probably more important for controlling material-induced blood trauma, however, is the development of surface treatments, for whatever material is used, that present a nonstimulating blood-material interface. Some possible treatments are discussed in another chapter in this book. Although little hard quantitative data is currently available that delineates the trans-

62

The Artificial Lung

port conditions that minimize material-induced trauma for any given material and surface treatment, it can be conjectured that one wants as little surface area as necessary and not let the flow rate and average shear rate go below some value, here expressed in terms of a shear stress of 4 dynes/cm2. The calculations reveal the surface area is minimized by selecting small diameter fibers and large bundle void fractions. Staying above the minimal shear rate, however, would be easier with large diameter fibers and/or small bundle void fractions. Using 200 ∝m fibers and bundle void fractions of about 0.7, however, may be the best compromise for minimizing the total surface area required while maintaining shear rates high enough to avoid accumulating any reaction products at the surfaces. The trauma factors, Tf,s and Tf,m, used in these calculations to design devices that induce minimal blood activation, it should be emphasized, are based on currently available data and should be considered tentative. As more detailed understanding of blood activation becomes available, these factors and the ‘safe zone’ will, undoubtedly, have to be modified. Choosing small diameter fibers and large bundle void fractions appear to result in an optimum fiber bundle design, all in all, with current technologies. These choices, for most sets of design specifications, result in devices that have minimal surface area, reasonable size and shape, very little, if any, potential for shear-induced trauma, and minimal potential for material-induced trauma. The above pulsatility calculations assume the right heart is pumping into a simple system represented by an effective resistance, Re, and an effective compliance, Ce. In reality, the system is more complicated. The effects of attaching a TAL to the pulmonary circulation and the performance of the device depend not only on the fiber bundle resistance and device compliance and the pulmonary system properties, but also on the attachment mode and the relative placement of the various resistances and compliances of the combined system. The inlet and outlet sections of the device and any bands around the main pulmonary artery (to direct the flow) can create local ‘minor losses.’ Because the fiber bundle is designed to have a very low resistance, these so-called ‘minor losses’ may not be so relatively minor and may significantly affect device function. Also, because the device inlet is attached just distal to the pulmonary valve, the impedance to the right heart may require that the inlet is also compliant. Some hemodynamic effects of attachment mode and the placement and magnitudes of the various device components are discussed in Boschetti et al.21 They studied the system responses in a model of the pulmonary circulation with a TAL attached. They consider a range of device properties and both a normal pulmonary circulation and one with an elevated pulmonary vascular resistance, as would be common in most applications of TAL use. They conclude that an inlet compliance is necessary to reduce required right heart power and impedance modulus and TAL compliance is necessary to reduce blood flow pulsatility. Reducing fiber bundle resistance results in a tradeoff by reducing right heart power requirements but increasing blood flow pulsatility. The hybrid implantation mode appears to be the best compromise between hemodynamic performance and preservation of some blood flow through the natural lungs for nonrespiratory functions. The hybrid implantation mode, however, would be the most surgically difficult. Without a compliance element proximal to the fiber bundle, the output pulse from the right ventricle would force a strong pulsatile flow through the device. When compared to the values that would occur with a steady flow through the fiber bundle, a largely pulsatile flow, for the same mean flow, decreases the rate of gas transfer, increases the time-average and peak work load for the right ventricle, and both increases and decreases

Engineering Design of Thoracic Artificial Lungs

63

the instantaneous shear stress within the fiber bundle. The latter could be conducive to both shear-induced and material-induced blood trauma.

Acknowledgement The work in this manuscript was supported, in part, by a grant from the National Institutes of Health: HL 59537. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

15. 16. 17. 18. 19.

Hooker TP, Hammond MD, Salem A. Adult respiratory distress syndrome: A review for the clinician. J Am Osteop Assoc 1992; 92:886-896. Eya K, Tatsumi E, Taenaka Y et al. Importance of metabolic function of the natural lung evaluated by prolonged exclusion of the pulmonary circulation. ASAIO J 1996; 42:M805-M809. Takewa Y, Tatsumi E, Taenaka Y et al. Hemodynamic and humoral conditions in stepwise reduction of pulmonary blood flow during venoarterial bypass in awake goats. ASAIO J 1997; 43:M494-M499. Cotes JE. Lung Function: Assessment and Application in Medicine. 2nd ed. Philadelphia: FA Davis Co., 1968. Mockros LF, Leonard R. Compact cross-flow tubular oxygenators. ASAIO Trans 1985; 31:628-632. Vaslef SN, Mockros LF, Cook KE et al. Computer-assisted design of an implantable intrathoracic artificial lung. Artificial Organs 1994; 18:813-817. Vaslef SN, Mockros LF, Anderson RW et al. Use of a mathematical model to predict oxygen transfer rates in hollow fiber membrane oxygenators. ASAIO J 1994; 40:990-996. Colton CK, Lowrie EG. Hemodialysis: Physical principles and technical considerations. In: Brenner BM, Rector FC, eds. The Kidney. 2nd ed. Philadelphia: Saunders, 1981:2425-2489. Hellums JD, Peterson DM, Stathopoulos NA et al. Studies on the mechanisms of shear-induced platelet activation. In: Hartmann A, Kuschinsky W, eds. Cerebral Ischemia and Hemorheology. Berlin: Springer-Verlag, 1987: 80-89. McIntire LV, Frangos JA, Rhee BG et al. The effect of fluid mechanical stress on cellular arachidonic acid metabolism. Ann NY Acad Sci 1987; 516:513-524. Chow TW, Hellums JD, Moake JL et al. Shear stress-induced von Willebrand factor binding to platelet glycoprotein Ib initiates calcium influx associated with aggregation. Blood 1992; 80:113-120. Wurzinger LJ, Opitz R, Blasberg P et al. Platelet and coagulation parameters following millisecond exposure to laminar shear stress. Thromb Haemost 1985; 54:381-386. Bluestein M, Mockros LF. Hemolytic effects of energy dissipation in flowing blood. Med Biol Eng 1969; 7:1-16. Salzman EW, Hirsh J. The epidemiology, pathogenesis, and natural history of venous thrombosis. In: Coleman RW, Hirsh J, Marder VJ, Salzman EW, eds. Hemostasis and Thrombosis: Basic Principles and Clinical Practice. 3rd ed. Philadelphia: Lippincott, 1994:1275-1296. Spaeth EE, Roberts GW, Yadwadkar SR et al. The influence of fluid shear on the kinetics of blood coagulation reactions. Trans Am Soc Artif Intern Organs 1973; 19:179-187. Basmadjian D, Sefton MV, Baldwin SA. Coagulation on biomaterials in flowing blood: Some theoretical considerations. Biomaterials 1997; 18:1511-1522. Goldstick TK. Oxygen transport. In: Brown JHU, Gann DS, eds. Engineering Principles in Physiology. New York: Academic Press, 1973:257-282. Fricke H. A mathematical treatment of the electric conductivity and capacity of disperse systems. I. The electric conductivity of a suspension of homogeneous spheroids. Physical Reviews 1924; 24: 575-587. Cook KE, Makarewicz AJ, Backer CL et al. Testing of an intrathoracic artificial lung in a pig model. ASAIO J 1996; 42:M604-M609.

64

The Artificial Lung

20. Bagley B, Bagley A, Henrie J et al. Quantitative gas transfer into and out of circulating venous blood by means of an intravenacaval oxygenator. ASAIO Trans 1991; 37:M413-M415. 21. Boschetti F, Perlman CE, Cook KE et al. Hemodynamic effects of attachment modes and device design of a thoracic artificial lung. ASAIO J 1999; 46:42-48.

CHAPTER 4

Biocompatibility of Artificial Lungs Keith E. Cook and Lyle F. Mockros

Introduction

T

he concept of extracorporeal circuit-induced blood activation is nearly as old as cardiopulmonary bypass itself. From 1951 to 1953, six different groups of surgeons made the first attempts at using cardiopulmonary bypass to repair congenital heart defects.1 Of 18 patients, only one survived. The prevailing opinion of the day was that the lack of success was due to the patients’ unhealthy, failing hearts and not with the perfusion techniques or heart-lung machines. In 1954, however, Dr. Lillehei and colleagues at the University of Minnesota began experimenting with cross-circulation, the process of using a donor animal as the oxygenator. Arterial blood from the donor animal was pumped into the recipient animal’s aorta while blood from the recipient’s vena cavae was pumped back to the donor’s saphenous vein, thus bypassing the recipient’s heart and lungs. During dog experiments, they were astonished to find a higher survival rate and more rapid recovery using cross-circulation rather than a heart-lung machine. At the time, no one knew what hematologic derangements were occurring and the methodology to measure these derangements was not yet established. Yet, for the first time, there was evidence that part of the difficulty with open-heart surgery was the patient’s reaction to the heart-lung machine itself. Cardiopulmonary bypass (CPB) is now a common and highly successful practice, and the use of heart-lung machines has been extended from hours during CPB to days or weeks during extracorporeal membrane oxygenation (ECMO). Much of the physiologic derangement induced by extracorporeal circulation has also been uncovered, if not fully explained. It is now understood that the blood-artificial surface contact and mechanical stress inherent in extracorporeal circulation cause activation of the coagulation and immune systems and, when severe, major organ dysfunction or failure. Despite this understanding, means have not been found to eliminate blood activation during CPB or ECMO. At best, activation can be masked or compensated for retroactively, and it continues to contribute to patient morbidity and mortality. An artificial lung (AL), whether implanted or used paracorporeally, would be expected to cause similar blood trauma. Like the oxygenators used in extracorporeal life support (ECLS), an AL has a large, artificial surface area. Unlike ECLS, however, an AL does not employ an extracorporeal circuit and has a very different hemodynamic environment. Use of an AL, furthermore, does not involve periods of hypothermia or heart and lung ischemia as occurs during CPB. No experiments have, as yet, sought to document the hematological derangements produced by an AL. In vivo 24-hour experiments, however,

The Artificial Lung, edited by Steven N. Vaslef and Robert W. Anderson. ©2002 Eurekah.com.

66

The Artificial Lung

indicate platelet consumption, hemodynamic derangements and pulmonary damage occur, suggesting significant activation of both the clotting and immune systems.2 Activation of blood systems is initiated by protein adsorption upon blood contact. Adsorption of contact system proteins causes their activation, initiates the intrinsic branch of the clotting cascade, and causes clot formation unless anticoagulants are utilized. At the same time, activated complement factors are adsorbed to the artificial surface, allowing them to escape inactivation by plasma inhibitors or water and cause further complement activation. Adsorbed proteins and their circulating products then activate platelets and leukocytes. Platelets bind and are activated by adsorbed plasma proteins, such as fibrinogen, and soluble thrombin that is generated near the artificial surface by contact activation of the coagulation cascade. Platelet activation then causes degranulation, which releases numerous procoagulant molecules into plasma and worsens an already procoagulant state. Blood leukocytes become activated by circulating by-products of contact and complement activation and possibly by proteins adsorbed to the artificial surface. Activation of leukocytes leads to release of proinflammatory molecules that amplify activation, expression of cell surface receptors that promote leukocyte binding to the endothelium, and release of various cytotoxic agents that damage the endothelium and increase capillary permeability. Contact and complement activation products cause further increases in capillary permeability, leading to excessive extravasation of fluid and major organ dysfunction. Blood activation has been combatted primarily through the use of heparin and other drug therapies that combat tail-end rather than initiating reactions, cripple blood systems necessary for hemostasis and immunity, and lead to further complications. Heparin-coated circuits have been used to reduce blood activation but have not proven effective enough to significantly reduce the need for anticoagulation. Use of protein resistant surfaces such as polyethylene oxide and immobilization of newly developed bioactive molecules may be able to reduce blood activation at its initiation, however, and minimize the need for drug therapy during long term use of an AL.

Initiation of Blood-Artificial Surface Reaction: Protein Adsorption Plasma proteins adsorb to artificial surfaces almost instantaneously upon blood contact, forming a protein layer approximately 200 angstroms thick.3 Adsorption can induce conformational changes and proteolysis and activate procoagulant and inflammatory mediators. The protein layer can also promote cell-artificial surface interactions, leading to platelet and leukocyte adhesion and activation. The adsorptive characteristics of a biomaterial are, therefore, of extreme importance in defining its thrombotic and inflammatory characteristics. Physical and chemical characteristics of artificial materials affect the relative amounts of specific adsorbed proteins. These characteristics include wettability, surface free energy, surface charge, crystallinity, molecular chain mobility, ordered structure, and locations of chemical radicals, amongst others.4 Most characteristics are of little use, however, in the prediction of general protein adsorption.5,6 Fibrinogen and albumin tend to adsorb preferentially to hydrophobic surfaces, for example, but this is not true of all plasma proteins. Surfaces with a net positive charge tend to be more adsorptive of most plasma proteins and cells, which are all negatively charged at physiologic pH.7 A surface with a net negative surface charge, on the other hand, is more active in adsorption and activation of the contact system proteins Factor XII and high molecular weight kininogen.7,8 The effect of most surface characteristics, therefore, must be defined only with respect to specific proteins.

Biocompatibility of Artificial Lungs

67

Protein adsorption is also a dynamic process.5,6 Adsorbed fibrinogen, for example, is preferentially adsorbed upon initial blood contact. Its surface concentration, however, falls over time as it is displaced by activated high molecular weight kininogen (HKa) and, to a lesser extent, Factor XII.9 Studies of this nature, however, are typically on the order of hours, and the effect of in vivo circulation of blood on the order of days is unknown. Contact and complement activation are two specific cases of protein adsorption and activation during ECLS. Activation of the contact system leads to initiation of the intrinsic branch of the clotting cascade. Activation of both the alternative and classical complement pathways leads to formation of the terminal complement complex (TCC) and the proinflammatory anaphylatoxins, C3a, C4a and C5a.

Contact Activation The contact system is comprised of four serine proteases, Factor XII (FXII), high molecular weight kininogen (HK), Factor XI (FXI), and prekallikrein (PK). Contact activation is initiated by FXII binding to a negatively charged surface where it is autoactivated (Fig. 4.1). Activation, however, may proceed on a surface of any net charge. Natural and artificial surfaces contain both positive and negative charges, and the net surface charge has too little resolution to predict molecular level events.7 Surfaces, furthermore, rarely retain their original charge. Instead, due to protein adsorption, they take on the net negative charge possessed by all plasma proteins and blood cells at physiologic pH.3 High molecular weight kininogen, complexed with either PK or FXI, is also adsorbed to negatively charged artificial surfaces. Adsorption of HK-PK and HK-FXI brings them in close association with FXIIa, which activates PK, FXI, and HK to kallikrein, activated FXI (FXIa), and activated HK (HKa), respectively. Activation of HK, however, proceeds slowly. Factor XIa initiates the intrinsic branch of the clotting cascade, leading to the formation of thrombin (Fig. 4.2), and thrombin converts fibrinogen to fibrin to form the solid clot. Kallikrein greatly accelerates these reactions by activating FXII and HK at a much faster rate than by autoactivation or FXIIa, respectively. Kallikrein also breaks FXIIa into two fragments, soluble FXIIf and an inactive, surface-bound domain. Factor XIIf retains the active site of FXIIa and is thus a potent fluid phase activator of PK. Activated high molecular weight kininogen, furthermore, binds more avidly to the artificial surface than HK and thus accelerates the activation of PK and FXI. Activated high molecular weight kininogen also participates in reciprocal activation of FXII. The contact system is progressively activated during ECLS. Plasma FXIIa concentration and FXII activity is elevated after initiation of CPB, increases progressively, peaks at the end of CPB, and decreases thereafter.10-14 During ECMO, the plasma level of FXIIa-C1 esterase inhibitor complex increases significantly within one hour after the initiation of ECMO, decreases thereafter, and returns to normal by 48 hours.15 Kallikrein-C1 inhibitor (C1INH) complex concentration is also increased during in vitro simulated extracorporeal circulation (SECC),16,17 and kallikrein inhibitor activity has been shown to decrease during and up to one day after CPB due, presumably, to elevated levels of kallikrein.18,19 Wachtfogel et al. have not seen elevated Kallikrein-C1INH during CPB,16 however, perhaps due to clearance of the enzyme-inhibitor complex. High molecular weight kininogen activity has also been shown to increase progressively during CPB,11 and the plasma concentration of bradykinin, a product of HK proteolysis by kallikrein, FXIa, and FXIIa, has been shown to increase during SECC.13 Normally, bradykinin is predominantly inactivated by pulmonary endothelial cells, which are

68

The Artificial Lung

Fig. 4.1. Contact activation at an artificial surface. FXII: Factor XII; FXIIa: activated Factor XII; FXIIf: soluble, active Factor XII fragment; FXIIi: surface-bound, inactivated FXII fragment; HK: high molecular weight kininogen; HKa: activated high molecular weight kininogen; PK: prekallikrein; K: kallikrein; FXI: Factor XI; FXIa: activated Factor XI.

taken out of circulation during CPB. It appears, however, that sufficient clearance is provided by nonpulmonary endothelial cells. Usui et al. found that bradykinin concentrations increased upon initiation of CPB but decreased gradually thereafter,13 and Ellison et al. found no bradykinin increase during CPB.20 Bradykinin concentrations appear to increase, furthermore, only during hypothermia, which causes spontaneous contact activation in vitro.21

Alternative Complement Pathway Activation Small amounts of complement factor C3 are cleaved spontaneously in plasma, forming fragments C3a and C3b (Fig. 4.3). Typically, C3b is a short-lived intermediate that is rapidly proteolytically inactivated by water. Fragment C3b, however, may be adsorbed to an artificial surface or covalently attached to nucleophilic groups (NH2-, OH-) on an artificial or cell surface. Adsorption of C3b slows its inactivation by plasma inhibitors Factor I and H, blood cell and endothelial plasma membrane inhibitors complement receptor 1 (CR1) and membrane cofactor protein (MCP), and water. Fragment C3b, if not inactivated, initiates the alternative complement pathway, leading to formation of C3a, C5a, and the terminal complement complex (TCC).

Biocompatibility of Artificial Lungs

69

Fig. 4.2. The coagulation cascade.

A surface’s activating efficiency depends on its relative affinity for soluble pro-activation cofactors B, D, and properdin and soluble inhibitors Factors I and H. Surfaces that act as efficient activators limit binding of Factor I and H and favor binding of C3b, Factors B and D, and properdin. Surfaces with nucleophilic groups tend to promote C3b binding and thus tend to be activators. Surfaces with negatively charged constituents, such as COO! and SO!3 , tend to promote factor H binding to C3b and tend to be inactivators.22

Classical Complement Pathway Activation The initiating classical complement protein is C1, which is composed of three subunits, C1q, C1r, and C1s. Subunit C1q itself consists of six identical subunits. When two C1q subunits bind simultaneously to two nonantigen-binding Fc regions of immunoglobulin G (IgG) or immunoglobulin M (IgM), C1q activates C1r, and C1r activates C1s. Immunoglobulin G, however, is monomeric, and thus, two or more IgG molecules must be present to activate complement. Significant IgG activation of complement can only occur, therefore, during IgG aggregation. This situation typically occurs at a multideterminant antigen, such as a bacterial surface. Immunoglobulin M, in contrast, is a pentamer. The Fc regions of IgM, however, are only exposed when IgM is bound to antigen, which induces a conformational change. After C1 activation, the classical complement pathway proceeds in an analogous fashion to the alternative pathway (Fig. 4.3), resulting in the formation of C3a, C4a, C5a, and the TCC. Activation of the classical pathway by artificial surfaces is not as clear or as well studied as alternative pathway activation. The majority of in vitro studies of extracorporeal

70

The Artificial Lung

Fig. 4.3. Alternative and classical complement activation.

circulation have examined formation of common products such as C3a, C5a, and the TCC rather than activation of factors unique to the classic pathway such as C1 and C4. Some clinical CPB studies have demonstrated classical pathway activation, but these studies alone cannot confirm the role of surface-induced complement activation. Other factors involved in CPB such as bacterial invasion and protamine infusion may be the cause. The prevailing thought is that direct, surface-induced complement activation proceeds primarily through the alternative pathway via C3b binding to an artificial surface.22 Some level of C4b binding may occur, however, in the same fashion. Both C3b and C4b share an internal thioester bond that allows both to bind to surfaces via nucleophilic attack by surface OH- and NH2- groups.23 Furthermore, activation and inhibition of the classical and alternative pathway C3 convertases is accomplished in a similar manner with many of the same factors. The adsorption of Ig at the surface of biomaterials also offers another possible means of classical pathway activation. Immunoglobulin G is known to adhere to artificial surfaces,5,6 and it is likely that this is also true of IgM. Adsorption of these pro-

Biocompatibility of Artificial Lungs

71

teins, like antigen-binding, may cause exposure of their Fc regions, allowing interaction with circulating or adsorbed C1q and activation of the classical pathway. Activation of the classical complement pathway by the activated contact system has been demonstrated. Factor XII fragments directly activate the C1r component of C1 and, to a lesser degree, C1s.8 The major plasma inhibitor of FXIIa, FXIIf, kallikrein, and C1 is C1INH, furthermore, and thus, complement and contact activation have a synergistic effect due to taxation of their mutual regulatory mechanisms.

Effect of Extracorporeal Circulation on Complement Activation Both alternative and classical complement pathways are activated during CPB, although alternative pathway activation is more severe. Plasma C3a concentration rises markedly and progressively during CPB and peaks at the end of bypass.12,24-31 Plasma C4a increases slightly or not at all during CPB and to a greater extent after the procedure,25,26,28 perhaps due to adverse reactions to protamine administration.32 Shigemitsu et al., however, found a progressive increase in C4a during CPB.11 Plasma levels of C1-C1INH complex also increase during SECC,16,17 but like kallikrein-C1INH levels, no increase is detected during CPB.16 Total complement complex levels also increase progressively during CPB,28,33 but plasma C5a levels increase only slightly or not at all.12,24,25,34 An increase in the TCC levels, however, indicates indirectly that C5a is being generated, and the lack of detectable C5a has been attributed to rapid binding to leukocytes. Complement activation has also been documented during ECMO. Plasma levels of C3a increase during the first 24 hours of ECMO, after which they return to near-baseline levels.15,35 Plasma levels of inactivated complement, C3 and C5, decrease accordingly within 30 minutes of ECMO initiation and remain low up to 48 hours.36 Surgery itself is also known to produce C3a, but without ECC, C3a levels increase only slightly37 and fall to normal by completion of the procedure.24 Patients undergoing thoracotomy without CPB, furthermore, show no increase in TCC levels.33

Platelet Activation

Platelets adhere to an extracorporeal circuit surface within approximately one minute3,7 and may remain discoid or spread. Platelet spreading precedes degranulation and may, therefore, be an indication of the thrombogenicity of the surface.7 An important component in the initial adherence is thought to be adsorbed fibrinogen. When soluble in plasma, fibrinogen will not bind to inactivated platelets. Upon adsorption, however, fibrinogen undergoes a conformational change, allowing it to bind selectively to the inactivated form of platelet receptor GpIIb/IIIa. Platelet adherence to biomaterials has, therefore, been correlated with the level of adsorbed fibrinogen.38,39 Other adsorbed proteins may also contribute to initial platelet activation, including fibronectin,40 von Willebrand Factor (vWF),41 thrombospondin,42 FXII,43 vitronectin,44 and IgG.45,46 Initial platelet activation may also be caused by surface-focused thrombin generation due to contact activation. Thrombin is a potent platelet activator, and it is likely that unbound thrombin contributes to platelet activation despite the presence of the heparin-antithrombin III complex. Platelet activation is soon followed by aggregation due to the release of granule contents ADP, platelet factor 4 (PF4), vWF, fibrinogen, thrombospondin, and fibronectin, as well as the secretion of thromboxane A2. Aggregation may also be due to changes in bound platelet receptors such as GpIIb/IIIa that make the platelets more adhesive.7 Over time, platelets cease adhering to the surface.47 Although there is no definitive cause for this, fibrinogen displacement by HKa on the surface and changes in platelet

72

The Artificial Lung

reactivity to soluble agonists have been proposed. Surface concentrations of platelet GpIIb/ IIIa, GpIb, and ∃-adrenergic receptors are reduced following CPB,48-51 and aggregation in response to epinephrine, thrombin, ADP, and collagen is decreased during CPB and grows progressively worse thereafter.52-54 The dysfunction is most likely due to damage by shear stress, surface adherence, and proteolytic removal of GpIb by plasmin. Intracellular stores of ∃-granules are also depleted, and the platelet is thus unable to respond when activated. Platelet turnover is increased during ECC,55 furthermore, and platelet counts are decreased accordingly during CPB.3

Leukocyte Activation and the Whole Body Inflammatory Response Leukocyte activation and the mounting of a “whole body inflammatory response” during ECC occurs in several distinct stages. Initial activation is thought to be induced predominantly by activated contact factors and complement fragments. Activated FXII causes in vitro neutrophil aggregation and degranulation56 and monocyte activation.57 Kallikrein causes in vitro neutrophil chemotaxis,58 aggregation, respiratory burst,59 and degranulation.60 High molecular weight kininogen binds specifically to neutrophils, furthermore, and increases the effectiveness of kallikrein activation.61 Neutrophils, monocytes, mast cells, and basophils also all possess a specific, high-affinity C5a receptor that has signal transduction capabilities.23,62 For the neutrophil, low concentrations of C5a can induce in vitro chemotaxis, increased CD11b/CD18 adhesiveness for endothelial ICAM-1, and, thus, margination. At higher concentrations, C5a induces the respiratory burst.3,23 For the monocyte, C5a induces chemotaxis,63 trans-endothelial migration,64-66 and leukotriene B4 (LTB4),67 interleukin-8 (IL-8), and interleukin-6 (IL-6) production and release.68,69 Interleukin-1 (IL-1) mRNA transcription is also induced by C5a, but a secondary stimulus is required for translation.70 On mast cells and basophils, C3a, C4a, and C5a binding causes degranulation and histamine release.23 Neutrophil and monocyte binding to adsorbed proteins on the extracorporeal circuit may also contribute to activation and lead to frustrated phagocytosis. Leukocyte adhesion to extracorporeal circuits has been demonstrated,12 yet there are few detailed studies of the mechanisms. Complement fragments C3b, C4b, and inactivated C3b (iC3b) adhere to artificial surfaces during complement activation, and thus, neutrophils and monocytes can adhere to the artificial surfaces via complement receptor 1 (CR1) and CD11b/CD18. Binding to CR1 and CD11b/CD18 initiates phagocytic processes and, in the case of microorganisms, leads to endocytosis. In the case of larger matter, however, such as an oxygenator surface, phagocytes will release their cytotoxic agents extracellularly, into the blood, in a process known as frustrated phagocytosis.71,72 Neutrophils and monocytes also possess fibrinogen and fibronectin receptors,23 and surface adsorption of these molecules may help mediate adhesion. Fibronectin, unlike C3b or iC3b, does not directly induce phagocytosis, however, but instead upregulates CR1 and CD11b/CD18 activity.71 Lim and Cooper have shown that neutrophil adhesion on polyether urethanes under static conditions is increased by surface hydrophobicity and the presence of plasma proteins and that adhesion ceases after 15 min of exposure.73 This suggests a possible role for fibrinogen, which is preferentially deposited on hydrophobic surfaces before displacement by HKa and FXII.

Biocompatibility of Artificial Lungs

73

Release of Pro-Inflammatory Mediators During CPB The initial activation of neutrophils and/or monocytes during ECLS leads to release of the rapidly synthesized, pro-inflammatory molecules, platelet activating factor (PAF) and LTB4. Platelet activating factor rises progressively during CPB, peaking at cross-clamp removal and declining thereafter.74,75 Leukotriene B4 production appears to increase during CPB,31,76-78 yet Jansen et al. saw no LTB4 changes during or post-CPB.27 The inflammatory response is also amplified due to cytokine production. Plasma concentration of IL-8, a chemotactic cytokine or chemokine, increases during CPB. Interleukin-8 initiates neutrophil chemotaxis and increases ICAM-1 affinity for CD11b/ CD18 and CD11a/CD18. IL-8 is made predominantly by mononuclear phagocytes but is also produced by the endothelium, fibroblasts, and megakaryocytes and carried by circulating platelets. The IL-8 increase during CPB is modest during hypothermic circulation but becomes large after cross-clamp removal and during rewarming.28,30,79-81 Interleukin-8 release during CPB is most likely due to activation of circulating monocytes by C5a. After cross-clamp release, however, IL-8 is most likely produced by monocytes and macrophages that were trapped within the ischemic heart and lungs. Hypoxia followed by hyperoxia has been shown to induce IL-8 release in monocytes.82 The roles of IL-1 and tumor necrosis factor-∃ (TNF) during ECLS-induced inflammation are still unknown. The majority of clinical studies show no presence of plasma IL-1 and TNF before or after CPB.28,79,81,83,84 Some researchers, however, believe that IL-1 and TNF are released in small quantities that are immeasurable due to rapid binding and are sufficient for local activation of leukocytes. Intracellular levels of IL-1 are elevated post-bypass,85,86 and this may indicate that small, immeasurable quantities of IL-1 are being released during CPB. Production, however, does not necessarily indicate release. The mechanism of IL-1 release is as yet unknown, and similar studies of TNF production have not been carried out.

Release of Cytotoxic Agents During CPB Extensive evidence indicates that preformed, cytotoxic neutrophil and monocyte granule contents such as elastase, myeloperoxidase, and lactoferrin are released progressively during ECC. Plasma concentrations of elastase,17,87 myeloperoxidase, and lactoferrin88 rise progressively during SECC. Elastase,12,27,30,89 myeloperoxidase, and lactoferrin90-98 also rise progressively during CPB and fall upon completion. Various techniques have been used to demonstrate reactive oxygen species production and activity during CPB. Direct measurement of reactive oxygen species in plasma have demonstrated increased H2O299,100 and O!2 101 during CPB. Neutrophil oxygen free radical producing capacity has also been shown to increase during CPB,101,102 but this is more a measure of priming than of actual activation. Evidence of lipid damage by free radicals during CPB has also been found by measuring plasma levels of malondialdehyde (MDA), a product of unsaturated fatty acid peroxidation. Malondialdehyde concentrations increase during CPB and begin to decrease during reperfusion.101,103 Levels of vitamin E, a free radical scavenger, decrease in right atrial tissue during CPB104 and in plasma after CPB,99 presumably due to consumption by free radicals. Increases in free-radical production and activity after CPB may, however, be due to ischemia-reperfusion injury rather than neutrophil and monocyte activation. During ischemia, intracellular ATP is broken down to hypoxanthine, and xanthine dehydrogenase is converted to xanthine oxidase. During reoxygenation, xanthine oxidase degrades

74

The Artificial Lung

hypoxanthine to urate, producing H202.105 Plasma hypoxanthine concentrations are elevated during CPB and increase further after reperfusion of the ischemic heart and lungs.97 Histamine levels increase progressively during adult106 and pediatric CPB.107 Histamine levels during pediatric surgery, however, vary greatly from patient to patient, with some patients experiencing no change while others experience increases of over 300%. Nitric oxide levels also appear to be elevated during CPB, as measured by increased plasma nitrite levels,108 but more studies need to be performed.

Upregulation of Adhesion Molecules During CPB Cardiopulmonary bypass increases the number of adhesion receptors, their binding affinities, and, perhaps, the effectiveness of their intracellular signaling mechanisms.71 This leads to increased margination and adhesion-induced activation. Expression of CD11b/ CD18 increases progressively during CPB, showing a different time course for neutrophils and monocytes.109,110 Neutrophil expression rises quickly upon CPB initiation and peaks at the end, while monocyte expression rises slowly and peaks 2 to 4 hours post CPB.

Leukocyte Activation During ECMO Very few studies of leukocyte activation during ECMO have been reported. They do, however, suggest patterns similar to CPB during the first 24 hours of extracorporeal circulation, at which point activation subsides. Plasma IL-8 and elastase concentrations and CD11b expression all increase significantly within 15 minutes of ECMO initiation. IL-8 and elastase levels rise progressively, begin to decrease by 12 hours, and return to normal post-ECMO.111 CD11b expression returns to normal by 12 hours and is significantly decreased by 24 hours post-ECMO. Neither IL-1 nor TNF levels show significant increases.15,111

Shear Stress Activation Mechanical forces within the extracorporeal circuit may also activate platelets or leukocytes. The effect of shear stress on platelet activation has been studied for over two decades, and there are many studies demonstrating a positive correlation between platelet activation and shear stress magnitude and duration.112-120 There has been controversy, however, about the mechanism of activation, with some researchers suggesting that activation is solely a product of platelet lysis, release of granule contents, and, thus, chemical activation.121 An increasing number of recent studies suggest, however, that shear-induced platelet activation is mediated by a signal transduction pathway completely separate from chemical activation. Early studies of shear-induced platelet activation focused on shear stress magnitudes greater than 150 dynes/cm2, application times greater than five minutes, or, typically, both.112-116,121 Due to excessive shear stress magnitude and duration, many of these became platelet lysis rather than activation studies. Blood flow within oxygenators of all types, however, is low-Reynolds number, low shear flow with residence times on the order of 1 to 10 seconds. These studies, therefore, fail to examine shear stress magnitudes and durations relevant to oxygenators. Chow et al, however, measured platelet activation under shear stresses from 15 to 120 dynes/cm2 and application times from zero to 100 seconds.118 The results indicate that there is a shear-stress dependent time below which no shear-induced platelet aggregation occurs (Fig. 4.4). For a shear stress below 15 dynes/cm2, there is no activation at all, and

Biocompatibility of Artificial Lungs

75

this time is effectively infinite. For shear stresses between 30 and 90 dynes/cm2, this time is on the order of 10 seconds. For shear stresses above 120 dynes/cm2, aggregation appears to begin within less than a second of shear application. Shear stress studies on leukocytes have focused on adhesion rather than activation, per se. Adhesion of neutrophils and monocytes to artificial surfaces and the endothelium is decreased with increasing shear stress,122-124 providing indirect evidence that shear stress does not cause activation and upregulation of cell surface receptors. Neutrophils and monocytes suspended in 20% serum show decreasing adherence to poly(etherurethane ureas) with increasing shear stress, and adherence above 17.4 dynes/cm2 is negligible.123 Adsorbed C3 and, to a lesser extent, fibronectin were important mediators in this adherence. Neutrophils or monocytes suspended in 20% serum that has been C3- or fibronectin-depleted show negligible adherence at shear stresses above 0.8 dynes/cm2.123 Initial platelet deposition may also play an important role in leukocyte adherence to extracorporeal circuits. Neutrophils suspended in platelet poor plasma will adhere only slightly to extracellular matrix under shear conditions from 0.2 to 3.2 dynes/cm2, above which there is no adherence.124 If platelets are allowed to initially adhere to the ECM, however, neutrophil adherence is increased an order in magnitude and is maintained at shear stresses greater than 6.4 dynes/cm2. Leukocytes suspended in platelet poor plasma, furthermore, do not adhere to nylon at shear stresses greater than 1.96 dynes/cm2.122

Organ and System Level Effects Activation of the contact and complement systems, platelets, and leukocytes all contribute to the “post-perfusion” or “post-pump” syndrome. This includes bleeding complications, increased susceptibility to infection, increased capillary permeability, accumulation of interstitial fluid, and dysfunction of the heart, lungs, kidneys, and the gut.

Coagulopathy There is a significant increase in the bleeding time during CPB that returns to normal within two to four hours after the procedure, and the average postoperative blood loss is approximately one liter.125 Three to five percent of patients, furthermore, have serious bleeding complications, defined as transfusion of more than 10 units of blood. More than half of these bleeding complications can be corrected in some part by surgical exploration, but the remainder are due solely to hemostatic derangements. Cardiopulmonary bypass is a relatively brief procedure, however, and bleeding complications are rare compared to ECMO. Thirty-five percent of neonatal ECMO patients have general hemorrhagic complications, and intracranial hemorrhage occurs in 36% of patients under 35 weeks of gestation and 12% over 35 weeks.126 The major cause of bleeding complications is a reduction in the platelets’ ability to recognize and react to soluble agonists (see above), but elevated fibrinolysis and reduced coagulation factor concentrations may contribute. Bradykinin and thrombin cause endothelial release of tissue plasminogen activator (tPA), which converts plasminogen to plasmin,3,8 and increased plasmin concentrations during ECLS induce a pro-fibrinolytic state. Tissue plasminogen activator levels rise at the initiation of bypass, remain elevated during the procedure, and fall at the completion.27,89 Plasmin levels increase significantly after heparin induction, fall to a less elevated level during CPB, and return to normal at the completion of CPB.127 Consequently, plasminogen and ∃2-antiplasmin concentrations decrease after the initiation of bypass and remain reduced at its completion.89 The concentration of plasmin-∃2-antiplasmin complexes increases accordingly after the

76

The Artificial Lung

Fig. 4.4. The effect of shear stress magnitude and duration on platelet intracellular calcium concentration and aggregation with (vWF) and without (control) the presence of von Willebrand Factor. Reprinted by permission of W. B. Saunders Co. Chow TW, Hellums JD, Moake JL et al. Shear stress-induced von Willebrand factor binding to platelet glycoprotein Ib initiates calcium influx associated with aggregation. Blood 1992; 80:113-120.

initiation of CPB, remains slightly less elevated during CPB, and falls to normal levels by 24 hours after the procedure.127 The concentration of fibrin degradation products rises progressively during CPB and remains elevated up to 2 days after completion of the procedure.11,12,89 Plasmin concentration remains constant during the initial 48 hours of ECMO, becomes significantly elevated by 72 hours, remains elevated until at least 96

Biocompatibility of Artificial Lungs

77

hours, and returns to normal after the procedure.21 Concentrations of fibrin degradation products mirror plasmin activation. A reduction in plasma concentrations of various coagulation factors during CPB may also contribute to bleeding. Hemodilution causes decreases in Factors II, VII, IX, X, XIII, and fibrinogen,52,128 but cannot account for further reductions in Factor V and vWF.52,129 All factors, except fibrinogen, return to normal within 12 hours after bypass. Despite excessive patient bleeding, excessive coagulation within the extracorporeal circuit also remains a problem during ECMO. Nineteen percent of ECMO cases have mechanical complications due to clots in the circuit, and 4% have oxygenator failure that is usually due to some coagulopathy.126

Susceptibility to Infection The exaggerated immune response during ECLS can lead to exhaustion of both cellular and humoral immune mechanisms. Blood activation reduces complement levels and, thus, the opsonins necessary for phagocytosis of bacteria. Neutrophil bactericidal capacity and natural killer cell activity is also impaired up to 2 and 3 days, respectively, following CPB.130-132 The number and interleukin-2 producing capacity of CD4 T cells are also reduced after CPB. Cell counts reach a nadir at the first postoperative day, improve thereafter, and return to normal or remain slightly reduced at the seventh post-operative day.132-135 Interleukin-2 producing capacity remains significantly reduced after seven days.133 Numbers of CD8 T cells increase following CPB,132-134 but T cell cytolytic activity is reduced, reaching a nadir on the first postoperative day and returning to normal by three days.132 General lymphocyte counts also drop by the second day of ECMO and return to normal within one week,136 suggesting similar reductions in immunoreactivity are present. The result of these changes is a reduction in natural immune response, an increased risk of infection that can be correlated with CPB duration,137,138 and possible septic shock or multiorgan failure. In one institution, septic multiorgan failure was responsible for 55% of CPB perioperative mortality.133

Endothelial Dysfunction Endothelial functional derangements occur during ECLS due to the short term effects of soluble agents and the longer term changes induced by structural cell damage or death. Bradykinin,3 thrombin,139 C5a23 and histamine23,140 cause reversible increases in capillary permeability, and release of neutrophil and monocyte cytotoxic agents may result in more permanent damage. Activated neutrophils cause elastase-mediated endothelial cell injury141 and detachment from cultured monolayers142 during in vitro coculture. The result of increased capillary permeability is a progressive extravasation of plasma proteins and an increase in extravascular fluid.31,140,143 Leukocyte adhesion molecules are upregulated during ECLS, as discussed previously, and it is likely endothelial cell receptors for leukocytes are as well. Plasmin, C5a, and TCC cause P-selectin expression on endothelial cells,144,145 and IL-1 and TNF cause up-regulation of E-selectin, ICAM-1, and VCAM-1.146 Up-regulation of these receptors increases leukocyte adherence and leads to release of cytotoxic substances directly at the endothelium. Coculture of endothelial cell monolayers with activated granulocytes causes an oxygen free-radical-mediated increase in endothelial albumin permeability.147 The increase in permeability appears to be dependent on leukocyte-endothelial binding and is eliminated when the two cell types are separated by a filter.

78

The Artificial Lung

Pulmonary and Cardiovascular Dysfunction Cardiopulmonary bypass is particularly damaging to the heart and lungs due to the compounding of artificial surface and ischemia-induced mediators. As discussed earlier, ischemia causes release of reactive oxygen species from all cell types and the neutrophil chemotactic agent, IL-8, from monocytes. The lung, furthermore, has a highly concentrated endothelial surface area, thus, a disproportional amount of endothelial damage. Significant pulmonary neutrophil and monocyte sequestration is evident during heart and lung reperfusion following bypass25,34 and is mediated by CD11b/CD18 binding.110,148 This sequestration has been associated with cell membrane injury due to lipid peroxidation149 and increased pulmonary permeability during CPB on dogs.150 Neutrophil adherence is inversely proportional with pulmonary blood flow,151 thus, lung damage is reduced with strong cardiac performance. The clinical result of CPB-induced lung injury resembles ARDS and includes increased capillary permeability,152-155 increased extravascular lung water, focal and diffuse atelectasis, occasional alveolar and bronchiolar hemorrhage,156 vascular congestion, increased alveolar-arterial oxygen gradient, decreased functional residual capacity, decreased lung compliance, reduced surfactant activity due to epithelial damage157,158 and bronchospasm due to histamine release.152,156 The result of these changes is reduced gas exchange through the lung and, in severe cases, respiratory failure. The effects of inflammation on the cardiovascular system include reduced myocardial contractility, changes in systemic vascular resistance, and hyper- or hypotension. Post-operative cardiac dysfunction is significantly related to length of bypass and higher C3a24 and IL-6 levels.159 Complement fragment C3a has been shown to cause tachycardia, arrhythmias due to impaired atrioventricular conduction, coronary vasoconstriction, and reduced contractility via activation of endogenous, cardiac leukocytes.160 The arrhythmias and tachycardia appear to be caused by histamine release from cardiac basophils and mast cells, and contractile failure appears to be caused by production of leukotrienes by these same cells.160,161 Coronary vasoconstriction is most likely due to production of thromboxane by platelets.160 Interleukin-6 is produced by mononuclear phagocytes and endothelial cells in response to IL-1 and TNF23 and has been shown to cause myocardial depression via activation of NO production in isolated hamster papillary muscle.162 Interleukin-6 levels increase or remain stable during CPB, increase more significantly during rewarming, and remain elevated up to seven days postoperatively.28,31,79,80,83,84,133 Adherence of neutrophils to coronary endothelial cells and myocytes via CD11b/ CD18 has also been implicated in cardiac dysfunction. Inhibition of CD11b/CD18 expression significantly reduced plasma myeloperoxidase activity and increases in coronary vascular resistance and helped preserve cardiac performance during cardiac surgery on pigs.163 Likewise, antibodies to CD18, CD11b, or ICAM-1 blocked neutrophil adherence to canine cardiac myocytes, subsequent respiratory burst, and myocyte death.164 Numerous vasoactive compounds are also generated during CPB (Table 4.1).165 Most frequently, the result of these appears to be increased systemic vascular resistance (SVR), but this depends on the patient and the stage of CPB. The majority of patients experience increased SVR post-bypass, but a significant portion experience unchanged or slightly reduced SVR.166 Despite a general increase, SVR is markedly decreased at the commencement of bypass, during rewarming, and immediately after cessation of bypass.167 Systemic vascular resistance, however, reflects changes in resistance due to changing blood viscosity and vasoconstriction or vasodilation. In order to isolate anatomic changes in resistance,

systemic systemic systemic systemic systemic systemic systemic systemic small vessels

/ / 1 / / / / / /

Atrial natriuretic factor Bradykinin Triiodothyronine C3a, C4a, C5a PAF PGE2 PGI2 NO Histamine

0 / 1 / 0 0 /

Region

Concentration Change During After CPB CPB

Vasodilator

Table 4.1. Vasoactive substances released during CPB165

Epinephrine Norepinephrine Renin C3a Thromboxane A2 LTB4, LTC4, LTD4 Endothelin-1 Serotonin

Vasoconstrictor

/ / 0 or / / / / or 0 / 0 or / / 0 or / / 0 0 / or 0 0or /

Concentration Change During After CPB CPB

renal and skin systemic systemic coronary systemic systemic and coronary systemic systemic

Region

Biocompatibility of Artificial Lungs

79

The Artificial Lung

80

the results of Kam et al. during CPB can be normalized by dividing SVR by blood viscosity. Blood viscosity, ∋, is calculated according to the formula: 5 2 1965 + 2.31 Hct 7 4 6 !5 3 273.15+ T

∋ = 2.205 ⋅ 10 e

,

(Eqn. 4.1)

in which T represents temperature in degrees Celsius and Hct represents fractional hematocrit.168 The results for normalized SVR (Fig. 4.5), SVRn, suggest that SVRn increases markedly during CPB and remains slightly elevated thereafter. This suggests that vasoconstriction dominates during CPB. There is less information available on cardiovascular dysfunction during ECMO. However, 11% of patients experience hypotension, 6% acquire myocardial stun, 4% have arrhythmias, and an additional 7% acquire other cardiopulmonary complications.

Gastrointestinal Dysfunction Gastric mucosal blood flow is reduced approximately 49% during hypothermic, nonpulsatile CPB169 and can be attributed to hypothermia, steady flow, and various vasoconstrictive substances generated during and after extracorporeal circulation (Table 4.1).170 The reduction in blood flow is particularly damaging due to hemodilution and a normal submucosal, arteriolar hematocrit that is 50-60% lower than systemic.171 Studies show, therefore, that the majority of patients develop some gastric mucosal hypoxia 4-5 hours after CPB.170 Even so, a reduction in perfusion during uncomplicated CPB may not be enough to damage the mucosal tissue of the gut independently.75,172 If cardiovascular complications reduce the cardiac output, however, marked gastric mucosal hypoxia, intramucosal or tissue reductions in pH, and resultant tissue damage may occur. This damage is most likely exacerbated by gut reperfusion injury during the rewarming phase.170 Intra-abdominal tissue damage can lead to high mortality rate complications such as gastrointestinal hemorrhage, but more insidious damage frequently occurs. Gut permeability increases an average of six to seven times immediately following CPB, and this increase is still evident five days after the procedure.169 Increased permeability reduces the gut’s barrier to bacterial translocation, leading to release of endotoxin into the blood stream. Endotoxin is a lipopolysaccharide derived from the walls of degraded gram-negative bacteria and is an activator of the coagulation cascade, classic and alternative complement pathway, leukocyte, and endothelial cell.

Kidney Dysfunction There is evidence of significant kidney damage resulting from CPB, but the clinical significance of this damage is small. Markers of glomerular and tubular injury increase sharply at the first post-operative day, and these markers remained elevated over five days.173 The normal, resting kidney, however, utilizes only a fraction of its filtration capacity, and thus, damage may not result in clinically significant kidney dysfunction. Renal functional reserve decreased significantly at nine days post-CPB, the first postoperative time point, and resolved by six months, the next time point.174 This reduction, however, was not sufficient to influence normal renal function. Nine percent of children, however, experience acute renal failure as a result of CPB. From three to five percent of children also require dialysis, and their mortality rate is from 58 to 79%.175,176 During neonatal ECMO, furthermore, 13% of patients require dialysis, and their mortality rate is 40%. Approximately 1.5% of adults experience renal failure, and their mortality rate is 27-57%.177-178 Evidence suggests, however, that failure in adults

Biocompatibility of Artificial Lungs

81

Fig. 4.5. Normalized systemic vascular resistance during cardiopulmonary bypass.

during CPB is attributed to some renal deficit preceding surgery. It can be concluded, therefore, that adult CPB causes functional deficits in kidneys that are not clinically significant without the presence of a preoperative deficit.

Central Nervous System Dysfunction

Postoperative neuropsychological dysfunction occurs in 40-79% of CPB patients179,180 and persists in up to 35% of patients for at least 12 months.181 During neonatal ECMO, 35% of patients experience neurologic complications. Traditionally, this dysfunction has been attributed predominantly to microemboli. Recent evidence demonstrating inflammation-induced endothelial damage and increases in vascular permeability, however, indicates a possible role for the whole body immune response. Magnetic resonance imaging indicates that the brain becomes edematous in all patients following CPB,182 and leukocyte activation and sequestration may be capable of causing neuronal damage due to release of cytotoxic agents.

Blood Activation During Artificial Lung Use There are, at present, no studies examining blood activation during use of an AL. Examination of the similarities and differences between CPB, ECMO, and AL use should, however, provide insight into the relative character and extent of blood activation with an AL. Like ECLS oxygenators, an AL has a large, artificial gas transfer surface that is designed for maximal blood volume contact in order to minimize oxygen boundary layers and increase gas transfer efficiency. Unlike ECLS, however, an AL utilizes no external pump, heat exchanger, tubing, or blood reservoir, which significantly reduces blood exposure to artificial surfaces. An AL should, therefore, experience less total protein adsorption and, thus, less complement and contact activation.

82

The Artificial Lung

This gain may, however, be offset by the low Reynolds number flow within an AL. Blood flow through the AL is driven by the right ventricle (RV), and the AL must have a very low blood flow resistance to avoid straining the RV. The blood flow through an AL thus has a significantly lower average Reynolds number, blood flow velocity, and shear stress than ECLS oxygenators. Convection of activated factors away from the device surface is reduced, consequently, and this may increase local concentrations of activated complement and contact factors over those within ECLS oxygenators. The reduced shear stress should also affect platelet and leukocyte adherence. The root mean square shear stress within an AL fiber bundle2 is 7 dynes/cm2 (unpublished results), significantly below platelet shear stress activation thresholds. Thus, an AL should not cause shear-induced activation of platelets. It may, on the other hand, increase leukocyte adherence and activation. Blood flow through the AL is also driven by the RV rather than a mechanical pump and is, therefore, pulsatile. The effect of cycling velocity and shear fields on blood activation, however, is unknown. Patient conditions during AL use are similar to those during ECMO. Like ECMO, use of an AL does not involve periods of hypothermia and heart or lung ischemia, and, as in veno-venous ECMO, systemic blood flow remains fully pulsatile. Normothermia may heighten neutrophil reactivity. Hypothermia has been shown to increase contact activation,21 but controlled studies indicate reduced neutrophil elastase release and no difference in p-selectin, e-selectin, or IL-8 concentrations during hypothermic CPB.30 Normothermia and the pulsatile left ventricular output should, however, protect the gut against ischemia. Lack of cardiopulmonary ischemia should reduce free radical generation, monocyte IL-8 production, and neutrophil adhesion and activation within the heart and lungs. Pulmonary blood flow may, however, be diverted to the AL, reducing pulmonary perfusion as during veno-arterial ECMO. Blood flow velocity and shear stresses would be reduced, thus, pulmonary neutrophil adhesion and subsequent pulmonary damage may be increased. The natural lung is also the site for conversion of various vasoactive molecules, and thus, systemic blood pressure could be affected. Studies indicate, however, that a 50% reduction in pulmonary blood flow does not significantly affect SVR, and a reduction between 25 and 50% may also be tolerable.242 In sum, blood activation within an AL should be most similar to blood activation during ECMO. The AL surface area and average Reynolds number is reduced, however, and the character and extent of blood activation will thus be different. Reduced shear stresses should, to some extent, decrease platelet activation while increasing leukocyte adhesion and activation. The sum of these effects must, however, be determined experimentally.

Combatting Blood Activation Blood activation during ECLS has been combatted primarily through pharmacological means and secondarily through surface immobilization of bioactive molecules. This hierarchy has been acceptable for short-term ECLS, but long-term use of an AL necessitates an adjustment in philosophy. Long-term pharmacological inactivation of the entire clotting cascade, platelets, complement, or leukocytes is unwise due to their functions in maintaining hemostasis and fighting infection. A wiser approach is warranted, therefore, based on the knowledge we now have about artificial surface-induced blood activation. Blood activation must be reduced primarily at its source, protein adsorption and activation. Secondarily, bioactive molecules can be used to inactivate proteins that do manage to become activated at the material surface. Lastly, pharmacological intervention should be utilized minimally to make up for the failings of the first two lines of defense.

Biocompatibility of Artificial Lungs

83

Pharmacological Intervention Blood activation has been combatted mainly through the elimination of fibrin clots using systemic heparinization. This technique has been highly successful overall, allowing ECLS to proceed without the formation of occluding thrombus and consequent failure of the entire extracorporeal circuit. Systemic heparinization, however, combats one specific consequence of blood activation, fibrin formation, rather than its direct cause. For this reason, heparin does little to combat activation and consumption of contact factors, complement, platelets and leukocytes and subsequent complications. Heparin, furthermore, directly activates platelets and may contribute to thrombocytopenia and platelet dysfunction caused by ECLS.183 Systemic heparinization, furthermore, inhibits both the intrinsic and extrinsic branches of the coagulation cascade. During surgery, both branches are activated and general inhibition is necessary to eliminate clotting within the surgical field. During post-surgical use of an AL, however, the vast majority of blood activation occurs within the circuit due to the intrinsic branch. Systemic heparinization at this stage, therefore, addresses an intrinsic branch coagulopathy by unnecessarily disabling clotting of both branches. The extrinsic branch of the coagulation cascade is necessary for the elimination of internal as well as post-operative bleeding and, thus, bleeding complications can occur. The inability of heparin to eliminate contact, platelet, complement, and leukocyte dysfunction has motivated the search for other drug therapies. A number of agents have proven effective at inhibiting various platelet activation pathways (Table 4.2).184 Prostacyclin (PGI2), PGE1, and Iloprost, however, cause vasodilation and unacceptable reductions in blood pressure, and ticlopidine and clopidogrel may cause bone marrow depression. Both types of agents are, therefore, unacceptable for long-term use. Thromboxane synthase inhibitors, thromboxane receptor antagonists, and disintegrins have not demonstrated such limiting side effects, and new thrombin or FXa inhibitors may reduce platelet activation during ECLS when used as a replacement or adjunct to unfractionated heparin. Leukocyte inhibition has been attempted using the steroids prednisolone, methylprednisolone, and dexamethasone. These steroids have been shown to reduce post-bypass IL-8,86,185 and LTB4 concentrations and t-PA activity.27 There was no effect on elastase concentrations, and complement activation was reduced86 or unchanged27 during CPB and increased during ECMO.36 Significant, long-term inhibition of these systems may lead, however, to increased complications due to bleeding and infection. Furthermore, attempted inhibition of any of these systems is an inefficient means of eliminating blood activation during ECLS because it does not remove the initial activating stimuli. A more promising approach is the use of contact inhibitors. These agents can reduce many avenues of blood activation indirectly by inhibiting FXIIa or kallikrein and may also have a direct inhibitory effect on other proteases involved in blood activation. A fully functioning contact system, furthermore, is not necessary for normal hemostasis.8 These agents include aprotinin, nafamostat, Bz-Pro-Phe-boroArg-OH, Arg15-aprotinin, Ala357-Arg358 ∃1-antitrypsin, soybean trypsin inhibitor, and ecotin. Aprotinin is primarily an inhibitor of plasmin and has been effective at reducing post-bypass bleeding.186 In addition, aprotinin is a kallikrein inhibitor at high doses. Aprotinin’s kallikrein inhibitory action is, however, one one-hundredth of its inhibition of plasmin, and aprotinin has, therefore, been ineffective at reducing contact activation during clinical CPB. Despite an increase in kallikrein inhibiting capacity, there was no difference in kallikrein activity and prekallikrein, FXII, HK, and FXI concentrations during

The Artificial Lung

84

Table 4.2. Non-heparin anticoagulants185 Agent

Inhibits

Platelet inhibitors Prostacyclin (PGI2), PGE2, Iloprost Ticlopidine, Clopidogrel Disintegrins Thromboxane synthase inhibitors Thromboxane receptor antagonists

Platelet activation ADP-induced platelet activation GpIIb/IIIa binding Platelet thromboxane production Platelet thromboxane-induced activation

Thrombin inhibitors Hiruden, hirulog, hirugen, argotroban, DUP714 PPACK Hirudisins

Thrombin, reversibly Thrombin, irreversibly Thrombin and GpIIb/IIIa binding

FXa inhibitors Tick anticoagulant peptide, DX-9065

FXa reversibly

clinical CPB.89,187 Aprotinin, furthermore, had no effect on platelet counts and activation,89 complement activation,188 and elastase release.89,188 Nafamostat mesilate (FUT-175) is an inhibitor of FXIIa, kallikrein, tissue factor-FVIIa activity, thrombin, plasmin, and complement factors C1r and C1s.189-191 The effect of nafamostat on contact activation, thrombin formation, platelet counts, complement activation, and leukocyte activation13,80,192,193 during SECC and CPB has, unfortunately, been mixed. Nafamostat does, however, appear to reduce platelet release of B-TG. Wachtfogel et al. have also tested the selective kallikrein inhibitors, Bz-ProPhe-boroArg-OH, Arg15-aprotinin, Ala357-Arg358 ∃1-antitrypsin, and soybean trypsin inhibitor during SECC. 17 Each demonstrated similar, significant reductions in kallikrein-C1INH concentration, but none significantly reduced C1-C1INH complex concentrations. Soybean trypsin inhibitor, furthermore, significantly increased C1-C1INH complex concentrations. Only Bz-Pro-Phe-boroArg-OH and Ala357-Arg358 ∃1-antitrypsin significantly reduced elastase release, and none affected platelet losses. Aprotinin and nafamostat are not potent enough to be effective inhibitors of contact and subsequent blood activation during ECLS. Aprotinin and nafamostat’s equilibrium dissociation constants, Ki, for kallikrein are 45 nM and 310 nM respectively.192,194 Nafamostat also inhibits FXIIa, but less effectively than it does kallikrein.195 The most effective contact inhibitor during the study of Wachtfogel et al., Bz-Pro-Phe-boroArg-OH, has a potent Ki for kallikrein, 150 pM, but fails to significantly inhibit FXIIa or other serine proteases involved in blood activation.196 Ecotin, however, is a potent inhibitor of both FXIIa and kallikrein and appears, therefore, to be more promising. Ecotin’s Ki for FXIIa and kallikrein are 89 and 163 pM, respectively.197 In addition, ecotin is a potent inhibitor of FXa and elastase, with Ki’s of 54 and 55 pM, respectively.197,198 Ecotin has not yet been tested during SECC or CPB but has the potential to minimize contact activation, exert an anticoagulant effect, and directly reduce damage done by activated neutrophils. Furthermore, it may indirectly reduce platelet, complement, and leukocyte activation.

Biocompatibility of Artificial Lungs

85

Heparin Coatings The problem of heparin-induced bleeding complications has also been addressed clinically by the creation of heparin-coated extracorporeal circuits. These circuits bind heparin directly to the material surfaces in order to focus the anticoagulant where it is needed most, at the site of activation. This allows systemic doses of heparin to be reduced, which, in turn, reduces the probability of bleeding complications. Two types of heparin coatings are currently available for clinical use, Duraflo II and the Carmeda Bioactive Surface (CBAS). The Duraflo II coating ionically binds unfractionated heparin to the surface via a carrier molecule, alkylbenzyldimethylammonium chloride containing an 18-carbon length alkyl chain.199 The Carmeda Bioactive Surface covalently binds nitrous-acid degraded heparin to the surface via a polyethylenimine spacer molecule.200,201 Experiments examining Duraflo II’s ability to reduce blood activation have had mixed results. Duraflo II has been shown to bind FXII and reduce plasma FXII activity, bind ATIII14 and reduce plasma concentrations of the thrombin-antithrombin III (TAT) complex,202-204 and improve the thrombin-inhibiting capacity of surfaces14 during clinical CPB. Duraflo II also reduces fibrinogen adsorption on polyester membranes and consumption during clinical CPB.205 Platelet losses and release of PF4 are reduced during SECC,206 but there was no difference in platelet numbers or release of B-TG during CPB.14,207 Duraflo II reduces formation of the TCC and C3 activation products, C3b, iC3b, and C3a, during CPB, 95,205,211 and the effect on neutrophil release of elastase,29,204 lactoferrin, and myeloperoxidase95,96,211,212 has been positive, yet somewhat mixed. Lastly, reductions in bleeding have been demonstrated during CPB with reduced systemic heparinization,208,209 but there is no benefit from Duraflo II coatings in the absence of reduced heparin doses and clotting times.204,207,210 The Carmeda Bioactive Surface (CBAS) has shown similar and somewhat more promising results. Carmeda-coated circuits demonstrate reduced consumption of FXII and reduced kallikrein activity during SECC,213 and reduced consumption of FXII, HK, and fibrinogen during clinical CPB.11 There was also less fibrin deposition on CBAS-coated arterial filters during clinical CPB.214 Despite these promising results, Gorman et al. have shown that CBAS does not affect prothrombin fragment F1.2 or TAT, indicating that thrombin formation is not affected.215 The Carmeda coating does reduce platelet losses213,216 and release of B-TG217,218 and PF-4218 during SECC and platelet losses214,215 during CPB. Platelet release of B-TG during CPB is either reduced98,219 or unchanged.215 A 168-hour study examining the effect of a CBAS-like coating during ECMO on sheep impressively demonstrated constant platelet counts and reduced B-TG release.220 Reductions in bleeding have also been demonstrated during clinical CPB using CBAS with reduced systemic heparinization.221 Carmeda reduces TCC and C3 activation products during SECC,222 formation of the TCC and C3 activation products during clinical CPB,95,205,223,224 and C3a formation during ECMO on sheep.220 Moen et al., in contrast, demonstrated reduced TCC formation but no significant difference in C3 activation products during CPB.98 Carmeda also reduces concentrations of ∃1-proteinase-elastase complex,213 myeloperoxidase,88,222 lactoferrin, and CD11b during SECC222 as well as lactoferrin94,95,217,223,224 and elastase219 during CPB. Concentrations of myeloperoxidase during CPB were either reduced94,223 or unchanged.95,224 The Carmeda surface also causes significantly less generation of TCC, C3 activation products, lactoferrin, and myeloperoxidase when compared to Duraflo II.95,211,212

86

The Artificial Lung

Coating Development for Long-Term AL Use Heparin coatings appear to provide some measure of protection against blood activation and, thus, may be of value within an AL. These surfaces, however, do not completely eliminate blood activation and still require significant pharmacological intervention. Duraflo II, furthermore, will lose its effectiveness over time due to leaching of its ionically bound heparin. Moreover, the original purpose of heparin coatings was to eliminate fibrin formation through the inactivation of FXa and thrombin, yet this function has not been proven. Heparin coatings, unlike systemic heparin, have, however, been shown to reduce contact activation. It is also likely that the heparin coatings affect adsorption and activation of other proteins, and this may be the cause of reduced platelet, complement, and leukocyte activation. The serendipitous nature of these benefits indicates that a more effective coating could be developed through purposeful minimization of protein adsorption and activation via surface grafting of long-chain molecules. When a protein approaches a surface grafted with long-chain molecules, steric hindrance from the protein reduces the motion and configurational freedom of the grafted molecules. The reduction in configurational entropy increases the configurational free energy of the grafted molecules and creates a repulsive force that pushes the protein away from the surface base.225-227 This force may be sufficient to overcome attractive van der Waals forces, thus, the protein may deform the grafted molecules but not attach to them or the surface material.228 There may also be an osmotic repulsion force that occurs due to the rise in osmotic pressure when long-chain molecules interpenetrate each other.227 Both of these forces are enhanced with increasing chain length and density of the grafted molecules and increasing solvent affinity for the grafted molecules and approaching proteins.229 Surface bound polyethylene oxide (PEO) and polyethylene glycol (PEG) (Fig. 4.6) have proved to be successful at protein repulsion. Both molecules are structurally the same, and PEG refers to PEO molecules of less than 20,000 g/mol. Polyethylene oxide is highly mobile, has a large exclusion volume,228 and thus, a greater repulsive force from configurational restriction. Polyethylene oxide, furthermore, has a high affinity for water due to its repeating ether bonds and a larger osmotic repulsive force. In addition to its mobility, polyethylene oxide also possesses no ionic charge, thus, an approaching protein finds few sites and little time to form a solid bond. Polyethylene oxide has been bound to surfaces via covalent grafting, plasma polymerization, and adsorption of PEO surfactants; incorporated within cast or extruded PEO-copolymers, and swollen into physical interpenetrating polymer networks (IPNs). Surface grafting of PEO significantly reduces protein adsorption,230,231 platelet adhesion,230,232 fibroblast230 and bacterial growth233, and thrombogenicity.234 Adsorption of PEO surfactants reduces protein adsorption235,236 and platelet adhesion,236 and PEO copolymers reduce leukocyte adhesion.73 Physical IPNs reduce protein adsorption, platelet adhesion, fibroblast adhesion, and thrombogenicity.237 Polyethylene oxide has also been incorporated within grafted semi- and sequential IPNs. Grafted semi-IPNs are formed by end-grafting linear PEO to the matrix of a second, cross-linked polymer that covers the base polymer.238 Sequential IPNs are formed by creating an initial, crosslinked, polymer network on the surface of the base material. Polyethylene oxide plus a cross-linker is then swollen into the first polymer network using a mutual solvent and polymerized in situ.229 This forms two overlapping, covalently bonded polymer matrices.239 Both types of IPNs are thought to be superior to other means of incorporating PEO. Surface grafting may leave microscopic areas of the surface uncoated due to diffusional or steric limitations, adsorbed PEO desorbs over time, and physical

Biocompatibility of Artificial Lungs

87

Fig. 4.6. Polyethylene oxide.

IPNs leach PEO in the presence of solvents such as water.238 The protein-resistant character of these surfaces, furthermore, can be easily damaged by shear or chemical attack during inflammation. Polyethylene oxide in grafted semi-IPNs and sequential IPNs is both stable and dense. The PEO cannot diffuse from the surface, and slight degradation does not expose the base polymer. The dense PEO matrix may also enhance the protein resistance due to increased steric hindrance. Semi-IPNs show large reductions in protein adsorption and fibroblast adhesion,238 and sequential IPNs have shown no protein adsorption from 15% fetal bovine serum and reduced endothelial cell adhesion.239 The thrombogenicity of polyethylene oxide surfaces can be further reduced using bound bioactive molecules. Surface grafted PEO with end-grafted heparin has reduced the thrombogenicity of Biomer vascular grafts during ex vivo rabbit studies240 and in vivo dog studies.241 The quest for effective surface-bound biological molecules should, however, be no more limited than the quest for effective systemic drug therapies. More potent anticoagulants and platelet, leukocyte, and contact system inhibitors could be immobilized upon PEO-coated surfaces without crippling natural hemostatic and immune function. Promising anticoagulants include PPACK, a potent, irreversible inhibitor of thrombin that also inhibits platelet activation, and hirudisins, thrombin inhibitors that also inhibit platelet GpIIb/IIIa activity. Still more promising is ecotin, which has the potential to: 1. inhibit all forms of blood activation by direct inhibition of FXIIa and kallikrein, 2. inhibit thrombin and fibrin formation and platelet and endothelial cell activation via its actions on FXa, and 3. reduce the harmful effects of neutrophil elastase release.

Continued development of protein resistant surfaces, surface bound bioactive molecules, and combinations of the two should reduced complications and the need for drug therapy during AL use.

Summary Knowledge of the mechanisms involved in blood activation has increased exponentially since the first attempts at extracorporeal circulation. It is now understood that blood activation is initiated by protein adsorption and activation. Adsorption causes autoactivation of contact system factors, allows activated complement to escape from plasma inhibitors, and induces conformational changes in fibrinogen that allow binding with inactivated platelets. Activated contact factors, complement, and fibrinogen then initiate platelet and leukocyte activation, formation of solid clot, and the whole body inflammatory response without pharmacological intervention. Activation of these systems can result in endothelial and organ dysfunction that leads to increased patient mortality and morbidity. Knowledge of this process should allow better directed inhibition of blood activation that focuses on the root of activation rather than the end effects. Current means of combatting activation focus on drug therapy that typically inhibits later stages of activation:

88

The Artificial Lung

fibrin formation and platelet and leukocyte activation. These systems are necessary, however, for normal hemostasis and immunity, and inhibition may lead to further complications. Surface-immobilized heparin has thus been applied to ECLS circuits to decrease fibrin formation and blood clotting but has not been successful enough to justify a significant reduction in anticoagulant therapy. Reduction of protein adsorption is a more efficient means of minimizing blood activation because it inhibits its initiation: contact factor, complement, and fibrinogen adsorption. One proven method of reducing adsorption is application of PEO interpenetrating polymer networks. These surfaces demonstrate excellent adsorption resistance, have sufficient durability for long-term use, and can be grafted with bioactive molecules that compensate for activation that the PEO is unable to prevent. Contact inhibitors’ actions are focused at a root of blood activation and are thus a promising choice for immobilization. Ecotin is one encouraging, potent example with additional anticoagulant and anti-elastase effects. Potent immobilized molecules with combined anticoagulant and platelet inhibitory effects, such as hirudisins and PPACK, may also be effective. Artificial lung use necessitates minimal drug therapy at all times such that significant hemodynamic and immune compromise do not occur. Blood activation within an AL, like ECMO, should be severe during surgery and the initial 24 hours due to the compounding of surgical and surface induced activation. Drug therapy may, therefore, be necessary during this period. After this period, however, blood activation subsides and drug therapy could be reduced accordingly. Development of protein resistant surfaces and creation of new, potent inhibitors of various blood systems continues and may lead to more effective, specific inhibition of surface reactions. Creation of endothelial-like artificial surfaces is not yet attainable, yet long-term use of large surface area devices with minimal drug therapy is certainly within our reach. References 1. Lillehei CW. History of the development of extracorporeal circulation. In: Arensman RM, Cornish JM, eds. Extracorporeal Life Support. Boston: Blackwell Scientific Publications, 1993:9-30. 2. Cook KE, Makarewicz AJ, Backer CL et al. Testing of an intrathoracic artificial lung in a pig model. ASAIO J 1996; 42:M604-M609. 3. Edmunds LH. Inflammatory and immunological response to cardiopulmonary bypass. In: Jonas RA, Elliot MJ, eds. Cardiopulmonary bypass in Neonates, Infants, and Young Children. Boston: Butterworth-Heinemann, 1994:224-241. 4. Edmunds LH. The sangreal. J Thorac Cardiovasc Surg 1985; 90:1-6. 5. Ziats NP, Pankowsky DP, Tierney BP et al. Adsorption of Hageman factor (factor XII) and other human plasma proteins to biomedical polymers. J Lab Clin Med 1990; 116:687-696. 6. Uniyal S, Brash JL. Patterns of adsorption of proteins from human plasma onto foreign surfaces. Thromb Haemost 1982; 47:285-290. 7. Salzman EW, Merrill EW, Kent KC. Interactions of blood with artificial surfaces. In: Colman RW, Hirsh J, Marder VJ, Salzman EW, eds. Hemostasis and Thrombosis: Basic Principles and Clinical Practice. 3rd ed. Philadelphia: J.B. Lippincott Company, 1994:1469-1485. 8. DeLa Cadena RA, Wachtfogel YT, Colman RW. Contact activation pathway: Inflammation and coagulation. In: Colman RW, Hirsh J, Marder VJ, Salzman EW, eds. Hemostasis and Thrombosis: Basic Principles and Clinical Practice. 3rd ed. Philadelphia: J.B. Lippincott Company, 1994:219-240. 9. Brash JL, Scott CF, ten Hove P et al. Mechanism of transient adsorption of fibrinogen from plasma to solid surfaces: role of the contact and fibrinolytic systems. Blood 1988; 71:932-939. 10. Irvine L, Sundaram S, Courtney JM et al. Monitoring of Factor XII activity and granulocyte elastase release during cardiopulmonary bypass. ASAIO Trans 1991; 37:569-571. 11. Shigemitsu O, Hadama T, Takasaki H et al. Biocompatibility of a heparin-bonded membrane oxygenator (Carmeda MAXIMA) during the first 90 minutes of cardiopulmonary bypass: Clinical comparisons with the conventional sytem. Artif Organs 1994; 18:936-941.

Biocompatibility of Artificial Lungs

89

12. Sundaram S, Courtney JM, Taggart DP et al. Biocompatibility of cardiopulmonary bypass: Influence on blood compatibility of device type, mode of blood flow and duration of application. Intl J Artif Organs 1994; 17:118-28. 13. Usui A, Hiroura M, Kawamura M et al. Nafamostat mesilate reduces blood-foreign surface reactions similar to biocompatible materials. Ann Thorac Surg 1996; 62:1404-1411. 14. van der Kamp KWJH, van Oeveren W. Contact, coagulation and platelet interaction with heparin treated equipment during heart surgery. Int J Artif Organs 1993; 16:836-842. 15. Plotz F, van Oeveren W, Bartlett RH et al. Blood activation during neonatal extracorporeal life support. J Thorac Cardiovasc Surg 1993; 105:823-832. 16. Wachtfogel YT, Harpel PC, Edmunds LH et al. Formation of C1s-C1-inhibitor, kallikreinC1-inhibitor, and plasmin-∃2-plasmin inhibitor complexes during cardiopulmonary bypass. Blood 1989; 73:468-471. 17. Wachtfogel YT, Hack CE, Nuijens JH et al. Selective kallikrein inhibitors alter human neutrophil elastase release during extracorporeal circulation. Am J Physiol 1995; 268:H1352-H1357. 18. de Smet AAEA, Joen MCN, van Oeveren W et al. Increased anticoagulation during cardiopulmonary bypass by aprotinin. J Thorac Cardiovasc Surg 1990; 100:520-527. 19. van Oeveren W, Harder MP, Roozendaal KJ et al. Aprotinin protects platelets against the initial effects of cardiopulmonary bypass. J Thorac Cardiovasc Surg 1990; 99:788-797. 20. Ellison N, Behar M, MacVaugh H et al. Bradykinin, plasma protein fraction, and hypotension. Ann Thorac Surg 1979; 29:15-19. 21. Pang LM, Stalcup SA, Lipset JS et al. Increased circulating bradykinin during hypothermia and cardiopulmonary bypass in children. Circulation 1979; 60:1503-1507. 22. Chenoweth DE. Complement activation in extracorporeal circuits. Ann NY Acad Sci 1987; 516:306-313. 23. Abbas AK, Lichtman AH, Pober JS. Cellular and Molecular Immunology. Philadelphia: WB Saunders Company, 1996. 24. Kirklin JK, Westaby S, Blackstone EH et al. Complement and the damaging effects of cardiopulmonary bypass. J Thorac Cardiovasc Surg 1983; 86:845-857. 25. Howard RJ, Crain C, Franzini DA et al. Effects of cardiopulmonary bypass on pulmonary leukostasis and complement activation. Arch Surg 1988; 123:1496-1501. 26. Tennenberg SD, Clardy CW, Bailey WW et al. Complement activation and lung permeability during cardiopulmonary bypass. Ann Thorac Surg 1990; 50:597-601. 27. Jansen NJG, van Oeveren W, van den Broek L et al. Inhibition by dexamethasone of the reperfusion phenomena in cardiopulmonary bypass. J Thorac Cardiovasc Surg 1991; 102:515-525. 28. Steinberg JB, Kapelanski DP, Olson JD et al. Cytokine and complement levels in patients undergoing cardiopulmonary bypass. J Thorac Cardiovasc Surg 1993; 56:1387-1389. 29. Gu YJ, van Oeveren W, Akkerman C et al. Heparin-coated circuits reduce the inflammatory response to cardiopulmonary bypass. Ann Thorac Surg 1993; 55:917-922. 30. Menasché P, Peynet J, Haeffner-Cavaillon N et al. Influence of temperature on neutrophil trafficking during clinical cardiopulmonary bypass. Circulation 1995; 92(Suppl II):334-340. 31. Tsang GM, Allen S, Pagano D et al. Pentoxyfylline preloading reduces endothelial injury and permeability in cardiopulmonary bypass. ASAIO J 1996; 42:M429-M434. 32. Weiler JM, Gellhaus MA, Carter JG et al. A prospective study of the risk of an immediate adverse reaction to protamine sulfate during cardiopulmonary bypass surgery. J Allergy Clin Immunol 1990; 85:713-719. 33. Fosse E, Mollnes TE, Ingvaldsen B. Complement activation during major operations without cardiopulmonary bypass. J Thorac Cardiovasc Surg 1987; 93:860-866. 34. Chenoweth DE, Cooper SW, Hugli TE et al. Complement activation during cardiopulmonary bypass: Evidence for generation of C3a and C5a anaphylatoxins. N Engl J Med 1981; 304:497-503. 35. Hocker JR, Wellhausen SR, Ward RA et al. Effect of extracorporeal membrane oxygenation on leukocyte function in neonates. Artif Organs 1991; 15:23-28. 36. Westfall SH, Stephens C, Kesler K et al. Complement activation during prolonged extracorporeal membrane oxygenation. Surgery 1991; 110:887-891. 37. Hahn-Pederson J, Sorenson H, Kehlet H. Complement activation during surgical procedures. Surg Gynecol Obstet 1978; 146:66-68. 38. Brash JL, Davidson VJ. Adsorption on glass and polyethylene from solutions of fibrinogen and albumin. Thromb Res 1976; 9:249-259.

90

The Artificial Lung

39. Brash JL. Hydrophobic polymer surfaces and their interactions with blood. Ann NY Acad Sci 1977; 283: 356-371. 40. Yamada KM, Olden K. Fibronectins: Adhesive glycoproteins of cell surface and blood. Nature 1978; 275:179-184. 41. Sakariassen KS, Bolhuis PA, Sixma JJ. Human blood platelet adhesion to artery subendothelium is mediated by factor VII-von Willebrand factor bound to the subendothelium. Nature 1979; 279:636-638. 42. Jaffe EA, Leung LL, Nachman RL et al. Thrombospondin is the endogenous lectin of human platelets. Nature 1982; 295:246-248. 43. Mosher DF. Influence of proteins on platelet-surface interaction. In: Salzman EW, ed. Interaction of the Blood with Natural and Artificial Surfaces. New York: Marcel Dekker, 1981:85-101. 44. Collins WE, Mosher DF, Tomasini BR et al. A preliminary comparison of the thrombogenic activity of vitronectin and other RGD-containing proteins when bound to surfaces. Ann NY Acad Sci 1987; 516:291-299. 45. Lindon JN, Kushner L, Shiba E et al. Platelet activation by a hydrophobic surface is mediated by cooperative interactions of fibrinogen with IgG. Circulation 1988; 78(Suppl II):621. 46. Shiba E, Lindon JN, Kushner L et al. Antibody-detectable changes in fibrinogen adsorption affecting platelet activation on polymer surfaces. Am J Physiol 1991; 260:C965-C974. 47. Addonizio VP, Macarak EJ, Nicolaou KC et al. Effects of prostacyclin and albumin on platelet loss during in vitro simulation of extracorporeal circulation. Blood 1979; 53:1033-1042. 48. Wachtfogel YT, Musial J, Jenkin B et al. Loss of platelet ∃2-adrendergic receptors during simulated extracorporeal circulation: prevention with prostaglandin E 1. J Lab Clin Med 1985; 105:601-607. 49. George JN, Pickett EB, Saucerman S et al. Platelet surface glycoproteins. Studies on resting and activated platelets and platelet membrane microparticles in normal subjects, and observations in patients during adult respiratory distress syndrome and cardiac surgery. J Clin Invest 1986; 78:340-348. 50. Dechavanne M, Ffrench M, Pages J et al. Significant reduction in the binding of a monoclonal antibody (LYP 18) directed against the IIb/IIIa glycoprotein complex to platelets of patients having undergone extracorporeal circulation. Thromb Haemost 1987; 57:106-109. 51. Wenger RK, Lukasiewicz H, Mikuta BS et al. Loss of platelet fibrinogen receptors during clinical cardiopulmonary bypass: a study of hemodynamic and humoral factors. J Thorac Cardiovasc Surg 1989; 97:235-239. 52. Harker LA, Malpass TW, Branson HE et al. Mechanism of abnormal bleeding in patients undergoing cardiopulmonary bypass: Acquired transient platelet dysfunction associated with selective ∃-granule release. Blood 1980; 56:824-834. 53. Zilla P, Fasol R, Groscurth P et al. Blood platelets in cardiopulmonary bypass operations. Recovery occurs after initial stimulation, rather than continual activation. J Thorac Cardiovasc Surg 1989; 97:379-388. 54. Moriau M, Masure R, Hurlet A et al. Haemostasis disorders in open heart surgery with extracorporeal circulation. Vox Sang 1977; 32:41-51. 55. Harker LA, Schlister SJ. Platelet and fibrinogen consumption in man. N Engl J Med 1972; 287:999-1005. 56. Wachtfogel YT, Pixley RA, Kucich U. Purified plasma factor XIIa aggregates human neutrophils and causes degranulation. Blood 1986; 67:1731-1737. 57. Toossi Z, Sedor JR, Mettler MA et al. Induction of expression of monocyte interleukin 1 by Hageman factor (FXII). Proc Natl Acad Sci USA 1992; 89:11969-11972. 58. Kaplan AP, Kay AB, Austen KF. A prealbumin activator of prekallikrein. III. Appearance of chemotactic activity for human neutrophils by the conversion of human prekallikrein to kallikrein. J Exp Med 1972; 135:81-97. 59. Schapira M, Despland E, Scott CF et al. Purified human plasma kallikrein aggregates human blood neutrophils. J Clin Invest 1982; 69:1191-1202. 60. Wachtfogel YT, Kucich U, James HL et al. Human plasma kallikrein releases neutrophil elastase during blood coagulation. J Clin Invest 1983; 72:1672-1677. 61. Gustafson EJ, Schmaier AH, Wachtfogel YT et al. Human neutrophils contain and bind high molecular weight kininogen. J Clin Invest 1989; 84:28-35. 62. Chenoweth DE, Hugli TE. Demonstration of a specific C5a receptor on intact human polymorphonuclear leukocytes. Proc Natl Acad Sci 1978; 75:3943-3947.

Biocompatibility of Artificial Lungs

91

63. Birdsall HH, Green DM, Trial J et al. Complement C5a, TGF-beta 1, and MCP-1, in sequence, induce migration of monocytes into ischemic canine myocardium within the first one to five hours after reperfusion. Circulation 1997; 95:684-692. 64. Issekutz TB, Chuluyan HE, Lopes N. CD11/CD18-independent transendothelial migration of human polymorphonuclear leukocytes and monocytes: Involvement of distinct and unique mechanisms. J Leuk Biol 1995; 57:553-561. 65. Issekutz TB. In vivo blood monocyte migration to acute inflammatory reactions, IL-1∃, TNF-∃, IFN-8, and C5a utilizes LFA-1, Mac-1, and VLA-4. The relative importance of each integrin. J Immunol 1995; 154:6533-6540. 66. Monk PN, Barker MD, Partridge LJ. Multiple signaling pathways in the C5a-induced expression of adhesion receptor Mac-1. Biochim Biophys Acta 1994; 1221:323-329. 67. Palmantier R, Surette ME, Sanchez A et al. Priming for the synthesis of 5-lipoxygenase products in human blood ex vivo by human granulocyte-macrophage colony-stimulating factor and tumor necrosis factor-∃. Lab Invest 1994; 70:696-704. 68. Ember JA, Sanderson SD, Hugli TE et al. Induction of interleukin-8 synthesis from monocytes by human C5a anaphylatoxin. Am J Pathol. 1994; 144:393-403. 69. Scholz W, McClurg MR, Cardenas GJ et al. C5a-mediated release of interleukin-6 by human monocytes. Clin Immunol Immunopathol 1990; 57:297-307. 70. Geiger T, Rordorf C, Galakatos N et al. Recombinant human C5a induces transcription but not translation of interleukin-1% mRNA in human monocytes. Res Immunol 1992; 143:117-123. 71. Edwards SW. Biochemistry and Physiology of the Neutrophil. New York: Cambridge University Press, 1994. 72. Janvier G, Baquey C, Roth C et al. Extracorporeal circulation, hemocompatibility, and biomaterials. Ann Thorac Surg 199; 62:1926-1934. 73. Lim F, Cooper SL. The effect of surface hydrophobicity on biomaterial-leukocyte interactions. ASAIO Trans 1991; 37:M146-M147. 74. Inui K. Participation of platelet activating factor in the pulmonary injury during cardiopulmonary bypass. Nippon Kyobu Geka Gakkai Zasshi 1993; 41:238-246. 75. Quigley RL, Caplan MS, Perkins JA et al. Cardiopulmonary bypass with adequate flow and perfusion pressures prevents endotoxaemia and pathologic cytokine production. Perfusion 1995; 10:27-31. 76. Gu YJ, van Oeveren W, Boonstra PW et al. Leukocyte activation with increased expression of CR3 receptors during cardiopulmonary bypass. Ann Thorac Surg 1992; 53:839-843. 77. Utoh J, Yamamoto T, Kambara T et al. Perioperative alteration of neutrophil functions after open heart and other major surgeries. J Cardiovasc Surg 1989; 30:692-695. 78. Gadaleta D, Fahey, AL, Verma M et al. Neutrophil leukotriene generation increases after cardiopulmonary bypass. J Cardiovasc Surg 1994; 108:642-647. 79. Frering B, Philip I, Dehoux M et al. Circulating cytokines in patients undergoing normothermic cardiopulmonary bypass. J Thorac Cardiovasc Surg 1994; 108:636-641. 80. Sawa Y, Shimazaki Y, Kadoba K et al. Attenuation of cardiopulmonary bypass-derived inflammatory reactions reduces myocardial reperfusion injury in cardiac operations. J Thorac Cardiovasc Surg 1996; 111:29-35. 81. Finn A, Naik S, Klein N et al. Interleukin-8 release and neutrophil degranulation after pediatric cardiopulmonary bypass. J Thorac Cardiovasc Surg 1993; 105:234-241. 82. Metinko AP, Kunkel SL, Standiford TJ et al. Anoxia-hyperoxia induces monocyte-derived interleukin-8. J Clin Invest 1992; 90:791-798. 83. Butler J, Chong GL, Baigrie RJ et al. Cytokine responses to cardiopulmonary bypass with membrane and bubble oxygenation. Ann Thorac Surg 1992; 53:833-838. 84. Inaba H, Kochi A, Yorozu S. Suppression by methylprednisolone of augmented plasma endotoxin-like activity and interleukin-6 during cardiopulmonary bypass. Br J Anaesth 1994; 72:348-350. 85. Kalfin RE, Engelman RM, Rousou JA et al. Induction of interleukin-8 expression during cardiopulmonary bypass. Circulation 1993; 88:II401-II406. 86. Engelman RM, Rousou JA, Flack JE et al. Influence of steroids on complement and cytokine generation after cardiopulmonary bypass. Ann Thorac Surg 1995; 60:801-804. 87. Kappelmeyer J, Bernabel A, Gikakis N et al. Upregulation of Mac-1 surface expression on neutrophils during simulated extracorporeal circulation. J Lab Clin Med 1993; 121:118-126. 88. Videm V, Nilsson L, Venge P et al. Reduced granulocyte activation with a heparin-coated device in an in vitro model of cardiopulmonary bypass. Artif Organs 1991; 15:90-95.

92

The Artificial Lung

89. Marx G, Pokar H, Reuter H et al. The effects of aprotinin on hemostatic function during cardiac surgery. J Cardiothorac Vasc Anesth 1991; 5:467-474. 90. Venge P, Nilsson L, Nystrom SO et al. Serum and plasma measurements of neutrophil granule proteins during cardiopulmonary bypass: A model to estimate human turnover of lactoferrin and myeloperoxidase. Eur J Hematol 1987; 39:339-345. 91. Nilsson L, Brunnkvist S, Nilsson U et al. Activation of inflammatory systems during cardiopulmonary bypass. Scand J Thor Cardiovasc Surg 1988; 22:51-53. 92. Riegel W, Spillner G, Schlosser V et al. Plasma levels of main granulocyte components during cardiopulmonary bypass. J Thorac Cardiovasc Surg 1988; 95:1014-1019. 93. Faymonville ME, Pincemail J, Duchateau J et al. Myeloperoxidase and elastase as markers of leukocyte activation during cardiopulmonary bypass in humans. J Thorac Cardiovasc Surg 1991; 102:309-317. 94. Borowiec J, Thelin S, Bagge L et al. Heparin-coated circuits reduce activation of granulocytes during cardiopulmonary bypass. J Thorac Cardiovasc Surg 1992; 104:642-647. 95. Ovrum E, Mollnes TE, Fosse E et al. Complement and granulocyte activation in two different types of heparinized extracorporeal circuits. J Thorac Cardiovasc Surg 1995; 110:1623-1632. 96. Ovrum E, Mollnes TE, Fosse E et al. High and low heparin dose with heparin-coated cardiopulmonary bypass: Activation of complement and granulocytes. Ann Thorac Surg 1995; 60:1755-1761. 97. Pesonen EJ, Korpela R, Peltola K et al. Cardiopulmonary bypass, myocardial management, and support techniques. J Thorac Cardiovasc Surg 1995; 110:768-773. 98. Moen O, Fosse E, Dregelid E et al. Centrifugal pump and heparin coating improves cardiopulmonary bypass biocompatibility. Ann Thorac Surg 1996; 62:1134-1140. 99. Cavarocchi NC, England MD, O’Brien JF et al. Superoxide generation during cardiopulmonary bypass: Is there a role for vitamin E? J Surg Res 1986; 40:519-527. 100. England MD, Cavarocchi NC, O’Brien JF et al. Influence of antioxidants (mannitol and allopurinol) on oxygen free radical generation during and after cardiopulmonary bypass. Circulation 1986; 74(Suppl III):III134-III137. 101. Prasad K, Kalra J, Bharadwaj B et al. Increased oxygen free radical activity in patients on cardiopulmonary bypass undergoing aortocoronary bypass surgery. Am Heart J 1992; 123:37-45. 102. Roccatello D, Formica M, Cavalli G et al. Changes in neutrophil oxidative potential in patients undergoing cardiopulmonary bypass with polypropylene hollow fiber oxygenators. Artif Organs 1990; 14:69-72. 103. Inal M, Alatas O, Kural T et al. Oxygen free radicals in erythrocytes during open heart operation. J Cardiovasc Surg 1994; 35:147-150. 104. Barsacchi R, Pelosi G, Maffei S et al. Myocardial vitamin E is consumed during cardiopulmonary bypass: indirect evidence of free radical generation in human ischemic heart. Intl J Cardiol 1992; 37:339-343. 105. Stryer L. Biochemistry. New York: WH Freeman and Company, 1988. 106. Valen G, Kaszaki J, Nagy S et al. Open heart surgery increases the levels of histamine in arterial and coronary sinus blood. Agents Actions 1994; 41:11-16. 107. Withington DE, Man WK, Elliot MJ. Histamine release during paediatric cardiopulmonary bypass. Can J Anaesth 1993; 40:334-339. 108. Ruvolo G, Speziale G, Greco E et al. Nitric oxide release during hypothermic versus normothermic cardiopulmonary bypass. Eur J Cardiothorac Surg 1995; 9:651-654. 109. Rinder CS, Bonan JL, Rinder HM et al. Cardiopulmonary bypass induces leukocyte-platelet adhesion. Blood 1992; 79:1202-1205. 110. Gillinov AM, Redmond JM, Zehr KJ et al. Inhibition of neutrophil adhesion during cardiopulmonary bypass. Ann Thorac Surg 1994; 57:126-133. 111. Fortenberry JD, Bhardwaj V, Niemer P et al. Neutrophil and cytokine activation with neonatal extracorporeal membrane oxygenation. J Pediatr 1996; 128:670-678. 112. Brown CH III, Lemuth RF, Hellums JD et al. Response of human platelets to shear stress. Trans Am Soc Artif Intern Organs 1975; 21:35-38. 113. Brown CH III, Leverett LB, Lewis CW et al. Morphological, biochemical, and fuctional changes in human platelets subjected to shear stress. J Lab Clin Med 1975; 86:462-471. 114. Colantuoni G, Hellums JD, Moake JL et al. The response of human platelets to shear stress at short exposure times. Trans Am Soc Artif Intern Organs 1977; 23:626-630. 115. Anderson GH, Hellums JD, Moake J et al. Platelet response to shear stress: Changes in serotonin uptake, serotonin release, and ADP induced aggregation. Thromb Res 1978; 13:1039-1047.

Biocompatibility of Artificial Lungs

93

116. Ramstack JM, Zuckerman L, Mockros LF. Shear-induced activation of platelets. J Biomech 1979; 12:113-125. 117. Peterson DM, Stathopoulos NA, Giorgio TD et al. Shear-induced platelet aggregation requires von Willebrand Factor and platelet membrane glycoprotein Ib and IIb-IIIa. Blood 1987; 69:625-628. 118. Chow TW, Hellums JD, Moake JL et al. Shear stress-induced von Willebrand factor binding to platelet glycoprotein Ib initiates calcium influx associated with aggregation. Blood 1992; 80:113-120. 119. Ikeda Y, Handa M, Kamata T et al. The role of von Willebrand factor and fibrinogen in platelet aggregation under varying shear stress. J Clin Invest 1991; 87:1234-1240. 120. Ikeda Y, Handa M, Kawano K et al. Transmembrane calcium influx associated with von Willebrand factor binding to GP Ib in the initiation of shear-induced platelet aggregation. Thromb Haemost 1993; 69:496-502. 121. Wurzinger LJ. Histology of the circulating platelet. Adv Anat Embryol Cell Biol 1990; 120:1-96. 122. Lawrence MB, McIntire LV, Eskin SG. Effect of flow on polymorphonuclear leukocyte/endothelial cell adhesion. Blood 1987; 70:1284-1290. 123. Kao WJ, Sapatnekar S, Hiltner A et al. Complement-mediated leukocyte adhesion on poly(etherurethane ureas) under shear stress in vitro. J Biomed Mater Res 1996; 32:99-109. 124. Kuijper PHM, Gallardo Torres HI, van der Linden JAM et al. Platelet-dependent primary hemostasis promotes selectin- and integrin-mediated neutrophil adhesion to damaged endothelium under flow conditions. Blood 1996; 87:3271-3281. 125. Woodman RC, Harker LA. Bleeding complications associated with cardiopulmonary bypass. Blood 1990; 76:1680-1697. 126. Zwischenberger JB, Nguyen TT, Upp JR et al. Complications of neonatal extracorporeal membrane oxygenation. J Thorac Cardiovasc Surg 1994; 107:838-849. 127. Saatvedt K, Lindberg H, Michelson S et al. Activation of the fibrinolytic, coagulation and plasma kallikrein-kinin systems during and after open heart surgery in children. Scand J Clin Lab Invest 1995; 55:359-367. 128. Brody JI, Pickering NJ, Fink GB. Concentrations of factor VIII-related antigen and factor XIII during open heart surgery. Transfusion 1986; 26:478-480. 129. Weinstein M, Ware JA, Troll J et al. Changes in von Willebrand factor during cardiac surgery: effect of desmopressin acetate. Blood 1988; 71:1648-1655. 130. Burrows FA, Steele RW, Marmer DJ et al. Influence of operations with cardiopulmonary bypass on polymorphonuclear leukocyte function in infants. J Thorac Cardiovasc Surg 1987; 93:253-260. 131. Ryhanen P, Huttunen K, Ilonen J. Natural killer cell activity after open heart surgery. Acta Anaesthesiol Scand 1984; 28:490-492. 132. Nguyen DM, Mulder DS, Shennib H. Effect of cardiopulmonary bypass on circulating lymphocyte function. Ann Thorac Surg 1992; 53:611-616. 133. Markewitz A, Faist E, Lang S et al. Successful restoration of cell-mediated immune response after cardiopulmonary bypass by immunomodulation. J Thorac Cardiovasc Surg 1993; 105:15-24. 134. Pollock R, Ames F, Rubio P et al. Protracted severe immune deregulation induced by cardiopulmonary bypass: a predisposing etiologic factor in blood transfusion-related AIDS? J Clin Lab Immunol 1987; 22:1-5. 135. Brody JI, Pickering NJ, Fink GB et al. Altered lymphocyte subsets during cardiopulmonary bypass. Am J Clin Pathol 1987; 87:626-628. 136. Zach TL, Steinhorn RH, Georgieff MK et al. Leukopenia associated with extracorporeal membrane oxygenation in newborn infants. J Pediatr 1990; 116:440-444. 137. Ulicny KS Jr, Hiratzka LF. The risk factors of median sternotomy infection: A current review. J Card Surg 1991; 6:338-351. 138. Loop FD, Lytle BW, Cosgrove DM et al. Sternal wound complications after isolated coronary artery bypass grafting: Early and late mortality, morbidity and cost of care. Ann Thorac Surg 1990; 49:179-187. 139. DeMichele MA, Minnear FL. Modulation of vascular endothelial permeability by thrombin. Sem Thromb Hemost 1992; 18:287-295. 140. Seghaye MC, Duchateau J, Grabitz RG et al. Inflammation reaction and capillary leak syndrome related to cardiopulmonary bypass in neonates undergoing cardiac operations. J Thorac Cardiovasc Surg 1996; 112:687-697. 141. Smedly LA, Tonnesen MG, Sandhaus RA et al. Neutrophil-mediated injury to endothelial cells. Enhancement by endotoxin and essential role of neutrophil elastase. J Clin Invest 1986; 77:1233-1243.

94

The Artificial Lung

142. Harlan JM, Killen PD, Harker LA et al. Neutrophil-mediated endothelial injury in vitro mechanisms of cell detachment. J Clin Invest 1981; 68:1394-1403. 143. Smith EEJ, Naftel DC, Blackstone EH et al. Microvascular permeability after cardiopulmonary bypass. J Thorac Cardiovasc Surg 1987; 94:225-233. 144. Foreman KE, Vaporciyan AA, Bonish BK et al. C5a-induced expression of P-selectin in endothelial cells. J Clin Invest 1994; 94:1147-1155. 145. Hattori R, Hamilton KK, McEver RP et al. Complement proteins C5b-9 induce secretion of high molecular weight multimers of endothelial von Willebrand factor and translocation of granule membrane protein GMP-140 to the cell surface. J Biol Chem 1989; 264:9053-9060. 146. Pober JS, Cotran RS. Cytokines and endothelial cell biology. Physiol Rev 1990; 70:427-451. 147. Shasby DM, Shasby SS, Peach MJ et al. Granulocytes and phorbol myristate acetate increase permeability to albumin of cultured endothelial monolayers and isolated perfused lungs. Role of oxygen radicals and granulocyte adherence. Am Rev Resp Dis 1983; 127:72-76. 148. Dreyer WJ, Michael LH, West MS et al. Neutrophil accumulation in ischemic canine myocardium. Insights into time course, distribution, and mechanism of localization during early reperfusion. Circulation 1991; 84:400-411. 149. Royston D, Fleming JS, Desai JB et al. Increased production of peroxidation products associated with cardiac operations: Evidence for free radical generation. J Thorac Cardiovasc Surg 1986; 91:759-766. 150. Braude S, Nolop KB, Fleming JS et al. Increased pulmonary transvascular protein flux after canine cardiopulmonary bypass: Association with lung neutrophil sequestration and tissue peroxidation. Am Rev Respir Dis 1986; 134:867-872. 151. Martin BA, Wright JL, Thommasen H et al. Effect of pulmonary blood flow on the exchange between the circulating and marginating pool of polymorphonuclear leukocytes in dog lungs. J Clin Invest 1982; 69:1277-1285. 152. Macnaughton PD, Braude S, Hunter DN et al. Changes in lung function and pulmonary capillary permeability after cardiopulmonary bypass. Crit Care Med 1992; 20:1289-1294. 153. Raijmakers PG, Groeneveld AB, Schneider AJ et al. Transvascular transport of 67Ga in the lungs after cardiopulmonary bypass surgery. Chest 1993; 104:1825-1832. 154. Sinclair DG, Haslam PL, Quinlan GJ et al. The effect of cardiopulmonary bypass on intestinal and pulmonary endothelial permeability Chest 1995; 108:718-724. 155. Messent M, Sinclair DG, Quinlan GJ et al. Pulmonary vascular permeability after cardiopulmonary bypass and its relationship to oxidative stress. Crit Care Med 1997; 25:425-429. 156. Komai H, Haworth SG. The effect of cardiopulmonary bypass on the lung. In: Jonas RA, Elliot MJ, eds. Cardiopulmonary bypass in Neonates, Infants, and Young Children. Boston: Butterworth-Heinemann, 1994:224-241. 157. Anywanwu E, Gieseking R, Enders HJ. Ultrastructural changes in the human lung following cardiopulmonary bypass. Bas Res Cardiol 1982; 77:309-322. 158. McGowan FX, Ikegami M, del Nido PJ et al. Cardiopulmonary bypass significantly reduces surfactant activity in children. J Thorac Cardiovasc Surg 1993; 106:968-977. 159. Hennein HA, Ebba H, Rodriguez JL et al. Relationship of the proinflammatory cytokines to myocardial ischemia and dysfunction after uncomplicated coronary revascularization. J Thorac Cardiovasc Surg 1994; 108:626-635. 160. del Balzo UH, Levi R, Polley MJ. Cardiac dysfunction caused by purified human C3a anaphylatoxin. Proc Natl Acad Sci USA 1985; 82:886-890. 161. Seghaye MC, Grabitz RG, Duchateau J et al. Histamine liberation related to cardiopulmonary bypass in children: Possible relation to transient postoperative arrhythmias. J Thorac Cardiovasc Surg 1996; 111:971-981. 162. Finkel MS, Oddis CV, Jacob TD et al. Negative inotropic effects of cytokines on the heart mediated by nitric oxide. Science 1992; 257:387-389. 163. Wilson I, Gillinov AM, Curtis WE et al. Inhibition of neutrophil adherence improves postischemic ventricular performance of the neonatal heart. Circulation 1993; 88:II372-II379. 164. Entman ML, Youker K, Shoji T et al. Neutrophil induced oxidative injury of cardiac myocytes. A compartmented system requiring CD11b/CD18-ICAM-1 adherence. J Clin Invest 1992; 90:1335-1345. 165. Downing SW, Edmunds LH. Release of vasoactive substances during cardiopulmonary bypass. Ann Thorac Surg 1992; 54:1236-1243. 166. Putnam EA, Manners JM. Vascular resistance during cardiopulmonary bypass. Anaesthesia 1983; 38:635-643.

Biocompatibility of Artificial Lungs

95

167. Kam PCA, Hines L, O’Connor E. Effects of cardiopulmonary bypass on systemic vascular resistance. Perfusion 1996; 11:346-350. 168. Mockros LF, Leonard R. Compact cross-flow tubular oxygenators. Trans Am Soc Artif Intern Organs 1985; 31:628-632. 169. Ohri SK, Bjarnason I, Pathi V et al. Cardiopulmonary bypass impairs small intestinal transport and increases gut permeability. Ann Thorac Surg 1993; 55:1080-1086. 170. Ohri SK. Systemic inflammatory response and the splanchnic bed in cardiopulmonary bypass. Perfusion 1996; 11:200-212. 171. Jodal M, Lundgren O. Plasma skimming in the intestinal tract. Acta Physiol Scand 1970; 80:50-60. 172. Baue AE. The role of the gut in the development of multiple organ dysfunction in cardiothoracic patients. Ann Thorac Surg 1993; 55:822-829. 173. Jorres A, Kordonouri O, Schiessler A et al. Urinary excretion of thromboxane and markers for renal injury in patients undergoing cardiopulmonary bypass. Artif Organs 1994; 18:565-569. 174. Mazzarella V, Galluci MT, Tozzo C et al. Renal function in patients undergoing cardiopulmonary bypass operations. J Thorac Cardiovasc Surg 1992; 104:1625-1627. 175. Gomez-Campdera FJ, Maroto-Alvaro E, Galiñanes M et al. Acute renal failure associated with cardiac surgery. Child Nephrol Urol 1988-89; 9:138-143. 176. Picca S, Principato F, Mazzera E et al. Risks of acute renal failure after cardiopulmonary bypass surgery in children: A retrospective 10-year case-control study. Nephrol Dial Transplant 1995; 10:630-636. 177. Gailiunas P Jr, Chawla R, Lazarus JM et al. Acute renal failure following cardiac operations. J Thorac Cardiovasc Surg 1980; 79:241-243. 178. Joachimsson PO, Stahle E, Nystrom SO et al. Incidence of acute renal failure in open heart surgery. J Cardiothorac Anesth 1989; 3(5 Suppl 1):58. 179. McLean RF, Wong BI, Naylor CD et al. Cardiopulmonary bypass, temperature, and central nervous system dysfunction. Circulation 1994; 90:II250-II255. 180. Shaw DA. Early intellectual dysfunction following coronary bypass surgery. Quart J Med 1986; 58:59-68. 181. Venn G, Klinger L, Smith P. Neuropsychological sequelae of bypass twelve months after coronary artery surgery. Br Heart J 1987; 57:565. 182. Harris DNF, Bailey SM, Smith PLC et al. Brain swelling in the first hour after coronary artery bypass surgery. Lancet 1993; 342:586-587. 183. Salzman EW, Rosenberg RD, Smith MH et al. Effect of heparin and heparin fractions on platelet aggregation. J Clin Invest 1980; 65:64-73. 184. Verstraete M. New developments in antiplatelet and antithrombotic therapy. Eur Heart J 1995; 16(Suppl L):16-23. 185. Tabardel Y, Duchateau J, Schmartz D et al. Corticosteroids increase blood interleukin-10 levels during cardiopulmonary bypass in men. Surgery 1996; 119:76-80. 186. Bidstrup BP, Royston D, Sapsford RN et al. Reduction in blood loss and blood use after cardiopulmonary bypass with high dose aprotinin (Trasylol). J Thorac Cardiovasc Surg 1989; 97:364-372. 187. Fuhrer G, Gallimore MJ, Heller W et al. Aprotinin in cardiopulmonary bypass-effects on the Hageman factor (FXII)-kallikrein system and blood loss. Blood Coagul Fibrinolysis 1992; 3:99-104. 188. Baufreton C, Velthuis HT, Jansen PGM et al. Reduction of blood activation in patients receiving aprotinin during cardiopulmonary bypass for coronary artery surgery. ASAIO J 1996; 42:M417-M423. 189. Fujii S, Hitomi Y. New synthetic inhibitors of C1r, C1 esterase, thrombin, plasmin, kallikrein, and trypsin. Biochim Biophys Acta 1981; 661:342-345. 190. Hitomi Y, Fujii S. Inhibition of various immunological reactions in vivo by a new synthetic complement inhibitor. Int Arch Allergy Appl Immunol 1982; 69:262-267. 191. Uchiba M, Okajima K, Abe H et al. Effect of nafamostat mesilate, a synthetic protease inhibitor, on tissue factor-factor VIIa complex activity. Thromb Res 1994; 74:155-61. 192. Sundaram S, Gikakis N, Hack CE et al. Nafamostat mesilate, a broad spectrum protease inhibitor, modulates platelet, neutrophil and contact activation in simulated extracorporeal circulation. Thromb Haemost 1996; 75:76-82. 193. Murase M, Usui A, Tomita Y et al. Nafamostat mesilate reduces blood loss during open heart surgery. Circulation 1993; 88:II432-II436.

96

The Artificial Lung

194. Tschesche H, Beckman J, Mehlich A et al. Semisynthetic arginine-15-aprotinin, an improved inhibitor for human plasma kallikrein. Adv Exp Med Biol 1989; 247B:15-21. 195. Hitomi Y, Ikari N, Fujii S. Inhibitory effect of a new synthetic protease inhibitor (FUT-175) on the coagulation system. Haemostasis 1985; 15:164-168. 196. DeLa Cadena RA, Stadnicki A, Uknis AB et al. Inhibition of plasma kallikrein prevents petidoglycan-induced arthritis in the Lewis rat. FASEB J 1995; 9:446-452. 197. Ulmer JS, Lindquist RN, Dennis MS et al. Ecotin is a potent inhibitor of the contact system proteases factor XIIa and plasma kallikrein. FEBS Letters 1995; 365:159-163. 198. Seymour JL, Lindquist RN, Dennis MS et al. Ecotin is a potent anticoagulant and reversible tight-binding inhibitor of Factor Xa. Biochemistry 1994; 33:3949-3958. 199. von Segesser L, Turina M. Cardiopulmonary bypass without systemic heparinization for 24 hours. Proc AmSECT Annual Meeting 1989:76-82. 200. Larm O, Larsson R, Olsson R. A new nonthrombogenic surface prepared by selective covalent binding of heparin via a modified reducing terminal residue. Biomater Med Devices Artif Organs 1983; 11:161-173. 201. Hsu LC. Principles of heparin-coating techniques. Perfusion 1991; 6:209-219. 202. Gu YJ, van Oeveren W, van der Kamp KWHJ et al. Heparin-coating of extracorporeal circuits reduces thrombin formation in patients undergoing cardiopulmonary bypass. Perfusion 1991; 6:221-225. 203. Nakajima T. Osawa S. Ogawa M et al. Clinical study of platelet function and coagulation/fibrinolysis with Duraflo II heparin coated cardiopulmonary bypass equipment. ASAIO J 1996; 42:301-305. 204. Pradhan MJ, Fleming JS, Nkere UU et al. Clinical experience with heparin-coated cardiopulmonary bypass circuits. Perfusion 1991; 6:235-242. 205. Yamanaka J, Takeuchi Y, Torii S et al. The evaluation of the bio-compatibility and the clinical usefulness of heparin-coated cardiopulmonary bypass circuits. Nippon Kyobu Geka Gakkai Zasshi 1996; 44:47-53. 206. Hsu LC. Biocompatibility in heparin-coated extracorporeal circuits. Perfusion 1996; 11:256-263. 207. Gravlee GP, Phipps J, Mills SA et al. Hematologic evaluation of a heparin-coated circuit for cardiopulmonary bypass. Anesthesiology 1992; 77:A99. 208. von Segesser LK, Weiss BM, Garcia E et al. Reduced blood loss and transfusion requirements with low systemic heparinization: Preliminary clinical results in coronary artery revascularization. Eur J Cardiothorac Surg 1990; 4:639-643. 209. von Segesser LK, Weiss BM, Garcia E et al. Reduction and elimination of systemic heparinization during cardiopulmonary bypass. J Thorac Cardiovasc Surg 1992; 103:790-799. 210. Ladowski JS, Schatzlein MH, Peterson AC et al. Clinical heparin coated cardiopulmonary bypass: Reduction of systemic heparin requirements for redo cardiac surgery. ASAIO J 1996; 42:34-36. 211. Moen O, Fosse E, Brockmeier V et al. Disparity in blood activation by two different heparin-coated cardiopulmonary bypass systems. Ann Thorac Surg 1995; 60:1317-1323. 212. Lundblad R, Moen O, Fosse E. Endothelin-1 and neutrophil activation during heparin-coated cardiopulmonary bypass. Ann Thorac Surg 1997; 63:1361-1367. 213. Wendel HP, Heller W, Gallimore MJ et al. Heparin-coated oxygenators significantly reduce contact system activation in an in vitro cardiopulmonary bypass model. Blood Coagul Fibrinolysis 1993; 5:673-678. 214. Borowiec JW, Bylock A, van der Linden J et al. Heparin coating reduces blood cell adhesion to arterial filters during coronary bypass: A clinical study. Ann Thorac Surg 1993; 55:1540-1545. 215. Gorman RC, Ziats N, Rao AK et al. Surface-bound heparin fails to reduce thrombin formation during clinical cardiopulmonary bypass. J Thorac Cardiovasc Surg. 1996; 111:1-12. 216. Stenach N, Korn RL, Fisher CA et al. The effects of heparin bound surface modification (Carmeda Bioactive Surface) on human platelet alterations during simulated extracorporeal circulation. J Am Soc Extracorporeal Technol 1992; 24:97-102. 217. Moen O, Fosse E, Braten J et al. Differences in blood activation related to roller/centrifugal pumps and heparin-coated/uncoated surfaces in a cardiopulmonary bypass model circuit. Perfusion 1996; 11:113-23. 218. Korn RL, Fisher CA, Livingston ER et al. The effects of Carmeda Bioactive Surface on human blood components during simulated extracorporeal circulation. J Thorac Cardiovasc Surg 1996; 111:1073-1084.

Biocompatibility of Artificial Lungs

97

219. Fukutomi M, Kobayashi S, Niwaya K et al. Changes in platelet, granulocyte, and complement activation during cardiopulmonary bypass using heparin-coated equipment. Artif Organs. 1996; 20:767-776. 220. Nojiri C, Hagiwara K, Yokoyama K et al. Evaluation of a new heparin bonding process in prolonged extracorporeal membrane oxygenation. ASAIO J 1995; 41:M561-M567. 221. Saenz A, Larranaga G, Alvarez L et al. Heparin-coated circuit in coronary surgery. A clinical study. Eur J Cardiothorac Surg 1996; 10:48-53. 222. Hogevold HE, Moen O, Fosse E et al. Effects of heparin coating on the expression of CD11b, CD11c and CD62L by leucocytes in extracorporeal circulation in vitro. Perfusion 1997; 12:9-20. 223. Fosse E, Moen O, Johnson E et al. Reduced complement and granulocyte activation with heparin-coated cardiopulmonary bypass. Ann Thorac Surg 1994; 58:472-477. 224. Ovrum E, Fosse E, Mollnes TE et al. Complete heparin-coated cardiopulmonary bypass and low heparin does reduce complement and granulocyte activation. Eur J Cardiothorac Surg 1996; 10:54-60. 225. Mackor EL. A theoretical approach to the colloid-chemical stability of dispersions in hydrocarbons. J Colloid Sci 1951; 6:492-495. 226. Mackor EL, van der Waals JH. The statistics of the adsorption of rod-shaped molecules in connection with the stability of certain colloidal despersions. J Colloid Sci 1952; 7:535-550. 227. Meier DJ. Theory of polymeric dispersants. Statistics of constrained polymer chains. J Phys Chem 1967; 71:1861-1868. 228. Harris JM. Introduction to biotechnical and biomedical applications of poly(ethylene glycol). In: Harris JM, ed. Poly(ethylene glycol) Chemistry. New York: Plenum Press, 1992:1-14. 229. Hesselink FT, Vrij A, Overbeek JTG. On the theory of the stabilization of dispersions by adsorbed macromolecules. II. Interaction between two flat particles. J Phys Chem 1971; 75:2094-2103. 230. Desai NP, Hubbel JA. Biological responses to polyethylene oxide modified polyethylene terephthalate surfaces. J Biomed Mater Res 1991; 25:829-843. 231. Gombotz, WR, Guanghui W, Horbett TA et al. Protein adsorption to poly(ethylene oxide) surfaces. J Biomed Mater Res 1991; 25:1547-1562. 232. Brinkman E, Poot A, van der Does L et al. Platelet deposition studies on copolyether urethanes modified with poly(ethylene oxide). Biomaterials 1990; 11:200-205. 233. Dunkirk SG, Gregg SL, Duran LW et al. Photochemical coatings for the prevention of bacterial colonization. J Biomater Appl 1991; 3:131-156. 234. Nagaoka S, Nakao A. Clinical application of antithrombogenic hydrogel with long poly(ethylene oxide) chains. Biomaterials 1990; 11:119-121. 235. Lee JH. Kopeckova P. Kopecek J et al. Surface properties of copolymers of alkyl methacrylates with methoxy (polyethylene oxide) methacrylates and their application as protein-resistant coatings. Biomaterials 1990; 11:455-464. 236. Amiji M. Park K. Prevention of protein adsorption and platelet adhesion on surfaces by PEO/ PPO/PEO triblock copolymers. Biomaterials 1992; 13:682-692. 237. Desai NP, Hubbel JA. Solution technique to incorporate polyethylene oxide and other water soluble polymers into surfaces of polymeric biomaterials. Biomaterials 1991; 12:144-153. 238. Drumheller PD, Hubbell JA. Densely crosslinked polymer networks of poly(ethylene glycol) in trimethylolpropane triacrylate for cell-adhesion surfaces. J Biomed Mater Res 1995; 29:207-215. 239. Bearinger JP, Castner DG, Golledge SL et al. P(AAm-coEG) interpenetrating polymer networks grafted to oxide surfaces: surface characterization, protein adsorption, and cell detachment studies. Langmuir 1997; 13:5175-5183. 240. Park KD, Okano T, Nojiri C et al. Heparin immobilization onto segmented polyurethaneurea surfaces—Effect of hydrophilic spacers. J Biomed Mater Res 1988; 22:977-992. 241. Nojiri C, Park KD, Okano T et al. In vivo protein adsorption onto polymers: A transmission electron microscopic study. ASAIO Trans 1989; 35:357-361. 242. Takewa Y, Tatsumi E, Taenaka Y et al. Hemodynamic and humoral conditions in stepwise reduction of pulmonary blood flow during venoarterial bypass in awake goats. ASAIO J 1997; 43:M494-M499.

CHAPTER 5

Testing and Performance Evaluation of Artificial Lungs Akhil Bidani, Weike Tao and Joseph B. Zwischenberger

Introduction

T

he Association for the Advancement of Medical Instrumentation (AAMI) publishes a working draft pertaining to “blood-gas exchange” devices with periodic revisions.1 The AAMI defines a “blood-gas exchange” device as “an extracorporeal device designed to supplement, or be a substitute for, the respiratory function of the lung.” This group has also provided recommended procedures and guidelines to be used for the comprehensive evaluation and testing of these devices. Since 1982, the working drafts have been revised in response to changes in both ideology and technology. “Oxygenator” standards and testing guidelines are now the formal responsibility of the International Organization for Standardization (ISO). Current ISO (and previous AAMI) guidelines1 offer device manufacturers and clinicians a standard testing format by which they can evaluate a “blood-gas exchanger’s” biological, physical, and performance characteristics. Up until recently, the primary focus of these guidelines and standards were almost exclusively on the role of blood-gas exchange devices in cardiopulmonary bypass. Over the past 15 years, there has been resurgence of interest in the support of gas-exchange in the care of critically ill patients. In this setting “blood-gas exchanger” devices must provide partial or total support beyond several hours to days, therefore some of the criteria and considerations for the performance of artificial lung devices must include both intracorporeal and extracorporeal devices. Likewise, because of the pioneering work of Kolobow and Gattinoni and their colleagues2-7 on the importance of CO2 removal, exclusive emphasis on “oxygenation” by blood-gas exchange devices must be broadened to emphasize CO2 clearance properties of these devices. This chapter is divided into three main sections. In the first section, we present some basic aspects of artificial lung (AL) design and mass transfer characterization of gas transfer. In the remaining two sections we focus exclusively on gas-exchange aspects of an intracorporeal AL (IVOX) and extracorporeal AL. In our own work over the past seven years, we have used a combination of mathematical modeling, in vitro testing and in vivo evaluations of ALs in animal models of acute respiratory failure. Selected studies were also undertaken in patients with ARDS. In sections 2 and 3, we summarize key aspects of our methodology and results. This brief review does not discuss other aspects of the

The Artificial Lung, edited by Steven N. Vaslef and Robert W. Anderson. ©2002 Eurekah.com.

Testing and Performance Evaluation of Artificial lungs

99

comprehensive evaluation of artificial lung (AL) such as hemodynamics, blood-membrane interactions and other hematological and cellular alterations.

Gas Exchange in “Artificial Lungs” Versus Human Lung

The estimated surface area of the adult lung is of the order of 100 m2. Even the most optimally designed AL has surface area that is <5% of this number, leading to a very low overall diffusing capacity compared to the adult human lung (Table 5.1). Despite this obvious limitation, there are, however, some notable “design” differences between the native human lung and an AL that can help overcome this shortcoming:

Design Advantages of AL Over Human Lung 1. The adult lung is adapted for maximal exercise with an estimated O2 uptake of 2500 ml/ min and CO2 excretion of 2000 ml/min. Critically ill afebrile patients, while often hypermetabolic, are at rest with VO2< 400 ml/min and VCO2< 350 ml/min. 2. The adult lung is over-designed so that the diffusing capacity has to be reduced to less than 20% before arterial oxygenation and CO2 excretion are compromised, suggesting that a reduced AL surface area may be acceptable. 3. The normal driving partial pressure gradient is about 60 mm Hg for oxygen and only 6 mm Hg for carbon dioxide. Depending on the gas flow rate in an AL, the partial pressure gradient for CO2 can be > 30 mm Hg under normocapnic conditions and as high as 80 mm Hg under conditions of significant “permissive hypercapnia”. Thus the AL is better suited for CO2 excretion than O2 uptake. 4. The normal V/Q ratio in an adult lung is 1, whereas this can be significantly increased to > 2 in the AL. 5. The normal capillary transit time is of the order of 1 sec in the adult lung, which can be substantially increased in the AL, allowing greater transfer of oxygen and CO2 between gas and blood. 6. It is possible to use either countercurrent or cross-flow between gas and blood flow in an AL, which is inherently more efficient than the exchange of O2 and CO2 between pulmonary capillary blood and mixed alveolar gas.

Design Disadvantages of AL Relative to Human Lung 1. Pulmonary capillaries have a diameter approximating that of a red cell. Thus, the diffusion distance between blood and gas is very small, minimizing the boundary layer mass transfer resistance. On the other hand, a major limitation of gas transfer between blood and gas is the large boundary layer mass transfer resistance in blood that provides the major impediment to mass transfer between gas and blood.8 Measures to increase bulk mixing in blood are therefore crucial to improving the efficiency of gas exchange in AL devices.8,9 2. The vascular endothelium is “designed” to prevent activation and damage to blood elements, particularly platelets and neutrophils. Despite vast improvements in biomaterial and heparin-coated surfaces used in AL, there is an inherent problem of cell activation and thrombus formation in AL, necessitating the use of systemic anticoagulation, which is associated with substantial complications and morbidity. 3. Endothelium of the human lung has extensive nonrespiratory functions including biotransformation of many constituents including bio-amines that is not possible in the current endothelial-poor AL.

The Artificial Lung

100

Table 5.1. Comparison of human adult lung versus "Artificial lung" Property/Characteristics

Human Adult Lung

"Artificial lung"

Surface area (m2) Blood path width (∝m) Blood path length (∝m) Membrane thickness (∝m) Blood transit time (seconds) Maximum oxygen transfer (ml/min) Maximum CO2 transfer (ml/min) Alveolar-venous blood PO Gradient (mm Hg) 2 Venous blood-alveolar PCO 2 Gradient (mm Hg)

70 8 200 0.5 0.3-1.0 2000 2000 40-50 3-12

Blood-membrane interactions

-hemo-compatible -no anticoagulation -"self-cleaning"

0.5-4.0 ~200 >200,000 100-150 3-30 50-400 50-250 ~600 30-90 (depending on level of hypercapnia) -not hemo-compatible -anticoagulation -protein build-up

Mass Transfer in Artificial Lungs During blood flow in an artificial lung, O2 uptake into blood from gas and excretion of CO2 from blood to gas involves both convective and diffusive processes. For artificial lungs incorporating gas flow through hollow fibers, intraluminal gas phase concentration is determined by the balance between convective flow of gas through the fibers and the radial diffusion of gas either from the fiber to the inner surface of the fiber (as in the case of O2) or from the fiber surface to the lumen (as in the case of CO2). The concentration of gas at the outer wall of the fiber (in contact) with blood is determined by the resistance to mass transfer through the membrane itself. Finally, the concentration of gas in the blood is determined by the rate of blood flow and the flux of gas either from the outer wall of the fiber to blood (as in the case of O2) or from blood to the outer wall of the fiber (as in the case of CO2). A convenient measure of the resistance to mass transfer across any barrier or interface is the mass transfer coefficient, Ki , defined as the amount of gas “i” transfer per unit time, per unit surface area, per unit mm Hg partial pressure gradient across the barrier, such that:

m˙ i (Eqn. 5.1) , As ∀( Pi ) where mi is the gas transfer rate (ml gas/min), As is the surface area for transfer (m2), ∀(Pi) is the partial pressure gradient of gas “i” across the barrier, and Ki is the mass transfer coefficient (ml gas/(min•m2•mm Hg). Since the partial pressure gradient may differ between device inlet and devices, the log mean partial pressure difference is used.10,11 The average mass transfer coefficient is a convenient and useful method of comparing the efficiency of gas transfer in different AL under comparable conditions. As defined in Equation 5.1, Ki represents a mass transfer conductance of species “i” (inverse of mass transfer resistance). A high value of Ki indicates more efficient mass transfer. Since, in biological systems there are at least three mass transfer resistances (gas phase, membrane, and blood phase) in series, the overall mass transfer resistance is dominated by the highest resistance, which, in most ALs, is the blood. The average mass transfer coefficient can be Ki =

Testing and Performance Evaluation of Artificial Lungs

101

calculated, using Equation 5.1, from measured rates of gas transfer in an AL, if the log mean pressure difference is known. The prime barrier to oxygen in an AL is the boundary layer adjacent to the membrane. This boundary layer consists of a thin layer of well-oxygenated blood that is not well mixed with the bulk of flowing blood, and acts as a “thick membrane”. To minimize this rate limiting mass transfer resistance, Bellhouse proposed secondary flow patterns mechanically introduced into the main flow path. For carbon dioxide, because of a substantially greater solubility in plasma, the apparent boundary layer thickness is much reduced, but still represents the limiting mass transfer resistance.9 Under normocapnic conditions, in which 100% humidified oxygen is used as the inflow gas to an AL, the initial driving force for CO2 is approximately (46-0) = 46 mm Hg, and for oxygen (713-40) ) 670 mm Hg. Thus, the ratio of driving force for CO2 to O2 is 46/670 ) 0.07. Consider the hypothetical case in which the predominant resistance to mass transfer is that due to the membrane, and that the mass transfer resistance due to chemical reactions in blood and that due to the boundary layer in blood is negligible. Under these circumstances, a ten-fold higher membrane permeability for CO2 relative to O2, the flux ratio of CO2 to O2 would be ) 0.7. Under conditions of severe systemic hypercapnia (blood PCO2 = 80 mm Hg), the initial driving force for CO2 may be increased to approximately (80-0) = 80 mm Hg, and the corresponding flux ratio of CO2 relative to O2 would be )1.4, indicating the potential benefit of permissive hypercapnia for CO2 removal in an AL. These calculations are illustrative but inaccurate because of the limiting role of boundary layer mass resistance, which affects both O2 uptake and CO2 excretion in an AL.

Reaction and Transport Processes Involved in CO2 Transfer

Quantitation of the efficiency of carbon dioxide transfer has received less attention in the literature, presumably related to the greater complexity of carbon dioxide transfer. CO2 transport involves a series of interrelated reaction and transport processes12 (Fig. 5.1). These include the hydration-dehydration of CO2 within plasma and red blood cells (RBCs) and the reversible combination of carbon dioxide with the amino groups on oxygenated hemoglobin and de-oxygenated hemoglobin.12 Additionally, HCO3- in plasma/ RBC must undergo one-for-one exchange with RBC/plasma Cl- mediated by RBC membrane-associated Band 3 anion exchanger.12 Following the addition to or removal of CO2 from blood, the rate of reestablishment of CO2 - HCO!3 - H + equilibrium within RBCs and plasma is rate-limited by the availability of the enzyme carbonic anhydrase (CA) (known to catalyze CO2 hydration-dehydration reactions). The availability of vascular CA activity on pulmonary capillary endothelium to plasma13 results in near-equilibrium of CO2 hydration-dehydration reactions by the end capillary transit in the pulmonary circulation14 and most tissues (specifically cardiac, skeletal muscle, brain and renal). Thus, in the absence of significant inhibition of either CA activity or RBC anion exchanger (as by drugs), blood may be considered to be in a state of near-equilibrium. Artificial lung devices are always associated with some RBC hemolysis, which would result in RBC CA being spilled into plasma and CA activity becoming available to plasma, akin to the in vivo availability of CA activity available to plasma via pulmonary vascular-associated endothelial CA activity. Under these conditions, use of an empirical CO2 dissociation curve can be used to relate the total blood concentration of CO2 (in all forms), [CO2]T , to the blood partial pressure of CO2, (PCO2). Such an approach avoids complexity associated with modeling of individual reaction pathways involved in CO2 transfer.14

The Artificial Lung

102

Fig. 5.1. Schematic of reaction and transport processes associated with O2 and CO2 exchange that occur within blood during its passage through an “artificial lung.”

Measurement of Gas Transfer in Artificial Lung Devices In the blood phase, gas transfer can be determined by the application of the law of conservation of mass. This calculation assumes that the transfer of O2 or CO2 across the membrane lung is responsible for the difference in gas content between blood flowing into and out of the AL. Such a calculation assumes that there are no leaks in the system. Thus, the gas content of inflow and outflow blood needs to be determined by continuous in-line monitors or intermittent blood gas samples for measurements in vitro. Currently, in-line oximeters may be used to measure blood oxygen saturations. However, there is currently no technology available for continuous blood CO2 content monitoring. Instead, one must resort to inflow and outflow blood samples to measure total CO2 content. The difference in inflow and outflow concentration multiplied by the rate of blood flow provides an estimate of gas transfer: m˙ o = Qb ([O2 ]o ! [O2 ] i ),

(Eqn. 5.2)

m˙ co 2 = Qb ([CO2 ]i ! [CO2 ]o ),

(Eqn. 5.3)

2

where mO2 and mCO2 represent rates of oxygen and carbon dioxide transfer, respectively; [O2]o and [CO2]o represent blood O2 and CO2 contents of out-flowing blood; and [O2]i and [CO2]i represent blood O2 and CO2 contents of in-flowing blood.

Testing and Performance Evaluation of Artificial Lungs

103

Gas transfer can also be determined by measurements in the gas phase, perhaps more rapidly, in some cases, than blood phase measurements. Gas flowing into and flowing out of the AL can be sampled continuously using a mass spectrometer, and gas phase fractions may be estimated. The difference between inflow and outflow gas fractions multiplied by the volumetric gas flow rate can be used to calculate gas transfer rates: m˙ o = Q g ([ Fo 2 ]i ! [ Fo 2 ]o ),

(Eqn. 5.4)

m˙ co 2 = Q g [ Fco 2 ]o ),

(Eqn. 5.5)

2

where [FO2]o and [FCO2]o represent fractional concentrations of O2 and CO2 in out-flowing gas and [FO2]i represents fractional concentration of O2 of the in-flowing gas. Calculations made using Equations 5.4 and 5.5 can be somewhat inaccurate. Since the rate of oxygen uptake is usually not equal to the rate of CO2 removal, the volumetric flow rate of gas entering the AL is not the same as the volumetric flow rate of gas leaving the AL. To prevent this error in calculations, an inert and insoluble gas, such as helium, may be added (5-10%) to the inflow gas. Helium in the gas phase rapidly equilibrates with helium that becomes physically dissolved in blood, so that after a very short time, the net transmembrane flux of helium across the gas-blood interface is zero. Since there is no net uptake of helium in the AL, the rate of helium inflow can be equated to that in the outflow gas, so that:

9 ( FHe ) i < (Q g )o = (Q g ) i : =, ;( FHe )o >

(Eqn. 5.6)

where (Qg)o and (Qg)i are the volumetric rates of gas outflow and inflow, respectively; and (FHe)i and (FHe)o are the fractional concentrations of helium in the inflow and outflow gas. The rate of oxygen transfer can be calculated by combining Equations 5.4 and 5.6 to account for the difference in the inlet and outlet gas flow rates: ? 9 ( FHe ) i <Β m˙ o = (Q g ) i Α( Fo 2 ) i ! ( Fo 2 )o : = ∆, 2 Α≅ ;( FHe ) o >∆Χ

(Eqn. 5.7)

Alternative Methods for Quantifying Gas Exchange Performance Many previously published clinical evaluations have reported on the use of a particular “oxygenator” for a specified number of procedures.15,16 Typically, such studies report performance parameters such as mean values and ranges for arterial oxygen tension (PaO2), arterial carbon dioxide tension (PaCO2), fractional inspired oxygen concentration (FIO2), gas flow, blood flow, hemoglobin concentration, arterial oxygen saturation (SaO2), mixed venous oxygen saturation (SvO2), and oxygen transfer.

Oxygen Transfer Slope

Fried et al.17 have developed and supported the use of the oxygen transfer slope (OTS) to clinically evaluate membrane “oxygenators”. The OTS is derived from a plot of the change in FI O2 required for a change in oxygen transfer, m O2, (OTS = ∀ FI O 2/ ∀ mΕ2 ), to achieve a specific range of resultant PaO2 values. A higher oxygen transfer slope

The Artificial Lung

104

is associated with low oxygen transfer capability, while a lower OTS is associated with higher levels of oxygen transfer. Such an approach has not been considered for CO2 exchange.

Device Diffusing Capacity Similar to the concept of diffusion capacity in the lung, one may define an overall device diffusion capacity (Dcap) as the product of the mean transfer coefficient, Ki (Equation 5.1), and the overall surface area for exchange: (Eqn. 5.8)

Dcap = K i As ,

The potential advantage of such an approach is its comparison of the overall efficiency of gas transfer of the device versus the native lung.18

Percentage Clearance of CO2

The percent of total blood CO2 content arriving at the device that is excreted can be expressed as:

m˙ co 2

(Eqn. 5.9) ⋅ 100%, Qb [CO2 ]i where mCO2 is CO2 transfer (ml/min), Qb is blood flow or cardiac output (l/min), [CO2]i is CO2 content (vol%) of blood arriving at the device. By definition, (%cl) is dimensionless, and expressed as a percentage.14,19

(% cl ) co 2 =

Transfer Capacity for CO2 Excretion

Of the total CO2 excreted in the device, mCO2, a significant fraction (as much as 40-45%) is due to the release of Bohr protons during oxygenation of hemoglobin (Fig. 5.1). Thus, overall CO2 excretion can be partitioned into O2-linked CO2 transfer and O2-independent CO2 excretion:

m˙ co 2 = (m˙ co 2 )o 2 ! linked + (m˙ co 2 )o 2 ! indep .

(Eqn. 5.10)

During systemic hypercapnia, the inlet blood-to-gas phase PCO2 gradient, (∀PCO2)i-g, increases, but the O2-linked component of CO2 excretion remains nearly constant. The increment in total CO2 excretion, mCO2, is due to an increase in (∀PCO2)i-g which results from O2-independent sources of CO2 generation, and is dependent on the rates of RBC anion exchange and RBC and lung CA activity. In order to examine the influence of changes in (∀PCO2)i-g on mCO2, one can define a “transfer capacity for CO2 excretion,” TCO2, as: ∀m˙ co 2

(Eqn. 5.11) ∀( ∀Pco 2 ) i ! g . TCO2 represents incremental changes in CO2 elimination by the device with increments in the partial pressure gradient between inlet blood and the gas phase, (∀PCO2)i-g. The transfer capacity, TCO2, has the units (ml CO2)/(min•mm Hg) akin to the widely used diffusing capacity of O2, DLO2.14, 19 T co 2 =

Testing and Performance Evaluation of Artificial Lungs

105

Table 5.2. Device properties and performance of IVOX Size (mm) a

Surface area (m2)

No. of fibers

O2 Transfer b

CO2 Removal b

7 8 9 10

0.21 0.32 0.41 0.52

589 703 894 1107

40.3 45.6 54.2 72.5

43.8 60.2 60.1 71.0

a Size in diameter at the site of potting of fibers. The overall size of IVOX is slightly greater. b

Units in ml/min. Data based on summarized clinical trials.26

Transfer Coefficient for CO2 Excretion

In order to normalize the transfer capacity, TCO2, for changes in blood flow, one can define the “transfer coefficient for CO2 excretion” as:

KΦ

T co

2 (Eqn. 5.12) Qb where Qb represents blood flow through the device, and KΦCO2has units of (ml CO2)/ (l blood•mm Hg) and represents a dynamic capacitance coefficient (or an overall mass transfer coefficient).14,19

=

co 2

Intracorporeal Gas Exchange A new concept in respiratory support was the development of an intravascular gas exchange device (IVOX) by Mortensen with CardioPulmonics (Salt Lake City, UT).20, 21 IVOX was a miniature membrane lung that consisted of multiple long, crimped hollow fibers placed within the vena cavae to provide CO2 removal and blood oxygenation without the need for extracorporeal circulation or blood transfusion. The hollow fibers were joined together in a potted manifold that communicated with the dual-lumen gas conduit at both its proximal and distal ends. The fibers were crimped to produce turbulent blood flow in the vena cava, thus increasing blood-membrane contact time and blood mixing. The fibers were spiralled, allowing the device to be compressed by furling while being inserted into the vena cava through right femoral or jugular venotomy and unfurled after insertion to perform gas exchange. The membrane was coated with thrombus-resistant silicone to which heparin was bonded covalently. IVOX devices were manufactured in sizes 7, 8, 9 and 10 (mm in transverse diameters) (Table 5.2). Our group initially tested the IVOX for safety and efficacy and described experimental and clinical use of the IVOX.22-24 Patients selected for IVOX implantation underwent evaluation of vascular access sites for size and patency with Doppler ultrasound. Following surgical exposure of the femoral vein, the IVOX was passed into the vena cava over a guide wire under fluoroscopy to ensure proper position. At the time of venotomy, the patient was given 400 U/kg of heparin as a bolus, followed by a continuous heparin infusion to maintain the activated clotting time at 200-250 seconds or partial thromboplastin time at 80-90 seconds. The patient was maintained on prophylactic antibiotics. A vacuum pump withdrew 100% oxygen into the multiple hollow fibers from an oxygen source via the gas inlet and gas flow was regulated by an in-line flowmeter. The exhaust gas was analyzed for carbon dioxide concentration by a capnometer. The estima-

The Artificial Lung

106

tion of O2 transfer was obtained by measuring the changes in mixed venous O2 saturation (SvO2) with the IVOX on versus a brief period with the IVOX off:

[

]

m˙ o 2 (ml / min) = 1.34 ⋅ [Hgb ] ⋅ CO ⋅ (S vo 2 )IVOXOn ! (S vo 2 )IVOXOff ⋅ 10

(Eqn. 5.13)

where [Hgb] is the hemoglobin concentration (g/dl), CO is cardiac output (l/min), SvO2 is mixed venous oxygen saturation (%), and 1.34 is the oxygen carrying capacity of hemoglobin (ml/g). CO2 removal by IVOX was estimated from the product of gas flow, Qg (ml/min), and the fractional concentration of CO2 in the outflowing gas, [FCO2]o: m˙ co (ml / min) = Q g ⋅ [ Fco 2 ]o . 2

(Eqn. 5.14)

Functional Assessment of Intracorporeal Lung Device (IVOX) We initially tested the IVOX for safety and efficacy in an ovine model to describe the design features and to delineate the experimental and potential clinical use of the IVOX.22-25 Implantation of IVOX did not adversely affect hemodynamic function, nor was there evidence of significant hemolysis, thrombosis/embolism, foaming in the blood, catheter migration or vena caval intimal injury. The significant reduction in foreign surface area as compared to extracorporeal membrane oxygenation (ECMO) resulted in fewer blood surface interactions as evidenced by reduced pulmonary leukosequestration and complement activation.22,24 In the initial design, IVOX was capable of removing up to 30% of CO2 production in an ovine model of severe smoke inhalation injury. The average CO2 exchange was approximately 40 ml/min for size 7 IVOX, and ranged from 30 ml/min to 55 ml/min. This amount of CO2 removal represented approximately 30% of the CO2 production of an adult sheep (150-180 ml/min). An international multicenter clinical trial of IVOX was conducted for a phase I and phase II Food and Drug Administration (FDA) study in major critical care centers in the United States and Europe. From February, 1990 to May, 1993, a total of 164 IVOX devices were utilized in 160 clinical trial patients as a means for temporary augmentation of gas exchange in patients with severe, but potentially reversible, acute respiratory failure.26 Basic entry criteria for the study in the United States and Europe were similar: 1) the requirement for mechanical ventilation for over 24 hours; 2) PaO2 < 60 mm Hg on FIO2 > 0.5 and PEEP > 10 cm H2O, PaCO2 > 40 mm Hg with minute ventilation > 150 ml/min/kg. All patients were adults and were selected for IVOX implantation for acute respiratory failure due to a variety of causes including lung infection, trauma, sepsis and ARDS. IVOX was implanted according to a predetermined protocol. The right common femoral or right jugular veins were sites for insertion. The O2 and CO2 exchange by IVOX were calculated by the formulae previously described. The amount of O2 transfer and CO2 removal varied from approximately 40 to 70 ml/min depending on the size of the implanted device. Use of IVOX was associated with immediate blood gas improvement in a majority of patients that allowed a reduction in ventilator settings: FIO2 , PEEP, mean or peak airway pressure and minute ventilation were decreased by > 10% in over 60% of patients, and by > 25% in over 40% of patients. Overall survival of reported patients receiving IVOX was 30%, however, survival was directly related to the severity of lung injury and patient selection—patients with an increasing severity of lung injury or pulmonary malfunction, as indicated by Murray score, oxygenation index or intrapulmonary shunt,

Testing and Performance Evaluation of Artificial Lungs

107

had a decreased rate of survival. Unfortunately, there was no control arm to this study to evaluate improvement of survival with IVOX. Complications or adverse events associated with use of IVOX included mechanical and/or performance problems (29%) and patient complications such as bleeding, thrombosis, infection, venous occlusion, arrhythmia, as well as user errors which reflected the learning curve with a new device. Seven clinically recognized adverse events were probably contributory to the deaths of these patients. In reporting individual experiences with functional performance of IVOX, Gentilello et al.27 treated nine adult patients with acute respiratory failure with IVOX for a mean duration of 5.6 days. Mean CO2 removal by IVOX (sizes 7-10) was 40-51 ml/min, and although application of IVOX was associated with an increase in PaO2 and a decrease in PaCO2, the quantity of gas transfer was not sufficient to allow a reduction in PEEP, FIO2 or minute ventilation. High et al.,28 by comparing the gas exchange of IVOX and lungs using mass spectrometry in five patients with ARDS, showed IVOX (sizes 7-9) transferred 13-84 ml/min of O2 and 14-82 ml/min of CO2, an amount not exceeding 29% of the total gas exchange, and only small changes in ventilator support were achieved in two patients. Jurmann et al.29 implanted IVOX in three patients with severe respiratory failure and, in their best experience with a size 9 IVOX, only partial gas exchange support (55-74 ml/min CO2 removal) and moderate reduction in ventilator settings were achieved. In their experience with eight patients, Kallis et al.30 achieved an average of 58 (40-106) ml/min of CO2 removal and 85 (68-140) ml/min of O2 transfer in eight patients. Conrad et al.31 treated two patients with size 9 and 10 IVOX, which transferred 43-92 ml/min of O2 and 33-86 ml/min of CO2, and achieved a significant reduction in FIO2 in both patients and minute ventilation in one. In an adult patient with extended use of IVOX, von Segesser et al. showed increased PaO2 that allowed a reduction in PEEP and FIO2, together with improved hemodynamic function.32 During our clinical and animal studies, we observed that as blood PaCO2 increased with progressive development of respiratory failure, IVOX CO2 exchange concomitantly increased.33 These findings provided the basis of the adjunctive strategy of permissive hypercapnia. With intentional gradual increase in systemic blood PaCO2 levels, the CO2 pressure gradient across the IVOX membrane can be increased, thus CO2 removal can be enhanced. Application of IVOX with permissive hypercapnia allows a further reduction in ventilatory settings that helps minimize barotrauma inflicted with conventional ventilatory treatment. To allow more precise measurements of gas exchange and define factors affecting the performance, we modified an ex vivo right atrium-to-pulmonary artery extracorporeal circuit34 to model intravascular gas exchange by size 7 IVOX (Fig. 5.2). IVOX O2 transfer was determined at blood flow rates ranging from 1.0 to 4.5 l/min, and at hemoglobin (Hgb) concentrations ranging from 3.0 to 9.0 g/dl. IVOX CO2 removal was measured at different blood PaCO2 levels. The bypass circuit allowed us to precisely describe IVOX performance with different changing variables. Total O2 transfer increased in proportion to the blood flow up to 41 ml/min at a blood flow of 3.5-4.0 l/min (Figs. 5.3A and 5.3B). Similarly, CO2 removal gradually increased from 17 to 42 ml/min as blood flow increased from 1.0 to 3.0 l/min. However, higher flow did not further increase CO2 removal (Fig. 5.4). Hgb concentration above 7.0 g/dl did not result in proportional increases in O2 transfer (Fig. 5.5). The ex vivo circuit also confirmed the benefit of permissive hypercapnia established in our in vivo experiments of IVOX: CO2 removal nearly doubled from 41 to 81 ml/min when arterial blood PaCO2 increased from 46 to 90 mm Hg.34

The Artificial Lung

108

Table 5.3. Design properties of different prototypes of IVOX (Size 7)

Surface area (m2) No. of fibers Fiber length (cm) Fiber crimp size (mm) Internal fiber diameter (∝m) External fiber diameter (∝m)

IVOX I

IVOX IIa

IVOX IIb

0.21 589 38 4.76 200 246

0.23 1400 31.5 4.76 120 167

0.26 1400 35 3.17 120 167

To further analyze the limiting factors in O2 and CO2 exchange by IVOX and to explore possible ways to enhance performance, we developed a detailed mathematical model.8 Comprising multiple hollow fibers separating gas from blood by a gas-permeable membrane, IVOX was theoretically modeled as a repeating unit of a single cylindrical IVOX fiber surrounded by a cylindrical sleeve of blood with closed boundaries and no net interaction between adjacent units (Fig. 5.6). This modeling approach is similar to the Krogh cylinder model used previously to quantify gas transport in vascular capillaries surrounded by tissue.35 The performance of the entire device could be deduced from the analysis of each unit structure. Blood was treated as a homogeneous fluid over a sample volume in which blood-gas reactions were assumed to occur instantaneously. Empirical dissociation curves were used to relate partial pressures of O2 and CO2 with their respective content in whole blood. Blood and gas velocities were assumed to be fully developed, laminar, at steady state, parallel to the fiber axis and independent of mass transfer. Model equations were obtained by writing mass balance equations for O2 and CO2 in the gas, membrane and blood phases. Boundary conditions of continuity of wall transfers were imposed at the gas-membrane interface as well as at the membrane-blood interface. Radial symmetry was assumed on the gas side at the center of the fiber. The diffusion-convection partial differential equations for both O2 and CO2, in each of the three phases, were discretized and numerically solved using an alternating-direction implicit finite difference procedure.8 The computed results obtained with the mathematical model are very similar to those obtained in our ex vivo bypass model. Figures 5.7 and 5.8 show the effect of changing blood and gas flow on CO2 removal. Additional results using this model indicate that most of the mass transfer resistance to O2 uptake and CO2 removal is in the blood phase, and can be diminished by enhanced mixing of blood in the venae cava. While an increase in fiber surface area could significantly enhance gas transfer, increased resistance to venous return would result. Newer generations of IVOX designed to increase blood mixing could significantly enhance the efficiency of IVOX gas exchange. After recognizing the limitations of the device and limited clinical utility of removing only 30% of the body CO2 production under normocapnic conditions, design changes were incorporated to improve the gas exchange capabilities of IVOX.36 New prototype IVOX designs (IIa and IIb, Table 5.3) include increased fiber number, decreased fiber length, decreased fiber diameter and increased crimping to enhance gas transfer. An experimental study was conducted in 17 sheep with smoke inhalation injury treated with one of three IVOX devices.36 Neither IVOX I, IIa or IIb significantly changed hemodynamics or hemoglobin concentration. No significant differences between groups were noted

Testing and Performance Evaluation of Artificial Lungs

109

Fig. 5.2. The ex vivo bypass circuit for testing intravenacaval oxygenators.34 The circuit is flow-controlled and temperature maintained and allows sampling of pre and post-device blood gases. Reprinted with permission from the Society of Thoracic Surgeons. Tao W, Zwischenberger JB, Nguyen TT et al. Performance of an intravenous gas exchanger (IVOX) in a venovenous bypass circuit. Ann Thorac Surg 1994; 57:1484-1490.

for insertion technique, thrombosis, emboli or bleeding complications. Oxygen transfer by IVOX IIa and IIb was 59 ± 4 ml/min and 57 ± 7 ml/min, respectively, with no significant difference between the two designs. When compared to IVOX I, there was only a small improvement in oxygen transfer. IVOX IIa and IIb, however, significantly improved CO2 removal over IVOX I (Fig. 5.9). The CO2 transfer efficiency can be expressed as an arbitrary “efficiency coefficient” by dividing the CO2 removal value for a given collection period by the arterial blood PaCO2 measured at the same time. The efficiency coefficient for each device is a useful index of the expected amount of CO2 removal for a given partial pressure of CO2 in the blood. In our ovine model of severe smoke inhalation injury, the efficiency coefficient of IVOX I CO2 removal/PaCO2 ratio was 1:1. Prototypes IIa and IIb have an efficiency coefficient of 1.63 ± 0.02 and 1.81 ± 0.03 respectively, representing an increased CO2 removal of 60-80%.37,38 To investigate the hypothesis that a reduction in mass transfer resistance by bulk mixing of blood in the vicinity of the IVOX fibers can enhance the O2 transfer and CO2 removal by IVOX, we used the ex vivo, veno-venous bypass circuit described above that incorporated an “intra-aortic” balloon pump as the mixing device.9 Size 9 IVOX (894

110

The Artificial Lung

Fig. 5.3 A. Overall O2 transfer by IVOX increases linearly with blood flow up to 41 ml/min (* p < 0.05 and ** p < 0.01 compared to results at 1.0 l/min flow). B. The net mass transfer coefficient for O2 increases in proportion to blood flow (** p < 0.01 compared to results at 1.0 l/min flow). Reprinted with permission from the Society of Thoracic Surgeons. Tao W, Zwischenberger JB, Nguyen TT et al. Performance of an intravenous gas exchanger (IVOX) in a venovenous bypass circuit. Ann Thorac Surg 1994; 57:1484-1490.

Testing and Performance Evaluation of Artificial Lungs

111

Fig. 5.4 A. Overall CO2 removal by IVOX increases from 17 ml/min at a flow of 1.0 l/min to 41 ml/min at 3.0 l/min, but at higher flows it remains unchanged (* p < 0.05 and ** p < 0.01 compared to results at 1.0 l/min flow). B. The net mass transfer coefficient of CO2 increases linearly with flow from 1.0-3.0 l/min, but tends to plateau at greater flows (* p < 0.05 and ** p < 0.01 compared to results at 1.0 l/min flow). Reprinted with permission from the Society of Thoracic Surgeons. Tao W, Zwischenberger JB, Nguyen TT et al. Performance of an intravenous gas exchanger (IVOX) in a venovenous bypass circuit. Ann Thorac Surg 1994; 57:1484-1490.

112

The Artificial Lung

Fig. 5.5. Overall O2 transfer increases proportionally with hemoglobin content from 3.0-7.5 g/dl, but does not increase further at higher hemoglobin levels (* p < 0.05 compared to results at 1.0 l/min flow). Reprinted with permission from the Society of Thoracic Surgeons. Tao W, Zwischenberger JB, Nguyen TT et al. Performance of an intravenous gas exchanger (IVOX) in a venovenous bypass circuit. Ann Thorac Surg 1994; 57:1484-1490.

fibers with 0.41 m2 surface area, n=5) was incorporated in the bypass circuit and the blood flow was controlled by a roller pump ranging from 1.0 to 4.0 l/min. The balloon was placed near the shaft of the IVOX and pulsated at the rate adjusted to best improve the CO2 removal (100-120 bpm). O2 transfer and CO2 removal were measured with balloon pulsation on and off at different flow rates. Results showed that blood mixing by pulsation of the balloon caused a 25-49% increase in O2 transfer by IVOX, with the greatest increase seen at the highest flow range. CO2 removal was also increased by up to 35%, but at flows between 3.5-4.0 l/min, the effect of mixing was diminished (Fig. 5.10). These results confirmed the rate limiting mass transfer resistance of blood. Since O2 is more diffusion-limited, it is more dependent on mixing of blood for gas exchange than CO2. Other centers working on related gas exchange devices showed significant improvement in gas exchange with mixing of the fluid phase by rotation or oscillation of the fibers,39-42 further proving the potential in achieving better gas exchange with active blood mixing.

Testing and Performance Evaluation of Artificial Lungs

113

Fig. 5.6. Schematic diagram of a Krogh-like cylindrical model of a single component fiber of the IVOX device.8 Reprinted with permission from the American Physiological Society. Niranjan SC, Clark JW, San KY et al. Analysis of factors affecting gas transfer in an intravascular blood oxygenator. J Appl Physiol 1994; 77:1716-1730.

Prediction of IVOX Surface Area Requirements In Vivo In a final study to evaluate IVOX, we utilized our cumulative experimental data to evaluate the potential cumulative effects of changes in blood flow, systemic blood levels, and “bulk mixing” on the “predicted” membrane surface area requirements for IVOX-like devices. O2 uptake and CO2 excretion were measured at different blood flow rates for devices of size 7-9, with and without “bulk mixing”. From this data, overall O2 and CO2 mass transfer coefficients (ml gas/(min•mm Hg•m2), KO2 and KCO2, were determined using the following equations:

Ko 2 = Kco 2 =

m˙ o

2

As ( ∀Pco 2 )lm m˙ co

(Eqn. 5.15)

,

2

As ( ∀Pco 2 )lm

,

(Eqn. 5.16)

where mO2 and mCO2 are the measured oxygen uptake and carbon dioxide excretion (ml gas/min) in the ex vivo veno-venous bypass circuit, (∀PO2)lm and (∀PCO2)lm are the log mean partial pressure gradients between gas and blood (mm Hg) for O2 and CO2 , respectively, and As is the device surface area (m2). Mean overall mass transfer coefficients were calculated as a function of device blood flow using Equations 5.2 and 5.3 in Equations 5.15 and 5.16. Both mass transfer coefficients display a linear dependence on blood flow, with KCO2 being approximately 30 times greater than KO2. Using these calculated values of the overall mass transfer coefficients for O2 and CO2, the surface area necessary for a given level of CO2 excretion, at specific levels of PaCO2, was estimated as a function of device blood flow:

114

The Artificial Lung

Fig. 5.7. Theoretical predictions8 (solid lines) and experimental data34 (symbols) of CO2 removal (top panel) and O2 uptake (bottom panel) versus blood flow in an IVOX device.

Testing and Performance Evaluation of Artificial Lungs

115

Fig. 5.8. Theoretical predictions8 (solid line) and experimental data34 (symbols) of CO2 removal as a function of blood PCO2 at inlet of an IVOX device.

m˙ co

2 , Kco 2( ∀Pco 2 ) lm where (∀PCO2)lm is defined as:

As =

(Eqn. 5.17)

(Eqn. 5.18)

in which the subscripts b and g indicate the blood and gas phases, respectively, and the subscripts i and o indicate the inlet and the outlet of the IVOX, respectively. A linear form of the CO2 dissociation curve11 was used to estimate the log mean partial pressure gradient between gas and blood. To analyze the effect of bulk blood mixing via an intra-aortic balloon pump and permissive hypercapnia on surface area requirements for a given CO2 removal, we used different sizes of IVOX in our ex vivo veno-venous bypass circuit to calculate overall mass transfer coefficients with associated surface area requirements using IVOX alone, IVOX with permissive hypercapnia, and IVOX with additional active blood mixing. The effect of active blood mixing was estimated with and without blood mixing via an augmentation factor, (Ka)i, defined as:

The Artificial Lung

116

Fig. 5.9. Gas exchange of different IVOX prototypes (IIA and IIB).38 Reprinted with permission from the American Society for Artificial Internal Organs. Nguyen TT, Zwischenberger JB, Tao W et al. Significant enhancement of carbon dioxide removal by a new prototype IVOX. ASAIO J 1993; 39:M719-M724.

(K a ) i =

( K i ) mix , ( K i ) stat

(Eqn. 5.19)

where (Ki)mix and (Ki)stat are mass transfer coefficients of species i with and without active blood mixing, respectively. Our data indicate that surface area requirements for different rates of CO2 removal can be reduced by increase in blood flow. Active blood mixing achieved a Ka of 1.47 to 1.04 for CO2 and 1.49 to 1.28 for O2 with a blood flow of 1.0 to 4.0 l/min. Increasing blood PaCO2 levels enhances CO2 removal in a linear fashion. For a targeted CO2 removal of 150 ml/min at 4.0 l/min blood flow, the surface area needed at normocapnia without blood mixing is 1.76 m2, which cannot be accommodated by the vena cavae. This surface area can be reduced to 0.47 m2 with permissive hypercapnia at PaCO2 of 80 mm Hg, and further reduced to 0.33 m2 with additional active blood mixing. Such a surface area has already been achieved in previous IVOX designs (Fig. 5.11). Thus, systemic hypercapnia with active blood mixing and an adequate blood flow can significantly reduce the surface area requirements of IVOX-like devices to a practical range while maintaining the capacity to remove 50-70% CO2 production during acute respiratory failure. The first generation intravenacaval gas exchangers, while initially promising, were limited in their ability to provide clinically significant amounts of gas exchange. The rate of gas transfer by IVOX is dependent on: 1. membrane surface area; 2. gas permeability of the siloxane coated fibers and the porous support; 3. mass transfer resistance of blood phase;

Testing and Performance Evaluation of Artificial Lungs

117

Fig. 5.10A. O2 transfer by IVOX at different blood flow rates with and without balloon pulsation. Blood mixing by balloon pulsation causes a parallel increase in O2 transfer at all flows, and at 3.5-4.0 l/min of flow, this increase is more significant. * p < 0.05 between nonpulsed and pulsed, r2 = 0.968 for nonpulsed and r2 = 0.974 for pulsed.9 B. CO2 removal by IVOX at different blood flow rates with and without balloon pulsation. Blood mixing causes an increase in CO2 removal at all flows, but at flows of 3.5-4.0 l/min, the increased removal is diminished. * p < 0.05 and ** p< 0.01 between nonpulsed and pulsed, r2 = 0.875 for nonpulsed and r2 = 0.977 for pulsed.

118

The Artificial Lung

Fig. 5.11. Predicted surface area required for CO2 removal as a function of the desired CO2 removal, during normocapnia, hypercapnia alone, and with hypercapnia and active blood mixing in an IVOX.

4. flow rate of sweep gas; 5. vena cava blood flow rate around the IVOX fibers; and 6. the partial pressure gradient across the IVOX fibers (between blood and gas).

Our mathematical analysis and experimental data (see above) indicate that 80% of the resistance to gas transfer resides in the blood phase, whereas that of the siloxane coated fibers contributes < 20%. Our calculations suggest that the vast majority of blood (80%) in the vena cava flows in a laminar flow pattern past the IVOX fibers without participating in gas transfer with the IVOX gas phase. Mechanical modalities that enhance the convective mixing of blood flow around the hollow fibers (such as with a central pulsating balloon) can potentially reduce the overall resistance to gas transfer by 10-40% (see above). Increases in membrane surface area may be accomplished by changes in fiber number, fiber diameter and fiber length, but are limited by the resulting increased resistance to blood flow in the vena cavae and reduction in venous return. Our mathematical model indicates that the bulk of gas transfer occurs at the proximal end where blood first contacts the fibers. Therefore, use of shorter, more numerous fiber lengths with a cross flow arrangement may allow increases in surface area and enhancement of trans-radial mixing of

Testing and Performance Evaluation of Artificial Lungs

119

blood while avoiding prohibitive decrements in venous return. Increases in gas flow through the fibers favors CO2 removal but adds little to O2 uptake, while increments in vena caval blood flow enhance both CO2 removal and O2 uptake. Finally, theoretical and experimental studies indicate that the IVOX device is more effective for carbon dioxide removal than oxygen uptake, in large part due to the higher diffusivity x solubility product for CO2 and the ability to significantly increase the trans-fiber partial pressure gradient for CO2 through permissive hypercapnia.

Extracorporeal Gas Exchange The extracorporeal membrane oxygenator (ECMO) circuit is a modified heart-lung machine consisting of a venous blood drainage reservoir, a servoregulated roller pump, a membrane lung to exchange oxygen and carbon dioxide, and a heat exchanger to maintain temperature. Extracorporeal circulation for respiratory failure was first attempted in newborns in the 1960s. Bartlett et al. began clinical trials in 1972 and reported the first successful use of ECMO in newborn respiratory failure in 1976. During the initial experience, ECMO had an overall survival rate of 75-95% and these results helped to establish the therapeutic effectiveness of ECMO in infants after having met criteria predicting 80-100% mortality. Although less well established than neonatal ECMO, the application of ECMO in adults with respiratory failure is undergoing continued development and refinements. The ELSO database on adults with severe respiratory failure demonstrates a cumulative short-term survival rate of 47%.43 Recent appreciation of the neonatal and pediatric populations with acute respiratory failure has also yielded new therapeutic strategies that may impact on the adult population. A number of therapies look promising, pending controlled studies, while others fade into disuse as no benefits are shown. Some of these promising therapies include pressure-limited ventilation (permissive hypercapnia), inverse ratio ventilation, high frequency jet ventilation, high frequency oscillatory ventilation, intratracheal pulmonary ventilation and prone position ventilation. In addition, “cutting edge” therapies such as partial liquid ventilation, inhaled nitric oxide (NO), extracorporeal carbon dioxide removal (ECCO2R), intravascular oxygenation, and arteriovenous carbon dioxide removal (AVCO2R) are being pursued by several groups, and their ultimate role is being defined (Fig. 5.12). Several concepts have been persistent in these efforts: 1. acute respiratory failure is multi-factorial, and may be the end organ response to a host of initiating events; 2. pressure controlled or “gentle” mechanical ventilation limits barotrauma and volutrauma during management of acute respiratory failure; 3. several modalities (surfactant, NO, pressure limited ventilation, and ECMO) appear to favorably impact on outcome in select patients; and 4. given the complexity of the problem, each case of acute respiratory failure is still unique and treatment should remain individualized. All new treatment modalities, however, must ultimately be subjected to prospective, randomized trials.

The hypothesis that “lung rest” can be accomplished using an artificial lung device with extracorporeal circulation was originally derived from animal experiments using a spiral-coil membrane lung designed for CO2 removal in chronic lung disease.2 During testing of these carbon dioxide membrane lungs (CDML) in spontaneous breathing animals in 1977, Kolobow et al. noted that spontaneous ventilation decreased proportionally to CO2 removal by CDML.3 By removal of 100% of metabolic CO2 production by CDML, the

120

The Artificial Lung

Fig. 5.12. Available treatment modalities for lung injury at different severity. MV: Mechanical Ventilation; NO: Nitric Oxide; PH: Permissive Hypercapnia; PLV: PressureLimited Ventilation; VV: Venovenous; VA: Venoarterial; ECMO: Extracorporeal membrane oxygenation

animals could apparently be kept completely apneic, while an amount of oxygen equal to the animal’s oxygen consumption was provided without ventilation (apneic ventilation).6 Thus, the normal physiological reaction to a reduced need for CO2 removal was a reduction in tidal volume and in minute ventilation, resulting in a reduced excretion of CO2 through the native lung while maintaining arterial blood PaCO2 at the normal level. Furthermore, any increase or decrease in extracorporeal CO2 within a few seconds was reflected by the animals by an increase or decrease in breathing and, more specifically, in an increase or decrease in alveolar ventilation.3 A major problem with the use of ECMO is that more than 60% of the cardiac output has to flow through the gas exchange device to achieve adequate arterial oxygenation. Many of the detrimental effects of ECMO are felt to be related to the increased transit time of blood through the gas exchanger (namely, reduction of pulmonary blood flow, increased tendency for bleeding, activation of blood leukocytes). With the technique of extracorporeal carbon dioxide removal (ECCO2R),44,45 it was possible to decrease mechanical ventilation to near-zero, while oxygenation was maintained by apneic oxygenation, intratracheal insufflation of oxygen, or low levels of positive-pressure ventilation in conjunction with moderate levels of positive end-expiratory pressure (PEEP). With ECCO2R, total CO2 removal can be accomplished with blood flow to the extracorporeal gas exchange device of approximately 20-30% of cardiac output. The objective in ECCO2R

Testing and Performance Evaluation of Artificial Lungs

121

is to reduce the need for alveolar ventilation, thereby minimizing the potential exacerbation of lung injury due to barotrauma and volutrauma.46

Physiological Basis of Extracorporeal CO2 Removal

The two main respiratory functions, oxygenation and CO2 removal, have different requirements: oxygenation depends on adequate pulmonary blood flow because the relatively high saturation of normal mixed venous blood (about 75%) limits the amount of oxygen that can be taken up by the blood. Ventilation, per se, is not required, because oxygenation can be accomplished by providing an adequate amount of oxygen, via supplemental FIO2, to an inflated lung, using the technique of apneic oxygenation. By contrast, CO2 excretion depends less on blood flow, since the content of CO2 in mixed venous blood is high, but on an adequate level of ventilation. During extracorporeal gas exchange, the two respiratory functions are dissociated. In fact, 70-80% of oxygenation may occur through the native lungs, while these are kept inflated with pressure sufficient to keep open the recruitable regions of the lung. Removal of total CO2 production (200-400 ml/ min) can occur predominantly through an AL. Normal arterial oxygen content (20 ml/dl), even with an inspired oxygen concentration of 100%, is very close to the maximal oxygen carrying capacity of blood of 22 ml/dl. Thus, it is not possible to super-oxygenate a fraction of the pulmonary blood flow to compensate for a larger fraction of pulmonary blood flow, the oxygenation of which is impaired. In contrast, normal central mixed venous blood total CO2 content is about 52 ml/dl compared with a normal arterial blood CO2 content of 48 ml/dl, which amounts to a percentage CO2 extraction of 7%.14 Thus, it is theoretically possible to remove a much larger fraction of CO2 from a fraction of the pulmonary blood flow to compensate for a much larger fraction of blood flow which has a limited CO2 excretion. Since CO2 excretion is inherently more efficient (approximately 20 fold, see above), it is theoretically possible to remove the majority of the normal CO2 excretion of 200 ml/min through a small fraction (10-20%) of blood flow. This would necessitate CO2 extraction of about 30-50%, such that: for normal lung: m˙ co = CO ⋅ ([CO2 ]mv ! [CO2 ]a ) = 5 ⋅ (52 ! 48) ⋅ 10 = 200ml / min (Eqn. 5.20) 2

for AL: m˙ co 2 = f ⋅ CO ⋅ ([CO2 ] mv ! [CO2 ]o ) = 1 ⋅ (52 ! 32) ⋅ 10 = 200ml/ min (Eqn. 5.21) where f is the fraction of blood flow (of the cardiac output, CO) through the extracorporeal AL (assumed to be 20% in the calculation above), [CO2]mv is the “normal” mixed venous blood CO2 content, [CO2]a is the “normal” arterial blood CO2 content and [CO2]o is the content of CO2 in blood leaving the AL. Based on the calculations shown above, blood leaving the AL would have a CO2 content of 32 ml/dl, corresponding to a partial pressure of about 15 mm Hg. This would allow removal of the entire normal metabolic production of carbon dioxide.

Veno-Venous Extracorporeal Gas Exchange The concept of low-flow veno-venous extracorporeal CO2 removal is not new and was proposed by Kolobow et al., in the late 1970s, to remove all of the metabolically produced CO2 while maintaining normal blood PaO2 via apneic oxygenation.6 Targeting CO2 removal, Gattinoni, together with Kolobow, developed the extracorporeal technique

122

The Artificial Lung

of ECCO2R in animals3,44,45 and for clinical application.44,47 Gattinoni et al.44 also showed that low frequency ventilation combined with ECCO2R provided sufficient gas exchange and improved survival in patients with ARDS. Gattinoni and his colleagues successfully used veno-venous extracorporeal CO2 removal in critically ill adults. Oxygenation was maintained by a modification of apneic oxygenation by using low frequency positive-pressure mechanical ventilation (about 4 breaths/min). With the success of venoarterial extracorporeal membrane oxygenation (ECMO) for the treatment of severe ARDS,48 especially in the neonatal population, a resurgence of popularity has led to a number of investigations into the application of ECCO2R and similar technologies in the treatment of ARDS. ECCO2R, however, still requires the use of a pump in the extracorporeal circuit, and the associated complications such as tubing rupture, cavitation and other life-threatening sequelae.49

Arteriovenous Extracorporeal CO2 Removal To overcome the need for a pump and simplify the extracorporeal circuit, Barthelemy et al.50 utilized a pumpless artery-to-vein extracorporeal system and combined it with the apneic oxygenation technique to satisfy all the gas exchange requirements of an experimental animal for up to 24 hours. Subsequently, Awad et al. demonstrated the feasibility of prolonged arteriovenous support for up to seven days in animals.51 Both of these studies were limited by high circuit resistance to spontaneous arteriovenous blood flow. We have developed a technique of simplified arteriovenous extracorporeal CO2 removal with a low resistance membrane gas exchanger52 to provide lung rest in the setting of severe respiratory failure. Arteriovenous carbon dioxide removal (AVCO2R) minimizes the foreign surface interactions and blood element shear stress inherent in an extracorporeal circuit with a pump and provides a gas exchange membrane of sufficient surface area for total CO2 removal. The procedure involves cannulation of the femoral artery and vein with a membrane oxygenator interposed in the circuit. Blood flows spontaneously through the oxygenator according to the pressure gradient between the artery and vein. The circuit is essentially identical to that used for continuous arteriovenous hemofiltration (CAVH), a commonplace procedure in the ICU. The difference is the use of a gas exchange device (membrane oxygenator) in place of a hemofilter, and larger vascular cannulae (approximately 17 French) to accommodate flows about five times greater. There are many theoretical advantages to this approach over veno-venous extracorporeal CO2 removal (ECCO2R). The first is the use of a small, highly efficient low resistance fiber membrane oxygenator. This allows the use of a much smaller priming volume (< 300 ml). In addition, this system does not require any type of mechanical pump, thereby greatly simplifying its application and providing greater degree of safety. It would be suitable for use in patients with primary hypercapnic ventilatory failure as well as for use in patients with permissive hypercapnia receiving ventilatory support for hypoxemic respiratory failure.

Functional Evaluation of AVCO2R

The collaborative efforts at University of Texas Medical Branch (UTMB) and Louisiana State University (LSU) include both small animal and large animal studies. The first series of studies in a piglet model combined with mathematical modeling of whole body gas exchange was used to demonstrate the efficacy of CO2 removal and provide estimates of arterial blood PaCO2 as a function of extracorporeal blood flow under conditions of total CO2 removal.53 In collaboration with Dr. Steve Conrad at LSU, we have developed a mathematical model of whole body gas-exchange that allows quantitative estimation of

Testing and Performance Evaluation of Artificial Lungs

123

Fig. 5.13. Block diagram of the mathematical model of whole body CO2 exchange incorporating arteriovenous CO2 removal (AVCO2R).53 The tissue compartments, represented as single compartments in the diagram, were implemented as five parallel compartments representing muscle, brain, kidneys, fat, and other. The AVCO2R device is modeled as a cross -flow arrangement of blood and gas phases. Reprinted with permission from the American Society for Artificial Internal Organs. Conrad SA, Brown EG, Grier LR et al. Arteriovenous extracorporeal carbon dioxide removal (AVCO2R): a mathematical model and experimental evaluation. ASAIO J 1998; 44:267-277.

extracorporeal gas exchange. In this model (Fig. 5.13), all compartments are assumed to be well mixed. Inter-compartmental blood flow is considered instantaneous, i.e., time delays for blood flow between compartments are not incorporated. Blood is assumed to be in internal equilibrium, so that the total carbon dioxide content of blood can be related to the partial pressure of CO2 using an empirical dissociation curve, as provided by Cherniack and Longobardo.54 These latter nonlinear equations implicitly include the Bohr and Haldane effects. In this model, the lungs are modeled as a single alveolar compartment associated with a distributed pulmonary capillary bed, discretized as N serial sub-compartments. Intrapulmonary shunting is included in the mathematical model as a direct venous admixture from central mixed venous blood to arterial (preoxygenator) blood, and parameterized as Qs/Qt. The peripheral tissue compartments are modeled as an adaptation of the multi-compartmental model of carbon dioxide stores proposed by Fahri and Rahn.55 The Fahri-Rahn model is composed of separate compartments representing muscle, heart, brain and ‘other’ tissues. Based on their observation, the heart muscle compartment was lumped into the ‘other’ compartment. Separate kidney and fat compartments were included. The extracorporeal gas exchange device has blood flow oriented in the radial direction and gas flow oriented in the axial direction. The device was modeled as a nb x ng multi-compartmental cross-flow matrix, in which blood flow is arranged through nb com-

124

The Artificial Lung

Fig. 5.14. Predicted steady-state arterial blood PaCO2 as a function of shunt flow (expressed as % of cardiac output).53

partmental columns (i), and gas flow through the ng compartment rows (j). Total extracorporeal blood and gas flow are divided equally among the respective flow paths. Conrad et al.53 used this model to simulate extracorporeal CO2 removal at various levels of shunt flow (ranging from 5 to 20% of cardiac output). The model predicts that extracorporeal flow of 10-15% could support total CO2 removal with modest to significant levels of systemic hypercapnia. The results of these calculations are shown in Figure 5.14. It can be seen that shunt flows of 10-15% of the cardiac output can support total CO2 removal at levels of PaCO2 that are physiologically tolerable. We applied AVCO2R to healthy animals to quantify the gas exchange capabilities of the system and establish ventilator management protocols for the subsequent studies in a large animal model of smoke inhalation injury-induced, severe respiratory failure. Increase in sweep gas flow increased CO2 removal up to a gas/blood flow ratio of 2:1. For a sweep gas flow of 3 l/min, CO2 removal increased proportionally to AVCO2R blood flow to a maximum of 1417 ± 26 ml/min (19% of cardiac output) (Fig. 5.16).56 Normal PaO2 and PaCO2 could be maintained with minimal ventilator support (MV=16% baseline MV with near-apneic oxygenation with 2 breath/min at dead-space tidal volume) at a blood flow of 500 ml/min or higher. At these maximally reduced ventilator settings, moderate hypercapnia (PaCO2 . 75 mm Hg) resulted only when AVCO2R blood flow was decreased to below 500 ml/min (Fig. 5.15). These findings indicate that optimizing AVCO2R blood and gas flow maximizes CO2 removal and allows a significant reduction in minute ventilation. In cases of severely limited blood flow, lung rest can still be realized at moderate hypercapnia. At flow rates achievable by percutaneous access, extracorporeal AVCO2R can be utilized to achieve lung rest during mechanical ventilation. AVCO2R also allowed reductions in mechanical ventilatory support to near apneic conditions while maintaining

Testing and Performance Evaluation of Artificial Lungs

125

Fig. 5.15. As AVCO2R flow is reduced following maximal reduction in ventilator support to 16% of baseline minute ventilation, moderate hypercapnia (40-70 mm Hg) results only below 500 ml/min.56

normocapnia at flow rates in the range of 500 ml/min (14.6 ml/kg/min = 7% of CO) or greater. Normocapnia can be maintained at blood flow of 500 ml/min while maintaining maximally reduced minute ventilations (Fig. 5.16). Based on our encouraging results, we have undertaken a phase I clinical study of AVCO2R in patients with ventilatory failure.

Shunt Physiology During AVCO2R

To evaluate the effects of arteriovenous shunt applied in AVCO2R on hemodynamics during ARDS, we applied prolonged AVCO2R to sheep with severe inhalation injury.57 Six adult female sheep were subjected to intratracheal cotton severe smoke insufflation to a mean carboxyhemoglobin level of 83 ± 3%. Twenty-four hours after injury, a low-resistance 2.5 m2 membrane oxygenator was placed in a carotid-to-jugular pumpless arteriovenous shunt at unrestricted flow to allow complete carbon dioxide (CO2) removal and reductions in ventilator support. Animals remained conscious and heart rate (HR), cardiac output (CO), mean arterial pressure (MAP), and pulmonary arterial pressure (PAP) were measured at baseline, after injury, and daily on AVCO2R support for a duration of seven days. All animals survived the study period. CO2 removal ranged from 99.7 ± 13.7 to 152.2 ± 16.2 ml/min, and five of the six animals (83%) were successfully weaned from the ventilator prior to day 7. During full AVCO2R support, shunt flow (Qb) ranged from 1.24

126

The Artificial Lung

Fig. 5.16. Prediction of CO2 removal as a function of extracorporeal blood flow during normocapnia and hypercapnia (arterial PaCO2 60-80 mm Hg) in AVCO2R .

± 0.06 to 1.43 ± 0.08 l/min and accounted for 20.1 ± 1.4% to 25.9 ± 2.4% of CO. No statistically significant changes in HR, CO, MAP, or PAP were demonstrated over the study course despite the extracorporeal shunt flow. AVCO2R as a simplified extracorporeal gas exchange support is relatively safe without adverse hemodynamic effects or complications. To evaluate the effects of different shunt levels on organ blood flow, we used colored microspheres in a conscious ovine model of AVCO2R. A low resistance 2.5 m2 oxygenator was placed in a simple carotid-to-jugular arteriovenous circuit. AVCO2R flow (Qb) was incrementally increased to 5%, 10%, 15%, 20%, and 25% of the baseline CO. Following equilibration, colored microspheres were injected into a left atrial catheter while reference blood was withdrawn from an arterial line at a constant rate. Organ blood flow obtained by quantifying microspheres in the tissues, showed approximately a 10-20% decrease at a 5% shunt, but remained relatively unchanged thereafter up to a 25% shunt. There was no evidence of adverse hemodynamic effects or end organ ischemia. AVCO2R can achieve lung rest during respiratory failure at flow rates of 10-25% of CO, with a resultant mild decrease in critical organ blood flow that appears well tolerated.

AVCO2R in Severe Respiratory Failure

We further studied the role of AVCO2R on ventilator reduction in sheep with severe respiratory failure secondary to smoke inhalation injury.58 Animals were instrumented with femoral and pulmonary arterial catheters and underwent an LD50 cotton smoke inhalation injury via a tracheostomy under halothane anesthesia. Twenty-four hours after

Testing and Performance Evaluation of Artificial Lungs

127

Fig. 5.17. Minute ventilation reduced from 10.3 ± 1.4 l/min to 0.5 ± 0.0 l/min at six hours on arteriovenous AVCO2R58 while maintaining PaCO2 (* p < 0.05 vs. baseline).

smoke inhalation injury, they were reanesthestized and systemically heparinized for cannulation of the left carotid artery and common jugular vein to construct a simple arteriovenous shunt. A commercial membrane gas exchanger (2.5 m2 surface area) was interposed within the arteriovenous shunt, and blood flow produced by the arteriovenous pressure gradient was unrestricted upon complete recovery from anesthesia. CO2 removal by the gas exchanger was measured as the product of the sweep gas flow (100% oxygen at 2.5-3.0 l/min) and its exhaust carbon dioxide content measured with an in-line capnometer. CO2 removed by the native lungs was determined by the expired gas CO2 content in a Douglas bag. A 20% reduction in ventilator support was made in a stepwise fashion every hour, first in tidal volume to achieve peak inspiratory pressure (PIP) below 30 cm H2O, and then in respiratory rate while maintaining normocapnia. Arterial PaO2 was maintained by adjusting the fraction of inspired oxygen (FIO2) and the level of positive end-expiratory pressure (PEEP). Mean AVCO2R blood flow ranged from 1154 ± 82 ml/ min (25% of cardiac output) to 1277 ± 38 ml/min (29% of cardiac output) over the six-hour study period. The pressure gradient across the gas exchanger was always lower than 10 mm Hg. Maximum AVCO2R CO2 removal was 102.0 ± 9.5 ml/min (96% of total CO2 production), allowing minute ventilation (MV) to be reduced from 10.3 ± 1.4 l/min before AVCO2R to 0.5 ± 0.0 l/min at six hours on AVCO2R while maintaining normocapnia (Fig. 5.17). Similarly, PIP decreased from 40.8 cm H2O to 19.7 ± 7.5 cm H2O. Arterial PaO2 was maintained above 100 mm Hg at maximally reduced ventilator support. Mean arterial pressure and cardiac output did not change significantly with the use of AVCO2R. Extracorporeal CO2 removal using a low-resistance gas exchanger in a simple arteriovenous shunt allows significant reduction in MV and PIP without hyper-

-in situ oxygenation of mixed venous blood -partial CO2 removal

Intracorporeal 50-70% of CO

None

Conventional PPV (Volume or Pressure Control) PIP <45 cm H2O PEEP 5-25 cm H2O Rate 10-25/min

100% of CO

Yes

<15 days

Goal

Blood-flow

Pump requirement

Ventilation

Tissue blood flow

Anticoagulation

Duration

IVOX

Table 5.4. Comparison of IVOX, ECCO2R and AVCO2R ECCO2R

<30 days

Yes

100% of CO

LFPPV (Volume or Pressure Control) PIP <35-40 cm H2O PEEP 5-25 cm H2O Rate 2-4/min

Yes

Extracorporeal (veno-venous) 30% of CO

-lung rest -reduction of ventilatory requirements (reduction of minute ventilation, PIP and FIO2)

<20 days

Yes

80-90% of CO

ALI: LFPPV (Volume or Pressure Control) PIP <35-40 cm Rate 2-4/min OLD: spontaneous respiration

None

-lung rest -reduction of ventilatory requirements (reduction of minute ventilation and PIP)

AVCO2R

128 The Artificial Lung

Testing and Performance Evaluation of Artificial Lungs

129

Table 5.5. Design and application considerations in AL to increase gas exchange—the "bottom line" Gas

Increase sweep gas: Increase transmembrane pressure gradient: hypercapnia

Blood-flow

Counter-current flow Low device resistance Optimize cardiac output

Fiber design

Short fibers Low gas phase resistance Thin membrane Microporous

Blood O2 carrying capacity

Optimize hemoglobin level

capnia or the complex circuitry and monitoring required for conventional extracorporeal membrane oxygenation (ECMO). AVCO2R can be applied as an easy and cost-effective treatment to minimize ventilator-induced barotrauma and volutrauma during severe respiratory failure. Recently, our group has developed a more clinically relevant ovine model of severe respiratory failure secondary to smoke inhalation and cutaneous burn injury. Percutaneous AVCO 2R proved to be capable of minimizing ventilator pressures, reducing ventilator-dependent days, and improving survival. A prospective randomized trial is being conducted to evaluate different treatment strategies, including percutaneous AVCO2R and gentle ventilation with permissive hypercapnia, on the outcome of severe respiratory failure using the smoke and burn model.

Clinical Application of AVCO2R

Recently, a series of patients with severe respiratory failure were treated with percutaneous AVCO2R at UTMB and LSU. Initially, patients with primary CO2 retention were selected. AVCO2R with size 10-14 F percutaneous cannulae proved to be capable of removing up to 353 ml/min CO2 removal, or up to 97% of CO2 production, which allowed significant reductions in ventilator pressures. Oxygenation was successfully managed with gentle or near-apneic oxygenation. As more cases accumulate, it has become more clear to the investigators that more patients can be selected with combined O2 and CO2 exchange failure, as well. A prospective, randomized clinical trial has been initiated in collaboration between UTMB and LSU.

Conclusion The past decade has seen a remarkable surge of interest in the use of AL in the management and treatment of patients with acute respiratory failure and ARDS (Table 5.4). Major innovations such as the Intravenous Oxygenator have rekindled interest in the design of more efficient intracorporeal gas-exchange devices (Table 5.5). Both veno-venous and arteriovenous CO2 removal have shifted the previous exclusive focus on oxygenation to carbon dioxide removal thereby promoting the concept of “lung rest,” as an important adjunctive strategy in the management of patients with ARDS.

130

The Artificial Lung

Acknowledgments The authors would like to acknowledge the support of this work from the Shriners Hospitals for Crippled Children (Grant No. 8530), the Constance Marcili Schafer Endowed Research Fund, and the Bioengineering Center of the University of Texas Medical Branch. References 1. American National Standard. Cardiovascular Implants and Artificial Organs—Blood-Gas exchangers. Arlington: Association for the Advancement of Medical Instrumentation, 1997. 2. Kolobow T, Gattinoni L, Tomlinson T et al. The carbon dioxide membrane lung (CDML): A new concept. Trans Am Soc Artif Intern Organs 1977; 23:17-21. 3. Kolobow T, Gattinoni L, Tomlinson T et al. Control of breathing using an extracorporeal membrane lung. Anesthesiology 1977; 46:138-141. 4. Gattinoni L, Agostoni A, Pesenti A et al. Treatment of acute respiratory failure with low-frequency positive-pressure ventilation and extracorporeal removal of CO2. Lancet 1980; 2:292-294. 5. Gattinoni L, Kolobow T, Tomlinson T et al. Control of intermittent positive pressure breathing (IPPB) by extracorporeal removal of carbon dioxide. Br J Anaesth 1978; 50:753-758. 6. Kolobow T, Gattinoni L, Tomlinson T et al. An alternative to breathing. J Thorac Cardiovasc Surg 1978; 75:261-266. 7. Borelli M, Kolobow T, Spatola R et al. Severe acute respiratory failure managed with continuous positive airway pressure and partial extracorporeal carbon dioxide removal by an artificial membrane lung. A controlled, randomized animal study. Am Rev Respir Dis 1988; 138:1480-1487. 8. Niranjan SC, Clark JW, San KY et al. Analysis of factors affecting gas transfer in an intravascular blood oxygenator. J Appl Physiol 1994; 77:1716-1730. 9. Tao W, Schroeder T, Bidani A et al. Improved gas exchange performance of the intravascular oxygenator by active blood mixing. ASAIO J 1994; 40:M527-M532. 10. Snider MT, Richardson PD, Friedman LI et al. Carbon dioxide transfer rate in artificial lungs. J Appl Physiol 1974; 36:233-239. 11. Tao W, Bidani A, Cardenas VJ Jr et al. Strategies to reduce surface area requirements for carbon dioxide removal for an intravenacaval gas exchange device. ASAIO J 1995; 41:M567-M572. 12. Bidani A, Crandall ED. Quantitative aspects of capillary CO2 exchange. In: Chang HK, Paiva M, eds. Respiratory Physiology: A Quantitative Approach. New York: Marcel Dekker, 1989:371-419. 13. Bidani A, Mathew SJ, Crandall ED. Pulmonary vascular carbonic anhydrase activity. J Appl Physiol 1983; 55:75-83. 14. Bidani A. Analysis of abnormalities of capillary CO2 exchange in-vivo. J Appl Physiol 1991; 70:1686-1699. 15. Zacour RK. Clinical evaluation of the Sarns membrane oxygenator. Perfusion 1988; 3:219-224. 16. Nowak J, Pol G, Brands W. Clinical evaluation of the Bentley Univox membrane oxygenator. Perfusion 1992; 7:59-65. 17. Fried DW. Performance evaluation of blood gas exchange devices. Int Anesthesiol Clin 1996; 34:47-60. 18. High KM, Snider MT, Bashein G. Principles of oxygenator function: gas exchange, heat transfer, and blood-artificial surface interaction. In: Gravlee GP, Davis RF, Utley JR, eds. Cardiopulmonary bypass: Principles and Practice. Baltimore: Williams & Wilkins, 1993:28-54. 19. Bidani A, DuBose TD Jr. Cellular and whole-body acid-base regulation. In: Arieff AI, Defronzo RA, eds. Fluid, Electrolyte and Acid-base Disorders. 2nd ed. New York: Churchill Livingstone, 1995:69-103. 20. Mortensen JD. An intravenacaval blood gas exchange (IVCBGE) device. A preliminary report. ASAIO Trans 1987; 33:570-573. 21. Mortensen JD, Berry G. Conceptual and design features of a practical, clinically effective intravenous mechanical blood oxygen/carbon dioxide exchange device (IVOX). Int J Artif Organs 1989; 12:384-389. 22. Cox CS Jr, Zwischenberger JB, Traber LD et al. Use of an intravascular oxygenator/carbon dioxide removal device in an ovine smoke inhalation injury model. ASAIO Trans 1991; 37:M411-M413. 23. Zwischenberger JB, Cox CS Jr. A new intravascular membrane oxygenator to augment blood gas transfer in patients with acute respiratory failure. Tex Med 1991; 87:60-63.

Testing and Performance Evaluation of Artificial Lungs

131

24. Zwischenberger JB, Cox CS Jr, Minifee PK et al. Pathophysiology of ovine smoke inhalation injury treated with extracorporeal membrane oxygenation. Chest 1993; 103:1582-1586. 25. Zwischenberger JB, Cox CS Jr, Graves D et al. Intravascular membrane oxygenation and carbon dioxide removal—A new application for permissive hypercapnia? Thorac Cardiovasc Surg 1992; 40:115-120. 26. Conrad SA, Bagley A, Bagley B et al. Major findings from the clinical trials of the intravascular oxygenator. Artif Organs 1994; 18:846-863. 27. Gentilello LM, Jurkovich GJ, Gubler KD et al. The intravascular oxygenator (IVOX): Preliminary results of a new means of performing extra-pulmonary gas exchange. J Trauma 1993; 35:399-404. 28. High KM, Snider MT, Richard R et al. Clinical trials of an intravenous oxygenator in patients with adult respiratory distress syndrome. Anesthesiology 1992; 77:856-863. 29. Jurmann MJ, Demertzis S, Schaefers HJ et al. Intravascular oxygenation for advanced respiratory failure. ASAIO J 1992; 38:120-124. 30. Kallis P, al Saady NM, Bennett ED et al. Early results of intravascular oxygenation. Eur J Cardiothorac Surg 1993; 7:206-210. 31. Conrad SA, Eggerstedt JM, Morris VF et al. Prolonged intracorporeal support of gas exchange with an intravenacaval oxygenator. Chest 1993; 103:158-161. 32. von Segesser LK, Schafner A, Stocker R et al. Extended (29 days) use of intravascular gas exchanger. Lancet 1992; 339:1536 33. Zwischenberger JB, Nguyen TT, Tao W et al. IVOX with gradual permissive hypercapnia: A new management technique for respiratory failure. J Surg Res 1994; 57:99-105. 34. Tao W, Zwischenberger JB, Nguyen TT et al. Performance of an intravenous gas exchanger (IVOX) in a venovenous bypass circuit. Ann Thorac Surg 1994; 57:1484-1490. 35. Krogh A. The number and distribution of capillaries in muscles with calculations of the oxygen pressure head necessary for supplying the tissue. J Physiol 1919; 52:409-415. 36. Nguyen TT, Zwischenberger JB, Cox CS et al. Marked enhancement of CO2 removal by second generation IVOX. ASAIO Abstracts 1993;(Abstract). 37. Cox CS Jr, Zwischenberger JB, Kurusz M. Development and current status of a new intracorporeal oxygenator. Perfusion 1991; 6:291-296. 38. Nguyen TT, Zwischenberger JB, Tao W et al. Significant enhancement of carbon dioxide removal by a new prototype IVOX. ASAIO J 1993; 39:M719-M724. 39. Hattler BG, Johnson PC, Sawzik PJ et al. Respiratory dialysis. A new concept in pulmonary support. ASAIO J 1992; 38:M322-M325. 40. Reeder GD, Hattler BG, Rawleigh J et al. Current progress in the development of an intravenous membrane oxygenator. ASAIO J 1993; 39:M461-M465. 41. Makarewicz AJ, Mockros LF, Anderson RW. A pumping intravascular artificial lung with active mixing. ASAIO J 1993; 39:M466-M469. 42. Vaslef SN, Mockros LF, Anderson RW. Development of an intravascular lung assist device. ASAIO Trans 1989; 35:660-664. 43. Tracy TF Jr, Delosh T, Stolar CJH. The registry of the Extracorporeal Life Support Organization. In: Zwischenberger JB, Bartlett RH, eds. ECMO. Extracorporeal Cardiopulmonary Support in Critical Care. Ann Arbor: Extracorporeal Life Support Organization, 1995:251-260. 44. Gattinoni L, Pesenti A, Mascheroni D et al. Low-frequency positive-pressure ventilation with extracorporeal CO2 removal in severe acute respiratory failure. JAMA 1986; 256:881-886. 45. Gattinoni L, Kolobow T, Tomlinson T et al. Low-frequency positive pressure ventilation with extracorporeal carbon dioxide removal (LFPPV-ECCO2R): An experimental study. Anesth Analg 1978; 57:470-477. 46. Hickling KG. High frequency ventilation. N Z Med J 1988; 101:50-51. 47. Gattinoni L, Kolobow T, Agostoni A et al. Clinical application of low frequency positive pressure ventilation with extracorporeal CO2 removal (LFPPV-ECCO2R) in treatment of adult respiratory distress syndrome (ARDS). Int J Artif Organs 1979; 2:282-283. 48. Ichiba S, Bartlett RH. Current status of extracorporeal membrane oxygenation for severe respiratory failure. Artif Organs 1996; 20:120-123. 49. Zwischenberger JB, Nguyen TT, Upp JR Jr et al. Complications of neonatal extracorporeal membrane oxygenation. Collective experience from the Extracorporeal Life Support Organization. J Thorac Cardiovasc Surg 1994; 107:838-849. 50. Barthelemy R, Galletti PM, Trudell LA et al. Total extracorporeal CO2 removal in a pumpless artery-to-vein shunt. Trans Am Soc Artif Intern Organs 1982; 28:354-358.

132

The Artificial Lung

51. Awad JA, Deslauriers J, Major D et al. Prolonged pumpless arteriovenous perfusion for carbon dioxide extraction. Ann Thorac Surg 1991; 51:534-540. 52. Brunston RL Jr, Zwischenberger JB, Tao W et al. Total arteriovenous carbon dioxide removal (AVCO2R): Simplifying extracorporeal support for severe respiratory failure. Ann Thorac Surg 1997; 64:1599-1605. 53. Conrad SA, Brown EG, Grier LR et al. Arteriovenous extracorporeal carbon dioxide removal (AVCO2R): A mathematical model and experimental evaluation. ASAIO J 1998; 44:267-277. 54. Cherniack NS, Longobardo GS, Staw I et al. Dynamics of carbon dioxide stores changes following an alteration in ventilation. J Appl Physiol 1966; 21:785-793. 55. Fahri LE, Rahn H. Dynamics of changes in carbon dioxide stores. Anesthesiology 1960; 21:604-614. 56. Brunston RL, Jr., Tao W, Bidani A et al. Determination of low blood flow limits for arteriovenous carbon dioxide removal. ASAIO J 1996; 42:M845-M849. 57. Brunston RL, Jr., Tao W, Bidani A et al. Prolonged hemodynamic stability during arteriovenous carbon dioxide removal (AVCO2R) for severe respiratory failure. J Thorac Cardiovasc Surg 1997; 114:1107-1114. 58. Tao W, Brunston RL, Jr., Bidani et al. Significant reduction in minute ventilation and peak inspiratory pressures with arteriovenous carbon dioxide removal during severe respiratory failure. Crit Care Med 1997; 25:689-695.

CHAPTER 6

Gas Exchange in the Venous System: Support for the Failing Lung Brack G. Hattler and William J. Federspiel

Introduction

A

cute respiratory failure from a variety of causes is one of the main concerns in the care of patients in intensive care units. One form of severe, acute respiratory failure, the acute respiratory distress syndrome (ARDS), although first described by Ashbaugh in 1967, continues to exhibit a mortality rate of approximately 40-50% in spite of a better understanding of the pathophysiology responsible for this condition.1,2 For the patient with acute respiratory failure, maintaining gas exchange at a level consistent with survival is associated frequently with progressively increasing levels of ventilator support. A review of our recent (1998) experience, with 1,839 patients who were placed on ventilators at the University of Pittsburgh, has indicated not surprisingly that mortality following intubation, oxygen delivery, and use of the respirator increases the longer the patient is ventilated (Fig. 6.1). Although a variety of diagnoses were responsible for initiating ventilator therapy, respiratory failure was listed as the primary cause of death in approximately 30% of patients. How use of the ventilator may contribute in a secondary manner to mortality is difficult to quantitate but may be more significant than has been appreciated and is addressed subsequently in this chapter. For the 799 patients within this group requiring ventilation for over 96 hours, a mortality of 37% was noted. These numbers are not unlike those found in other medical centers. The inability to accurately predict which patients will live and which patients will die from multiple organ failure and sepsis, versus those whose respiratory failure will contribute primarily to death, has led to an aggressive approach to respiratory support in this entire patient population. This approach is also based on the fact that mortality is the same regardless of whether the patient is classified as having simply acute lung failure or the more severe acute respiratory distress syndrome.2-5 Of the 600,000 people in the United States who undergo ventilator therapy on an annual basis, approximately 240,000 patients require ventilation for 96 hours or more.6,7 Costs to families, hospitals, and insurance carriers continue to escalate. Combined data for 90 medical centers in the United States for the year 1996 indicate a total cost of $3.8 billion with a mean mortality of 30.2% in patients on ventilators for 96 hours or greater at these institutions (Table 6.1).8 Taking these circumscribed numbers and expanding them on a national scale indicate that intensive care costs for patients on a ventilator reflect a disproportionately large amount of the health care dollar. Any new therapies that can reduce the need for or intensity of ventilator support, including earlier weaning from mechanical The Artificial Lung, edited by Steven N. Vaslef and Robert W. Anderson. ©2002 Eurekah.com.

134

The Artificial Lung

Fig. 6.1. Beginning at 48 hours after being placed on a respirator, a progressively increasing mortality is noted for the 1839 patients evaluated in the year 1998.

ventilation, should result in significant benefits in terms of patient suffering, patient ICU days, resource utilization, and earlier hospital discharge. These findings underscore the need for improved approaches to this patient population and constitute the basis for the continuing interest in the development of better ventilator strategies, new drug therapies, and artificial lung devices in the treatment of acute lung failure and in the management of patients on ventilators.9-15 This chapter describes progress in one of these areas as it relates to our development of an intravascular hollow fiber membrane lung assist device, the Hattler Respiratory Support Catheter or Hattler Catheter (HC) * that began with early prototype design and animal testing in 1984. These initial efforts were primitive compared to devices now being tested and involved bulky bundles of hollow fibers through which 100% oxygen would flow (Fig. 6.2). Both ends of each hollow fiber were potted into a single proximal manifold so that O2 and CO2 exchange would occur as the oxygen traversed the full length of the fiber. Kinking of the fibers at their distal tip was prevented by a wire loop. To avoid gas emboli, oxygen was vacuumed through the fibers. In vivo testing of these devices following surgical placement in the vena cava of dogs revealed that O2 and CO2 exchange occurred, but at levels (8-12 ml/minute) that were considered not to be of clinical value. It became apparent during these early experiments that passive flow of blood in the vena * Also reported in previous publications as the Intravenous Membrane Oxygenator (IMO) for use in intracorporeal membrane oxygenation (ICMO). The catheter based device, however, exchanges both oxygen and carbon dioxide. Simply describing it as a membrane oxygenator is not sufficient. It is intended for respiratory support in cases of acute respiratory failure.

Gas Exchange in the Venous System

135

Table 6.1. Mortality, charges, and length of stay for patients from 90 medical centers on continuous mechanical ventilation for greater than 96 hours 1 Length of Stay (days)

Charges ($)

Patients (N)

3,831,379,805

26,809

34.0

61,796,448

426

30.2

5.9

44,612,280

178

6.4

Total Mean

Mortality (%)

Standard Deviation

1Data compiled for the year 1996 from University Health System Consortium; UHC Services

Corporation; Oak Brook, IL.

cava, flow that was largely parallel to the hollow fibers, did not provide adequate gas exchange with these early prototypes. With this understanding, our efforts have evolved over the last 16 years with continued experimental work defining conditions that allow for improved gas exchange in the venous system.16-21 We have emphasized our own experience having previously reviewed the contributions of others, beginning with the first published report by JD Mortensen in 1987 detailing his work in the development of the IVOX (intravenous oxygenator).15, 22 The underlying concepts that are fundamental to gas exchange in blood have been instrumental in guiding research in this new field and have formed the basis for our approach in the development of the Hattler Catheter (HC), which stresses the science of bio-engineering as it applies to oxygen and carbon dioxide exchange in the venous system.

Background Patients who are placed on ventilators for over 24 to 48 hours have, by definition, some form of acute respiratory failure that may occur in a setting of a de novo pulmonary disorder (i.e., pneumonia) or against a background of chronic lung disease. A prime example of the latter would be patients with chronic obstructive pulmonary disease (COPD) who develop an exacerbation of airflow obstruction with acute CO2 retention (hypercapnic respiratory failure). The goals of mechanical ventilation in these two examples would be different with return of the lungs to their normal state in the case of the patient with pneumonia and support of the patient with COPD through the acute respiratory failure episode until he can be returned to his stable, chronically impaired pulmonary state. Modalities chosen for mechanical ventilation might vary depending upon whether the clinical presentation involved primarily hypoxemia or hypercapnia. The goal of mechanical ventilation, however, would be the same regardless of the patient on the respirator — that being reversal of the acute pulmonary process under conditions most favorable for healing of the lungs. In spite of this understanding, what has become only gradually apparent — initially through the early work of Kolobow and Gattinoni — is that iatrogenic injury caused by mechanical ventilation contributes to the morbidity and mortality of many ventilated patients.23 Further contributions over the past 20 years by other scientists in the field have culminated in the recent publication of a trial sponsored by the National Institutes of Health, demonstrating an improved survival for patients on ventilators suffering from acute respiratory failure.9 Patients were randomized to either “small breaths” (6 ml/

136

The Artificial Lung

Fig. 6.2. Device manufactured and tested in dogs in 1984. See text for details.

kg tidal volume) or “large breaths” (12 ml/kg tidal volume) with all other respiratory parameters remaining the same. A 22% improvement in mortality was noted in the “small breath” group when compared with the “large breath” group (31% vs. 40% respectively). Permissive hypercapnia was seen in the “small breath” group. Some of these critically ill patients could not tolerate the reduced ventilation because of the increased PCO2 levels. Respiratory acidosis with the use of lower tidal volumes has been noted by others.24,25 A PCO2 level greater than 70 mm Hg from a norm of 35-40 mm Hg can lead to an acute respiratory acidosis that is predominantly within the cell, potentially threatening cellular integrity. As tidal volumes fall, so does alveolar ventilation with a concomitant increase in the physiologic dead space. Bicarbonate administration and an increase in the respiratory rate do not fully correct for the progressive rise in PCO2 which can reach dangerously high levels. Hypercapnia aggravates and worsens hypoxemia as CO2 displaces O2 at the alveolar level. The patient experiences various cardiorespiratory changes including an increased heart rate, an increased respiratory drive, and elevated pulmonary artery pressures.26 Reducing the negative effects of acute hypercapnia would make reductions in tidal volume more universally feasible. In addition, the cellular effects of permissive hypercapnia may be partially compensated for by the improved hemodynamics associated with decreased tidal volumes and reduced airway pressures. The results of the NIH trial imply, therefore, that a further reduction in ventilation, in a setting where all patients would tolerate the therapy and the acute changes in PCO2 levels would not occur, should lead to a consistent and even greater reduction in ventilator requirements and in mortality for patients in acute respiratory failure. This statement is strengthened by the fact that prior trials examining decreased tidal volumes, but not the very low tidal volumes per ideal body weight used in the NIH trial, did not demonstrate a difference between the treatment groups.27-29 It was not until the early 1960’s, with the establishment of the first pulmonary intensive care units, that the pathology associated with respiratory barotrauma and pulmonary damage began to be clearly recognized, along with the role of oxygen in high concentrations as a toxic molecule often leading to progressive respiratory failure and death. Increasing concentrations of oxygen and delivered ventilator volumes and pressures produce in the patient and in experimental animals a noncardiogenic pulmonary edema. Histologically, damage occurs at the epithelial and endothelial cell level with mitochondrial swell-

Gas Exchange in the Venous System

137

ing, cytoplasmic disruptions, and nuclear degeneration. Platelet plugging and polymorphonuclear leukocyte (PMN) margination in the pulmonary capillary bed become increasingly prominent with gradual PMN spreading into the pulmonary parenchyma. Increased alveolar edema provides a setting for bacterial invasion, as does decreased ciliary action, tracheal-bronchial particle clearance, and surfactant inactivation.30 Thus, the over zealous use of the very therapy intended for the treatment of acute respiratory failure can itself promote further lung damage and eventual death. A return to reduced inspired oxygen concentrations and ventilator settings before a lethal stage is reached can result in healing, survival, and, in later stages, avoidance of pulmonary fibrosis. Patients who survive severe forms of acute respiratory failure (i.e., ARDS) and its treatment frequently do well and resume normal levels of activity while others demonstrate progressive pulmonary fibrosis.30 These findings have stressed the need for direct treatment strategies for acute lung failure that protects the lung not only from barotrauma and volutrauma but also from high partial pressures of oxygen. A lung-protective strategy may have benefits in areas other than reducing direct injury to the lungs. Studies both in animals and in man have shown that over-distension of atelectatic lung segments, with their contained alveoli, elicits an inflammatory response marked by increased pulmonary and systemic cytokine concentrations.31-33 Clinical studies support an association between elevated cytokine levels and an increased mortality in patients with acute respiratory failure on ventilators.31 A recent randomized, controlled trial in patients with the acute respiratory distress syndrome has shown that mechanical ventilation can induce an elevated cytokine response (IL-6, TNF-∃, IL-1%) in both broncho-alveolar lavage and plasma samples that is attenuated by a strategy to minimize overdistension and recruitment/derecruitment of the lung.9, 32 Similar results have been noted in animal studies.34 A correlation between elevated cytokine levels and multiple organ failure (MOF) has been reported.31 It is difficult however, in the human, to differentiate between elevated cytokine levels because of multiple organ failure or multiple organ failure contributed to, and exacerbated by, a primary pulmonary source for cytokine production.35,36 In the animal, however, these confounding variables are easier to control in a setting where the lungs begin in a normal state and are subjected to experimental manipulation while other organ systems are left relatively undisturbed. Here, time sequence-dependent variables can be discerned and indicate that mechanical ventilation can initiate an inflammatory response with ultimately systemic manifestations. These studies suggest that mechanical ventilation can contribute to the de novo development of multiple organ failure.9, 31 An equally important concern as it relates to multiple organ failure is the effect that mechanical ventilation has on the hemodynamic status and the microcirculation of these critically ill patients on respirators.37 Ventilators and ventilatory maneuvers, especially in the setting of severe respiratory failure where part of the therapy involves vigorous diuresis and a lowered intravascular volume, can have profound effects on the heart and circulation. Positive end expiratory pressure (PEEP) levels (10 cm H2O or greater) commonly employed in patients with acute respiratory failure lead to an increased intrathoracic pressure and a decrease in venous return with a concomitant drop in cardiac output and ejection efficiency. Increased pulmonary vascular resistance with an increased right ventricular afterload and a decrease in left ventricular compliance have also been reported under these circumstances. A primary concern surrounding changes in these parameters relates to what is happening with oxygen transport and oxygen content at the capillary level. Levels of PEEP of 10 cm of water have been reported to decrease oxygen transport in patients with

138

The Artificial Lung

acute respiratory failure on ventilators.38 The use of drugs to maintain the hemodynamic and cardiac status of the patient (i.e., epinephrine) becomes increasingly necessary and can lead to further vasoconstriction in the capillary beds, with a further decrease in tissue perfusion. A vicious cycle is, therefore, set in motion, where the end result is multiple organ failure and most frequently death. In Figure 6.3, the direct and indirect results of persistently high levels of mechanical ventilation are depicted. The patient on a ventilator, therefore, represents a highly complex set of physiological variables that are usually multi-factorial and reflect organ-system and patient-ventilation interactions. The ability to consistently and dramatically reduce ventilator support should reduce the incidence of death and its most frequent cause, multiple organ failure. Other detrimental effects of prolonged intubation and mechanical ventilation are more difficult to quantitate and fall into the realm of hospital acquired pneumonias.39 Prolonged intubation leads to airway microbial colonization and the risk of hospital acquired lower respiratory tract pneumonias. The risk of pneumonia increases the longer the patient is ventilated and mandates early extubation when clinically feasible. While pathological changes in the kidney, liver, and gastrointestinal tract can occur concomitantly with prolonged ventilator support, the most frequent and potentially lethal complications remain those of lower respiratory tract infections. A mortality greater than 10% has been reported with these ventilator associated pneumonias which lead to prolonged hospitalizations and greatly increased medical costs. The ability of the patient to be extubated, leave the intensive care unit, survive, and be discharged is significantly dependent, therefore, on the severity and number of complications associated with mechanical ventilation and the length of time a patient is on a ventilator. Extracorporeal membrane oxygenation (ECMO) arose out of this need to rest the lungs. Early experience with ECMO however, was disappointing and multi-center trials demonstrated a 90% mortality in patients receiving joint ECMO and mechanical ventilation—no better than the group treated with mechanical ventilation alone.40 For ECMO, an arteriovenous circuit was used at greater than 50% of cardiac output for blood flow to and from the externally positioned oxygenator, with an emphasis on oxygenation. A modification of ECMO, ECCO2R, incorporated a veno-venous circuit with flows at 20 to 30% of the cardiac output, with an emphasis on CO2 removal.41 Veno-venous ECMO allows all blood that is removed and oxygenated to be returned to the patient via the venous-return cannula. This setting may provide a unique opportunity not present with veno-arterial ECMO for the injured lung to be favorably exposed to an increased level of oxygenated blood at the tissue level. Maintaining blood flow to the lungs would also favor metabolic and endocrine functions ascribed to these organs. Continued blood flow to the lungs and heart chambers as determined by the patient’s cardiac output should also reduce any opportunity for thrombus formation within the vasculature of these organs. In the last 10 years, veno-venous access has been the support mode of choice at most major medical centers that are ECMO active. Intravenous gas exchange devices that are under development are essentially a simpler intravascular form of veno-venous ECMO and offer the same benefits. Both ECMO and ECCO2R have been included under the broader term extracorporeal life support (ECLS) and allow a reduction in ventilator settings, including ventilator frequency, ventilator volumes, and airway pressures. A combination of ECLS with a reduction in ventilator settings offers greater benefit than either therapy alone.42-45 With these newer approaches to ECLS, expected survival has increased from 10% to 50% in adults with acute respiratory failure. Recent lessons learned in caring for patients with acute lung failure teach that the earlier ECMO is instituted, the greater the opportunity

Gas Exchange in the Venous System

139

Fig 6.3. Direct and indirect detrimental effects of mechanical ventilation on patients. Adapted from references 31, 32, 33, 37.

140

The Artificial Lung

for a successful outcome. Bartlett and his colleagues have reported that the number preECLS days of mechanical ventilation is a significant predictor of outcome for survivors vs. nonsurvivors [2.7 days vs. 4.4 days respectively (P=0.0003)].45 A similar approach to early intervention with lung failure before multiple organs become affected would be applicable to lung assist devices such as the Hattler Catheter. Extracorporeal life support, however, is a complex, personnel intensive, costly therapy that can only be carried out in the intensive care units of major medical centers where these resources are available and where complications from the therapy can be addressed. It does not serve, therefore, the majority of patients in community hospitals who are in need of a safer, relatively simple, inexpensive, and readily accepted therapy to treat patients with acute respiratory failure who are at risk for progressive lung injury while on the ventilator. Our approach to this dilemma has been to define new forms of therapy as they pertain to intravenous gas exchange that can benefit the patient with acute respiratory failure no matter whether he is admitted to a University or a community hospital. Here it has been important to note that even severe forms of respiratory failure are characterized by areas of both relatively normal and diseased tissues interspersed throughout the lungs.46, 47 Intravenous devices currently under development are designed, therefore, to supplement without further injuring those areas of the lung that are functioning normally while diseased lung is allowed to heal. It is not anticipated at present that these devices will need to replace the entire lung function (mean resting exchange rates of 270 ml O2/min and 240 ml CO2/min). Based on the average percentage of the lung involved as seen on CT scans in the adult with ARDS and animal studies indicating the amount of residual lung required to meet basal requirements, the goal for the Hattler Catheter (HC) has been to provide at least 50% of basal O2 consumption and CO2 elimination (135 ml O2/min; 120 ml CO2/min).46-48 This might appear difficult because of the anatomical constraints imposed by those sites where an intravenous device will affect gas exchange (i.e., inferior and superior vena cava and right atrium) without compromising hemodynamics. This limits the total hollow fiber membrane that can be introduced into the venous system to about 0.5 m2 of fiber surface area. Certain conditions, however, can be created in a device that the natural lungs would not tolerate. The ability to maximize O2 diffusion by introducing 100% oxygen at atmospheric pressure and to precisely control a rapid gas flow rate for CO2 removal in hollow fiber membrane mats exposed to actively mixed blood over an extended blood path length has allowed us to reach a goal of 50% basal gas exchange rates and to establish certain basic principles as they apply in practice to the development of an intravenous membrane oxygenator. Oxygenators have been available for over 40 years as external devices developed for gas exchange — O2, CO2, and other gases, including anesthetics — during extracorporeal circulation as it pertains to heart surgery. To minimize blood trauma and the incidence of gaseous and particulate emboli, hollow fiber membranes as the gas exchange component have replaced bubble oxygenators over the last 20 years in their use in the operating room. Although the oxygenator provides a gas exchange function similar to the natural lungs, the natural lungs function at a level that far exceeds the capability of any artificial lung and provide a total surface area for gas exchange that cannot be approached by present engineering solutions. The role of pulmonary support, however, is not to provide the extraordinary level of gas exchange achieved by the natural lungs. Rather, a gas exchange device, whether it be external as used in ECMO and open heart surgery, or internal, as used in an intravenous device, is intended to meet those gas exchange requirements necessary to maintain the patient’s metabolic needs. For the patient during open-heart surgery, the oxygenator provides full gas exchange support during the cardiac procedure. For an

Gas Exchange in the Venous System

141

intravenous gas exchange device (the HC), only 50% of basal requirements can be met. It is therefore important to reemphasize that even in severe respiratory failure, injured tissue is interspersed with areas of normal lung, providing a setting where the patient can contribute to gas exchange through these functioning lung segments.46, 47 Partial support with the intravenous device is therefore available as a supplement to the patient while normal lung segments are gently ventilated, and the damaged and diseased areas of lung are allowed to heal.

Clinical Trial of an Intravenous Respiratory Support Device The only intravenous device for O2 and CO2 exchange that has undergone human clinical trials is the intravascular oxygenator (IVOX) developed by J. D. Mortensen.22,49 The IVOX underwent human implantations in the early 1990’s but was withdrawn from FDA consideration, partially because of the lack of concurrent controls and randomization in the trial, which failed to show improvement in survival compared with historical controls. The company that manufactured the IVOX is no longer operative. The device consists of a bundle of microporous hollow fiber membranes which are joined at the distal end of the device to the inner lumen of a double-lumen gas conduit and at the proximal end to the outer lumen of the gas conduit. An oxygen source is attached to the inner lumen and a vacuum pump to the outer lumen of the gas conduit, thus drawing oxygen through the hollow fibers. Devices tested clinically ranged in membrane surface area from 0.21-0.52 m2. The hollow fiber membranes of the IVOX are crimped along their length, a process which permanently bends a straight fiber in the pattern of a sine wave. The crimping provides an opportunity for secondary currents around the fibers. Otherwise, the device sits in the central venous system and is dependent upon passive flow past the fibers to promote gas exchange. The fibers are not constrained and randomly assume their position once placed in the venous system. The polypropylene microporous hollow fiber membranes of the IVOX have an ultra-thin siloxane coating which functions as a true membrane. It is permeable to oxygen and carbon dioxide but impermeable to water, thus preventing the membrane pores from being subject to plasma leakage. A second coating is applied to all components of the IVOX and is a covalently bonded heparin derivative that is intended to increase the thrombo-resistance of the device. The intravascular oxygenator (IVOX) is the only intravenous device to have undergone Phase I and Phase II human clinical trials in the early 1990’s.50 The Phase I trials established the safety for introducing an intravascular oxygenator in humans, and the Phase II trial examined the clinical efficacy and gas exchange performance for this device. A total of 160 patients from the United States and Europe were studied. Criteria for inclusion or exclusion in the Phase II trial are given in Table 6.2. Once entry criteria were met, patients underwent a right internal jugular or femoral vein cut-down for device implantation. The largest size IVOX (0.21-0.51 m2 membrane surface area) was chosen for implantation according to the vena cava size determined by ultrasound and the suitability of the access vein. The IVOX was furled during insertion and guided over a wire into position in the inferior vena cava, right atrium, and superior vena cava where unfurling occurred to fully expose the hollow fiber membranes to the returning venous blood. Heparinization (ACT 180-200 sec) was maintained during use of the device. After activation of the IVOX, attempts were made to reduce FIO2 and minute ventilation as long as blood gases could be maintained with arterial oxygen saturations over 90%.

The Artificial Lung

142

Table 6.2. Criteria for IVOX trial Inclusion

FIO 0.50 for 24 h or more with PaO2<60 mm Hg and one or more of the following: 2 Positive end-expiratory pressure Γ10 cm H2O Peak inspiratory pressure Γ45 cm H2O Mean airway pressure Γ30 cm H2O Minute ventilation >150 ml/min per kg with PaCO2>40 mm Hg Exclusion Uncontrolled multiple organ failure or patient in extremis Contraindication to systemic anticoagulation Uncontrolled bacteremia/fungemia Low cardiac output refractory to inotropes Lack of usable access vein (jugular or femoral) Existence of thrombi in major veins or vena cava Abnormal anatomy of the access veins or vena cava

The IVOX clinical trials confirmed earlier animal experiments and proved the feasibility of extended intravenous respiratory gas exchange in severe acute respiratory failure patients. In some patients, peak O2 and CO2 transfer rates varying between 40-70 ml/min were noted. During IVOX utilization, some clinical trial patients showed improvements in blood gas partial pressures and the ability to reduce the intensity of mechanical ventilation. The IVOX functioned in some patients for weeks without adverse effects. In these severely ill patients with acute respiratory failure, 60% survived to have the device removed, but only 30% improved to a point where mechanical ventilation could be discontinued. These same patients were hospital survivors and were discharged for an average 30% survival rate. Further trials of the IVOX were not pursued. Nevertheless, as the first and only clinical trial of an intravenous oxygenator, the IVOX provided valuable evidence that gas exchange (O2 and CO2) occurs in humans, but at a level that at the very best met only 30% of patients needs and was frequently not reproducible. Those involved in the clinical trial of the IVOX felt that in order to be clinically useful, an intravenous respiratory assist device would have to supply 50% of basal metabolic needs, be relatively easily insertible with minimal bleeding, and demonstrate consistent and reproducible gas exchange in patients with acute respiratory failure. These recommendations for improvements have been instrumental in our approach to designing devices dedicated to gas exchange in the venous system.

The Hattler Respiratory Support Catheter The Hattler Respiratory Support Catheter or Hattler Catheter (HC) is an intravenous hollow fiber membrane gas exchange device intended for support of the acutely compromised lung. The catheter is introduced at the common femoral vein or internal jugular vein site and placed within the venous system, spanning the superior vena cava, right atrium, and inferior vena cava, as illustrated in Figure 6.4. A console outside the body connected to an exterior fitting of the Hattler Catheter provides a sweep gas flow of 100% O2 through the hollow fiber membranes of the device. Early prototypes (Fig. 6.2) exchanged O2 and CO2 at levels that were considered too low to be of any clinical significance. To enhance gas exchange, a mechanism to control blood flow past the hollow fibers

Gas Exchange in the Venous System

143

Fig. 6.4. Insertion and placement of the Hattler Catheter in the venous system. An external drive console provides a vacuum for driving sweep gas flow through the catheter and also pulsates the balloon with helium gas.

144

The Artificial Lung

positioned in the vena cava was developed and consisted of a centrally placed balloon, around which the hollow fibers were circumferentially mounted. Inflation and deflation of the balloon created secondary flow profiles around the fibers. A shift in flow direction relative to the fibers from one that was predominantly parallel at zero balloon pulsation to one that was substantially transverse at 60-120 beats/minute was also noted.16 Biocompatibility of the HC is dependent on reducing shear stress and the potential for blood clotting and fibrin and cellular deposition on the surface of the hollow fiber membranes. A heparin bond linked to a < 1 ∝m thick siloxane coating covering the pores of the hollow fiber membrane (true membrane) has been used to enhance biocompatibility at the blood-membrane interface. In addition, early work in our laboratory has demonstrated that nitric oxide, as part of the sweep-gas within the hollow fibers, markedly diminishes cellular and platelet deposition on the membrane surface.51 Other than the use of specialized coatings and controlling the composition of the oxygen mixture being vacuumed through the hollow fibers, a general principle applied to the design of the Hattler Catheter has been to minimize shear stress as a means of promoting biocompatibility. Flow in the vena cava is largely laminar where shear stresses occur because adjacent fluid layers travel at different speeds. Fluid in the center of the stream moves more quickly than fluid near the vessel wall. Therefore, shear rate is highest at the wall, the point of maximum shear being where blood cells adjacent to the wall are not moving and lowest at the center where friction is minimal. Flow velocity profiles visualized in the laboratory have shown that balloon pulsation can disrupt the layer of fluid next to the vessel wall (area of high shear stress), therefore reducing overall shear and potential damage to the formed elements of blood.16 In addition, since, for maximum efficiency, the Hattler Catheter depends on blood flowing freely around the fibers where the blood is captured and directed towards the fibers by balloon pulsation, very low pressure drops (2-3 mm Hg) have been noted in large animal studies. By reducing the pressure drop, a reduction in shear at the same flow rates is noted, which promotes overall biocompatibility. The initial fibers employed with this device in our research laboratories were not constrained and consisted of up to a thousand hollow fiber membranes, each floating within the blood. Testing in vivo and in vitro demonstrated that gas exchange with balloon pulsation and active mixing had improved significantly. However, results with identical devices (same membrane surface areas) tested under identical conditions frequently varied significantly. The tendency of free fibers to clump together in bundles soon became apparent, especially when testing in blood. The end result of clumped fibers was a decrease in effective membrane surface area, as fibers in the center of each clump were not exposed to blood. An increase in the shunt fraction past the device would therefore occur with a decrease in gas exchange efficiency. Clumping, as a random, noncontrolled process occurred also with crimped fibers and could only be overcome by changing the hollow fiber membranes used with the device from nonconstrained, free fibers to fully constrained hollow fiber mats. The present configuration of the Hattler Catheter is depicted in Figure 6.5. The balloon occupies a central position and is surrounded by concentric layers of hollow fiber membrane mats that are potted into proximal and distal manifolds at each end of the device. The functioning components of the device sit at the end of a catheter through which 100% oxygen flows to the proximal manifold and is vacuumed from the distal manifold. A separate port provides helium to the balloon and is driven by a console capable of completely inflating and deflating the balloon at 300-600 beats/min. The device is inserted at present through a cut-down on the common femoral or internal jugular vein. When positioned, it occupies the inferior vena cava, right atrium, and superior vena

Gas Exchange in the Venous System

145

Fig. 6.5. Configuration of the Hattler Catheter. A pulsating balloon resides within the hollow fiber bundle, which is made using fiber fabric. The pneumatic delivery shaft directs the flow of incoming and outgoing sweep gas to the fibers and the oscillating flow of helium to the balloon.

cava. Reduction in the size of the catheter without losing gas exchange efficiency will permit percutaneous insertion in the future. In cases other than severe acute respiratory failure there exists the possibility that the ventilator can be dispensed with altogether. Since the resting CO2 production of a 70 kg adult is approximately 240 ml/min, the majority of CO2 can be readily removed by an intravenous respiratory assist catheter. Breathing for CO2 removal will be reduced proportionally, leading to a reduction in tidal volume, peak inspiratory pressures, and the respiratory rate. In the extubated awake patient, this level of alveolar ventilation can lead to arterial hypoxemia but a small increase in the inspired FIO2 should maintain normal alveolar oxygen concentrations. The ultimate effect on arterial oxygen tension however would depend on the degree of pulmonary arterio-venous shunting which becomes more severe with prolonged ventilatory support. This underscores the fact that the earlier a respiratory assist catheter is inserted the better the chance for a beneficial effect as it relates to both CO2 removal and oxygenation. From a different perspective, the ability of the intravenous respiratory assist catheter to increase its CO2 removal capacity as venous PCO2 rises under the influence of a reduced tidal volume is therefore important if the device is to be used to its full potential. The bioengineering principles that have been utilized in arriving at our present state of development are discussed in the following section.

Basic Principles of Gas Exchange in the Hattler Catheter The fundamental exchange units of the Hattler Respiratory Support Catheter are the hollow fiber membranes that compose its fiber bundle. Standard blood oxygenators used clinically during cardiopulmonary bypass employ thousands of these same fiber membranes in ensemble arrangements or bundles that vary among the oxygenator designs. Hollow fiber membranes are essentially small polymer tubes, typically with outer diameters from 200 to 400 ∝m, and with microporous walls of 20 to 50 ∝m thickness (see Fig. 6.6). The pores within the walls have characteristic dimensions typically below about 0.1∝m, and the porosity or volume fraction of pores within the fiber wall can vary from about 30% to 50%. The polymers selected for oxygenator fiber membranes are hydrophobic, or liquid repelling, so that normally the pores remain gas filled and allow for ready diffusion of respiratory gases through the fiber wall.

The Artificial Lung

146

The hollow fibers used in the Hattler Catheter exchange gas with blood following the same general principles as in other artificial lungs composed of fiber membranes. Oxygen gas flows through the insides of the hollow fibers while blood flows on the outsides of the hollow fibers through the interstices of the hollow fiber bundle. Oxygen then moves from the gas to the blood side of the membrane, while carbon dioxide moves from the blood to the gas side of the membrane, both down their respective diffusion gradients determined by the partial pressures of these gases on both sides of the membrane (i.e., the walls of the hollow fibers). The gas exchange permeance * or mass transfer coefficient, K, of an artificial lung represents the rate of gas exchange, for either O2 or CO2, normalized to fiber surface area and to the gas partial pressure difference driving exchange. Considering O2 as the species of interest, the gas exchange permeance is related to the overall O2 exchange rate, mO2 , by

[

]

m˙ o = K o 2 As ( Po 2 ) g ! ( Po 2 ) b , 2

(Eqn. 6.1)

where ( Po 2 ) g and ( Po 2 )b are the average O2 partial pressures in the gas and blood phases, respectively, contacting the membrane, and As is the total surface area of the fiber membranes. Accordingly, the gas exchange permeance provides a measure of the gas exchange efficiency of an artificial lung.52 For the Hattler Catheter, gas permeance has been a useful index for characterizing device performance and tracking progress during device development. The gas exchange permeance of an artificial lung for CO2 can be defined through a similar relationship. Overall gas exchange for O2 or CO2 in an artificial lung like the Hattler Catheter is dictated by diffusional resistances encountered serially as each species moves from the gas to blood phase, or vice versa. Indeed, permeance and resistance are inversely related, and hence the overall transfer resistance of an artificial lung is given by 1/K. This overall transfer resistance has two principal components: 1 1 1 (Eqn. 6.2) = + , K K m Kl where 1/Km represents the diffusional resistance of the hollow fiber membranes and 1/Kl represents the transfer resistance associated with diffusional boundary layers on the liquid-side (i.e., blood-side) of the fibers, which will be further discussed below. The permeances Km and Kl are the membrane and liquid-side gas permeances of the artificial lung, respectively. The general relationship of the membrane and liquid-side resistances to gas exchange in artificial lungs and in the Hattler Catheter is illustrated in Figure 6.7. The general gradients in partial pressure associated with these resistances are also illustrated for the case of O2. First, transfer resistance in the gas phase within the fibers is considered negligible because of substantial bulk gas flow through the fiber lumens, which minimizes the PO2 gradients within the fiber. Most of the transfer resistance and hence the gradient in PO2 is associated primarily with the liquid-side diffusional boundary layer, and second-

* "Permeability" is often used in the perfusion literature in place of "permeance" to describe gas transfer features of blood oxygenators or the membranes composing them. "Permeability" is more correctly used to describe a material property of polymers relating to their gas transmissibility, and is independent of geometrical configuration and thickness. "Permeance" is the most appropriate term used to describe the gas transmission properties of a given material or system with a given geometrical configuration and membrane thickness.

Gas Exchange in the Venous System

147

arily with diffusion through the membrane itself. Thus, liquid-side and membrane gas permeances dictate overall gas exchange in the Hattler Catheter and represent serial transport processes, in which their resistances add directly to determine overall resistance (Eq. 6.2). The serial resistance relationship means that the overall gas exchange permeance, K, is always less than the smallest of the permeances, Km and Kl. More simply, the transport process with the smallest permeance or largest resistance controls gas exchange in the Hattler Catheter, as it does in any artificial lung. In what follows, we introduce some basic transport concepts to help understand the nature of the membrane and liquid-side gas permeances of an artificial lung like the Hattler Catheter.

Gas Permeance of Microporous Hollow Fibers Microporous hollow fiber membranes are the most common hollow fibers used for artificial lung design, especially in the standard blood oxygenators used in cardiopulmonary bypass. Microporous fiber membranes have fixed submicron pores within the wall that are contiguous from outer to inner lumen (see Fig. 6.6), and gas exchange occurs by gas diffusion through these pores. Generally, the polymer material does not dictate the exchange performance of the fiber as much as the pore characteristics and the fiber wall porosity, as will be further discussed below. An important polymer-related property of hollow fibers is their hydrophobic versus hydrophilic nature. Hydrophobic fibers are used in artificial lungs because they prevent intrusion of blood plasma into the fiber pores under normal conditions, a process known as plasma or liquid wetting.53,54 The principal manufacturers of microporous hollow fiber membranes for artificial lungs are Celgard (Charlotte, NC), Akzo-Nobel in Germany, and Mitsubishi Rayon in Japan. Currently, the Hattler Catheter under research development uses the Celgard X30-240 microporous fiber, with slit pores of 0.03 ∝m by 0.2 ∝m, and a nominal porosity of 40%. The Hattler Catheter made for ultimate human implantation will use similar microporous hollow fibers modified to prevent plasma wetting and to improve biocompatibility, as will be discussed below. The resistance to gas exchange offered by a microporous hollow fiber membrane, and its potential effect on overall gas exchange in an artificial lung like the Hattler Catheter, depends on the permeance, Km, of the fiber membrane (see Eqs. 6.1 and 6.2). Fiber permeance values, however, are not usually reported in the specifications provided by the fiber manufacturers. A likely reason is that Km values for microporous hollow fibers are relatively large, Km >> Kl, so that membrane resistance is negligible compared to liquid-side resistance in an artificial lung, thus obviating the need for a design specification and value for Km. A simple theoretical relation can be derived to estimate Km values for microporous fibers. Gas diffusion through the pores of oxygenator fibers is governed by Knudsen diffusion, where interactions of gas molecules with the pore walls dominate molecular motion.55 Assuming Knudsen diffusion, a microporous membrane of porosity, Η, composed of idealized cylindrical pores of diameter, dp, would have a membrane permeance given by:

2 dp 2 (Eqn. 6.3) , 3 +h #MRT where h is the fiber wall thickness, + is the pore tortuosity, M is the molecular weight of the gas, R is the gas constant, and T is absolute temperature. For parameter values approximating the Celgard X30-240 fiber: Η = 0.40, dp = 0.03 ∝m, and h = 30 ∝m, the estimated membrane permeance* for O2 at 37˚ C is Km = 2.3 x 10-2 ml/(cm2•s•cm Hg), assuming a Km = Η

148

The Artificial Lung

Fig. 6.6. Scanning electron micrograph of microporous hollow fiber membranes (top panel). The walls of hollow fibers contain submicron size pores (bottom panel) for ready transmural diffusion of respiratory gases.

Gas Exchange in the Venous System

149

pore tortuosity of + = 3. For these same conditions, the estimated membrane permeance for CO2 is Km = 2.0 x 10-2 ml/(cm2•s•cm Hg). A membrane permeance of Km ~ 10-2 ml/(cm2•s•cm Hg) represents an enormous gas exchange capacity which is never realized in an artificial lung. For example, an artificial lung with 0.5 m2 of membrane area, as in the largest anticipated Hattler Catheter device, and with a 700 mm Hg PO2 difference held between the gas and blood phases, could hypothetically transfer approximately 200 l/min of O2 across the membrane if Km dictated exchange. This exchange rate is not only much greater than needed to saturate blood flowing at physiological flow rates, but is several thousand fold greater than actually occurs because the predominant resistance to gas exchange resides within the blood phase, in the diffusional boundary layers adjacent to the fiber wall surfaces. Hence, the overall gas exchange permeance of a Hattler Catheter, as for any artificial lung, is substantially less than the permeance of the hollow fiber membranes comprising the device. The permeance of microporous hollow fibers can be measured with the fibers immersed in a gas rather than liquid phase (i.e., a gas-gas system), so that liquid-side boundary layers are not present and all the transfer resistance is associated with the membrane. If Knudsen diffusion conditions exist within the pores, convective gas flow across the fiber wall due to a transmural pressure difference can be ignored.55 Kamo et al.56 measured the oxygen permeance of the Mitsubishi KPF fiber in a gas-gas system and reported a Km = 6.7 x 10-2 ml/(cm2•s•cm Hg). Lund57 determined Km values of 1.72 x 10-2 and 1.47 x 10-2 ml/(cm2•s•cm Hg) for O2 and CO2, respectively, at room temperature for the Celgard X30-240 fiber. In the Lund studies, the measured Km were independent of the absolute pressure difference applied across the fiber walls, indicating that convective gas flow through pores was not important and that Knudsen diffusion dominated transfer. The ratio of membrane permeance to CO2 versus O2 was 1.47/1.72 = 0.85, which is also consistent with Knudsen diffusion of these gases through the fiber (see Eq. 6.3). Thus, even with microporous fibers immersed within a gas rather than liquid phase, gas movement across the fiber wall occurs by Knudsen diffusion, the same mechanism that would occur if the fibers were immersed in a liquid. Under these conditions fiber permeance measured in a gas-gas system should reflect that when fibers are immersed within a liquid. Can membrane permeance be reduced or otherwise changed by the fibers being immersed in a liquid, especially a more complex liquid like blood, in which proteins and cellular elements may adsorb to the fiber surface and affect membrane permeance? While an interesting question, the large Km of microporous hollow fibers makes it difficult to measure Km when the fibers are immersed in a liquid phase (i.e., a gas-liquid system). In a gas-liquid system, the Km measurement must account for the liquid-side permeance, Kl, due to diffusional boundary layers on the fiber surfaces, even in the presence of substantial liquid-side mixing. Lund et al.58 developed a diffusion chamber to measure hollow fiber membrane permeance in various liquid media, including water, plasma, and blood. By systematically increasing the stirring rate of the liquid bathing the fibers, the effect of the liquid boundary layer can be sequentially reduced, and a value of Km can be estimated by extrapolating to an infinite level of stirring. Lund57 demonstrated that Km can be determined in this manner provided its value is no more than 10 times the maximum Kl engendered by the applied mixing. In the diffusion chamber developed by Lund et al., the largest Kl value reached at a stirring rate of 3000 RPM was less than 1 x 10-3 ml/(cm2•s•cm Hg) for CO2 and less than 1 x 10-4 ml/(cm2•s•cm Hg) for O2. Hence, reliable determina* The units used for fiber permeance are conventional, where gas volume is referenced to STP conditions.

150

The Artificial Lung

Fig. 6.7. Membrane and liquid-side gas permeances, Km and Kl, respectively, dictate gas exchange from the gas phase flowing inside hollow fibers to the blood phase flowing outside the fibers. The "membrane" refers to the walls of the hollow fibers composing an artificial lung.

tion of a microporous fiber Km in the gas-liquid system would be difficult for CO2 and impossible for O2 unless liquid immersion causes substantial reductions in Km. Plasma wetting of the fiber pores is one potential mechanism for substantially reducing the membrane permeance of microporous hollow fiber membranes immersed in blood. Plasma wetting and the composite hollow fibers used to prevent plasma wetting are discussed later, after the principal concepts underlying the liquid-side permeance of the Hattler Catheter are introduced.

Liquid-Side Boundary Layer Permeance Fluid elements flowing adjacent to a stationary surface experience a decrease in velocity, eventually becoming stagnant immediately at the surface. The liquid or blood-side permeance of an oxygenator, Kl, accounts for gas diffusion away from or towards the fiber surface through these convection-poor and imperfectly-mixed boundary layers adjacent to the fiber surface. The gas species must traverse the boundary layer by molecular diffusion before it can be exposed to sufficient convection to be carried away by the flowing stream of blood past the fiber. The boundary layer concept is illustrated in Figure 6.8, which shows the diffusional boundary layer associated with uniform laminar flow past a flat membrane surface. The boundary layer adjacent to the surface, where transport occurs predominantly by diffusion, grows with distance along the surface in the flow direction. Even a submicron-sized diffusional boundary layer can result in a Kl appreciably below the

Gas Exchange in the Venous System

151

Fig. 6.8. Illustration of growing diffusional boundary layer on flat membrane surface with convective flow past the surface. Transfer to and from the membrane surface is predominantly by diffusion within the boundary layer, while convection by bulk flow predominantly carries gas species outside the boundary layer. (ϑ=fluid velocity past membrane surface; L=length of surface along direction of flow; Ι(x)=boundary layer thickness with position, x, along membrane surface; Pb and Pw=gas partial pressures in bulk fluid stream and at the membrane surface, respectively.)

Km for a microporous hollow fiber membrane, and can limit overall gas exchange, as will be further discussed below. The permeance of a liquid-side boundary layer can be expressed generally as

∃ l Dl (Eqn. 6.4) , Ι bl where ∃1 and D1 are the effective solubility and diffusion coefficient, respectively, of the diffusing gas in the liquid phase and Ιbl is the average boundary layer thickness (more specifically, the harmonic-mean thickness). Equation 6.4 is exact for simple liquids like water, but also describes gas diffusion within blood if the solubility coefficient effectively accounts for increased solubility due to hemoglobin binding for O2 or storage by the bicarbonate system for CO2. Essentially, the liquid boundary layer thickness, Ιbl, represents an average distance that O2 must diffuse away from the fiber before being swept away by the bulk flow of blood. For CO2, Ιbl represents the average distance for diffusion to the fiber surface after being carried to the fiber by the flow of blood. The value of Ιbl depends principally on the local-scale interaction between diffusional fields and flow dynamics in the blood phase subjacent to the fibers, the latter being complicated in the Hattler Catheter. Active mixing of blood by the pulsating balloon in the Hattler Catheter drives additional convective blood flow across the fibers to increase convective mixing, minimize the diffusional boundary layer thickness, and increase the liquid-side gas permeance, Kl. How the liquid-side permeance depends on boundary layer growth and thickness has important implications for the design of artificial lungs like the Hattler Catheter, and the theory for the diffusional boundary layer arising along a flat surface (as in Figure 6.8) provides important insight. The diffusional boundary layer on a flat surface grows as the square root of distance, x, along the surface according to: Kl =

The Artificial Lung

152

Fig. 6.9. Comparison of longitudinal flow (parallel flow) and transverse flow (perpindicular or cross flow) past hollow fibers. Diffusional boundary layers are smaller in cross-flow than in parallel flow leading to greater liquidside gas permeance, Kl. (ϑt and ϑl=fluid velocity past fiber in either transverse or longitudinal flow direcions, respectively; d=fiber outer diameter; l=fiber length.)

(Eqn. 6.5) where Κ is the liquid kinematic viscosity, Dl is the gas diffusion coefficient in the liquid and vb is the bulk flow velocity past the surface, essentially that of the fluid elements undisturbed by the surface. An important implication of boundary layer growth is that considerable advantage can be gained from orienting the exchange surfaces (i.e., the hollow fibers) perpendicular rather than parallel to the bulk convective blood stream (Fig. 6.9). With perpendicular or transverse flow to the hollow fibers, the fiber surface along the flow direction scales as the diameter, d, of the fibers, while that with parallel or longitudinal flow to the hollow fibers scales as the length, l, of the fiber, as illustrated in Figure 6.9. Accordingly, for a given gas species and liquid-side velocity, boundary layer theory (i.e., Eqs. 6.4 and 6.5) predicts that Kl for transverse versus longitudinal flow would scale as: K ltran

l (Eqn. 6.6) . d For fiber lengths and diameters characteristic of an intravenous artificial lung like the Hattler Catheter, with l/d ~1000, Equation 6.6 suggests that transverse blood flow, or cross-flow to the hollow fibers, improves gas exchange permeance approximately thirty-fold. K llong

)

Gas Exchange in the Venous System

153

The pulsating balloon in the Hattler Catheter improves gas exchange partly by generating more cross-flow to the fiber membranes than might otherwise exist in the vena cava and thereby capitalizing on the mass transfer advantages afforded by cross-flow. The boundary layer paradigm also suggests that balloon enhancement of gas exchange will increase as the square root of balloon pulsation, since the (balloon-engendered) flow velocity depends linearly on balloon pulsation rate, and boundary layer thickness depends inversely on the square root of flow velocity. The liquid-side gas permeance, Kl, associated with diffusional boundary layers usually limits gas exchange in artificial lung devices, and it does so in the Hattler Catheter. This is not surprising if one considers that Kl for O2 diffusing in water through a 10 ∝m thick boundary layer is about 9 x 10-6 ml/(cm2•s•cm Hg) (at 37°C for O2 in water: ∃1 = 3.2 x 10-4 ml/(cm3•cm Hg) and Dl = 2.8 x 10-5 cm2/s). The Kl in blood may be several times greater due to increased O2 solubility provided by hemoglobin, but still remains in the range of 10-5 ml/(cm2•s•cm Hg). The Kl value for CO2 diffusing through a boundary layer of the same thickness is on the order of 10-4 ml/(cm2•s•cm Hg). In contrast, a microporous hollow fiber membrane has a Km value in the range of 10-2 ml/(cm2•s•cm Hg), as discussed previously. Thus, a diffusional boundary layer of just 10 ∝m, only a fraction of the fiber size, yields O2 and CO2 liquid-side permeances appreciably below those of the fiber membrane, hence, Kl << Km. Moreover, increasing Kl by generating smaller-scale mixing is difficult, even with rapid balloon pulsation in the Hattler Catheter. Thus, under normal conditions in an artificial lung and in the Hattler Catheter, the overall exchange permeance is dictated by the diffusional boundary layer, or K Λ Kl,

(Eqn. 6.7)

Only under conditions of substantially reduced Km would the gas exchange permeance of the hollow fibers have an appreciable influence on overall artificial lung performance. Plasma wetting of the fiber pores is one potential mechanism for substantially reducing the membrane permeance of microporous hollow fiber membranes. For this reason, the permeance of fiber membranes assumes greater importance in the development of an implantable artificial lung like the Hattler Catheter, which must effectively resist plasma wetting. The fibers ultimately used in the Hattler Catheter will be composite hollow fiber membranes, with thin nonporous polymer coatings applied to the fibers to prevent plasma wetting. The nonporous polymer layer of a composite fiber reduces the membrane permeance substantially. If Km becomes comparable to Kl the overall gas exchange permeance of the artificial lung will also be reduced (see Eq. 6.2). For these reasons, studying the membrane permeance of composite hollow fibers has been central to development of the Hattler Catheter. The following section further discusses membrane permeance in the Hattler Catheter in relation to plasma wetting and the use of composite hollow fibers designed to prevent wetting.

Composite Fiber Membranes and Membrane Permeance Microporous fiber membranes are made from hydrophobic polymers (e.g., polypropylene) so that pores stay gas filled and membrane permeance remains high. Nevertheless, membrane permeance degrades significantly when these fibers are exposed to blood due to plasma wetting, whereby blood plasma infiltrates the microporous walls of hollow fibers. Plasma wetting is a common problem in the extended use of extracorporeal oxygenators,53,54 which can lead to device failure within days. Plasma wetting results primarily from phos-

The Artificial Lung

154

Fig. 6.10. Membrane permeance, Km, associated with dry (gas-filled) fiber wall pores versus liquid-wetted fiber wall pores. Pore wetting markedly reduces Km, even with partial wetting of fiber pores.

pholipids, lipoproteins and/or proteins in blood53 which can adsorb onto the fiber polymer surfaces at the plasma interface, rendering the interface hydrophilic and allowing for wetting of the pores by either partial or complete plasma infiltration. In contrast, the membrane permeance of hollow fibers immersed in pure water has been shown to remain constant for days under flow and pressure conditions applicable to blood oxygenators, thus indicating that fiber pores remain gas filled in simple liquids like water.59 Liquid infiltration into gas filled pores markedly diminishes the membrane permeance, Km, because rapid gas phase diffusion is replaced by diffusion in a liquid, as illustrated in Figure 6.10. For entirely liquid filled pores (i.e., complete wetting), the membrane permeance, Km, is given by:

∃ l Dl (Eqn. 6.8) , +h where, ∃1 and D1 are the solubility and diffusion coefficient of gas in the liquid filling the pores, and Η, h and + are the fiber wall porosity, thickness and pore tortuosity, which also affect the Km of gas filled microporous fibers (see Eq. 6.3). Thus, complete wetting reduces Km markedly because the gas diffusion coefficient in any liquid is substantially less than in a gas, and because the solubility of the gas in a liquid (i.e., the dissolved gas mass per unit mass of liquid) is small. A typical Km value for a completely wetted fiber, assuming waterfilled pores and the same fiber parameters used previously, is in the range of 10-7 ml/ (cm2•s•cm Hg) for O2, a 100,000-fold decrease compared to Km for gas-filled pores. This marked effect of wetting means that even partial liquid infiltration into fiber pores can significantly reduce Km. In the previous example, a 1% penetration of liquid into the gas filled pores would reduce Km to about 10-5 ml/(cm2•s•cm Hg), still a marked reduction in membrane permeance compared to that if the pores were gas-filled. Km in this situation would be comparable to liquid-side permeance, thus affecting overall gas exchange. The fiber membranes selected for the Hattler Catheter, as for other implantable artificial lungs, must effectively resist plasma wetting and the associated reduction in gas exchange performance. Unlike extracorporeal artificial lungs, replacing the Hattler Catheter after plasma wetting is not feasible nor desirable. Although a strategy to resist plasma Km = Η

Gas Exchange in the Venous System

155

Fig. 6.11. Composite hollow fibers incorporate a nonporous polymer layer, or true membrane, with a microporous polymer wall to block fiber wetting. The microporous polymer provides structural support for the true membrane, which dictates diffusion through the fiber wall.

wetting might involve using fibers with reduced pore size, preventing plasma wetting requires using composite hollow fiber membranes. Composite hollow fibers incorporate a thin nonporous polymer layer, or “true” membrane that typically coats the outside of a microporous fiber membrane, as indicated in Figure 6.11. The nonporous polymer layer, if contiguous on the outside of the fiber, effectively blocks liquid intrusion into the microporous wall. Composite hollow fiber membranes are made by either coating an existing microporous fiber with a thin nonporous polymer (a true composite hollow fiber), or by modifying the fabrication of the microporous fiber itself to seal-off pores at the surface (an asymmetric hollow fiber). Figure 6.12 shows scanning electron micrograph images of a composite hollow fiber as compared to an asymmetric hollow fiber, both shown in cross-section of the outer fiber wall. Although asymmetric hollow fiber membranes have manufacturing advantages, the same polymer constitutes the nonporous and microporous sections of the fiber wall, and the nonporous polymer may not have suitable gas transmission properties. Conversely, composite hollow fibers can use highly permeable nonporous polymers for the true membrane, but layer contiguity and potential delamination become important issues. Figure 6.11 also illustrates a composite hollow fiber membrane with the true membrane sandwiched within the microporous wall, a composite fiber previously available from Mitsubishi Rayon (MHF fiber series). This composite fiber may prevent plasma seepage into the fiber lumen, but if partial wetting of the outer

156

The Artificial Lung

Fig. 6.12. Scanning electron micrographs showing showing cross-sections: the outer fiber wall for a composite (coated) hollow fiber membrane (top panel), and an asymmetric hollow fiber membrane (bottom panel).

Gas Exchange in the Venous System

157

microporous wall segment can occur, the membrane gas permeance will be substantially reduced, as described previously. The very strategy meant to block plasma wetting in composite hollow fiber membranes diminishes membrane permeance, since the nonporous polymer layer presents an appreciable relative impediment to gas diffusion. Indeed, the membrane permeance of a composite hollow fiber is generally dictated by the true membrane, since nonporous or dense polymers have low gas permeances compared to the large permeance of the gas-filled microporous wall with which the nonporous polymer layer is in series. Like liquid layers, gas permeation across a nonporous polymer layer involves gas solubility within the polymer molecular matrix and subsequent diffusion through its matrix. Thus, the macromolecular properties of the polymer material used for the true membrane play a dominating role in membrane permeance. The membrane permeance of a composite hollow fiber is given by:

∃ p Dp

Pm (Eqn. 6.9) , Ιp Ιp where ∃p and Dp are the solubility and diffusivity of the gas within the dense polymer layer and Ιp is its thickness. Most often, polymer manufacturers report the product of polymer solubility and diffusivity, the polymer permeability, Pm, to specific gases rather than the solubility and diffusivity individually. Typical gas permeabilities of “normal” permeability polymers are in the range of a Barrier, which is the commonly used unit of polymer permeability, where 1 Barrier = 10-10 ml/(cm2•s•cm Hg). Thus, according to Equation 6.9, the membrane permeance of a composite hollow fiber with a 1 ∝m thick layer of a 1 Barrier nonporous polymer is Km = 10-6 ml/(cm2•s•cm Hg), a relatively small and potentially gas exchange limiting membrane permeance. Composite hollow fiber membranes suitable for artificial lung applications ideally require nonporous polymers with relatively high gas permeabilities in the range of 100 Barriers or greater, and which can be coated in stable contiguous layers in the range of 1 ∝m thickness or less. Many of the siloxane polymers, which have permeabilities of 100 Barriers or greater, have been and still remain attractive candidates for fabricating composite hollow fiber membranes. The selection or development of an appropriate composite hollow fiber membrane for the Hattler Catheter requires a fiber Km large enough compared to liquid-side gas permeance, Kl, so that the impact on overall gas exchange is minimal. Appropriate initial Km target values can be estimated using a criterion that the transfer resistance offered by the fiber membranes be less than 10% of the overall gas transfer resistance, or in permeance terms: Km . 10K. To use this criterion, the appropriate overall permeance, K, of the Hattler Catheter is estimated by rearranging Equation 6.1 and applying our design target for gas exchange for O2 and CO2 along with estimates of average gas-side and blood-side partial pressures. For example, targeting a gas exchange rate for O2 and CO2 of 125 ml/min with a 0.5 m2 Hattler Catheter, and assuming the following gas and liquid-side partial pressures for O2: ( Po 2 ) g = 660 mm Hg and ( Po 2 )b = 40 mm Hg, and for CO2: -6 ( Pco 2 ) g = 0 mm Hg and ( Pco 2 )b = 50 mm Hg, yields overall permeances of KO2 = 7 x 10 2 -5 2 ml/(cm •s•cm Hg) and KCO2 = 8 x 10 ml/(cm •s•cm Hg). Thus, the adopted criterion for membrane permeance suggests target Km values of : Km Γ 7x10-5 ml/(cm2•s•cm Hg) for O2 Km Γ 8x10-4 mlml/(cm2•s•cm Hg) for CO2 . Composite hollow fiber membranes with permeances appreciably below the target Km will engender significant reductions in overall gas exchange performance of the Hattler CathKm =

=

The Artificial Lung

158

Table 6.3. Membrane permeance of candidate composite hollow fiber membranes Km in ml/(cm2•s•cm Hg) Fiber

Oxygen

Carbon Dioxide

Senko TMCTS DIC PMP CMS 7 HC Target Values

1.7x10-4 1.8x10-4 2.3x10-3 >7x10-5

4.2x10-4 2.1x10-4 4.3x10-3 >8x10-4

eter for the specific gas. Ideally, given the approximate nature of the estimated Km targets above, appropriate composite hollow fiber membranes for the Hattler Catheter should incorporate a design safety factor and possess membrane permeances 2 to 3 times larger than given above. Not surprisingly, considering our previous estimate of composite fiber membrane permeance associated with a “normal” permeability nonporous polymer, the identification and/or development of a suitable composite hollow fiber for the Hattler Catheter, along with associated studies of membrane permeance, have been central components of our catheter development effort. Despite its central importance, not much information exists on membrane permeances for composite hollow fiber membranes suitable for the Hattler Catheter. Kamo et al.56 measured the O2 membrane permeance for the Mitsubishi Rayon MHF series of composite hollow fibers and reported Km values from 0.9 x 10-6 to 1.6 x 10-6 ml/(cm2•s•cm Hg), significantly below the membrane permeance target for the Hattler Catheter. Although the MHF composite fibers would not be suitable for the Hattler Catheter and would significantly affect overall gas exchange, the relatively small Km of the MHF series is not surprising given the “normal” permeability (~ 1 Barrier) of polyurethane used for the nonporous polymer layer. Our laboratory has conducted its own extensive studies of membrane permeance of candidate composite hollow fibers for the Hattler Catheter, using a gas-liquid measurement system described previously along with a simpler gas-gas measurement system.57, 58 Candidate composite hollow fibers include 1) Senko TMCTS Coated Fiber (Senko Medical Instruments Manufacturing Company, Japan), a Celgard x30-240 microporous fiber coated with a plasma polymerized layer of tetramethylcyclotetrasiloxane (TMCTS); 2) DIC PMP Fiber (Dainippon Ink and Chemical, Japan), an asymmetric fiber constructed from a microporous poly-4-methyl-pentene-1 (PMP) with sealed outer pores (see Figure 6.12); and 3) CMS 7 Fiber (Compact Membrane Systems, Wilmington, DE), a proprietary coating of a Teflon copolymer on a Celgard x30-240 fiber. Table 6.3 compares membrane permeance values determined from gas-gas measurements on these composite hollow fibers57 with the target membrane permeances for the Hattler Catheter. All candidate composite fibers achieve suitable Km for O2, but the target Km value for CO2 is more difficult to achieve and becomes the limiting constraint. The membrane permeance of the Senko fiber for CO2 is about half the Km target, which suggests that the Senko fiber in the Hattler Catheter could effect a 10%-20% reduction in overall CO2 gas exchange. The only composite hollow fiber satisfying both the O2 and CO2 targets for membrane permeance is the CMS fiber, with Km values at least 4 times the targets. Ultimately, the selection of a suitable composite hollow fiber for the Hattler Catheter will not only consider membrane gas permeance, but will also weigh manufacturing issues, biocompatibility (both natural biocompatibility and the suitability for surface modification of the fiber, e.g., with heparin attachment), and long term stability.

Gas Exchange in the Venous System

159

Fig. 6.13. Effect of sweep gas flow rate on carbon dioxide exchange. Solid lines based on predictions from simple model and symbols are experimental data on an extracorporeal blood oxygenator.

Role of Gas-Side Flow and Pressure Gas flow through the Hattler Catheter does not directly affect the overall gas permeance, K, of the device because, as discussed previously, the overall permeance is dictated principally by liquid-side gas permeance, conditioned by diffusional boundary layers at the fiber surfaces, and secondarily by membrane permeance. Nevertheless, the dynamics of sweep gas flow through the Hattler Catheter do affect overall gas exchange by a direct effect on gas-side partial pressures of O2 and CO2, and by the resulting impact on the partial pressure driving force for exchange of these gases (see Eq. 6.1). The rate of gas flow through the Hattler Catheter principally affects the CO2 exchange rate, while the pressure drop across the fiber bundle arising from the gas flow principally affects the O2 exchange rate. The dependence of CO2 exchange on gas flow rate arises because the partial pressure of CO2 in blood available for driving exchange is relatively low, approximately 40 to 60 mm Hg, and hence even a small buildup in CO2 within the sweep gas pathway can markedly diminish the PCO2 gradient between blood and gas flowing through the device. Thus, a direct relation exists between the actual rate of CO2 exchange, the gas flow rate through the device, and the degree to which insufficient gas flow may be reducing CO2 exchange. We previously developed a simple model describing the interrelationship between gas flow rate and CO2 exchange rate in an artificial lung like the Hattler Catheter.60 The model uses appropriate normalization to arrive at a universal relation between gas flow rate and CO2 exchange that is independent of device specific mass transfer features of the artificial

The Artificial Lung

160

lung, e.g., the overall device permeance K. The effect of sweep gas flow rate, Qg, on CO2 exchange, (m˙ co )max , is given by: 2

(Eqn. 6.10)

where (m˙ co )max is the maximum possible rate of CO2 exchange in the limit of infinite gas 2 flow, Pg,o is the outlet pressure of the sweep gas, and ( Pco )b is the mean blood side partial 2 pressure of CO2 within the fiber bundle.61 This simple theoretical relation, shown graphically in Figure 6.13, helps delineate regimes between relatively gas flow dependent and gas flow independent CO2 exchange. At a ( Pco )b of 50 mm Hg, the CO2 exchange rate is 2

within 15% of the maximum attainable exchange when

which has become

a useful criterion for specifying required gas flow rates of the Hattler Catheter. For example, if a CO2 exchange rate of 125 ml/min is a nominal target value for a specific Hattler Catheter, then the gas pathways of the device need to be designed for sweep gas flows exceeding 50 x 125 ml/min or 6.3 l/min. Otherwise, the reduction in CO2 exchange by insufficient gas flow will be appreciable. The theoretical prediction of the effect of gas flow rate on CO2 exchange (i.e., Eq. 6.10) was recently validated experimentally for standard extracorporeal blood oxygenators.61 As indicated in Figure 6.13, the experimental relation between CO2 exchange and gas flow rate (squares) agreed well with the theoretical predictions based on the ( Pco )b of 60.2 mm Hg measured during the CO2 exchange ex2 periment. In the same gas flow rate range where CO2 exchange is dependent on gas flow rate, the O2 exchange rate is essentially independent of gas flow rate per se. This is so because the buildup in CO2 within the sweep gas, although significant from the perspective of the gradient in PCO2 between blood and gas phase, has a much smaller impact on the gradient in PO2 between blood and gas phase. In this gas flow rate range (essentially the one shown in Figure 6.13), the sweep gas remains more than 95% O2. Hence, the principal determinant of the partial pressure of O2 in the gas-phase, and the PO2 gradient between blood and gas, is the average total gas pressure in the fibers. Thus, although gas flow rate does not directly affect O2 exchange, it does so indirectly because of the pressure drop engendered by the gas flow through the fiber bundle. The Hattler Catheter operates at subatmospheric (vacuum) pressures and has been designed with a negligible inlet sweep gas pressure drop, so that the sweep gas enters the fiber bundle at approximately atmospheric pressure. Accordingly, the average gas-side partial pressure of O2 in the Hattler Catheter (vacuum-driven gas flow), ( Po 2 ) g can be approximated by: ( ∀P ) g

(Eqn. 6.11) , 2 where (∀P)g is the gas-side (intrafiber) pressure drop, which is dependent on gas flow rate, fiber lumen size and number of fibers within the bundle. The fiber bundle pressure drop can be predicted accurately based on compressible Poiseuille flow theory through parallel tubes.62 ( Pco 2 ) g Λ Patm !

Gas Exchange in the Venous System

161

Fig. 6.14. Mock vena cava (MVC) flow loop used for bench testing of gas exchange in the Hattler Catheter.

Hattler Catheter Development Constraints associated with insertion and vena cava placement place limits on the fiber surface area of an intravascular artificial lung like the Hattler Catheter. Thus, the gas exchange efficiency (gas exchange per unit surface area) of the Hattler Catheter must exceed that of standard extracorporeal blood oxygenators, which can increase membrane area to accomplish a desired level of gas exchange. Our nominal design goal for gas exchange in the Hattler Catheter is 125 ml/min with a 0.5 m2 device, which represents a gas exchange efficiency of 250 ml/min•m2, over twice that of typical extracorporeal blood oxygenators. The Hattler Catheter uses balloon pulsation as a simple approach to improve its gas exchange efficiency for O2 and CO2 exchange. Balloon pulsation drives additional convective blood flow, beyond that which would otherwise exist in the vena cava, across the hollow fibers in a mainly cross-flow fashion to the fibers, as opposed to the mostly parallel flow that normal vena cava blood flow would impart. As described above, the altered flow pattern and the increase in convective flow velocity past the fibers reduces the diffusional boundary layers on the fiber surfaces, thus increasing the liquid-side gas permeance Kl and improving gas transfer efficiency. The development work towards improving balloon pulsation and gas exchange in the Hattler Catheter made extensive use of a simple and repeatable bench test of gas exchange, the experimental setup of which is as shown in Figure 6.14. The bench test setup consists principally of a mock vena cava (MVC: a one inch internal diameter Plexiglas or Tygon tube) within which the Hattler Catheter is placed, a pump for providing steady controlled flow rate through the MVC, a deoxygenator for establishing appropriate levels of inlet gas tension to the MVC, compliance chambers for accommodating balloon pulsation, and pre and post MVC sampling ports. A Hattler Catheter under test is placed within the MVC and perfused with distilled water at 3 l/min and 37° C. Pure O2 sweep gas is run through the Hattler Catheter under vacuum at 3 l/min, and the O2 and CO2 exchange rates, m˙ o and m˙ co , are determined over a full range of balloon pulsation rates. The O2 2 2 exchange rate is computed from the product of the PO2 difference in water across the test section (determined from samples at the pre and post MVC ports), water flow rate, and the solubility of O2 in the water, while the CO2 exchange is computed from the product of

162

The Artificial Lung

Fig. 6.15. Improvements in oxygen exchange associated with development of the Hattler Catheter, as measured in the MVC bench test using water at 3 l/min. Principal improvements came from greater achievable balloon pulsation rates (beats per minute, bpm, for each device shown within data bars) and use of fiber bundles made from fiber fabric (constrained fiber devices) rather than free fiber.

gas flow rate and the gas-side CO2 fraction leaving the device. The bench tests use water rather than blood as the test fluid to simplify testing and to facilitate multiple experiments on a given Hattler Catheter device. The CO2 and O2 exchange rates in water are less than would occur in blood due to effective solubility differences. Indeed, the O2 exchange in water is several-fold smaller than in blood because of the marked increase in effective solubility provided by hemoglobin. A mathematical model has been developed for O2 exchange in the Hattler Catheter to help relate gas exchange performance in water to that in blood.63,64 For physiologically relevant conditions of flow and inlet , the model analysis suggests that O2 exchange rates in blood are 2-3 fold greater than those in water. Thus, our design goal for gas exchange efficiency of 250 ml/min•m2 conservatively becomes a target of 125 ml/min•m2 for O2 exchange in our water bench test in the MVC. The recent development history of the Hattler Catheter, from the perspective of the maximum O2 exchange rate achieved by select devices in the MVC bench test, is summarized in Figure 6.15. Significant improvements in gas exchange occurred as optimization of the gas flow paths in the Hattler Catheter allowed for balloon pulsation at larger frequencies. Each of the devices shown had 40 ml balloons. The maximum pulsation frequency for each device is given within their respective data bars. The ability to effectively pulsate the balloon of a Hattler Catheter is studied with the device placed within a sealed-chamber plethysmograph, in which the swings in chamber pressure transduce balloon volume changes. In this manner, dynamic volume excursions of the balloon can be

Gas Exchange in the Venous System

163

Fig. 6.16. Dynamic balloon volume swings measured for Hattler Catheters with 13 ml, 25 ml and 40 ml balloons using pneumatic balloon drive system. Effective balloon pulsation, with complete filling and emptying of balloon can be achieved approximately up to 500 bpm for the smallest ballon (13 ml), 300 bpm for the intermediate-sized balloon (25 ml) and 200 bpm for the largest balloon (40 ml).

accurately determined over a broad range of pulsation frequencies. Early Hattler Catheter devices could only pulsate effectively, with full filling and emptying of the 40 ml balloon, up to 50-60 beats per minute (bpm), as evidenced by the D03 device in Figure 6.15. Our current gas pathway design for the Hattler Catheter allows a 40 ml balloon to pulsate effectively up to 180 bpm, as seen for the S2 devices in Figure 6.15. The actual dynamic balloon volumes measured using the balloon plethysmograph for several recent Hattler Catheter devices with 40 ml, 25 ml, and 13 ml balloons are shown in Figure 6.16 as a function of pulsation rate. Frequencies of 180 to 200 bpm can be consistently reached for Hattler Catheters with 40 ml balloons and frequencies of 300-400 bpm can be reached for Hattler Catheters with 10-25 ml balloons. Another significant improvement in gas exchange for the Hattler Catheter occurred with use of constrained fiber bundles made from fiber fabrics, rather than free fiber bundles (Fig. 6.15).65 Constrained fiber bundles help keep fibers fixed so as to increase the relative velocity of blood flow past the fibers by not allowing fiber movement by the blood. Moreover, the fiber fabrics enforce a more uniform spacing between fibers (see Fig. 6.17). As a result, the fibers are kept more evenly distributed around the pulsating balloon, thereby increasing the uniformity of balloon generated blood flow through the fiber bundle. The increase in blood flow velocity past the fibers, occurring more uniformly throughout the fiber bundle, reduces the average size of the diffusional boundary layers on the fiber surfaces and increases the liquid-side gas permeance of the Hattler Catheter. Bench tests of gas exchange for Hattler Catheter devices with a free fiber bundle and a constrained fiber

164

The Artificial Lung

Fig. 6.17. Fiber fabric used to make the fiber bundle for the Hattler Catheter. The fabric uses Celgard x30-240 hollow fiber membranes knit using support threads into a uniform spaced fiber array.

Fig. 6.18. Oxygen exchange as a function of balloon pulsation rate for otherwise comparable Hattler Catheters made using fiber fabric (constrained fiber device) versus free fiber. Data from bench test in MVC using water at 3 l/min.

Gas Exchange in the Venous System

165

Fig. 6.19. Oxygen exchange (top panel) and carbon dioxide exchange (bottom panel) for the same Hattler Catheter device perfused with blood versus water in the MVC bench test.

166

The Artificial Lung

Fig. 6.20. Mock vena cava flow loop used for ex vivo animal tests.

bundle, but otherwise identical, showed that the free fiber device tended to plateau in oxygen exchange with increasing balloon pulsation rate, while the constrained fiber device did not (Fig. 6.18). Furthermore, gas exchange for the free fiber Hattler Catheter exhibited more variability associated with device position and orientation within the mock vena cava (MVC) test section. The combined effect of balloon pulsation at higher rates and more effective exploitation of balloon pulsation using constrained fiber bundles have resulted in Hattler Catheters which have reached the targeted level of O2 exchange in our water bench tests, as can be seen in the development summary graph (Fig. 6.15). The bench tests of gas exchange in the Hattler Catheter can also be done using the same MVC setup but using prepared bovine blood and AAMI blood test conditions.52 The bench tests in blood help establish a benchmark for the gas exchange levels that might be achieved during animal implantation, and also help further establish the correspondence between bench tests in water versus the same tests in blood. Since blood tests are more intensive and difficult, they are reserved for select Hattler Catheter designs, while all Hattler Catheters within and across design series undergo the water bench test for gas exchange. The gas exchange levels achieved during bench tests in water and blood are compared for the same Hattler Catheter device in Figure 6.19. Gas transfer efficiency for O2 and CO2 reached 360 and 350 ml/min•m2, respectively, in blood under maximal pulsation conditions (300 bpm for these devices, which had 25 ml balloons). These same pulsation conditions yielded O2 and CO2 exchange rates of 120 and 280 ml/min•m2 for O2 and CO2, respectively, in water. Overall, these bench tests of the Hattler Catheter in blood help confirm that current devices produce gas exchange levels at or above our design target. Accordingly, recent efforts have focused more intensely on establishing the performance of

Gas Exchange in the Venous System

167

Fig. 6.21. Oxygen exchange (top panel) and carbon dioxide exchange (bottom panel) for the Hattler Catheter in the ex vivo animal tests. Gas exchange is shown for the Hattler Catheter without balloon pulsation and with maximal balloon pulsation at 180 bpm. Also shown for comparison is the reported gas exchange achieved by the IVOX in a similar ex vivo animal test using sheep.69 All gas exchange values normalized to device surface area.

the Hattler Catheter in animal tests, and on using the animal test plan to further define and refine the Hattler Catheter design for ultimate human implantation.

Animal Tests of the Hattler Catheter The Hattler Catheter has been tested in calves (of approximately 70-90 kg weight) using ex vivo (extracorporeal) circuit tests and in vivo implantations, both acute and chronic.

168

The Artificial Lung

Fig. 6.22. Carbon dioxide exchange for the Hattler Catheter determined during acute animal implantation (average and standard deviation shown for N=4 calves). Also shown for comparison are reported data for IVOX in sheep implantation70 and a summary of ex vivo O2 and CO2 exchange at 2 l/min blood flow for the Hattler Catheter and the IVOX from previous figure.

The ex vivo circuit tests involve placing the Hattler Catheter within a mock vena cava (as used in the bench test setup) that is attached to a bypass-like blood circuit.66 While the extracorporeal tests allow for the tight control, monitoring and variation of key covariates, including the blood flow rate past the catheter, they do not mimic the implantation setting within the vena cava. In the vena cava the Hattler Catheter must accommodate to the venous anatomy and to venous blood flow patterns, where numerous venous tributaries adjacent to the device increase blood exposure to the fiber bundle even without balloon pulsation. Moreover, blood flow past the Hattler Catheter within the vena cava may be conditioned by or be dependent on the flow resistance offered by the device. Finally, ex vivo experiments are, by necessity, short term (5-8 hours) and cannot be used to evaluate the response of the biological and physiological systems of the animal to the device, either acutely or chronically, in which case the calves are allowed to awaken, stand up and otherwise function normally. The ex vivo circuit test in the mock vena cava is an important evaluative assessment of gas exchange for the Hattler Catheter.66 The circuit consists of a mock vena cava or MVC (a 1 inch Tygon tube) and compliance bags connected to a blood circuit that is perfused off an anesthetized calf using a Biomedicus blood pump, as shown in Figure 6.20. The circuit perfuses the MVC with fresh bovine blood at controlled flow rates and is otherwise similar to the bench test setup (Fig. 6.14), but without the deoxygenator and heat exchanger. Blood is drawn by a drainage catheter passed into the inferior vena cava from the right atrium and returned through a separate catheter positioned in the upper superior vena cava from an insertion site at the jugular vein. The ex vivo test enables evaluation of the gas exchange performance of the Hattler Catheter under conditions of known flow and vessel geometry. More importantly, the ex vivo test is comparable to that used previously to evaluate the clinically utilized IVOX device. Thus, direct comparisons can be made

Gas Exchange in the Venous System

169

under controlled conditions between the gas exchange performance of the Hattler Catheter and the IVOX device. The O2 exchange rate measured in the ex vivo test of the Hattler Catheter is compared to that for the IVOX device in Figure 6.21. The comparisons are done over a physiologically relevant range of blood flow rates through the mock vena cava. Gas exchange results for the Hattler Catheter are shown both without and with maximal balloon pulsation (180 bpm for the HC device tested in this study). Across the blood flow rate range studied, the maximum O2 exchange rate of the Hattler Catheter varied from 250 to 310 ml/min•m2, which is consistent with projections based on bench characterization tests of gas exchange in water, although slightly less than achieved during actual bench tests using blood (Fig. 6.19). Overall, oxygen exchange for the Hattler Catheter in the ex vivo test was 2-4 times greater than that reported for the IVOX in comparable ex vivo testing. Similar results were found for the CO2 exchange rate of the Hattler Catheter, and the comparisons with the IVOX during ex vivo testing are shown in Figure 6.21. Across the blood flow rate range studied, the CO2 exchange rate of the Hattler Catheter varied from 280 to 340 ml/ min•m2 and was 1.5 to 3 times greater than that for the IVOX device. In both the O2 and CO2 exchange performance, balloon pulsation appears to have a significant effect on the dependence of gas exchange on blood flow rate. Whereas the IVOX shows variability in gas exchange with blood flow rate, as does the Hattler Catheter with no balloon pulsation (0 bpm), balloon pulsation in the Hattler Catheter at 180 bpm eliminates much if not all dependence of gas exchange on blood flow rate. An important potential implication of this result is that with variations in vena caval blood flow rate within and among patients (e.g., due to differences in cardiac output), the Hattler Catheter may show more consistent gas exchange levels clinically than did the IVOX device. Implantation of the Hattler Catheter in calves provides another important evaluative test of device performance and is a key test of device interaction with a real biological/ physiological system.67 In the implantation tests, the calves are instrumented so that arterial, pulmonary, central venous, and inferior vena caval pressures can be simultaneously recorded and those sites used for blood sampling. A Swan-Ganz thermal dilution catheter is connected to a cardiac output monitor. The calf is heparinized and the device inserted with a surgical cut-down into the jugular vein and positioned within the inferior vena cava, right atrium, and terminal superior vena cava. While O2 exchange rates are difficult to determine during implantation tests, the CO2 exchange rate can be reliably determined by measuring the gas flow rate and CO2 fraction of the exhaust gas leaving the Hattler Catheter. We recently completed a series of acute implants of the Hattler Catheter in four calves, with the devices placed within the right atrium and inferior vena cava.67 As shown in Figure 6.22, the maximum CO2 exchange rate (normalized to the device surface area) averaged 327 ± 15 ml/min•m2 over the series of calf implants. Moreover, the CO2 transfer rate was over 70% greater than reported for the clinically tested IVOX implanted acutely in sheep. The calf and sheep are analogous models for CO2 exchange, hence, the difference in CO2 exchange rate reflects differences in the gas exchange performance of the Hattler Catheter compared to the IVOX. The gas exchange levels accomplished by the Hattler Catheter during implantation are also consistent with those from the ex vivo tests (shown for comparison in Figure 6.22), as is the improvement compared to the IVOX. The CO2 exchange rate shown in Figure 6.22 is for our current Hattler Catheter design with constrained fiber bundles and is significantly greater than the CO2 exchange rate we reported of 140 to 180 ml/min•m2 based on an earlier implantation study using a previous Hattler Catheter design (D07 series) with free fiber bundles and reduced balloon pulsatility.68 The

The Artificial Lung

170

Table 6.4. Effect of permissive hypercapnia on carbon dioxide exchange** Venous PCO2 (mm Hg) 49.1 (baseline) 54.3 60.2 63.3 66.9

m˙ co 2 normalized to m˙ co 2 at 49.1 mm Hg

1 1.13 1.33 1.44 1.48

PCO2/49.1 1 1.11 1.23 1.29 1.36

** Hattler Catheter device 10-7 during acute implantation in calf.

improvement in in vivo gas exchange attributable to constrained fiber bundles and improved balloon pulsation in the Hattler Catheter is approximately twofold and is generally consistent with the improvement seen during bench tests of these devices (see Fig. 6.15). The CO2 exchange rate of the Hattler Catheter can also be increased naturally by use of permissive hypercapnia, in which venous levels of PCO2 are allowed to increase above normal. As the venous level of PCO2 rises, so does the gradient for CO2 diffusion from blood to sweep gas, thus increasing the CO2 exchange rate within the Hattler Catheter according to Equation 6.1 (written for CO2). Table 6.4 shows the increase in CO2 exchange rate measured during permissive hypercapnia from a baseline venous PCO2 of 49 mm Hg to an elevated PCO2 of 67 mm Hg during implantation of the Hattler Catheter in a calf. The CO2 exchange rate is shown as the fractional increase relative to that at the baseline PCO2. The Hattler Catheter increased its CO2 exchange rate by 48% in response to the permissive hypercapnia, an increase somewhat greater than the percent increase in PCO2 associated with the hypercapnia. Thus, permissive hypercapnia represents a promising strategy for further augmenting CO2 removal with the Hattler Catheter. Implantation of the Hattler Catheter in calves does not engender significant changes in cardiac output nor in other measured hemodynamic or hematological parameters. In our recent series of acute implantations of the Hattler Catheter,67 the decrease in cardiac output with insertion relative to baseline (preinsertion) was less than 10-15% for devices with 25 ml balloons (N=4), although larger changes were seen when implanting devices with larger diameter, 40 ml balloons (unpublished observation). Plasma-free hemoglobin rose no higher than 10-15 mg% after 6-8 hours of acute implantation and platelet count dropped an average of 30% in the early phase of implantation (0-2 hours) but stabilized thereafter. In our chronic experiments (N=7), plasma-free hemoglobin peaked at an average high of 11.5 ± 4.9 mg%, but by the end of the experiment (2-3 days) dropped to an average of 4.9 ±3.6 mg% (unpublished observation). The Hattler Catheter has progressed from basic bench research through small and large animal studies to a point where it is now ready for human implantation. Although the application of the technology in the research setting has evolved over the last 16 years, the basic bioengineering principles upon which it is founded have not changed and have allowed progress to a point where 50% of basal requirements for O2 and CO2 exchange in man can be met in a manner that is both dependable and reproducible. There is reason to anticipate, therefore, that direct intravenous gas exchange as a new approach for the support of the acutely injured lung will be a reality in the near future.

Gas Exchange in the Venous System

171

Acknowledgements 1. The work described in this chapter from our group was supported by the U.S. Army Medical Research, Development, Acquisition and Logisitcs command under prior Contract No. DAMD17-94-C-4052 and current Grant No. DAMD17-98-1-8638. The views, opinions and/or findings contained in this report are those of the authors and should not be construed as an Official Department of the Army position, policy or decision unless so designated by other documentation. 2. The authors wish to thank the research staff and students of the Artificial Lung Laboratory: Joseph Golob, Brian Frankowski, Thomas Merrill, Laura Lund, Heide Russian, Michael Lann, Tamara Tulou, Mariah Hout and Monica Garcia. We also appreciate the assistance of Ms. Alison Bushey in preparing the final manuscript.

References 1. Ashbaugh DG, Bigelow DB, Petty TL et al. Acute respiratory distress in adults. Lancet 1967; 2:319-323. 2. Zilberberg MD, Epstein SK. Acute lung injury in the medical ICU: Comorbid conditions, age, etiology, and hospital outcome. Am J Respir Crit Care Med 1998; 157:1159-1164. 3. Bernard GR, Artigas A, Brigham KL et al. The American-European Consensus Conference on ARDS: Definition, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149:818-824. 4. Doyle RL, Szaflarski N, Modin GW et al. Identification of patients with acute lung injury: Predictors of mortality. Am J Respir Crit Care Med 1995; 152:1818-1824. 5. 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. Am J Respir Crit Care Med 1999; 159:1849-1961. 6. Linde-Zwirble WT, Clermont G, Coleman MB et al. Incidence of ARDS in the US, Europe, and Japan. Intensive Care Med 1996; 22 (Suppl. 3):272. 7. Angus DC, Linde-Zwirble WT, Clermont G et al. What’s the incidence and mortality of ARDS in the US? Am J Respir Crit Care Med 1996; 153;4:A592. 8. University Health System Consortium. Oak Brook, Illinois, 1996. 9. 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. 10. Hirschl RB, Pranikoff T, Wise C et al. Initial experience with partial liquid ventilation in adult patients with the acute respiratory distress syndrome. JAMA 1996; 275:383-389. 11. Fort P, Farmer C, Westerman J et al. High-frequency oscillatory ventilation for adult respiratory distress syndrome—Pilot study. Crit Care Med 1997; 25:937-947. 12. Anzueto A, Baughman RP, Guntupalli KK et al. Aerosolized surfactant in adults with sepsis-induced acute respiratory distress syndrome. N Eng J Med 1996; 334: 1417-1421. 13. 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. 14. The Acute Respiratory Distress Syndrome Network. Ketoconazole does not reduce mortality in patients with the acute respiratory distress syndrome. JAMA 2000; 283:1995-2002. 15. Federspiel W, Sawzik P, Borovetz H et al. Temporary support of the lungs: the artificial lung. In: Cooper DKC, Miller LW, Patterson GA., eds. Transplantation and Replacement of Thoracic Organs. 2nd ed. Dudrecht: Kluwer Academic Publishers, 1996:717-728. 16. Hattler BG, Johnson PC, Sawzik PJ et al. Respiratory dialysis: A new concept in pulmonary support. ASAIO J 1992; 38:M322-M325. 17. Reeder GD, Hattler BG, Rawleigh J et al. Current progress in the development of an intravenous membrane oxygenator. ASAIO J 1993; 39:M461-M465. 18. Hattler BG, Reeder GD, Sawzik PJ et al. Development of an intravenous membrane oxygenator (IMO): Enhanced intravenous gas exchange through convective mixing of blood around hollow fiber membranes. Artif Organs 1994; 18:806-812.

172

The Artificial Lung

19. Hattler BG, Reeder GD, Sawzik, PJ et al. Development of an intravenous membrane oxygenator: A new concept in mechanical support for the failing lung. J Heart and Lung Trans 1995; 13:1003-1007. 20. Federspiel WJ, Hewitt T, Hout MS et al. Recent progress in engineering the Pittsburgh intravenous membrane oxygenator. ASAIO J 1996; 42:M435-M442. 21. Macha M, Federspiel WJ, Lund LW et al. Acute in vivo studies of the Pittsburgh intravenous membrane oxygenator. ASAIO J 1996; 42:M609-M615. 22. Mortensen JD. An intravenacaval blood gas exchange (IVCBGE) device: A preliminary report. Trans Am Soc Artif Intern Org 1987; 33:570-573. 23. Gattinoni L, Kolobow T, Tomlinson T et al. Low frequency positive pressure ventilation with extracorporeal carbon dioxide removal (LFPPV-ECCO2R): An experimental study. Anesth Analg 1978; 57:470-476. 24. 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; 22:1568-1578. 25. 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. 26. Weinberger SE, Schwartzstein RM, Weiss JW. Hypercapnia. N Engl J Med 1989; 321:1223-1227. 27. Stewart TE, Meade MO, Cook DJ et al. Evaluation of a ventilation strategy to prevent barotrauma in patients at high risk of acute respiratory distress syndrome. N Engl J Med 1998; 338:355-361. 28. Brochard L, Roudot-Thoraval F, Roupie E et al. Tidal volume reduction for prevention of ventilator-induced lung injury in acute respiratory distress syndrome. Am J Respir Crit Care Med 1998; 158:1831-1838. 29. 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; 27:1492-1498. 30. Ware LB, Matthay MA. The acute respiratory distress syndrome. N Engl J Med 2000; 342:1334-1349. 31. Slutsky AS, Tremblay LN. Multiple system organ failure: Is mechanical ventilation a contributing factor? Am J Respir Crit Care Med 1998; 157:1721-1725. 32. 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. 33. Slutsky AS. Lung injury caused by mechanical ventilation. Chest 1999; 116:9S-15S. 34. Tremblay L, Valenza F, Riberio SP et al. Injurious ventilation strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99:944-952. 35. Pugin J, Verghese G, Widmer MC et al. 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. 36. Pugin J, Ricou B, Steinberg KP et al. 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. 37. Pinsky MR. Heart-lung interactions. In: Marini JJ, Slutsky AS, eds. Physiological Basis of Ventilatory Support. New York: Marcel Dekker, 1998; 491-531. 38. Borelli M, Fumagalli R, Bernasconi F et al. Relief of hypoxemia contributes to a reduction in cardiac index related to the use of positive end-expiratory pressure. Intensive Care Med 1996; 22:382-386. 39. Fagon JY, Chastre J, Domont Y et al. Nosocomial pneumonia in patients receiving continuous mechanical ventilation. Am Rev Respir Dis 1989; 139:877-884. 40. Zapol WM, Snider MT, Hill JD et al. Extracorporeal membrane oxygenation in severe acute respiratory failure: a randomized prospective study. JAMA 1979; 242:2193-2196. 41. Gattinoni L, Pesenti A, Mascheroni D et al. Low-frequency positive pressure ventilation with extracorporeal CO2 removal in severe acute respiratory failure. JAMA 1986; 256:881-886. 42. Morris AH, Wallace CJ, Menlove RL et al. Randomized clinical trial of pressure controlled inverse ratio ventilation and extracorporeal CO2 removal for adult respiratory distress syndrome. Am J Respir Crit Care Med 1994; 149:295-305.

Gas Exchange in the Venous System

173

43. Brunel F, Belgheth M, Mira J et al. Extracorporeal carbon dioxide removal and low-frequency positive-pressure ventilation: improvement in arterial oxygenation with reduction of risk of pulmonary barotrauma in patients with adult respiratory distress syndrome. Chest 1993; 104:889-898. 44. Manert W, Haller M, Briegel J et al. Venovenous extracorporeal membrane oxygenation (ECMO) with a heparin-lock bypass system: An effective addition in the treatment of acute respiratory failure (ARDS). Anesthetist 1996; 45:438-448. 45. Kolla S, Awad SS, Rich PB et al. Extracorporeal life support for 100 adult patients with severe respiratory failure. Ann Surg 1997; 226:544-566. 46. Goodman LR. Congestive heart failure and adult respiratory distress syndrome: New insights using computed tomography. Radiol Clin North Am 1996; 34:33-46. 47. 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. 48. Muller EE, Kolobow T, Mandava S et al. How to ventilate lungs as small as 12.5% of normal: The new technique of intratracheal pulmonary ventilation. Ped Res 1993; 34:606-610. 49. Mortensen JD. Intravascular oxygenator: A new alternative method for augmenting blood gas transfer in patients with acute respiratory failure. Artif Org 1992; 16:75-82. 50. Conrad SA, Bagley A, Bagley B et al. Major findings from the clinical trials of the intravascular oxygenator. Artif Org 1994; 18:846-863. 51. Konishi R, Shimizer R, Firestone L et al. Nitric oxide prevents human platelet adhesion to fiber membranes in whole blood. ASAIO J 1996; 42:M850-M853. 52. AAMI Standards and Recommended Practices. Association for the Advancement of Medical Instrumentation. Volume 2.1 Biomedical Equipment, Part 1, Equipment Therapy and Surgery. Cardiovascular implants and artificial organs: blood gas exchangers. AAMI 1996; 7199:633-648. 53. Montoya JP, Shanley CJ, Merz SI et al. Plasma leakage through microporous membranes: Role of phospholipids. ASAIO J 1992; 38:M399-M405. 54. Mottaghy K, Oedekoven B, Starmans H et al. Technical aspects of plasma leakage prevention in microporous capillary membrane oxygenators. ASAIO J 1989; 35:640-643. 55. Brodkey RS, Hershey HC. Transport Phenomena: A Unified Approach. New York: McGraw-Hill Book Company, 1988:182-186. 56. Kamo J, Uchida M, Hirai T et al. A new multilayered composite hollow fiber membrane for artificial lung. Artif Org 1990; 14:369-372. 57. Lund LW. Measurement of Hollow Fiber Membrane permeance in a Gas-Liquid System. Ph.D. Thesis. University of Pittsburgh, 2000. 58. Lund LW, Federspiel WJ, Hattler BG. Gas permeability of hollow fiber membranes in a gas-liquid system. J Membr Sci 1996; 117:207-219. 59. Lund LW, Hattler BG, Federspiel WJ. Is condensation the cause of plasma leakage in microporous hollow fiber membrane oxygenators? J Membr Sci 1998; 147:87-93. 60. Federspiel WJ, Hattler BG. Sweep gas flowrate and CO2 exchange in artificial lungs. Artif Org 1996; 20:1050-1056. 61. Hout MS, Hattler BG, Federspiel WJ. Validation of a model for flow-dependent carbon dioxide exchange in artificial lungs. Artif Org 2000; 24:114-119. 62. Federspiel WJ, Williams JL, Hattler BG. Gas flow dynamics in hollow fiber membranes. AIChE 1996; 42:2094-2099. 63. Hewitt TJ, Hattler BG, Federspiel WJ. A mathematical model of gas exchange in an intravenous membrane oxygenator. Ann Biomed Eng 1998; 26:166-178. 64. Federspiel WJ, Hewitt TJ, Hattler BG. Experimental evaluation of a model for oxygenation exchange in a pulsating intravascular artificial lung. Ann Biomed Eng 2000; 28:160-167. 65. Federspiel WJ, Hout MS, Hewitt TJ et al. Development of a low flow resistance intravenous oxygenator. ASAIO J 1997; 45:M725-M730. 66. Federspiel WJ, Golob JF, Merrill TL et al. Ex-vivo testing of the intravenous membrane oxygenator (IMO). ASAIO J 2000; 46:261-267. 67. Golob JF, Federspiel WJ, Merrill TL et al. Acute in-vivo testing of an intravascular oxygenator. ASAIO J. In press. 68. Macha M, Federspiel WJ, Lund L et al. Acute in-vivo studies of the Pittsburgh intravenous membrane oxygenator. ASAIO J 1996; 42:M609-M615. 69. Tao W, Zwischenberger JB, Nguyen TT et al. Performance of an intravenous gas exchanger (IVOX) in a venovenous bypass circuit. Ann Thorac Surg 1994; 57:1484-1490.

174

The Artificial Lung

70. Zwischenberger JB, Cox CS Jr, Graves D et al. Intravascular membrane oxygenation and carbon dioxide removal—A new application for permissive hypercapnia? Thorac Cardiovasc Surg 1992; 40:115-120.

Index A

G

Acute respiratory distress syndrome (ARDS) 2, 3, 25, 78, 98, 106, 107, 121, 125, 129, 133, 137, 140, 144, 150, 161 Acute respiratory failure 1, 25, 28, 33, 98, 106, 116, 119, 129, 135-138, 140, 142, 145 Arteriovenous extracorporeal CO2 removal 122 Aspect ratio 36, 41, 46, 47, 55, 56 AVCO2R 3, 119, 122, 124-126, 129

Gas exchange 2-5, 7, 10-12, 19, 20, 22, 24-28, 31, 33, 34-41, 46, 47, 55, 57, 58, 59, 62, 78, 81, 98-108, 110-113, 116, 118, 119-122, 124-126, 128, 133, 135, 138, 140-142, 144-147, 149, 151, 153, 154, 157, 158, 161-163, 166-169

H

Balloon pulsation 112, 117, 144, 153, 161-164, 166-170 Bench tests 162, 166, 169, 170 Blood trauma 35, 38, 39, 42-46, 56, 60, 61, 63, 65, 140 Boundary layer 5, 81, 99, 101, 146, 149, 150, 151, 153, 159, 161, 163

Hattler Catheter 134, 135, 140, 142, 144-147, 150, 151, 153, 154, 157, 158, 160, 162-169 Heparin 12, 15, 25, 27, 31, 66, 71, 75, 83, 85, 87, 88, 99, 105, 126, 141, 144, 158, 169 Heparin coating 85, 86 Hollow fibers 5, 9, 12, 14, 20, 26, 27, 35, 59, 61, 100, 105, 107, 118, 135, 141, 142, 144, 146-149, 151, 152, 154, 155, 158, 161

C

I

Cardiopulmonary bypass 5, 12, 24, 25, 75, 78, 81, 98, 145, 147 Chronic respiratory failure 24, 29, 30 Coagulation 2, 12, 25, 28, 33, 35, 42, 63, 65, 66, 68, 75, 77, 80, 83, 99, 101, 129, 143 Complement 35, 66, 67, 68, 70-72, 75, 77, 78, 80, 82, 84, 86, 87, 106 Composite hollow fiber membrane 153, 155, 157, 159 Contact activation 66, 67, 69, 71, 82, 84, 86 Cytokine 73, 137

Implantable artificial lung 1-6, 8, 10, 12, 21, 22, 24, 25, 27-29, 31, 153, 154 Intravascular oxygenator 12, 141 IVOX 12, 27, 61, 98, 104-111, 113-118, 129, 135, 141, 142, 167-169

E

M

ECCO2R 119, 120, 121, 122, 129, 138 Extracorporeal carbon dioxide removal 3, 33, 119, 123 Extracorporeal membrane oxygenation (ECMO) 3, 12, 24, 25, 27, 28, 31, 61, 65, 67, 71, 74-76, 80, 81, 83, 85, 88, 106, 119, 120, 122, 127, 138, 140

Mass transfer coefficient 40, 51, 100, 105, 110, 111, 113, 116, 146 Membrane permeance 149, 153, 154, 157-159 Microporous membranes 27

B

L Leukocytes 35, 43, 66, 71, 73-75, 77, 78, 82, 120 Lung transplantation 1, 2, 4, 24, 28, 29, 31

The Artificial Lung

176

O

R

Oxygenator 5, 12, 18, 22, 25-27, 31, 35, 55, 61, 65, 72, 74, 77, 81, 98, 103, 109, 113, 119, 123, 126, 129, 135, 138, 140, 142, 145, 147, 150, 153, 160, 161, 168

Radial flow 19, 20, 21 Respiratory failure 1, 24, 26, 28-30, 33, 78, 98, 106, 107, 116, 119, 122, 124, 126, 129, 135-138, 140-142, 145 Right ventricular failure 21, 31 Right ventricular power 21, 41, 42, 57, 59

P PEG 86 PEO 86, 88, 133 Permeance 146, 147, 149-151, 153, 154, 157-159, 161, 163 Plasma wetting 147, 150, 153, 154 Platelets 43, 66, 71, 73-75, 78, 82, 87, 99 Polyethylene glycol 86 Polyethylene oxide 66, 86 Pressure drop 3, 8-11, 20, 35, 40, 41, 44, 47, 48, 51, 52, 55, 56, 144, 159, 160 Protein adsorption 66, 67, 81, 82, 86, 87 Pulmonary artery 2-4, 6, 7, 11, 12, 20, 21, 31, 37, 42, 57, 58, 62, 107, 136 Pulsatility 11, 39, 56, 57, 62, 169

S Shear stress 35, 42-44, 47, 49, 53, 55-58, 61, 63, 72, 74, 76, 82, 122, 144 Solid membranes 26 Surface area 2, 3, 5, 7, 9, 10, 12, 14, 20, 26, 36, 40, 41, 46, 48, 51, 52, 56, 60, 62, 65, 78, 82, 88, 99-101, 104, 106, 108, 109, 113, 115, 116, 118, 122, 126, 140, 141, 144, 146, 161, 167, 169

V Void fraction 7, 8, 11, 14, 20, 40-42, 46-48, 51, 52, 55-57, 60, 62