1 ORIGINAL ARTICLES: GENERAL THORACIC Percutaneous Extracorporeal Arteriovenous CO 2 Removal for Severe Respiratory Failure Joseph B. Zwischenberger, MD, Steven A. Conrad, MD, PhD, Scott K. Alpard, MD, Laurie R. Grier, MD, and Akhil Bidani, MD, PhD Departments of Surgery, Medicine, and Radiology, University of Texas Medical Branch and Shriners Burns Institute, Galveston, Texas; and Division of Pulmonary and Critical Care Medicine, Louisiana State University Medical Center, Shreveport, Louisiana Background. In previous animal studies, arteriovenous CO 2 removal (AVCO 2 R) achieved significant reduction in ventilator pressures and improvement in the PaO 2 to fraction of inspired oxygen ratio during severe respiratory failure. For our initial clinical experience, 5 patients were approved for treatment of severe respiratory failure and CO 2 retention to evaluate the feasibility and safety of percutaneous AVCO 2 R. Methods. Patients were anticoagulated with heparin (activated clotting time, 260 to 300 seconds), underwent percutaneous femoral cannulation (10F to 12F arterial and 12F to 15F venous catheters), and then were connected to a low-resistance, 2.5-m 2 hollow-fiber oxygenator for 72 hours. Results. Mean AVCO 2 R flow at 24, 48, and 72 hours was , , and ml/min, respectively, with no vascular complications and no significant change in heart rate or mean arterial pressure. Removal of CO 2 plateaued at an AVCO 2 R flow of 1086 ml/min with 208 ml/min CO 2 removed. Average CO 2 transfer at 24 and 48 hours was and ml/min. Use of AVCO 2 R allowed a significant decrease in minute ventilation from L/min at baseline to L/min at 24 hours. Conclusions. All patients survived the experimental period without adverse sequelae. Percutaneous AVCO 2 R can achieve approximately 70% CO 2 removal in adults with severe respiratory failure and CO 2 retention without hemodynamic compromise or instability. (Ann Thorac Surg 1999;68:181 7) 1999 by The Society of Thoracic Surgeons Despite recent advances in critical care, the overall mortality of adult respiratory distress syndrome (ARDS) remains approximately 40% to 50% . Until recently, ARDS was treated primarily with mechanical ventilation aimed at restoring normal blood gases by aggressive volume-controlled ventilation and high ( 50%) oxygen concentration while the lungs were recovering from the initial injury or disease process. Volume-controlled ventilation, however, causes iatrogenic injury to the lungs, exacerbating ARDS by inflicting barotrauma and volume trauma [2, 3] to both the injured and uninjured portions of the lung. To reduce the high airway pressures and associated trauma, recent trends in ventilator management limit inflation pressure and tidal volume at the cost of a rise in systemic arterial carbon dioxide (CO 2 ) levels. This technique, termed permissive hypercapnia, using low tidal volumes and low-pressure pulmonary ventilation, reduces the incidence of barotrauma and volume trauma and possibly improves survival in ARDS [4 8]. Unfortunately, permissive hypercapnia is limited by respiratory acidosis causing substantial changes in hemodynamics, organ blood flow, Presented at the Forty-fifth Annual Meeting of the Southern Thoracic Surgical Association, Orlando, FL, Nov 12 14, Address reprint requests to Dr Zwischenberger, Cardiothoracic Surgery, University of Texas Medical Branch, Galveston, TX ; and intracranial pressure unless arterial ph is controlled [6, 9]. We have previously shown, in a smoke inhalation model of severe ARDS in adult sheep, that arteriovenous carbon dioxide removal (AVCO 2 R) allows significant reductions in ventilator pressures without the need for hypercapnia or the complex circuitry and monitoring required for extracorporeal membrane oxygenation (ECMO) . At a shunt flow of up to 29% of cardiac output, AVCO 2 R did not significantly change cardiac output  or compromise organ blood flow . We have also shown that arterial percutaneous cannulas of 10F or larger allow adequate blood flow to achieve nearly total CO 2 removal without incurring hypercapnia in adult sheep [13, 14]. In this study we evaluated the initial safety and feasibility of percutaneous AVCO 2 R in adults with severe respiratory failure and CO 2 retention. Material and Methods Our purpose was to evaluate the safety and feasibility of percutaneous AVCO 2 R in patients with severe respiratory failure and CO 2 retention, in a collaborative study between the University of Texas Medical Branch and Louisiana State University Medical Center. For this initial experience, we selected patients with unresponsive, severe ARDS with CO 2 retention despite maximum medical management using the lung protective strategy of 1999 by The Society of Thoracic Surgeons /99/$20.00 Published by Elsevier Science Inc PII S (99)
2 182 ZWISCHENBERGER ET AL Ann Thorac Surg PERCUTANEOUS AVCO 2 R FOR RESPIRATORY FAILURE 1999;68:181 7 permissive hypercapnia. These patients were either not eligible for or had failed other therapies including inhaled nitric oxide, high-frequency ventilation, prone positioning, partial liquid ventilation, or ECMO. The study was approved by the Institutional Review Boards at each institution, and informed consent was obtained before each study. We applied a commercially available, low-resistance oxygenator (Affinity, Avecor Cardiovascular, Plymouth, MN) to a femoral-femoral arteriovenous shunt by means of percutaneous access. The AVCO 2 R technique includes a femoral-femoral arteriovenous shunt using 10F to 12F arterial and 12F to 15F venous percutaneous cannulas introduced by a modified Seldinger technique. The Affinity oxygenator was chosen for CO 2 removal because of its low-resistance blood flow characteristics. The Affinity membrane oxygenator was primed with normal saline and connected to the vascular cannulas after removing all the air. The patients were systemically anticoagulated with heparin to maintain an activated clotting time (Hemochron 400, International Technidyne, Edison, NJ) between 260 and 300 seconds throughout the study. The AVCO 2 R blood flow was monitored by an ultrasonic flow probe (model H6X, Transonic Systems, Ithaca, NY) placed on the arterial cannula and interfaced with a real-time flowmeter (model HT 109, Transonic Systems) with digital display. The AVCO 2 R blood flow was always at the maximum flow achieved by the arteriovenous pressure gradient. Sweep gas flow (100% oxygen) was controlled by an in-line regulator and set at 2 to 4 times AVCO 2 R blood flow. Removal of CO 2 by the device was calculated as the product of sweep gas flow and its exhaust CO 2 concentration measured by an in-line capnometer (SaraTrans, Lenexa, KS). Removal of CO 2 by the native lungs was calculated as the product of minute ventilation and the CO 2 concentration in the expired gas collected in a Douglas bag. Blood and expired gases were measured with a blood gas system (System BG3 and Co-Oximeter 482, Instrumentation Laboratory, Lexington, MA). The ventilator tidal volume or pressure control and respiratory rate were decreased as determined by the investigators at each participating institution. The goal of ventilator management was to decrease the peak inspiratory pressure (PIP) below 35 cm H 2 O and continue permissive hypercapnia yet keep the ph greater than 7.2. Each patient had one-on-one nursing care in an intensive care unit. The patients received AVCO 2 R for 72 hours, then were returned to mechanical ventilation for total gas exchange while continuing a lung-protective strategy with permissive hypercapnia. Results All patients were successfully cannulated for AVCO 2 Rat the bedside. Patient number 1 (Table 1) had 10F arterial and 12F venous cannulas, which allowed only 600 ml/ min of arteriovenous shunt flow. The subsequent patients, numbers 2 through 5, had 12F arterial and 15F venous cannulas inserted for an average arteriovenous shunt flow of 940 ml/min. All patients completed the 72-hour trial, and 3 of 5 were discharged from the hospital. At 48 hours, a review of AVCO 2 R performance demonstrated that AVCO 2 R blood flow ranged from 600 to 1100 ml/min. Figure 1 shows total CO 2 production simultaneous with CO 2 removal by the AVCO 2 R circuit. Total CO 2 production decreased from 200 ml/min to approximately 125 ml/min as CO 2 removal closely paralleled production and ranged from 75 to 150 ml/min (60% to 85% of CO 2 production). Figure 2 shows that Paco 2 at baseline was approximately 95 mm Hg and, on initiation of AVCO 2 R, decreased to 70 mm Hg and was maintained at that average. Likewise, AVCO 2 R removed approximately 70% of total CO 2 production throughout the 72 hours of the study. Changes in tidal volume, minute ventilation, and PIP during the 72 hours of AVCO 2 R are shown in Figure 3. The changes in ventilator parameters (average, n 5) from baseline to 48 hours included a decrease in tidal volume from 416 to 284 ml, a decrease in PIP from 33 to 29 cm H 2 O, a decrease in minute ventilation from 7.2 to 4.6 L/min, and a decrease in respiratory rate from 17 to 14 breaths per minute. Two patients improved during the 72 hours with decreased fraction of inspired oxygen requirements. The Pao 2 to fraction of inspired oxygen ratio (Fig 4) at 72 hours of AVCO 2 R reflects the rapid improvement of some (patients 2 and 4; eventual survivors) and deterioration of others. Rapid improvement was noted in patient 2 (152.5 at baseline to at 64 hours) and patient 4 (137.5 at 8 hours to at 72 hours); both patients survived and were eventually discharged. Deterioration was noted in patient 1 (94.3 at baseline to 71.6 at 72 hours) and patient 5 (89.0 at baseline to 72.0 at 72 hours); both patients eventually died. Patient 3 (92.8 at baseline to 72.7 at 72 hours) had a prolonged course but ultimately survived and was discharged from the hospital. Throughout the 72 hours, the heart rate and mean arterial pressure did not significantly change (Fig 5). No major complications or adverse events were attributable to the AVCO 2 R circuit. Two minor complications were noted: (1) an oxygenator thrombosis noted by decreased gas exchange, and (2) cannula-site bleeding that could be controlled with a pressure bag. Patients were weaned from AVCO 2 R after 72 hours. Individual patient data are summarized in Table 1. Comment For many years, mechanical ventilation attempted to simulate the normal respiratory cycle, targeting normal blood gases while using tidal volumes of 10 to 15 ml/kg. Conventional mechanical ventilation with high tidal volumes and high PIP causes iatrogenic lung injury in the setting of severe respiratory failure . Kolobow and associates  reported that positive airway pressure ventilation with high peak pressures and tidal volumes increases vascular filtration pressures and causes stress fractures of the capillary endothelium, epithelium, and basement membrane. This leads to leakage of fluid,
3 Ann Thorac Surg ZWISCHENBERGER ET AL 1999;68:181 7 PERCUTANEOUS AVCO 2 R FOR RESPIRATORY FAILURE 183 Table 1. Patient Data at Baseline and After 48 Hours of AVCO 2 R Patient 1 Patient 2 Patient 3 Patient 4 Patient 5 Variables BL 48 hrs BL 48 hrs BL 48 hrs BL 48 hrs BL 48 hrs ARDS cause sepsis/pneumonia sepsis/bowel perforation restrictive lung disease COPD exacerbation sepsis/pneumonia Other treatments Abx, NO, prone Abx, NO, prone Abx, mechanical ventilation Bipap, mechanical ventilation Abx, mechanical ventilation, permissive hypercapnia HR MAP (mm Hg) TV (ml/min) RR MV (L/min) PIP (cm H 2 O) Fio ABG (Po 2 /Pco 2 /ph) 66/103/ /110/ /81/ /97/ /90/ /62/ /71/ /38/ /123/ /45/7.38 Pao 2 /Fio Total CO 2 production (ml/min) Qb (ml/min) CO 2 removal (ml/min) CO 2 removal (% CO 2 production) Outcome Died Survived Survived Survived Died ABG arterial blood gas; Abx antibiotics; ARDS adult respiratory distress syndrome; AVCO 2 R arteriovenous CO 2 removal; Bipap non-invasive bi-nasal positive airway pressure; COPD chronic obstructive pulmonary disease; Fio 2 fraction of inspired oxygen; HR heart rate; MAP mean arterial pressure; MV minute ventilation; NO nitric oxide; PIP peak inspiratory pressure; Qb shunt flow; RR respiratory rate; TV tidal volume.
4 184 ZWISCHENBERGER ET AL Ann Thorac Surg PERCUTANEOUS AVCO 2 R FOR RESPIRATORY FAILURE 1999;68:181 7 Fig 1. Total CO 2 production simultaneous with CO 2 removal by the arteriovenous CO 2 removal (AVCO 2 R) circuit. Production of CO 2 ranged from to ml/min as CO 2 removal closely paralleled production and ranged from to ml/min. protein, blood, and air, resulting in pulmonary edema, hemorrhage, atelectasis, pneumatoceles, or pneumothorax . Healthy dogs ventilated with PIP of 30 cm H 2 O for as little as 2 hours show lung damage (atelectasis, alveolar edema, and hemorrhage) . Tsuno and coworkers  ventilated pigs at a PIP of 40 cm H 2 O for 24 hours and found histologic changes similar to those seen in early stages of respiratory failure. When volume-controlled ventilation is used in patients with ARDS, the risk of regional lung overdistention is high, because tidal volumes used to maintain normal blood gases invariably inflict high PIPs to the small fraction of compliant lung capable of gas exchange . To minimize volume distention of healthy alveoli and potentiation of the lung injury [16, 18], pressure-limited ventilation techniques have been proposed. In a prospective, randomized clinical trial , patients treated with pressure-limited ventilation had a more rapid increase in static lung compliance and normalization of blood Pco 2 than those with volume-controlled ventilation. Such Fig 3. Changes in airway pressures from baseline to 48 hours included a decrease in tidal volume from to ml, a decrease in peak inspiratory pressure (PIP) from to cm H 2 O, and a decrease in minute ventilation from to L/min. (AVCO 2 R arteriovenous CO 2 removal.) Fig 2. At baseline Paco 2 was mm Hg and decreased to mm Hg on initiation of arteriovenous CO 2 removal (AVCO 2 R). Arteriovenous CO 2 removal removed approximately 70% of total CO 2 production throughout the 72 hours of the study. changes in ventilator strategies to avoid high airway pressures may contribute to the lower mortality recently reported in patients with ARDS [4, 5, 8]. Limiting PIP, and therefore tidal volume depending on lung compliance, may not provide adequate minute ventilation to excrete the produced CO 2 and leads to an increase in systemic Paco 2, termed permissive hypercapnia. Use of ECMO provides complete gas exchange and improves survival in neonates , pediatric patients , and selected adult patients [21, 22]. More than 10,000 cases of ECMO to date reveal an 81% survival in neonates, 49% in children, and 38% in adults in patient populations estimated to have a greater than 80% mortality . Treatment with ECMO, however, involves intensive monitoring, high cost in labor and equipment, and frequent complications [23, 24]. Targeting CO 2 re-
5 Ann Thorac Surg ZWISCHENBERGER ET AL 1999;68:181 7 PERCUTANEOUS AVCO 2 R FOR RESPIRATORY FAILURE 185 Fig 4. Ratio of Pao 2 to fraction of inspired oxygen (Pao 2 /Fio 2 ) of all patients throughout the 72-hour study. (AVCO 2 R arteriovenous CO 2 removal.) moval, Kolobow and colleagues  developed the venovenous extracorporeal technique of extracorporeal CO 2 removal in animals. Gattinoni and associates  also reported that low-frequency ventilation combined with extracorporeal CO 2 removal provided sufficient gas exchange and improved survival in patients with ARDS. Although extracorporeal CO 2 removal was effective in reducing ventilatory requirements, extracorporeal CO 2 removal is essentially low-flow venovenous ECMO and has all the inherent disadvantages. Arteriovenous CO 2 removal was developed to minimize the foreign surface interactions and blood element shear stress inherent in an extracorporeal circuit with a pump and allow a gas exchange membrane of sufficient surface area for nearly total CO 2 removal. Barthelemy and associates  initially achieved significant CO 2 removal using a large membrane in a pumpless circuit with flows in the range of 1200 to 2000 ml/min. More recently, Young and colleagues  evaluated AVCO 2 R in animals in both a pumped and a pumpless circuit with a large oxygenator membrane and found that despite excellent CO 2 removal, resistance was a limiting factor at low flow rates. To evaluate the potential long-term use of AVCO 2 R, Awad and coworkers  demonstrated the feasibility of chronic arteriovenous support for 7 days in awake normal sheep without sequelae. However, the major limitation of all these studies was high circuit resistance. Recent developments in computational fluid dynamics led to a newly designed low-resistance oxygenator commercially available as the Affinity oxygenator . We selected this low-resistance gas exchanger to use with percutaneous arterial and venous cannulas matched to the flow ranges necessary for total CO 2 removal during ARDS induced by severe smoke inhalation and cutaneous flame burn injury in adult sheep . The extremely low resistance of the circuit and gas exchanger ( 10 mm Hg) allowed a blood flow of up to 13% of the Fig 5. Throughout the 72-hour study, the heart rate and mean arterial pressure did not significantly change despite the arteriovenous shunt. Heart rate ranged from to beats/min and mean arterial pressure ranged from to mm Hg. (AVCO 2 R arteriovenous CO 2 removal.) cardiac output at a mean arterial blood pressure of 90 mm Hg . During this initial patient experience, AVCO 2 R achieved approximately 870 ml/min of flow with a transdevice pressure gradient of consistently less than 10 mm Hg to achieve approximately 70% CO 2 removal of measured CO 2 production. All patients on AVCO 2 R achieved either a decrease in ventilator pressures (patients 1 through 3) or a decrease in Paco 2 (patients 3 through 5) depending on the management strategy. Use of AVCO 2 R in patient 1 was associated with an immediate reduction in ventilator requirements and peak airway pressure. Although arterial Pco 2 ranged from 99 to 121 mm Hg during AVCO 2 R support, it was still significantly less than it would be without AVCO 2 R, as evidenced by a rapid increase of Pco 2 to 214 mm Hg immediately after removal of the cannulas. In patient 2, CO 2 removal up to 208 ml/min, or 94% CO 2 production, was achieved under conditions of moderate hypercapnia. This patient and those subsequent tolerated the 12F arterial cannula without vascular complications. We now
6 186 ZWISCHENBERGER ET AL Ann Thorac Surg PERCUTANEOUS AVCO 2 R FOR RESPIRATORY FAILURE 1999;68:181 7 recommend a 12F or larger arterial cannula to achieve shunt flows capable of adequate CO 2 removal during periods of high CO 2 production. Although CO 2 removal was consistent, as shown in Figures 1 and 2, the impact of AVCO 2 R on oxygen transfer was patient dependent. Removal of CO 2 allowed reductions in barotrauma and volume trauma from the ventilator; however, the Pao 2 to fraction of inspired oxygen ratio best reflected patient improvement or deterioration during ARDS as shown in Figure 4. In our large animal model of ARDS induced by a smoke inhalation and cutaneous burn injury , the Pao 2 to fraction of inspired oxygen ratio showed significant improvement during AVCO 2 R . We speculate this improvement in arterial oxygenation with AVCO 2 R is likely caused by two separate effects. With resolution of noncardiogenic pulmonary edema associated with the lung injury, one would expect a decrement in the right-to-left shunt fraction. Second, the use of AVCO 2 R would be expected to improve the central mixed venous oxygen tensions because of the admixture of well-oxygenated blood from AVCO 2 R mixing with the venous return to the right atrium. Percutaneous AVCO 2 R involves much simpler monitoring and maintenance than conventional ECMO. Our technique of AVCO 2 R still requires systemic anticoagulation with heparin, increasing the risk of bleeding. Heparin-coated or biopassive circuits and gas exchangers may decrease the blood foreign surface interactions and decrease the need for anticoagulation. Management issues encountered throughout this feasibility study include two minor complications: (1) an oxygenator thrombosis that required oxygenator change-out, and (2) cannula-site bleeding that could be controlled with a pressure bag. Questions encountered throughout the study included the range of heparin dosing that was ideal for this type of circuit, cannula size selection that would allow minimal access to the femoral artery and vein with adequate blood flow, and best ventilator management in the presence of 70% to 75% CO 2 removal by the AVCO 2 R circuit. We conclude, from our initial clinical experience, that percutaneous AVCO 2 R is feasible with only minor complications and achieves approximately 70% CO 2 removal in adults with severe respiratory failure and CO 2 retention to allow decreased barotrauma and volume trauma without hemodynamic compromise. In the future, we will conduct large animal prospective, randomized survival studies of AVCO 2 R to evaluate ideal heparin dosing, heparin-coated circuits, and biopassive-coated circuits. Human prospective, randomized, clinical trials will address populations of patients with ARDS with CO 2 retention, severe smoke inhalation and burn-associated ARDS, and CO 2 retention syndromes to avoid intubation. Supported in part by the Constance Marsili Shafer Research Fund and Avecor Cardiovascular, Inc. References 1. Demling RH. The modern version of adult respiratory distress syndrome. Annu Rev Med 1995;46: Tsuno K, Miura K, Takeya M, Kolobow T, Morioka T. Histopathologic pulmonary changes from mechanical ventilation at high peak airway pressures. Am Rev Respir Dis 1991;143: Dreyfuss D, Soler P, Basset G, Saumon G. High inflation pressure pulmonary edema. Respective effects of high airway pressure, high tidal volume, and positive endexpiratory pressure. Am Rev Respir Dis 1988;137: Hickling KG, Walsh J, Henderson S, Jackson R. Low mortality rate in adult respiratory distress syndrome using lowvolume, pressure-limited ventilation with permissive hypercapnia: a prospective study. Crit Care Med 1994;22: Amato MBP, Barbas CSV, Medeiros DM, et al. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998;338: Bidani A, Tzouanakis AE, Cardenas VJ Jr, Zwischenberger JB. Permissive hypercapnia in acute respiratory failure. JAMA 1994;272: Rappaport SH, Shpiner R, Yoshihara G, Wright J, Chang P, Abraham E. Randomized, prospective trial of pressurelimited versus volume-controlled ventilation in severe respiratory failure. Crit Care Med 1994;22: Milberg JA, Davis DR, Steinberg KP, Hudson LD. Improved survival of patients with acute respiratory distress syndrome (ARDS): JAMA 1995;273: Cardenas VJ Jr, Zwischenberger JB, Tao W, et al. Correction of blood ph attenuates changes in hemodynamics and organ blood flow during permissive hypercapnia. Crit Care Med 1996;24: Tao W, Brunston RL Jr, Bidani A, 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: Brunston RL Jr, Tao W, Bidani A, Alpard SK, Traber DL, Zwischenberger JB. Prolonged hemodynamic stability during arteriovenous carbon dioxide removal (AVCO 2 R) for severe respiratory failure. J Thorac Cardiovasc Surg 1997; 114: Brunston RL Jr, Tao W, Bidani A, Traber DL, Zwischenberger JB. Organ blood flow during arteriovenous carbon dioxide removal. ASAIO J 1997;43:M Brunston RL Jr, Tao W, Bidani A, Cardenas VJ Jr, Traber DL, Zwischenberger JB. Determination of low blood flow limits for arteriovenous carbon dioxide removal (AVCO 2 R). ASAIO J 1996;42:M Frank BA, Tao W, Brunston RL Jr, Alpard SK, Bidani A, Zwischenberger JB. High flow/low resistance cannulas for percutaneous arteriovenous carbon dioxide removal (AVCO 2 R). ASAIO J 1997;43:M Kolobow T, Moretti MP, Fumagalli R, et al. Severe impairment in lung function induced by high peak airway pressure during mechanical ventilation. An experimental study. Am Rev Respir Dis 1987;135: Parker JC, Hernandez LA, Peevy KJ. Mechanisms of ventilator-induced lung injury. Crit Care Med 1993;21: Barsch J, Birbara C, Eggers GWN. Positive pressure as a cause of respiratory failure induced lung disease. Ann Intern Med 1970;72: Hickling KG. Ventilatory management of ARDS: can it affect the outcome? Intensive Care Med 1990;16: Field D, Davis C, Elbourne D, Grant A, Johnson A, Macrae D. UK collaborative randomized trial of neonatal extracorporeal membrane oxygenation. UK Collaborative ECMO Trial Group. Lancet 1996;348: Zwischenberger JB, Bartlett RH. Extracorporeal circulation and oxygenation. In: Civetta JM, Taylor RW, Kirby RR, eds. Critical care. Philadelphia: JB Lippincott, 1992: Kolla S, Awad SS, Rich PB, Schreiner RJ, Hirschl RB, Bartlett
7 Ann Thorac Surg ZWISCHENBERGER ET AL 1999;68:181 7 PERCUTANEOUS AVCO 2 R FOR RESPIRATORY FAILURE 187 RH. Extracorporeal life support for 100 adult patients with severe respiratory failure. Ann Surg 1997;226: Conrad SA, Rycus PT. Extracorporeal life support. ASAIO J 1998;44: 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: Becker JA, Short BL, Martin GR. Cardiovascular complications adversely affect survival during extracorporeal membrane oxygenation. Crit Care Med 1998;26: Kolobow T, Gattinoni L, Tomlinson T, Pierce JE. An alternative to breathing. J Thorac Cardiovasc Surg 1978;75: Gattinoni L, Pesenti A, Mascheroni D, et al. Low-frequency positive-pressure ventilation with extracorporeal CO 2 removal in severe acute respiratory failure. JAMA 1986;256: Barthelemy R, Galletti PM, Trudell LA, et al. Total extracorporeal CO 2 removal in a pumpless artery-to-vein shunt. Trans Am Soc Artif Intern Organs 1982;28: Young JD, Dorrington KL, Blake GJ, Ryder WA. Femoral arteriovenous extracorporeal carbon dioxide elimination using low blood flow. Crit Care Med 1992;20: Awad JA, Deslauriers J, Major D, Guojin L, Martin L. Prolonged pumpless arteriovenous perfusion for carbon dioxide extraction. Ann Thorac Surg 1991;51: Goodin MS, Thor EJ, Haworth WS. Use of computational fluid dynamics in the design of the Avecor Affinity oxygenator. Perfusion 1994;9: Brunston RL Jr, Zwischenberger JB, Tao W, Cardenas VJ Jr, Traber DL, Bidani A. Total arteriovenous carbon dioxide removal (AVCO 2 R): simplifying extracorporeal support for severe respiratory failure. Ann Thorac Surg 1997;64: Alpard SK, Zwischenberger JB, Tao W, Deyo DJ, Traber DL, Bidani A. Dose dependent development of severe respiratory failure in an ovine model of smoke inhalation and cutaneous flame burn injury. Crit Care Med 1999 (in press). 33. Alpard SK, Zwischenberger JB, Tao W, Deyo DJ, Bidani A. Reduced ventilator pressure and improved P/F ratio during percutaneous arteriovenous carbon dioxide removal (AVCO 2 R) for severe respiratory failure. Ann Surg 1999 (in press). DISCUSSION DR GEOFFREY M. GRAEBER (Morgantown, WV): When you had those patients and they continued for a longer period of time, since you were removing about 70% of the CO 2, did you notice a progressive increase in the concentration of CO 2 in the blood, and if you did, were you able to manage it at a steady level over time? DR ZWISCHENBERGER: As you would expect, for the 3 patients who survived, the 72-hour window of AVCO 2 R seemed to buy them decreased barotrauma and decreased ventilator exposure that could exacerbate lung injury. We were able to maintain lower ventilator settings for both peak inspiratory pressure and minute ventilation yet maintain Paco 2 at levels below pre- AVCO 2 R. The 2 patients who eventually died were end-stage ARDS. During AVCO 2 R, their Paco 2 and ph were stable but became unstable immediately after stopping AVCO 2 R. The entry criteria were designed to identify patients at the end of therapy after every other effort had been made to treat their ARDS. As a result, we were really surprised that any patient survived. DR GRAEBER: So, obviously, the ideal application would come earlier in the course of ARDS after verification in prospective, randomized clinical trials. DR ZWISCHENBERGER: Yes by The Society of Thoracic Surgeons Ann Thorac Surg 1999;68: /99/$20.00 Published by Elsevier Science Inc PII S (99)