An Update on Interventional Lung Assist Devices and Their Role in Acute Respiratory Distress Syndrome

Transcription

1 Lung (2006) 184: DOI /s An Update on Interventional Lung Assist Devices and Their Role in Acute Respiratory Distress Syndrome Marc-Alexander von Mach Æ Joachim Kaes Æ Babatunde Omogbehin Æ Ingo Sagoschen Æ Jascha Wiechelt Æ Kristina Kaiser Æ Oliver Sauer Æ Ludwig Sacha Weilemann Accepted: 31 October 2005 Ó Springer Science+Business Media, Inc Abstract In recent years, pumpless arteriovenous systems for extracorporeal gas exchange have become a new therapeutic option for the treatment of patients suffering from acute respiratory failure. Experiences with the pumpless extracorporeal membrane lung in animal experiments and in patients with adult respiratory distress syndrome published in the current literature are reviewed. In addition this article presents a case of varicella pneumonia with persistent hypoxemia and hypercapnia under mechanical ventilation that showed a significant improvement with treatment with a pumpless extracorporeal lung assist using an arteriovenous shunt for eight days. The patient made a complete recovery. This is the first report of a patient with a life-threatening varicella pneumonia successfully treated with pumpless extracorporeal lung assist device. This review provides an update on interventional lung assist devices and a critical discussion of their advantages and limitations. Keywords Adult respiratory distress syndrome Æ Extracorporeal lung assist Æ Pumpless Æ Varicella zoster Æ Necrotizing pneumonia Introduction Despite developments of new techniques for mechanical ventilation with lung-protective concepts, the treatment of patients suffering from adult respiratory distress syndrome M.A. von Mach (&) Æ J. Kaes Æ B. Omogbehin Æ I. Sagoschen Æ J. Wiechelt Æ K. Kaiser Æ O. Sauer Æ L.S. Weilemann II. Medical Department, University Hospitals, Langenbeckstr. 1, Mainz, Germany marcm@giftinfo.uni-mainz.de (ARDS) is still a major problem. When pulmonary carbondioxide elimination is significantly limited, lung-protective concepts of ventilation cannot be followed because of the development of serious acidosis. The reduced oxygenation usually requires increased end-expiratory pressures which further limits carbondioxide elimination. In such cases a small number of specialized centers apply pump-driven venovenous extracorporeal membrane oxygenation (ECMO) with great technical effort. In order to reduce the technical expenses and also the accompanying complications, pumpless interventional lung assist devices have been developed. With these systems membrane perfusion is achieved passively by the blood pressure gradient after cannulation of the femoral artery and vein. Technical Aspects of an Interventional Lung Assist Device Pumpless extracorporal lung assist was enabled by the development of a membrane lung with minimal blood flow resistance. This was achieved with membranes based on heparin-coated hollow-fiber technology with reduced inner surface area (1.3 m 2 ) and optimized blood flow characteristics. By reduction of the blood flow resistance, these membrane lungs produced a pressure gradient of approximately 15 mmhg between in- and outflow [21]. After animal experiments and tests in highly selected patients, this membrane lung became commercially available in 2001 and was called Nova Breath Ò (Jostra Medizintechnik, Hirrlingen, Germany); since 2002 it is called termed Novalung Ò (Novalung GmbH, Hechingen, Germany) [9, 10]. The complete system is coated with heparin and therefore treatment can be performed without therapeutic anticoagulation. After percutaneous cannulation of the arteria and

2 170 Lung (2006) 184: vena femoralis using Seldinger s technique with large cannulas, the membrane lung is interposed. Its low flow resistance results in a decrease in arterial blood pressure of a maximum of 15 mmhg with a maximal blood flow of 3 L/ min. Therefore, the patient s heart can compensate for the extracorporeal pump of conventional lung assist devices. The main problems regarding a sufficient shunt flow are good cardiac function with a cardiac index (CI) of more than 3 L/min/m 2 and a mean arterial pressure (MAP) of more than 70 mmhg [22]. The blood flow through the membrane lung is determined by the MAP, the resistance of the arterial cannula, the membrane lung, the venous cannula, and the central venous pressure in a series connection [22]. The selection of the cannula is limited by the diameter of the cannulated blood vessels and results in blood flow rates of ml/min depending on the MAP and CJ. This represents 20% 30% of the cardiac output. With these flow rates elimination of more than 90% of the total amount carbondioxide produced, depending on the diffusion gradient and on the flow of fresh gas sent through the membrane lung, was demonstrated in sheep and pig models [12, 23]. This enabled a significant reduction of the respiratory minute volume and of the peak respiratory pressure [32]. The first use on humans could demonstrate similar effects, but to a lesser extent [35]. While the carbondioxide that is excreted can be eliminated almost completely by the membrane lung, the immediate impact of the assist device on oxygenation is quite limited. This phenomenon is a result of the times higher Krogh s diffusion coefficient of carbondioxide compared with oxygen [8]. However, the lower the arterial oxygen saturation of the patient, the greater the improvement of oxygenation with the membrane lung. In any case, mechanical ventilation needs to be continued. The efficiency of the pumpless extracorporeal lung assist in providing additional oxygenation to the patient can be calculated as [28] Efficacy ðmembrane lungþ ðs after O 2 SaO 2 Þ F ml = ðsao 2 S venous O 2 ÞCO where CO is cardiac output, F ml is blood flow through membrane lung, SaO 2 is arterial oxygenation saturation, S after O 2 is oxygenation saturation after the membrane lung, and S venous O 2 is mixed venous oxygenation saturation. Mechanical Ventilation When Using a Pumpless Extracorporeal Lung Assist Device The decoupling of ventilation and oxygenation allows several ventilation strategies. All strategies have a sufficient PEEP level in common to achieve an almost complete recruitment of the lung and an as-small-as-possible pressure amplitude. Whether continuous positive airway pressure (CPAP) with spontaneous breathing, bilevel positive airway pressure (BIPAP) with a low respiratory frequency and a reduced peak airway pressure (see case below), or high-frequency oscillation ventilation (HFOV [11]) is applied should be decided for each individual case. Systematic and controlled investigations are not available at present. The longest reported time a pumpless extracorproeal lung assist device Novalung was used 100 days [13]. When the modifications of the ventilator lead to an improvement of oxygenation with a stable PaO 2 of at least approximately 70 mmhg under a reduced peak airway pressure of less than 30 cmh 2 O and an oxygen delivery of the ventilator of 50% or less, the stepwise recoupling of oxygenation and ventilation may be initiated [21]. Weaning from the pumpless extracorporeal lung assist device is easier when at least some degree of spontaneous breathing can be established [8]. The weaning process is accompanied by a stepwise reduction of the gas flow in the membrane lung and by additional modifications of the mechanical ventilator, if required. This stepwise procedure is especially advantageous when acute alterations of the PaCO 2 should be prevented, for instance, after cerebral trauma or when signs of an increased right ventricular load are present. Clamping of the extracorporeal circulation should not be performed abruptly to prevent acute arterial hypertension. The cannulas are usually removed manually followed by sufficient and continuous compression of the insertion site for at least 30 min. Only in exceptional cases is surgical support (vascular surgery) required [28]. Pumpless Extracorporeal Lung Assist in Adult Respiratory Distress Syndrome The pumpless extracorporeal lung assist may be used in most patients with ARDS when a sufficient CI and MAP can be achieved. ARDS is characterized by impaired oxygenation accompanied by reduced pulmonary compliance; it infiltrates born lungs. A number of different diseases could potentially cause ARDS, in particular, pneumonia, sepsis, and polytrauma. Up to now no increase in survival of ARDS patients could be demonstrated in controlled trials with the use of pumpless extracorporeal lung assist devices. The devices represent the last option when other sophisticated mechanical ventilation techniques fail [6, 16, 26, 34]. The most common techniques used are positive end-expiratory pressure (PEEP) ventilation with reduced tidal volume and restricted peak respiratory pressure [1]. The pumpless extracorporeal lung assist increases the therapeutic options for patients with ARDS [13]. So far the most number of patients who have used the device are those with sepsis and pneumonia, the most common causes

3 Lung (2006) 184: of ARDS. In the study population of Reng et al. [28], no critically hemodynamic instability was observed with pumpless extracorporeal lung assist, even in serious diseases like pneumococci sepsis and Waterhouse Friedrichsen syndrome. The maximal required increase of catecholamines was 5%. Invasive monitoring of blood pressure is obligatory. In particular, in the initial phase monitoring of cardiac output may be indicated for patients with unstable hemodynamics because the relationship between cardiac output and blood flow in the membrane lung determines the efficacy of the pumpless extracorporeal lung assist. Whether cardiac output measurements are performed using catheterization of the pulmonary artery, continuous pulse contour analysis, or echocardiography depends on the individual case. Pumpless extracorporeal lung assist should not be used in septic shock because the hemodynamic sequelae cannot be estimated. Patients with polytrauma with respiratory failure should particularly benefit from the treatment with pumpless extracorporeal lung assist because they will not be exposed to an increased risk of bleeding. Therapeutic heparinization is not required and therefore alterations of the coagulation system, which are caused by an extracorporeal pump, are absent [7]. This is particularly advantageous to patients with brain injury [5]. For this group of patients a protective ventilation strategy can be pursued despite reduced cerebral compliance because hypercapnia can be prevented by carbondioxide elimination via the pumpless extracorporeal lung assist [3]. However, when the patient suffers from irreversible cerebral damage or extended disease of the peripheral arteries, pumpless extracorporeal lung assist is contraindicated. Previously diagnosed heparin-induced thrombocytopenia type II (HIT II) is a contraindication because the system is coated with heparin. For the treatment of decompensated chronic obstructive pulmonary disease (COPD), an extracorporeal assist device is technically a therapeutic option. However, this indication should be assessed extremely critically because the prognosis regarding recovery of the lungs may be quite poor [8]. Pumpless Extracorporeal Lung Assist in Varicella Pneumonia Data from Europe and North America have shown that the incidence of chickenpox in adults has doubled in recent years, paralleled with an increase in hospital admissions [25]. Varicella pneumonia represents a severe complication of varicella zoster infection and typically occurs in adults. It has an overall mortality rate between l0% and 30% which increases to 50% when mechanical ventilation is necessary [24]. We present the case of a 23-year-old man who was admitted with a five-day history of the characteristic rash of varicella and two days of cough, hemoptysis, and increasing dyspnea. On presentation, the patient was in respiratory distress with an oxygen saturation of 89% with 8 L of oxygen per minute via a facial mask. Chest X-ray revealed bilateral interstitial infiltrates (Fig. 1). Microbiological testing of the tracheal fluid revealed varicella zoster DNA and serology showed a strongly positive varicella zoster IgM. Varicella pneumonia was suspected and intravenous treatment with 800 mg of acyclovir every 6 hours was initiated. Sixteen hours after admission the patient developed acute respiratory failure and required intubation. Computed tomography of the chest revealed necrotizing pneumonia with consolidation in the right upper lobe (Fig. 1). During the following 12 days the blood gases could not be sustainably improved, there was a fluctuating FiO 2 requirement between 65% and 100%, and hypercapnia was up to 90 mmhg. As conventional pressure controlled ventilation with increasing airway pressures of up to 36/20 cmh 2 O and low tidal volumes with a respiration frequency of 30 min )1 failed to improve blood gases, a pumpless extracorporal lung assist device (Novalung) was installed between the right femoral artery and the left femoral vein (Fig. 2). After the pumpless extracorporeal lung assist device was started, mechanical ventilation (bilevel positive airway pressure, BIPAP) was modified to minimize alveolar shear forces. The peak airway pressure of 32 was reduced to 24cm H 2 O and the respiration frequency decreased to 10 min )1 with a maintained positive endexpiratory pressure of 18 cmh 2 O. This resulted in a reduction of the respiratory volume from 19 to 3.5 L min )1 with sufficient oxygenation and carbon dioxide elimination. During eight days of pumpless extracorporeal lung assist without complications, the clinical situation steadily improved with decreasing infiltrates in chest X-ray investigations. After removal of the pumpless extracorporeal lung assist device, the patient was rapidly weaned from the ventilator and extubated after 23 days of mechanical ventilation. The patient made a complete recovery and was transferred from the intensive care unit on day 31. We present a case in which there was the need for mechanical ventilation because of ARDS [19], which has recently been reported to be observed in about 20% of patients with adult varicella pneumonia [14]. In particular necrotizing pneumonia with consolidation of a whole pulmonary lobe as diagnosed by computed tomography has recently been described to be a fatal complication [30]. As described previously, a pumpless extracorporeal lung assist using an arteriovenous shunt enabled less aggressive ventilator therapy in ARDS due to aspiration pneumonia [11]. No data regarding varicella pneumonia and extracorporal lung assist have been published, just animal data and some limited studies in patients with ARDS caused by different

4 172 Lung (2006) 184: Fig. 1 Chest X-ray with bilateral interstitial infiltrates (upper panel). Computed tomography of the chest with necrotizing pneumonia and consolidation in the right upper lobe. etiologies ([2, 15], Table 1). Typical complications associated with the pumpless extracorporeal lung assist included oxygenator failure, cannula problems, and thrombus formation, which were not observed in the present patient. We hypothesize that in the present case the reduction of the peak airway pressure and respiration frequency under the pumpless extracorporeal lung assist might have helped minimize alveolar damage and, hence, support pulmonary recovery. Therefore, this new technique could represent a useful option in critical cases of varicella pneumonia in which there is significant pulmonary consolidation. Practical Application of Pumpless Extracorporeal Lung Assist The low flow resistance of the membrane lung is achieved by sophisticated production of the applied capillary membrane. In addition, the surface of the membrane is coated with a protein matrix to which heparin is bound covalently and ionized. The filling volume of the system is only 175 ml so that the application of blood products is not required. Before the cannulation of the femoral artery and vein, the diameters of these vessels are determined

5 Lung (2006) 184: Fig. 2 Pumpless interventional lung assist device placed between the right femoral artery and the left femoral vein. using ultrasonography. The cannulas, which are available in the sizes 13, 15, 17, 19, and 21, are chosen according to the diameters. No more than 80% of the arterial vessel volume should be filled by the cannula. To optimize the outlet flow the venous cannula should be larger than the arterial cannula [8]. After sterile preparation of both groins and extensive covering, the percutaneous puncturing of the artery and the vein is performed. Using the minimally invasive Seldinger s technique stepwise dilatations are performed until the previously selected cannula size can be inserted. After insertion the cannula is rinsed, clamped, and fixed with a suture. Monitoring of leg perfusion with the arterial cannula by clinical examination and pulse oximetry is extremely important [4]. After this preparation the membrane lung, which has been filled with 0.9% sodium chloride solution and bled is interposed [4]. After loosening the clamps and establishing the blood flow, the gas inflow into the membrane lung is connected to oxygen (10 20 L/min, Fig. 2, [28]. The blood flow through the membrane lung is measured continuously by an ultrasonography sensor, which was installed at the venous tube of the extracorporeal system (Fig. 2). Besides the continuous blood flow measurement, the efficacy of the membrane lung can also be monitored by intermittent blood gas analyses from the venous tube. Because the tubes as well as the membrane lung are coated with heparin, only a low-dose heparinization is recommended during treatment with a pumpless extracorporeal lung assist [27, 33]. Complications of Pumpless Extracorporeal Lung Assist Maintenance of hemodynamic stability during application of the pumpless extracorporeal lung assist has been demonstrated in animal experiments [9, 17]. Supporting arterial blood pressure, if required, should be performed with a 1 -sympathomimetics rather than with b-sympathomimetics. The former leads to an increase in shunt volume through the membrane lung, while the latter causes an increase only in the cardiac output and, therefore, only the efficacy of the pumpless extracorporeal lung assist with unchanged or even reduced shunt volume is decreased [18]. Hemodynamic instability during initiation (hypotension) and completion (hypertension) of the pumpless extracorporeal lung assist can be prevented by stepwise loosening or fixing of the clamps. Using a standardized monitoring protocol, system-dependent complications like thrombosis should be recognized. Particular attention should be paid to the leg with the arterial cannula [21]. Besides continuous pulse oximetric monitoring, inspection (no covering of the leg!) and palpation as part of a regular examination for signs of reduced perfusion and compartment syndrome are obligatory [laboratory monitoring of the creatine kinase (CK)]. If doubt, Doppler sonography and tissue PO 2 /PCO 2 measurements can be helpful [8]. Blood counts should be controlled daily and a hiparin-induced thrombocytopenia type II (HIT II) represents a critical complication in which the system should be stopped immediately. This complication might be reduced by the development of systems coated with low-molecular-weight heparins instead of unfractionated heparin [29]. Conclusion Extracorporeal lung assist represents a newly therapeutic option for patients with adult respiratory distress syndrome. The device is easy to operate and no additional pumps or heating systems are required. Patients who have had cardiac failure should not use this device. Worldwide only a few hundred patients have been treated with pumpless extracorporeal lung assist so far (Table 1). The pumpless extracorporeal membrane lung could be used in hospitals where extracorporeal membrane oxygenation (ECMO) is not available [6]. Nevertheless, whether pumpless extracorpereal lung assist is beneficial still not clear [20, 31]. Of particular interest would be the determination of the optimal ventilation and weaning strategy. In addition the complication rate of pumpless extracorporeal lung assist (infections, ischemia, thrombus formation, embolization) should be determined in large clinical studies. Furthermore, the application of pumpless extracorporeal lung assist during pulmonary surgery or as a bridging procedure before lung transplantation should be evaluated. In catastrophies such as terror attacks, with high numbers of patients with pulmonary failure, the use of this device could rapidly be established [8]. Future

6 174 Lung (2006) 184: Table 1 Human studies with pumpless extracorporeal lung assist except case reports Reference Number of patients System Survival rate Observations Zwischenberger 5 Non-commercial 100% No complications observed et al Reng et al Nova Breath Ò 70% Not specified number of patients with clotting in the assist device; 1 patient who was switched from conventional pump-driven to pumpless lung assist due to hemolysis Liebold et al Nova Breath Ò 60% Clotting in assist device in 7 patients, plasma leakage of the device in 2 patients, candida contamination in the assist device in 1 patient Liebold et al Nova Breath Ò 36% Complications in 15 patients: Clotting in the assist device in 8 patients, HIT II in 1 patient, limb ischemia in 3 patients, 5 patients with plasma leakage of the device Bein et al Novalung Ò 50% Complications in 10 patients: difficult cannulation in 4 patients, hemodynamic instability in 4 patients, clotting or plasma leakage of the device in 3 patients, limb ischemia in 3 patients, surgical intervention after removal of the cannulas in 3 patients Bein et al Novalung Ò 80% Only patients with brain injury and adult respiratory distress syndrome; 1 patient with ischemia of the leg after removal of the arterial cannula due to a stenosis of the femoral artery which was treated surgically (venous patch) 6 studies 140 Nova Breath Ò 3 Novalung Ò % Complication rate: 36/130 = 27.7% HIT II: heparin-induced thrombocytopenia type II investigations should reveal whether this procedure can be used in the initial stage of treatment of patients with pulmonary failure. References 1. Amato MB, Barbas CS, Medeiros DM, et al. (1998) Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 338: Awad JA, Deslauriers J, Major D, Guojin L, Martin L (1991) Prolonged pumpless arteriovenous perfusion for carbon dioxide extraction. Ann Thorac Surg 51: Bein T, Kuhr LP, Metz C, et al. (2002) ARDS and severe brain injury. Therapeutic strategies in conflict. Anaesthesist 51: Bein T, Prasser C, Philipp A, et al. (2004) Pumpless extracorporeal lung assist using arteriovenous shunt in severe ARDS. Experience with 30 cases. Anaesthesist 53: Bein T, Scherer MN, Philipp A, Weber F, Woertgen C (2005) Pumpless extracorporeal lung assist (Pecla) in patients with acute respiratory distress syndrome and severe brain injury. J Trauma 58: Bensberg R, Dembinski R, Kopp R, Kuhlen R (2005) Artificial lung and extracorporeal gas exchange. Panminerva Med 47: Brederlau J, Anetseder M, Wagner R, et al. (2004) Pumpless extracorporeal lung assist in severe blunt chest trauma. J Cardiothorac Vase Anesth 18: Brederlau J, Anetseder M, Mullenbach R, et al. (2005) The present role of interventional lung assist (ILA) in critical care medicine. Anasthesiol Intensivmed Notfallmed Schmerzther 40: Brunston RL Jr, Tao W, Bidani A, et al. (1997) Prolonged hemodynamic stability during arteriovenous carbon dioxide removal for severe respiratory failure. J Thorac Cardiovasc Surg 114: Brunston RL Jr, Zwischenberger JB, Tao W, et al. (1997) Total arteriovenous CO 2 removal: simplifying extracorporeal support for respiratory failure. Ann Thorac Surg 64: David M, Heinrichs W (2004) High-frequency oscillatory ventilation and an interventional lung assist device to treat hypoxaemia and hypercapnia. Br J Anaesth 93: De Somer F, van Belleghem Y, Foubert L, et al. (1999) Feasibility of a pumpless extracorporeal respiratory assist device. J Heart Lung Transplant 18: Dschietzig T, Laule M, Melzer C, Baumann G (2005) Long-term treatment of severe respiratory failure with an extracorporeal lung assist a case report. Intensivmed 42: Frangides CY, Pneumatikos I (2004) Varicella-zoster virus pneumonia in adults: report of 14 cases and review of the literature. Eur J Intern Med 15: Gattinoni L, Kolobow T, Agostoni A, et al. (1979) Clinical application of low frequency positive pressure ventilation with extracorporeal CO 2 removal (LFPPV-ECCO2R) in treatment of adult respiratory distress syndrome (ARDS). Int J Artif Organs 2: Gattinoni L, Pesenti A, Mascheroni D, et al. (1986) Low-frequency positive-pressure ventilation with extracorporeal CO 2 removal in severe acute respiratory failure. JAMA 256: Jayroe JB, Alpard SK, Wang D, et al. (2001) Hemodynamic stability during arteriovenous carbon dioxide removal for adult respiratory distress syndrome: a prospective randomized outcomes study in adult sheep. ASAIO J 47: Jayroe JB, Wang D, Deyo DJ, et al. (2003) The effect of augmented hemodynamics on blood flow during arteriovenous carbon dioxide removal. ASAIO J 49: Leiba A, Apter S, Leiba M, Thaler M, Grossman E (2004) Acute respiratory failure in a patient with sarcoidosis and immunodeficiency an unusual presentation and a complicated course. Lung 182: Lewandowski K, Rossiant R, Pappert D, et al. (1997) High survival rate in 122 ARDS patients managed according to a clinical algorithm including extracorporeal membrane oxygenation. Int Care Med 23:

7 Lung (2006) 184: Liebold A, Reng CM, Philipp A, Pfeifer M, Birnbaum DE (2000) Pumpless extracorporeal lung assist experience with the first 20 cases. Eur J Cardiothorac Surg 17: Liebold A, Philipp A, Kaiser M, et al. (2002) Pumpless extracorporeal lung assist using an arterio-venous shunt. Applications and limitations. Minerva Anestesiol 68: Lynch WR, Montoya JP, Brant DO, et al. (2000) Hemodynamic effect of a low-resistance artificial lung in series with the native lungs of sheep. Ann Thorac Surg 69: Mer M, Richards GA (1998) Corticosteroids in life-threatening varicella pneumonia. Chest 114: Mohsen AH, McKendrick M. (2003) Varicella pneumonia in adults. Eur Respir J 21: Morris AH, Wallace CJ, Menlove RL, et al. ( 1994) Randomized clinical trial of pressure-controlled inverse ratio ventilation and extracorporeal CO 2 removal for adult respiratory distress syndrome. Am J Respir Crit Care Med 149: Murphy JA, Savage CM, Alpard SK, et al. (2001) Low-dose versus high-dose heparinization during arteriovenous carbon dioxide removal. Perfusion 16: Reng M, Philipp A, Kaiser M, et al. (2000) Pumpless extracorporeal lung assist and adult respiratory distress syndrome. Lancet 356: Schinzel H, Berghoff K, Beuermann I, et al. (in press) Anticoagulation with low-molecular-weight heparin (dalteparin) in plasmapheresis therapy initial experience. Transfusion (DOI: /j x) Published online 30. Stride PJ, Campher MJ, Geary JM, Coulter C, Duhig EE (2004) Adult chickenpox complicated by fatal necrotising pneumonia. Med J Aust 181: Suter M (1997) Therapeutic strategies leading to a better lung function in severe acute pulmonary failure. Intensivmed 34: Tao W, Brunston RL Jr, Bidani A, et al. (1997) Significant reduction in minute ventilation and peak inspiratory pressures with arteriovenous CO 2 removal during severe respiratory failure. Crit Care Med 25: Wahba A, Philipp A, Behr R, Birnbaum DE (1998) Heparincoated equipment reduces the risk of oxygenator failure. Ann Thorac Surg 65: Zapol WM, Snider MT, Hill JD, et al. (1979) Extracorporeal membrane oxygenation in severe acute respiratory failure. A randomized prospective study. JAMA 242: Zwischenberger JB, Conrad SA, Alpard SK, Grier LR, Bidani A (1999) Percutaneous extracorporeal arteriovenous CO 2 removal for severe respiratory failure. Ann Thorac Surg 68: