Secondary Logo

Journal Logo

Adult Circulatory Support

Clinical Outcomes of Advanced Heart Failure Patients with Cardiogenic Shock Treated with Temporary Circulatory Support Before Durable LVAD Implant

Shah, Palak*; Smith, Sara; Haft, Jonathan W.; Desai, Shashank S.*; Burton, Nelson A.§; Romano, Matthew A.; Aaronson, Keith D.; Pagani, Francis D.; Cowger, Jennifer A.

Author Information
doi: 10.1097/MAT.0000000000000309
  • Free


Implantation of durable ventricular assist devices (dVADs) has grown exponentially over the past decade.1 Parallel to this growth has been an increased level of sophistication with regards to candidate selection. When applying the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) patient profiles to classify severity and acuity of illness,2 mortality and morbidity are greatest in the INTERMACS profile 1 (“Critical Cardiogenic Shock”).3 Recognition of a poor survival (ranging from 65% to 76% at 1 year) in the most ill patients has resulted in a shift from implanting dVADs in patients characterized by INTERMACS profile 1 toward patients characterized by a lesser acuity and severity of illness, i.e., INTERMACS profiles 2 or 3.1,3 As such, INTERMACS profile 1 patients represented 30% of dVAD implants in the INTERMACS registry in 2008 and only 15% of dVAD implants in 2013.1

The concept of “bridging” a patient with temporary circulatory support (TCS) to stabilize hemodynamics and improve end-organ function before implantation of a dVAD is increasingly being employed.4 Coincident with the reduction in dVAD implants for critical cardiogenic shock, the use of TCS devices between 2008 and 2013 has increased at least 10-fold.5 This bridge strategy allows for varying degrees of ventricular support through rapidly deployable TCS devices. Ideally the bridge strategy would stabilize hemodynamics, improve end-organ function, and reduce surgical risk before dVAD implant. The most commonly employed TCS device is an intraaortic balloon pump (IABP), which has limited ability to augment cardiac output.6 Other TCS devices employed in a bridge strategy include extracorporeal membranous oxygenation (ECMO), extracorporeal ventricular assist devices (VADs) placed in the operating room, and more recently, extracorporeal VADs placed with percutaneous techniques.

Given the significant morbidity and costs of dVAD therapy, an ideal TCS device would not only provide sufficient cardiac support to improve hemodynamics before dVAD implant but would also improve outcomes after dVAD implant. Potentially, a TCS strategy could favorably change the risk associated with dVAD implant in a patient presenting with critical cardiogenic shock (INTERMACS profile 1) to the risk of dVAD implant associated with a patient with profile 2 or 3 characteristics. The purpose of the current investigation is to discern the impact of TCS before dVAD implant. We hypothesized that patients supported with TCS would have superior survival to that of profile 1 patients without TCS (65–76%) and similar to that of Profile 2–3 patients because of improved cardiopulmonary and end-organ status imparted by TCS.

Materials and Methods

The study was a two-center (Inova Heart and Vascular Institute, Falls Church, VA; University of Michigan, Ann Arbor, MI), retrospective, observational study, using prospectively collected data of patients who received dVAD therapy with or without prior TCS. Only patients who received a TCS device after presenting with critical cardiogenic shock evidenced by end-organ dysfunction and hemodynamic instability and then survived to dVAD were included in the study. The TCS devices included ECMO and the TandemHeart (TH; CardiacAssist Inc., Pittsburgh, PA). We excluded CentriMag (Thoratec Corp., Pleasanton, CA) and Impella (Abiomed Inc., Danvers, MA) because of very low numbers at both institutions (n = 4) to obtain a more homogenous population. Intraaortic balloon pumps were not considered a TCS device for the purposes of this analysis, given the aforementioned limitations of these devices to provide full hemodynamic support. Device selection was driven by institutional and clinician preference.

Patients were included in the study from March 2009 to May 2014 at Inova Fairfax Hospital and from September 2004 to September 2012 at the University of Michigan with follow-up continuing until August 1, 2014, for all patients. Patients who received a TCS device but did not receive a continuous-flow dVAD were excluded as the intent of the analysis was to investigate clinical stability before dVAD and analyze post-dVAD outcomes. Comparator groups included patients with profile 1 characteristics not receiving TCS and patients with profile 2–3 characteristics at the time of dVAD implant.

Baseline demographics, laboratory values, implant characteristics, and follow-up events for dVAD recipients were obtained from each institution’s mechanical circulatory support registry. Additional clinical characteristics before initiation of the TCS device were collected by systematic review of electronic health records. Interagency Registry for Mechanically Assisted Circulatory Support profile was assigned at the time of dVAD implant by the implanting surgeon or heart failure cardiologist. HeartMate II risk score was calculated at the time of dVAD implant according to methods previously published.7 The protocol was approved by both institutional review boards and complied with ethical guidelines outlined in the 1975 Declaration of Helsinki and the Health Insurance Portability and Accountability Act.

Definition of Post-dVAD Outcomes

The effect of TCS devices on post-dVAD operative and long-term mortality was assessed. Operative mortality was defined as death within 30 days or during the index implant hospitalization. Renal failure was defined by the need for renal replacement therapy after dVAD implant. Right-sided circulatory failure was defined according to INTERMACS: 1) need for right ventricular assist device (RVAD; temporary or permanent); 2) prolonged inotropic support (>7 days), or 3) prolonged inhaled nitric oxide (>7 days) with clinical signs of right heart failure.

Statistical Analysis

Baseline data are presented as the mean with standard errors or median with interquartile ranges for continuous variables based on normality of the distribution. Normality testing of continuous variables was determined using the Shapiro–Wilk test. Missing variables (n < 5) were derived by inputting the median of the remaining values. Mean values were compared using the independent samples t-test or analysis of variance method and medians with the Mann–Whitney or Kruskal–Wallis test. For categorical values, numbers with proportions are presented. Proportions were compared using Fisher’s exact test.

Multivariable Cox proportional hazard models were used to compare survival among the various TCS groups. Kaplan–Meier methods were used to estimate event-free survival, and survival was compared with Breslow testing.8 Patients were censored at the time of transplant or explant for recovery. A two-sided p value less than 0.05 was considered to represent statistical significance. All data were analyzed using the SPSS 22.0 (IBM Corp., Armonk, NY).


Patient Population

During the study period, 427 patients received a dVAD implant (Figure 1). Seventeen percent of these patients (n = 73) were identified as being in critical cardiogenic shock (INTERMACS profile 1) at the time of dVAD implant. Forty patients received TCS before dVAD (ECMO, n = 14 and TH, n = 26). An additional 29 profile 1 and 269 profile 2–3 patients without TCS formed comparator groups. Patient characteristics are outlined in Table 1. Profile 1 patients with and without TCS were more likely to be implanted as bridge to transplant and less likely to have an implantable defibrillator when compared with Profile 2–3 (p < 0.01).

Table 1
Table 1:
Baseline Patient Characteristics
Figure 1
Figure 1:
Patient allocation from clinical registries. Patients were selected from two institutions based on the use of TCS. Patients without TCS were classified according to the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) profile 1 or profile 2–3. Outcomes after dVAD are reported in follow-up for all patients. dVAD, durable ventricular assist device; TCS, temporary circulatory support.

Hemodynamics and End-Organ Function Before TCS

Hemodynamics and end-organ function before TCS support are presented in Table 2. The duration of TCS device support before dVAD was similar between ECMO (median, 5 days; range, 4–7) and TH (median, 4 days; range, 3–6; p = not significant [NS]). Patients were characterized by tachycardia with a depressed cardiac index and evidence of biventricular dysfunction. These hemodynamic derangements persisted despite the use of multiple inotropes in 40% of patients receiving TCS. There was a higher utilization of mechanical ventilation and vasopressors in ECMO patients when compared with TH (p < 0.01; Table 3). Serum sodium was reduced (133 mmol/L), and renal function (creatinine, 1.8 mg/dl) and liver function were impaired (Model for End-Stage Liver Disease score, 16 points). Intraaortic balloon pumps were used before TCS in 21% of patients receiving ECMO and 69% of patients receiving TH. Cardiac index and international normalized ratio (INR) were significantly higher in the ECMO group compared with TH (p = 0.02 and 0.04, respectively). Platelet counts were reduced in TH patients compared with ECMO (p = 0.01). No other clinically significant differences were observed between groups with respect to hemodynamic derangements or end-organ function before TCS.

Table 2
Table 2:
Hemodynamics and End-Organ Function of Patients Before Receiving Temporary Circulatory Support Device
Table 3
Table 3:
Hemodynamics, Cardiopulmonary Support, and End-Organ Function Before Durable VAD Implant

Patient Characteristics After TCS and Before dVAD

After TCS device insertion, heart rate was consistently higher in ECMO, TH, and profile 1 patients compared with profile 2–3 patients (Table 3, p < 0.01). Compared with TH, cardiac output was lower, and pulmonary capillary wedge pressure (PCWP) was higher in ECMO-supported patients (p < 0.01). The central venous pressure:PCWP pressure ratio was lower in the ECMO-supported patients compared with TH (p < 0.01). The use of an IABP was higher in profile 1 patients compared with profile 2–3 patients (83 vs. 45%; p < 0.01).

When comparing hemodynamic parameters between ECMO, TH, and profile 1 patients without TCS, cardiac output was higher and PCWP and mean pulmonary artery pressure lower in those on TCS (p < 0.01). There was no significant difference in pre-dVAD hemodynamics (cardiac output, right atrial pressure, PCWP) between profile 2–3 patients and those patients on ECMO (p > 0.05). TandemHeart was associated with improved pre-dVAD hemodynamics (cardiac output, mean pulmonary pressure, PCWP) when compared with profile 2–3 patients (p < 0.05).

Hematocrit and platelet counts were lower in TCS patients compared with profile 1–3 patients without TCS. Biochemical markers of end-organ function (aspartate aminotransferase, bilirubin, creatinine, and INR) showed greater impairment in profile 1 patients with or without TCS compared with profile 2–3 (p < 0.01). Among profile 1 patients, end-organ function was similar between TH and profile 1 patients without TCS (p = NS). When comparing ECMO with TH or profile 1 patients without TCS, end-organ function was consistently impaired (p < 0.01). The pre-dVAD HeartMate II risk score was highest in the ECMO (median, 2.36), indicating a higher severity of illness in ECMO patients immediately before dVAD implant when compared with TH (1.35), profile 1 (1.66), or profile 2–3 (1.13) patients (p < 0.01).

Patient Outcomes After dVAD Implant

The incidence of INTERMACS-defined right-sided circulatory failure was highest with ECMO, largely because of the use of inotropes >7 days (79%, p < 0.01; Table 4). However, no patients were discharged on inotropes, and the rate of temporary or permanent RVAD was similar among ECMO, TH, and profile 1 patients (p = NS). Interagency Registry for Mechanically Assisted Circulatory Support profile 2–3 patients had an RVAD implant cumulative incidence of 0.8% (p < 0.01). The frequency of postoperative renal failure after dVAD was similar between profile 1 patients with and without TCS, but significantly higher than patients in profile 2–3 (12–29 vs. 2%, p < 0.01). Intensive care unit length of stay after dVAD implant was a median of 7 days in the profile 2–3 group compared with a median of 13 and 22 days in the two TCS groups (p < 0.01). Overall hospital length of stay was similarly longer in the TCS and profile 1 patients.

Table 4
Table 4:
Postdurable Ventricular Assist Device Outcomes

Operative mortality after dVAD implant was highest in the ECMO group at 43% compared with all other groups (p < 0.01). Overall 90 day survival for ECMO (65%), TH (78%), and profile 1 patients (84%) was inferior to that of profile 2–3 patients 96% (p < 0.001; Figure 2). When comparing short-term survival estimates in profile 1 patients with and without TCS using Breslow testing, which provides more weight to deaths at early time points, we found no difference in early survival (p > 0.5; Figure 2).8

Figure 2
Figure 2:
Patient survival after dVAD by TCS strategy and Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) profile. Survival is compared among the two investigated TCS strategies (ECMO and TH) and the INTERMACS profile 1 and profile 2–3 groups without TCS. dVAD, durable ventricular assist device; TCS, temporary circulatory support; TH, TandemHeart.

Controlling for known clinically significant risk factors for VAD mortality (age, bridge to transplant (BTT), TCS, INTERMACS profile, and HeartMate II risk score (HMRS)) the adjusted risk for TCS use was evaluated. Compared with profile 2–3 patients, profile 1 status (without TCS) affords a 2.2 (95% confidence interval [CI], 1.1–4.3, p = 0.02) higher adjusted risk for mortality. The use of ECMO was associated with a 3-fold higher risk of death after VAD (hazard ratio, 3.1; 95% CI, 1.2–8.0; p = 0.02) compared with profile 2–3. TandemHeart did not predict mortality (p > 0.05). There was no significant difference in adjusted mortality in patients on ECMO compared with profile 1 patients without TCS. These data suggest that ECMO affords survivals equivalent to profile 1 and not similar to profile 2–3.


A significant portion of the mortality associated with dVAD placement occurs in the immediate postoperative period. Studies have cautioned the implantation of dVADs in patients with critical cardiogenic shock as 1 year mortality rates approach 35%.3 There has been a migration away from placing durable, costly devices in the sickest patients.1 Untreated, there is high mortality without mechanical circulatory support, and alternatives to dVADs are increasingly being used to manage this patient population.5,9 Temporary circulatory support devices appear to be an ideal strategy as these devices are easily deployed, less invasive, and have significantly lower costs compared with dVADs.10 We investigated TCS strategies used to improve the hemodynamic derangement associated with critical cardiogenic shock and compared clinical stability before dVAD implant as well as postimplant outcomes in patients with profile 1 and profile 2–3 characteristics without TCS. Our analysis indicates that 1) patients on TCS before dVAD implant are sicker at baseline then profile 1 patients without TCS, with a higher use of mechanical ventilation and vasopressors; 2) TCS with ECMO and TH improves hemodynamic parameters before dVAD at levels similar to profile 2–3 patients and better than profile 1 patients without TCS; 3) despite an improvement in hemodynamic parameters with TCS, end-organ function was still impaired with ECMO patients having the highest severity of illness pre-dVAD; and 4) outcomes after dVAD in TCS patients are similar to profile 1 patients without TCS.

Although TCS was able to successfully stabilize patients before durable VAD implantation, post-dVAD outcomes paralleled profile 1 patients without TCS. The incidence of right ventricular failure necessitating RVAD implantation, renal failure requiring renal replacement therapy, and mortality after dVAD implant was consistently higher in patients with TCS and profile 1 patients without TCS compared with profile 2–3 patients. When analyzing survival after dVAD, there was no statistical difference between TCS strategies and profile 1 patients, although ECMO patients tended to have a higher 90 day and 1 year mortality. Thus, the higher preoperative HeartMate II risk score likely translated into inferior survival both short- and long-term with ECMO.

The current INTERMACS protocol allows for any profile 1, 2, or 3 patient to be assigned the TCS modifier at the time of dVAD implant. If a clinician believes that hemodynamics and end-organ function have been stabilized with TCS support from profile 1 status, the clinician can upgrade the patient to a less-ill profile (profile 2–3) at the time of dVAD implant, simultaneously ascribing the TCS modifier. The rationale is that the hemodynamic improvement reflects an improved risk compared with the pre-TCS patient profile. With overall patient stabilization, the expected outcome by many would be a reduction in postoperative morbidity and mortality compared with profile 1 patients without TCS. We would expect their surgical risk to be lower and their outcomes to be more akin to an INTERMACS profile 2 or 3 patient. Despite the appearance of clinical stability and “improved” dVAD candidacy, operative mortality remained high in our study for TCS patients, paralleling that of profile 1 without TCS. On the basis of on our findings, patients receiving TCS devices for clinical deterioration should be designated as INTERMACS profile 1 regardless of the degree of hemodynamic improvement afforded by TCS. Patients receiving TCS before dVAD implant do not appear to have less risk compared with patients in profiles 2–3. Given the increased risk for biventricular failure with TCS, we avoid TCS implantation in destination therapy patients who are unlikely to be transplanted and for whom prolonged biventricular support may not be a good treatment option.

The failure to improve post-dVAD outcomes in advanced heart failure patients with TCS is similar to shock associated with acute myocardial infarction. In a meta-analysis of cardiogenic shock patients who were randomized to receive a percutaneous ventricular assist device (pVAD) or IABP, patient clinical outcomes were similar between groups despite improvement in hemodynamics on pVAD support.11Table 5 outlines selected major studies in the field of TCS. The majority of the studies include patients with acute myocardial infarction, which is a patient population distinctly different from our predominantly end-stage chronic cardiomyopathy patients with little potential for short-term myocardial recovery and simultaneously greater potential for acute and chronic end-organ instability. Further, these previous studies typically enrolled very few patients who received a dVAD, and none of the studies report detailed outcomes post-dVAD aside from survival.

Table 5
Table 5:
Major Recent TCS Studies

Clinical criteria for initiation of TCS are hard to determine. There are certain absolute indications (cardiac arrest, circulatory compromise despite inotropes, increased lactate level, vasopressor support) and contraindications (severe peripheral arterial disease, advanced age, frailty) that most clinicians would agree on. In our analysis, we were unable to determine which criteria led to each individual being initiated on TCS, but it was likely a combination of many of the aforementioned criteria. TandemHeart does require some element of patient stability as the device requires an intraatrial septal puncture under transesophageal or intracardiac echocardiographic guidance to be performed to place the inflow cannula into the left atrium, whereas ECMO can be started peripherally at bedside in a crashing patient. Of note, the majority of our ECMO patients came from Inova Heart and Vascular Institute and TH patients from the University of Michigan, reflecting institutional expertise with each TCS strategy.

It could be argued that inferior outcomes of patients on TCS may be caused by short support durations, allowing for only partial recovery of end-organ function. Although hemodynamic parameters and end-organ function did improve with TCS, values were consistently impaired before dVAD when compared with the profile 2–3 group. Thus, a longer duration of support may allow for complete end-organ recovery that could translate into improved post-dVAD outcomes. However, increased TCS support duration must be weighed against the risks of prolonged exposure to TCS devices and associated concomitant therapies (ventilator support, indwelling lines) and risks of immobility/deconditioning. The duration of support for our bridged patients was approximately 4–5 days, which is similar to most other bridge studies.4,18 Interestingly a recent article by Durinka et al.10 supported patients with ECMO for an average of 12 days. These investigators reported a 92% survival to discharge for ECMO patients who were transitioned to a dVAD in less than 14 days but reported inferior survival for patients supported >14 days, indicating that prolonged durations of support are associated with an increased risk.

Study Limitations

The largest limitation of our analysis is our small TCS sample size that predisposes the study to the risk of a type II error when comparing TCS strategies. The same is true for comparing TCS with the profile 1 patients without TCS. This is the largest series to date that has evaluated the effect of TCS devices on post-dVAD outcomes. We are unable to comment on TCS patients who did not go on to receive a dVAD or the adverse events (e.g., bleeding, hemolysis, thrombotic and limb complications) associated with the various TCS devices as we did not track such data. The study also included patients from two different VAD centers with varied clinical management styles and expertise with TCS devices, but this is reflective of the heterogeneity of the real world. Finally, we did not include the other forms of TCS because of the small number of other TCS devices used at our institutions. We also did not include IABP, which INTERMACS allows to be coded as TCS by providers. Despite the widespread use of IABPs, there is a significant body of literature through meta-analyses19 and a recent randomized trial that show a lack of mortality benefit with IABPs, and the focus of this study was on devices with the capability of significantly increasing cardiac output.12


Temporary circulatory support with ECMO and TH provided a consistent improvement in hemodynamics and end-organ function before dVAD implant. Despite improved clinical stability to levels comparable with patients with profile 2–3 characteristics, post-dVAD outcomes were consistently inferior and paralleled patients in profile 1 without TCS. These data suggest that reclassifying a patient in profile 1 receiving TCS after improvement in hemodynamics to a higher patient profile (i.e., profiles 2 or 3) at the time of dVAD implant may inappropriately suggest less risk at the time of dVAD implant. These findings warrant validation in a larger cohort to help guide more accurate risk prediction and INTERMACS modifier assignment in patients presenting with cardiogenic shock who are stabilized with TCS.


1. Kirklin JK, Naftel DC, Pagani FD, et al. Sixth INTERMACS annual report: A 10,000-patient database. J Heart Lung Transplant. 2014;33:555–564
2. Stevenson LW, Pagani FD, Young JB, et al. INTERMACS profiles of advanced heart failure: The current picture. J Heart Lung Transplant. 2009;28:535–541
3. Boyle AJ, Ascheim DD, Russo MJ, et al. Clinical outcomes for continuous-flow left ventricular assist device patients stratified by pre-operative INTERMACS classification. J Heart Lung Transplant. 2011;30:402–407
4. Pagani FD, Lynch W, Swaniker F, et al. Extracorporeal life support to left ventricular assist device bridge to heart transplant: A strategy to optimize survival and resource utilization. Circulation. 1999;100(19 suppl):II206–II210
5. Stretch R, Sauer CM, Yuh DD, Bonde P. National trends in the utilization of short-term mechanical circulatory support: Incidence, outcomes, and cost analysis. J Am Coll Cardiol. 2014;64:1407–1415
6. Seyfarth M, Sibbing D, Bauer I, et al. A randomized clinical trial to evaluate the safety and efficacy of a percutaneous left ventricular assist device versus intra-aortic balloon pumping for treatment of cardiogenic shock caused by myocardial infarction. J Am Coll Cardiol. 2008;52:1584–1588
7. Cowger J, Sundareswaran K, Rogers JG, et al. Predicting survival in patients receiving continuous flow left ventricular assist devices: The HeartMate II risk score. J Am Coll Cardiol. 2013;61:313–321
8. Breslow N. Covariance analysis of censored survival data. Biometrics. 1974;30:89–99
9. McCarthy F, McDermott K, Xie D, Gutsche J.. Trends in U.S. extracorporeal membrane oxygenation utilization and outcomes: 2002–2012. Semin Thorac Cardiovasc Surg. published online July 22, 2015. DOI: 10.1053/j.semtcvs.2015.07.005.
10. Durinka JB, Bogar LJ, Hirose H, et al. End-organ recovery is key to success for extracorporeal membrane oxygenation as a bridge to implantable left ventricular assist device. ASAIO J. 2014;60:189–192
11. Cheng JM, den Uil CA, Hoeks SE, et al. Percutaneous left ventricular assist devices vs. intra-aortic balloon pump counterpulsation for treatment of cardiogenic shock: A meta-analysis of controlled trials. Eur Heart J. 2009;30:2102–2108
12. Thiele H, Zeymer U, Neumann FJ, et al.IABP-SHOCK II Trial Investigators. Intraaortic balloon support for myocardial infarction with cardiogenic shock. N Engl J Med. 2012;367:1287–1296
13. Burkhoff D, Cohen H, Brunckhorst C, O’Neill WWTandemHeart Investigators Group. . A randomized multicenter clinical study to evaluate the safety and efficacy of the TandemHeart percutaneous ventricular assist device versus conventional therapy with intraaortic balloon pumping for treatment of cardiogenic shock. Am Heart J. 2006;152:469.e1–469.e8
    14. Thiele H, Sick P, Boudriot E, et al. Randomized comparison of intra-aortic balloon support with a percutaneous left ventricular assist device in patients with revascularized acute myocardial infarction complicated by cardiogenic shock. Eur Heart J. 2005;26:1276–1283
    15. Lauten A, Engström AE, Jung C, et al. Percutaneous left-ventricular support with the Impella-2.5-assist device in acute cardiogenic shock: Results of the Impella-EUROSHOCK-registry. Circ Heart Fail. 2013;6:23–30
    16. Lamarche Y, Cheung A, Ignaszewski A, et al. Comparative outcomes in cardiogenic shock patients managed with Impella microaxial pump or extracorporeal life support. J Thorac Cardiovasc Surg. 2011;142:60–65
    17. Griffith BP, Anderson MB, Samuels LE, Pae WE Jr, Naka Y, Frazier OH. The RECOVER I: A multicenter prospective study of Impella 5.0/LD for postcardiotomy circulatory support. J Thorac Cardiovasc Surg. 2013;145:548–554
    18. Hoefer D, Ruttmann E, Poelzl G, et al. Outcome evaluation of the bridge-to-bridge concept in patients with cardiogenic shock. Ann Thorac Surg. 2006;82:28–33
    19. Sjauw KD, Engström AE, Vis MM, et al. A systematic review and meta-analysis of intra-aortic balloon pump therapy in ST-elevation myocardial infarction: Should we change the guidelines? Eur Heart J. 2009;30:459–468

    temporary circulatory support; cardiogenic shock; ventricular assist device; intraaortic balloon pump; extracorporeal membranous oxygenation

    Copyright © 2016 by the American Society for Artificial Internal Organs