ECPELLA in Advanced Cardiogenic Shock: Venting for Anyone?

Utilization of temporary mechanical circulatory support (t-MCS) devices has risen steeply in recent years for the treatment of medically refractory cardiogenic shock (CS) of diverse etiologies. More and more patients with CS related to acute on chronic or de novo cardiomyopathy are receiving support with t-MCS, and for patients with CS related to acute myocardial infarction, transvalvular microaxial flow pumps (Impella, Abiomed, Inc., Danvers, MA) and extracorporeal membrane oxygenators (VA-ECMO) are increasing used and replacing the previously favored intra-aortic balloon pump (IABP).1–3 By draining venous blood and returning oxygenated blood into the systemic circulation, most commonly via the femoral artery, VA-ECMO pressurizes the arterial tree and increases left ventricular (LV) afterload.4 The growing use of VA-ECMO has spurred interest in several methods employed to decompress, or “vent,” the LV to address increased LV afterload caused by VA-ECMO.4 The primary goal of LV venting is to reduce LV cardiac filling pressures and prevent stasis in the LV and aortic root. In contrast, the primary goal of LV unloading is to reduce LV pressure and volume, thereby lowering native LV wall stress and myocardial oxygen consumption. Venting strategies include inotropes to increase LV contractility, atrial septostomy to reduce LV preload via passive shunting of blood from the left atrium to right atrium or via active removal of left atrial blood using an inflow cannula (eg, TandemHeart), insertion of an IABP to lower LV afterload, or placement of an Impella to displace blood from the LV (a configuration known as ECPELLA or ECMELLA).5 Each of these methods, however, carries important additional risks: inotropes may provoke atrial or ventricular arrhythmias and increase LV work and thus worsen myocardial ischemia; atrial septostomy requires transseptal puncture, which is of limited durability, and an additional left atrial cannula alters circuit inflow; IABP may not offer sufficient LV venting, is of limited use with atrial tachyarrhythmias, and introduces additional risk of vascular injury; and transvalvular microaxial flow pumps introduce added risk of vascular injury and hemolysis and may potentiate differential hypoxemia (ie, the north-south syndrome).5 Thus, the decision to implant other t-MCS for LV venting during VA-ECMO is not trivial. In this issue of the Journal, Piperata et al.6 performed a retrospective analysis entitled EVACS (ECMO alone versus ECPELLA in patients affected by CS) of patients with CS at four international academic centers (two in France, one in Japan, and one in Turkey) managed with VA-ECMO alone or ECPELLA between 2011 and 2021. The investigators performed propensity score matching to compare patients supported with VA-ECMO alone versus patients supported with ECPELLA. In total, 438 patients were included in this analysis, 319 (73%) with ECMO and 119 (27%) with ECPELLA, and 95 patient pairs were identified for the propensity score matching analysis. The principal finding of EVACS is that VA-ECMO alone and ECPELLA have similar mortality rates at 30 (49.5 vs. 43.2%) days and 1 (46.6 vs. 51.3%) year. The procedural complications, after matching, of peripheral ischemia, bleeding, renal replacement therapy, and hemolysis did not differ between groups, although sepsis was more common in the ECPELLA group. EVACS contradicts the findings of a larger, prospective study of ECPELLA, which found lower 30 day mortality among CS patients supported with ECPELLA versus VA-ECMO alone.7 Furthermore, prior systematic reviews and meta-analyses of LV venting strategies for VA-ECMO beyond Impella, including IABP, found reduced short-term mortality with LV unloading during VA-ECMO use (Table 1).7–11 Several explanations may account for these conflicting findings. First, EVACS was a small retrospective study, and only 95 patient pairs were identified for the propensity score-matched analysis. Thus, EVACS likely has insufficient power to confidently identify a difference in outcomes if one were to exist between these two groups. Indeed, with observational datasets, it is common for some analyses to find a benefit of an intervention, whereas other analyses do not. Second, the observed mortality rate in the EVACS cohort was less (approximately 45–50% both at 30 days and 1 year for both VA-ECMO alone and ECPELLA) compared to that seen in prior similar analyses (generally 50–65%). The comparatively lower mortality rate in the EVACS cohort suggests that patients may have had slightly less advanced forms of CS and thus stand to benefit less from ECPELLA. Table 1. - Prior Analyses of LV Venting During VA-ECMO Study Number of Patients (N) Type of Venting Follow-Up Duration Mortality p Value/Odd Ratio Bleeding Grandin et al. 8 (retrospective observational study) 12,734 3,399 (26.7%) received mechanical unloading: 2,782 (82.9%) IABP and 580 (17.1%) p-LVAD 2010–2019 OR: 0.84; 95% CI: 0.77–0.92; p < 0.001 OR: 1.25; 95% CI: 1.11–1.40; p < 0.001 Russo et al. 9 (meta-analysis of 17 studies) 3,997 1,696 (42%) received p-LVAD) 2014–2017 RR: 0.79; 95% CI: 0.72–0.87; p < 0.00001) RR:1.12 (0.84–1.49); p value 0.44 Schrage et al. 7 (retrospective cohort study) 686 49% used p-LVAD 2005–2019 Hazard ratio, 0.79 (95% CI: 0.63–0.98); p = 0.03 98 (38.4%) vs. 45 (17.9%); p value < 0.01 Ikeda et al. 10 J-PVAD (observational registry data) 593 70.2% used p-LVAD 2017–2020 63.1% (95% CI: 59.1–66.8%) 9.3%; p value 0.105 CI, confidence interval; IABP, intra-aortic balloon pump; LV, left ventricular; PVAD, percutaneous ventricular assist device; VA-ECMO, extracorporeal membrane oxygenators. Notably, EVACS reported similar rates of peripheral ischemia, bleeding, stroke, renal replacement therapy, and hemolysis between the ECPELLA and VA-ECMO only groups both before and after matching. Sepsis was the only complication that occurred at a higher frequency with ECPELLA. These findings stand in contrast to other prior observational analysis of ECPELLA, which found higher rates of limb ischemia, moderate or severe bleeding, and renal replacement therapy with ECPELLA compared to VA-ECMO alone.7 Indeed, in a large registry of CS patients, rates of limb ischemia were highest among patients supported with VA-ECMO and IABP or Impella compared to patients supported with VA-ECMO, Impella, or IABP alone.12 In a subanalysis of the EVACS cohort, the investigators compared ECPELLA patients who had simultaneous deployment of Impella and VA-ECMO (within 3 hours) to ECPELLA patients who had implantation of VA-ECMO and Impella more than 3 hours apart. About equal proportions (approximately 44%) of patients had Impella placed at the time of VA-ECMO cannulation versus Impella placement after VA-ECMO cannulation (after on average 1 day of ECMO support), whereas only 12% of patients had an Impella placed before VA-ECMO. These findings indicate heterogeneity in the ECPELLA cohort under study. Patients with Impella placement after VA-ECMO may have one or more complications of VA-ECMO, including the development of increased afterload compelling Impella placement. Patients with Impella and VA-ECMO placed at the same time may represent a sicker CS patient group that needed more support upfront. Indeed, as noted by the investigators, ECPELLA patients with simultaneous implantation had higher lactate levels at baseline. A recent analysis, moreover, demonstrated reduced mortality of early Impella implantation during VA-ECMO use.13 Finally, the addition of VA-ECMO to Impella support in the small group of patients likely indicates escalation of support and thus use of VA-ECMO as a bailout strategy in a rapidly deteriorating population. Thus, not all ECPELLA use occurs during similar clinical situations, and these three distinct ECPELLA scenarios indicate the timing of Impella implantation in relation to VA-ECMO is important to consider in any analysis of ECPELLA and may explain differences in clinical outcomes. The period over which patients were included was long, from 2011 to 2021. During this time, Impella devices evolved significantly. About one-third of patients included in this analysis were supported with the Impella 2.5, which is no longer available for clinical use and has been replaced by the slightly more powerful Impella CP. Furthermore, the Impella 5.0 has been improved to and replaced by the Impella 5.5 device. Indeed, the Impella CP was the least frequently used Impella device in this analysis, and in current clinical practice in the United States, many ECPELLA uses include Impella CP. The LV unloading capabilities of the 2.5/CP and 5.0/5.5 are significantly different and thus may not result in equal effects. This heterogeneity in the type of Impella device employed the ECPELLA cohort highlights the highly evolving nature of t-MCS in CS management. Additionally, it also likely introduces unmeasured confounding due to differences in CS patient phenotypes and limits this study’s generalizability to contemporary practice, which now includes primarily use of Impella CP and 5.5. Fundamentally, this study calls into question the need for routine LV unloading with Impella in all patients with advanced stages of CS who are supported with VA-ECMO or are candidates for VA-ECMO cannulation. However, as the findings of EVACS conflict with that of prior analyses, it remains unclear whether early or prophylactic mechanical LV unloading with Impella, IABP, or another device improves myocardial recovery and survival. A more definitive answer can come only from randomized controlled trials. UNLOAD ECMO (NCT05577195) is a randomized single-center trial in actively recruiting Germany comparing ECPELLA and VA-ECMO only strategies in severe CS. REMAP ECMO (NCT05913622) is an actively recruiting randomized trial at multiple centers in the Netherlands comparing VA-ECMO with IABP to VA-ECMO only in CS patients. Thus, at present, the decision to implant an additional t-MCS during VA-ECMO use is complex and must be individualized to the clinical scenario and patient.