Extracorporeal membrane oxygenation versus cardiopulmonary bypass during transcatheter aortic valve implantation: a meta-analysis of survival benefits

Since its approval by the Food and Drug Administration in 2011, transcatheter aortic valve implantation (TAVI) has rapidly evolved to become the preferred ultimate intervention for high- and intermediate-risk patients with severe symptomatic aortic stenosis.[1] This is due to its non-open-heart, minimally invasive and off-pump advantages.[1] Nevertheless, as a result of the frequent frailty and comorbidity profiles of patients undergoing TAVI, such as advanced cardiac dysfunction and extensive coronary artery disease, or technically difficult anatomy for the procedure itself,[2-4] it is common for these patients to experience critical circulatory collapse perioperatively. These factors are linked to elevated mortality rates, necessitating suitable mechanical circulatory support (MCS) to reverse the disastrous situations.[5]

Both extracorporeal membrane oxygenation (ECMO) and cardiopulmonary bypass (CPB) are potent MCS devices for circulatory collapse during TAVI, as they are independent of cardiopulmonary function and therefore allow for more sufficient support than other MCSs in cases of circulatory failure.[6-9] While there are some similarities between CPB and ECMO, there are also a number of distinctions.[10] No guidelines have recommended which techniques, either ECMO or CPB primarily used, would contribute to greater clinical benefit when circulatory collapse occurs. Herein, we conducted a pooled analysis to compare the survival outcomes of patients who underwent TAVI with either ECMO or CPB as their primary MCS.

This meta-analysis was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.[11] We conducted a systematic search of PubMed, Web of Science and Embase databases, supplemented by manually searching reference lists from relevant articles, from inception until August 12, 2023, without language restriction. Search terms: (extracorporeal membrane oxygenation) AND (transcatheter aortic valve replacement) AND (transcatheter aortic valve implantation) AND (cardiopulmonary bypass). Final inclusion was based on the consensus of two independent investigators, or a third independent reviewer served as the referee in case of disagreements. Studies were included according to the PECOS framework: (population, P) adult patients (>18 years) who underwent TAVI procedures; and (exposure, E) utilization of intraoperative CPB or VA-ECMO as primary MCS; (comparator, C) lacked CPB or VA-ECMO usage during TAVI; (outcomes, O) had valid in-hospital, 30-day, 6-month or 1-year mortality endpoint data; and (study type, S) had case-control, cohort, or cross-sectional study designs. Studies were excluded if they were ongoing or not in English. In case of overlap in the patient populations of different studies from the same registry or group, only the largest study with relevant outcomes was included. Ethical approval was not necessary for this pooled analysis of publicly available data. The quality and susceptibility to bias of the included studies were assessed using the Quality in Prognosis Studies tool.[12] Publication bias was evaluated by visually examining funnel plots.

Two independent investigators extracted and cross-checked the data from the included studies. The data abstracted included study year, sample size, location, type of study, CPB/VA-ECMO-associated parameters, TAVI-associated parameters, preoperative echocardiography data and survival outcomes. Any inconsistencies were resolved through discussion.

The primary endpoint was all-cause 30-day mortality, and the secondary endpoints were all-cause in-hospital, 6-month, and 1-year mortality (supplementary Table 1). For the primary endpoint, we further performed subgroup analyses based on the degree of urgency of MCSs (emergent vs. prophylactic). Emergent CPB/VA-ECMO was primarily utilized for the rescue treatment of sudden incidents during TAVI, such as ventricular rupture, implanted aortic valve displacement, and cardiac arrest, to facilitate the completion of TAVI or conversion to surgery in cases of anticipated hemodynamic instability.

The odds ratio (OR) and its 95% confidence interval (CI) were extracted from the publications of each individual study and converted to log (OR) and its standard error (SE) before the pooled analysis. The meta-analysis was directly implemented on the natural log (OR) scale, with results exponentiated and expressed on the original OR scale. For overall effect estimates, dichotomous variables are expressed as pooled ORs with 95% CIs, while continuous outcomes are expressed as pooled weighted mean differences (WMDs) with 95% CIs. The overlapping 95% CIs represented no statistical difference between the two groups.[13] The fixed-effect model or random-effect model was used based on the statistical heterogeneity among the included studies. Heterogeneity was assessed using the Chi-square test and I2 test. Heterogeneity was considered to be statistically significant when P<0.05, leading to the application of the random effect model; otherwise, the fixed-effect model was used. Statistical analyses were conducted using RevMan software (version 5.4, UK). All tests were two-sided and statistical significance was set at P<0.05.

A total of 16 observational trials of CPB/VA-ECMO support during TAVI were identified [7,8,14-27] (Supplementary Figure 1), with cohort characteristics summarized in supplementary Tables 2–4. The results of the current analyses were collected from 8 prospective reports, as well as from 8 retrospective institutional or national databases research. Overall, 70,802 patients were pooled, including 61,983 from seven VA-ECMO studies and 8,819 from nine CPB studies. Publication bias of involved studies is shown in Supplementary Figure 2.

Of these 70,802 participants, 1,076 accepted CPB or ECMO support during TAVI procedures, and the mean age at entry varied between 71 and 87 years with 24% to 83% male patients. After accounting for missing data, patients who used VA-ECMO or CPB presented similar ages (P=0.10), gender ratio (P=0.19), and New York Heart Association (NYHA) III/IV scores (P=0.46), but higher cardiac surgery risk was detected in the CPB population, as demonstrated by consistent Society of Thoracic Surgeons (STS) score ([12.50±9.13]% vs. [30.90±28.00]%; P<0.01) and Logistic Euro SCORE scores (33.96±23.15 vs. 56.95±7.73; P<0.01). The aortic valve echocardiographic results were roughly reflective of more severe aortic stenosis through higher mean gradient ([43.91±16.20] mmHg vs. [35.88±16.68] mmHg; P=0.01), also higher maximum gradient ([73.22±22.12] mmHg vs. [53.00±23.00] mmHg; P<0.01) for the ECMO group, while no significant difference was observed for LVEF (P=0.63) or AVA (P=0.69). The transfemoral-approach was the most frequently applied approach in both groups (P=0.77), but the device success of TAVI was higher in patients who underwent CPB (79.7% vs. 96.2%; P=0.02); furthermore, similar procedural times were showed (P=0.47).

Regarding the primary endpoints, VA-ECMO support during TAVI exhibited an apparent association with an increased OR for 30-day all-cause death (OR 6.36; 95%CI: 2.99–13.53; P<0.001); likewise, a consistent 30-day survival-harm was observed in patients receiving the CPB device (OR 5.16; 95%CI: 3.81–6.97; P<0.001). Similar findings were also shown in additional prespecified subgroup analyses, including those with emergent VA-ECMO (OR 19.17; 95%CI: 6.05–60.79; P<0.001), prophylactic ECMO (OR 4.26; 95%CI: 1.61–11.24; P<0.01) and prophylactic CPB (OR 4.68; 95%CI: 1.98–11.08; P<0.001), but not in the subset of patients with emergent CPB use (P=0.25). Furthermore, there was no significant difference in the 30-day survival outcome between patients who received VA-ECMO and those who received CPB during TAVI (95%CI: 2.99–13.53 vs. 3.81–6.97). Similarly, no modification of the treatment effect was shown for 30-day mortality in the prespecified prophylactic-MCS (VA-ECMO vs. CPB) subgroup (95%CI: 1.61–11.24 vs. 1.98–11.08). However, for emergent-MCS subgroup, CPB outperformed VA-ECMO in terms of the risk of 30-day mortality (95%CI: 0.68–1.11 vs. 6.05–60.79) (Figure 1).

Figure 1.

Pooled analyses of 30-day mortality in patients with ECMO or CPB support. Forest plot of (A) total, (B) emergent, and (C) prophylactic application of ECMO or CPB. ECMO: extracorporeal membrane oxygenation; CPB: cardiopulmonary bypass.

Regarding the secondary endpoints, there was no significant correlation of in-hospital mortality with neither the use of VA-ECMO (P=0.91) nor the CPB use (P=0.21) during TAVI procedures, also no difference of in-hospital death observed between those using VA-ECMO and CPB (95%CI: 0.04–34.19 vs. 0.43–49.16) (supplementary Figure 3). Overall, CPB use during TAVI was significantly associated with higher risk of 6-month death (OR 3.28; 95%CI: 1.31–8.22; P<0.05); whereas these survival risks did not persist in those using VA-ECMO (P=0.20). No difference in survival at 6-month was showed between patients who received VA-ECMO and those who received CPB (95%CI: 0.72–4.83 vs. 1.31–8.22) (supplementary Figure 4). Furthermore, neither VA-ECMO support (P=0.28) nor CPB support (P=0.40) during TAVI was significantly correlated with all-cause mortality at 1-year. No significant difference in 1-year mortality was found between patients who received VA-ECMO and those who received CPB (95% CI: 0.68–3.83 vs. 0.53–4.76) (supplementary Figure 4).

In the present systemic review and meta-analysis, which included 7 comparative studies evaluating VA-ECMO support and another 9 assessing CPB support during TAVI, we demonstrated that there was no significant difference in short-, as well as long-term survival between VA-ECMO and CPB post-TAVI. Additionally, we observed that both CPB and VA-ECMO were associated with increased risks for 30-day mortality, but the mortality risk associated with VA-ECMO disappeared at long-term (6-month and 1-year) follow-up. In contrast, the elevated mortality for patients receiving CPB persisted significantly at 6 months.

In this setting, both CPB and VA-ECMO were preferred for use as advanced-support MCS devices during TAVI procedures, emergently or prophylactically, due to their extreme efficiency in establishing rapid support for circulatory collapse in patients with complications of catheter-based interventions.[28] ECMO and CPB are conceptually similar but differ in many aspects and finalities;[29] for instance, the advantages of relatively low cost, open-heart surgery support and blood collection for CPB, and the advantages of rapid bedside access, lower anticoagulation, high-flow circuit and concomitant pulmonary support for ECMO; all of which may influence the prognosis following TAVI. Our results revealed significantly higher risks for short-term mortality at 30-day in patients receiving VA-ECMO or CPB support during TAVI. Although the well-tolerated and therapeutic benefits of CPB/VA-ECMO for high-risk TAVI populations identified in recent pooled reviews,[6] AlKhalil et al[26] used the National Inpatient Sample Registry to demonstrate that MCS (including intra-aortic balloon pump, ECMO, ventricular assist device, etc.) during TAVI was relevant to an approximate 10-fold increase in in-hospital mortality. This was consistent with our findings, and one could expect that the requirements for MCSs are likely reflective of higher frailty or high comorbid and thus represent more high-risk states for TAVI to some extent. Furthermore, due to the need for anticoagulation, invasive operations or any other treatment, both CPB and VA-ECMO have inherent risks of complications such as major strokes, cardiac damage, haemorrhage and infections. This could partially account for worse early postoperative outcomes in MCS patients.[29] Importantly, this higher risk of 30-day death did not negatively impact long-term survival for patients receiving VA-ECMO during TAVI; in contrast, this risk persisted for patients receiving CPB, which could be partially due to the inherently higher cardiac surgery risk for patients receiving CPB, as observed in this review.

Notably, CPB versus VA-ECMO only showed a significant short-term survival benefit in patients with hemodynamic emergencies, while no significant differences were shown regarding either overall short- or long-term mortality. The reason is that CPB is frequently employed in emergency situations resulting from multiple operational complications, therefore, the baseline status of patients necessitating CPB support may differ from that of patients requiring ECMO support, while the long-term prognosis of TAVI is more closely associated with patient’s inherent frailty. Currently, no guidelines recommended the use of CPB or VA-ECMO for circulatory collapse during TAVI. In clinical practice, the choice between CPB and ECMO is primarily based on the estimation of the length of circulatory assistance, whether the patient needs conversion to surgery, and whether the hospital has adequate ECMO resources; which also reflects the length of hospital stay and intensive care unit stay according to our findings (supplementary Figure 5). Herein, we deemed that CPB might be recommended as a preferred MCS during TAVI, especially for emergent situations, in cases where operators predict that only short-time support with MCS will be needed.

It should not be neglected that the choice of MCSs must be individualized after assessment of clinical features and technical aspects by a multidisciplinary heart team. For instance, CPB may be the norm for high-risk patients who need short-time support, whereas short-time VA-ECMO might also be preferred for those with a high bleeding risk of bleeding. Additionally, there are often contraindications and clinical scenarios that require appropriate adjustment of the MCS plan.

Our present study has several limitations. First, there is a certain restriction posed by the apparent publication biases of the included CPB studies in this meta-analysis. This study included a total of 9 CPB studies, most of which were published before 2018. Second, all studies were observational, and there were no head-to-head studies comparing the survival outcomes of patients who received CPB with those of patients who received VA-ECMO during TAVI; this made it difficult to specify the inclusion and exclusion criteria. Third, CPB was more likely to be applied in patients who experienced intraoperative circulatory failure necessitating concurrent surgical interventions or patients scheduled to be converted to open aortic valve replacement surgery. This difficulty in clinically matching the CPB and ECMO populations could contribute to selection bias. Fourth, there are limited data on the prediction of long-term outcomes between the two MCSs. Finally, events such as inter-institutional differences in perioperative management have prognostic implications for patients requiring CPB/ECMO, which weakens the meaningfulness of the conclusions in this review.

In conclusion, both CPB and VA-ECMO are feasible and valuable options for the treatment of circulatory collapse during TAVI, with promising long-term survival benefits. In addition, CPB might be the preferable option for TAVI patients experiencing sudden hemodynamic emergencies. Future studies with greater clarity on long-term survival are necessary to evaluate the differences in efficacy between CPB and ECMO use during TAVI.