Extracorporeal Membrane Oxygenation Combined With Mechanical Thrombectomy Versus Extracorporeal Membrane Oxygenation Alone for High-Risk Pulmonary Embolism: A Systematic Review and Meta-analysis.

OBJECTIVES

To compare clinical outcomes of venoarterial (VA) extracorporeal membrane oxygenation (ECMO) combined with mechanical thrombectomy (MT) versus VA-ECMO alone in high-risk pulmonary embolism (PE).

DESIGN

Systematic review and meta-analysis.

SETTING

PubMed and Embase search from inception through October 2025.

PARTICIPANTS

Adults with high-risk PE requiring VA-ECMO support.

INTERVENTIONS: VA-ECMO combined with catheter-based MT versus VA-ECMO alone.

MEASUREMENTS AND MAIN

RESULTS

Four observational studies (n = 2,824) met the inclusion criteria; two (n = 47) provided extractable count data for meta-analysis. Pooled odds ratios (ORs) for in-hospital mortality (0.36; 95% confidence interval [CI], 0.01-13.59) and 90-day mortality (0.43; 95% CI, 0.02-10.66) were imprecise and discordant, with very wide CIs reflecting the small number of pooled studies. A nationwide database study (n = 2,285) reported an adjusted OR of 0.75 (95% CI, 0.47-1.19) for in-hospital mortality, and an international multicenter cohort (n = 492) reported a 90-day hazard ratio of 0.68 (95% CI, 0.45-1.03) with early mechanical reperfusion; neither of the two more methodologically rigorous studies reached statistical significance. Hospital and intensive care unit lengths of stay were shorter in the ECMO-MT group, while major bleeding did not differ between groups. The certainty of evidence was very low for all mortality outcomes.

CONCLUSIONS

Comparative outcome data on adding MT to VA-ECMO for high-risk PE remain scarce. Pooled estimates were imprecise, and the most methodologically robust studies available did not establish a statistically significant mortality benefit. The pooled estimates should therefore be interpreted as exploratory and hypothesis generating. Given the practical and ethical barriers to randomized trials in this critically ill population, well-designed multicenter registries with standardized outcome definitions represent the most realistic path to higher-quality evidence.