Higher ECMO Flow Index And Hyperoxia Are Associated With Increased Mortality In Adults With Cardiogenic Shock Receiving Venoarterial Extracorporeal Membrane Oxygenation: An ELSO Registry Analysis

Abstract

Introduction

Among adults with cardiogenic shock (CS) undergoing venoarterial extracorporeal membrane oxygenation (VA-ECMO), recent observational studies suggest increased mortality for patients with early exposure to higher ECMO flows and hyperoxia. The interaction between ECMO flows and arterial hyperoxia with outcomes in this population is unknown.

Hypothesis

Compared to patients with early exposure to full ECMO flow and hyperoxia, those receiving partial flow and normoxia will have lower in-hospital mortality.

Methods

We queried the ELSO Registry from 2018-2023 for adults with CS receiving VA-ECMO, excluding patients with ECPR or a concomitant mechanical left ventricular unloading device. Patients were stratified at 24 hours of support into four groups based on ECMO flow index (circuit flow/body surface area), with partial flow defined as <2.0 L/min/m², and the partial pressure of arterial oxygen (PaO₂), with normoxia defined as PaO₂ ≤150 mmHg. We compared the primary outcome of 90-day in-hospital mortality using Kaplan-Meier time-to-event analysis and multivariable Cox proportional hazards modeling.

Results

Among 5274 adults with CS undergoing VA-ECMO, at 24 hours of support there were 2974 (56.4%) patients receiving partial flow and 2805 (53.2%) patients exposed to normoxia. The median duration of VA-ECMO support was 117 hours. Patients exposed to partial flow and normoxia were more often male, had a higher BSA, were less likely to have post-cardiotomy shock and central cannulation, were more likely to receive a smaller arterial cannula (≤15 Fr), had a lower incidence of pre-ECMO renal failure, and a higher rate of pre-ECMO cardiac arrest. (Figure, Panel A). Compared to patients exposed to partial flow and normoxia, those receiving full flow and hyperoxia had significantly higher 90-day in-hospital mortality, 56% versus 39%, log-rank p<0.001 (Figure, Panel B). This relationship persisted in multivariable Cox modeling, where compared to the referent group of patients with partial flow and normoxia: full flow & normoxia, adjusted hazard ratio (adj-HR) 1.15 (95% CI: 1.03-1.29), p=0.016; partial flow & hyperoxia, adj-HR 1.38 (95% CI: 1.24-1.54), p<0.001; full flow & hyperoxia, adj-HR 1.49 (95% CI: 1.33-1.67), p<0.001.

Conclusions

In adults with CS supported with VA-ECMO, patients with early exposure to partial ECMO flow (<2.0 L/min/m²) and normoxia (PaO₂ ≤150 mmHg) had significantly lower 90-day in-hospital mortality compared to those receiving full flow and hyperoxia.