Phenotype analysis reveals a “global collapse” cardiogenic shock signature despite VA-ECMO support in acute myocardial infarction – a subanalysis from ECLS-SHOCK
T Thevathasan, H Thiele, J Poess, A Freund, U Zeymer, S DeschAbstract
Background
Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) is increasingly applied in acute myocardial infarction-related cardiogenic shock (AMI-CS), yet mortality despite VA-ECMO remains high and patient selection remains unresolved. Conventional analyses focus on treatment effects in subgroups, whereas biological heterogeneity of shock phenotypes is less well understood.
Purpose
To identify baseline characteristics defining opposite clinical extremes, recovery without VA-ECMO versus death with VA-ECMO, in order to characterise phenotypic signatures associated with and without refractory AMI-CS.
Methods
Patients from the randomised ECLS-SHOCK trial in AMI-CS were analysed. Two predefined extreme phenotypes were evaluated: (1) patients surviving to 30 days without VA-ECMO and (2) patients treated with VA-ECMO who died within 30 days (Figure). Baseline demographic, clinical, haemodynamic and laboratory variables available at presentation were compared. Continuous variables were dichotomised using receiver operating characteristic-derived thresholds. Univariable logistic regression quantified associations with the refractory phenotype. Multidimensional phenotype structure was explored using integrated radar plot visualisation.
Results
Marked differences between extreme phenotypes were observed across all physiological domains. Death with VA-ECMO was consistently associated with indicators of advanced systemic hypoperfusion and organ dysfunction, including elevated lactate, prolonged cardiopulmonary resuscitation duration, impaired renal function, increased inflammatory activation and reduced haemoglobin levels. Lower left ventricular ejection fraction and increasing age further characterised the refractory phenotype, alongside greater comorbidity burden. Effect estimates demonstrated a coherent severity gradient rather than isolated predictors. Multidimensional phenotype mapping revealed uniform expansion across metabolic, haemodynamic and clinical domains among VA-ECMO non-survivors, suggesting the presence of a “global collapse signature” already detectable at baseline.
Conclusion
Phenotype analysis of both shock extremes identified a distinct “collapse signature” of AMI-CS characterised by systemic metabolic failure and multiorgan dysfunction despite advanced mechanical support. These findings highlight biological heterogeneity within AMI-CS and support phenotype-based risk stratification as a potential framework for future personalised mechanical circulatory support strategies.