Background: Cardiogenic shock (CS) is associated with high in-hospital mortality rates of 30-60%. Heterogeneity of CS patients has complicated clinical trial design. Recently, three distinct CS phenotypes have been identified and externally validated in cohorts from Version 1 of the multi-center national Cardiogenic Shock Working Group (CSWG) registry, the Danish Retroshock Registry and a single-center cohort of a mixed cardiac intensive care unit population, namely phenotype I: “non-congested”, phenotype II: “cardiorenal”, and phenotype III: “cardiometabolic” shock. These CS phenotypes appeared to be compatible with the Society for Cardiovascular Angiography and Interventions (SCAI) CS stages. However, little is known about the clinical course of patients within these CS phenotypes.

Aim: The aim of this study was (i) to confirm the external applicability of CS phenotypes in a mixed CS population, (ii) to assess the relevance of CS phenotypes for prognosis and outcomes, (iii) to confirm their compatibility with the SCAI classification system for CS, and (iv) to assess differences in treatment decisions in current clinical practice among CS phenotypes.

Methods: We included all-cause CS patients from Version 2 of the CSWG registry. Machine learning-based CS phenotypes were assigned using a nearest centroid classification based on centroids of the three initially reported CS phenotypes from the CSWG registry V1 derivation cohort. Baseline and maximum SCAI stage during hospitalization were calculated.

Results: Out of 1890 of all-cause CS patients included in the CSWG V2 registry, 796 met all requirements for retrospective phenotype assignment without imputation. Mortality in phenotype I (non-congested), II (cardiorenal) and III (cardiometabolic) was 23%, 41% and 52%, respectively, closely matching the initially reported 21%, 45% and 55% in the V1 derivation cohort. Phenotype-related demographic, hemodynamic and metabolic features resembled the initially reported properties (Figure). The proportion of patients receiving any mechanical circulatory support devices differed in the phenotypes with 58.8%, 45.7%, and 51.9% in phenotype I, II, and III, respectively (p=0.013). Receiving any mechanical circulatory support device was associated with increased mortality in cardiorenal CS (OR [95%-CI]: 1.82 [1.16 – 2.84; p=0.008), but not significantly in non-congested or cardiometabolic CS (OR [95% CI]: 1.26 [0.64 – 2.47], p=0.51 and OR [95% CI]: 1.39 [0.86 – 2.25], p=0.18, respectively). The odds of reaching SCAI Stage E (“Extremis”) during the hospitalization were 5.26 times higher in patients with cardiometabolic CS at baseline (95% CI: [3.53 – 7.83], p<0.001) compared to patients with non-congested CS at baseline. Patients with baseline cardiorenal and non-congested CS were similarly likely to reach SCAI stage E during their hospitalization (OR [95% CI]: 1.10 [0.76 – 1.61], p=0.609). Admission phenotype II and III as well as initial SCAI stage E were each independently associated with more than twofold increases in the odds of mortality compared to non-congested CS patients in baseline SCAI stage C in a multivariable logistic regression model (p<0.001, respectively).

Conclusion: Our findings support the universal applicability of phenotypes in all-cause CS using supervised machine-learning. CS phenotypes may inform future clinical trial design to tailor interventions to a specific CS phenotype.