Compatibility of Novel Cardiogenic Shock Phenotypes from the Cardiogenic Shock Working Group (CSWG) with the SCAI Staging System

Document Type

Conference Proceeding

Publication Date


Publication Title

J Heart Lung Transplant


Purpose: Cardiogenic shock (CS) is a heterogeneous syndrome that represents an acute and fulminant form of heart failure (HF). We (1) employed machine learning (ML) to identify distinct CS phenotypes which could help define treatment algorithms based on individual risk and (2) tested the correlation of the SCAI staging system.

Methods: We included data from 1957 CS patients from 2 cohorts: CSWG Registry, further grouped for myocardial infarction (CSWG-MI, n=408) and acute on chronic HF (CSWG-HF, n=480); and the Danish Retroshock Registry containing MI patients (DRR, n=1069). Independent consensus k means clustering derived phenotypes at admission in the CSWG-MI cohort that were then validated in the CSWG-HF and DRR cohorts. Patients were also categorized by the most severe SCAI stage reached during the hospitalization.

Results: The ML algorithms revealed 3 distinct clusters that we designated: ‘non-congested (I)’, ‘cardio-renal (II)’ and ‘cardio-metabolic (III)’. In-hospital mortality was 21% vs 29% vs 10%, 42% vs 46% vs 32%, and 55% vs 57% vs 54% among the CSWG-MI vs DDR vs CSWG-HF for Clusters I, II and III, respectively. Despite baseline differences among the overall cohorts, clusters presented similarly across the 3 cohorts. The risk of escalating to stage D or E shock was lowest in Cluster I and highest in Cluster III for both CS-MI and CS-HF patients. Within each phenotype, the SCAI staging (C-E) further stratified mortality (fig).

Conclusion: Using ML, we derived and externally validated 3 distinct CS phenotypes. SCAI stages and CSWG CS phenotypes identify patients at risk for in-hospital mortality.







First Page


This document is currently not available here.