Background: Shock, encompassing septic and cardiogenic etiologies, is a life-threatening condition associated with systemic inflammation, metabolic dysregulation, and high mortality in intensive care units. Traditional clinical markers often fail to capture the complexity of this syndrome, limiting personalized therapeutic approaches. Advances in metabolomics enable comprehensive analysis of metabolic disruptions, providing novel insights into shock pathophysiology. This study aimed to cluster critically ill patients with shock into metabolic phenotypes and investigate their associations with clinical severity.

Results: We analyzed metabolomic profiles from 60 critically ill patients with shock at ICU admission using Uniform Manifold Approximation and Projection (UMAP) for dimensionality reduction and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) for clustering. Three distinct clusters were identified: Cluster 1 (n = 13) exhibited the highest severity (median APACHE II: 29) and mortality (54%), with elevated biogenic amines, sugars, and sphingolipids, reflecting intense metabolic activation. Cluster 2 (n = 24), despite having low initial severity (median APACHE II: 25), demonstrated high mortality (38%) and was characterized by elevated glycerophospholipids and sphingolipids as in cluster 1, without enhanced biogenic amines and sugars, indicating inadaptive metabolic responses. Cluster 3 (n = 23) showed the lowest severity (median APACHE II: 22) and mortality (9%), with uniformly reduced metabolite levels, suggesting an adaptive metabolic profile.

Conclusions: Shock patients exhibit distinct metabolic phenotypes associated with clinical severity and outcomes. Metabolomic profiling offers a promising avenue for precision medicine in critical care by uncovering pathophysiological insights. Future research should validate these findings, identify practical biomarkers, and explore therapeutic interventions tailored to specific metabolic profiles.