Data-driven cluster analysis identifies distinct types of cardiovascular-kidney-metabolic syndrome

Abstract Introduction The current staging criteria for cardiovascular-kidney-metabolic (CKM) syndrome demonstrate extensive clinical heterogeneity, imposing constraints on risk prognostication and precision medicine. Objectives This study aimed to refine the subtyping of CKM syndrome using clinical biomarkers to enhance risk assessment and personalized prevention. Methods This study included patients from the UK Biobank cohort who were classified in stages 1-3 of CKM syndrome without concomitant organ-specific complications. K-means clustering was performed to identify phenotypically distinct subgroups using clinical biomarkers. Multivariate Cox regression analysis assessed the risk of complications with a median follow-up period of 15.88 years. The genetic risk factors and plasma proteomic signatures were analyzed across different clusters. Results A total of 44,200 individuals were included, with 34,487 participants designated as the training cohort and 9,713 participants designated as the validation cohort in the UK Biobank. Five clusters were identified. Low-risk (LR) cluster demonstrated the most favorable prognosis across all outcome measures. Liver high-risk (LHR) cluster was characterized by the highest risk of chronic liver disease. Cerebrovascular high-risk (CBHR) cluster exhibited a predominant susceptibility to cerebrovascular events. Age-driven high-risk (ADHR) cluster displayed elevated risks for both stroke and chronic kidney disease. Cardiorenal high-risk (CRHR) cluster demonstrated the highest vulnerability to both cardiovascular events and renal dysfunction. Each cluster exhibited unique plasma proteomic characteristics and genetic risk patterns. Conclusions Our findings provide insights into the patients with CKM syndrome, aiding in the identification of high-risk patients who may benefit from targeted interventions.