Abstract 4338227: Glycemic Status Modifies the Prognostic Value of the Cardiometabolic Index for Cardiorenal Outcomes: Evidence From a Population-Based Cohort Study

Background: The Cardiometabolic Index (CMI) is linked to cardiovascular and metabolic diseases, but its prognostic value for cardiorenal syndrome (CRS) and the modifying role of glycemic status remain unclear. Methods: We analyzed 205,830 participants from a UK Biobank cohort, excluding individuals with type 1 diabetes or preexisting cardiorenal disease. CMI was evaluated using unadjusted and adjusted Cox proportional hazards models, including interactions with glycemic status (normoglycemia, prediabetes, and type 2 diabetes [T2D]). To identify robust predictors, we applied LASSO Cox regression with cross-validation and performed stability selection. Variables with high selection probability (≥75%) were reintroduced into multivariable Cox models. Nonlinearity was assessed using restricted cubic splines (RCS). Model performance was evaluated using multiple approaches. Results: During a median follow-up of 14.25 years, 3,554 incident CRS events were observed. Higher CMI was consistently associated with elevated risk (HR : 1.51[1.48–1.54] unadjusted; 1.34[1.31–1.37] fully adjusted; all p < 0.001). A cutoff of 0.603 stratified participants into high- and low-risk groups (log-rank p < 0.0001). The association between CMI and CRS was attenuated in individuals with T2D (HR 1.13), and was stronger in prediabetes (HR 1.38) and normalglycemic (HR 1.36) individuals compared to T2D ( p for interaction < 0.001) . LASSO and stability selection consistently identified CMI and its interaction with glycemic status as important predictors. RCS modeling revealed a significant nonlinear dose–response (Chi-square = 92.7, p < 0.001), with risk rising steeply at higher CMI levels. Although absolute risks differed by glycemic state, the shape of the CMI-risk curve remained similar. Adding CMI modestly improved model discrimination (C-index from 0.813 to 0.820) and significantly improved individual risk reclassification (NRI = 15.4%, p < 0.001). Time-dependent AUCs were consistently higher with CMI across 1–15 years, with the greatest gain at 1 year (+0.043), indicating strong early risk stratification potential. Conclusion: CMI is a robust and independent predictor of CRS. Its prognostic strength is more evident in individuals with normoglycemia or prediabetes than in those with T2D, highlighting its clinical utility for early identification of high-risk individuals before overt diabetes develops.