Objective:To identify circulating proteins associated with cardiovascular, renal, and cardiorenal comorbidity events in individuals with metabolic syndrome, to construct a predictive model incorporating these proteins to improve prediction accuracy and to investigate their mediating effects on the interplay between cardiovascular and renal diseases.Methods:Data from the UK Biobank cohort were utilized. Cox proportional hazards models were applied to identify circulating proteins associated with various outcomes, followed by time-truncated sensitivity analyses. A predictive model incorporating protein scores was then developed using the LightGBM algorithm and compared with other models. Gene Ontology(GO) functional enrichment analysis was performed to explore the biological pathways of the identified proteins. Finally, mediation effect analysis was conducted to assess the role of circulating proteins in cardiorenal interactions. Results:The Cox analysis identified 180, 275, and 322 circulating proteins associated with cardiovascular events, renal events, and cardiorenal comorbidity events, respectively. Incorporating protein scores significantly improved model performance; the areas under the curve(AUC) for cardiovascular, renal, and cardiorenal events were 0.833, 0.907, and 0.890, respectively. GO functional enrichment analysis demonstrated significant enrichment in pathways such as cytokine activity(GO: 0005125), glycosaminoglycan binding(GO: 0005539), and humoral immune response(GO: 0006959) among all outcome-related proteins. Notably, EDA2R, GDF15, and WFDC2 exhibited significant mediating effects, each with mediation proportions exceeding 10%. Conclusions:A predictive model incorporating circulating protein scores can substantially improve prediction accuracy for cardiovascular and renal outcomes in individuls with metabolic syndrome.

Objective:To identify circulating proteins associated with cardiovascular, renal, and cardiorenal comorbidity events in individuals with metabolic syndrome, to construct a predictive model incorporating these proteins to improve prediction accuracy and to investigate their mediating effects on the interplay between cardiovascular and renal diseases.Methods:Data from the UK Biobank cohort were utilized. Cox proportional hazards models were applied to identify circulating proteins associated with various outcomes, followed by time-truncated sensitivity analyses. A predictive model incorporating protein scores was then developed using the LightGBM algorithm and compared with other models. Gene Ontology(GO) functional enrichment analysis was performed to explore the biological pathways of the identified proteins. Finally, mediation effect analysis was conducted to assess the role of circulating proteins in cardiorenal interactions. Results:The Cox analysis identified 180, 275, and 322 circulating proteins associated with cardiovascular events, renal events, and cardiorenal comorbidity events, respectively. Incorporating protein scores significantly improved model performance; the areas under the curve(AUC) for cardiovascular, renal, and cardiorenal events were 0.833, 0.907, and 0.890, respectively. GO functional enrichment analysis demonstrated significant enrichment in pathways such as cytokine activity(GO: 0005125), glycosaminoglycan binding(GO: 0005539), and humoral immune response(GO: 0006959) among all outcome-related proteins. Notably, EDA2R, GDF15, and WFDC2 exhibited significant mediating effects, each with mediation proportions exceeding 10%. Conclusions:A predictive model incorporating circulating protein scores can substantially improve prediction accuracy for cardiovascular and renal outcomes in individuls with metabolic syndrome.