Beyond lipids: sarcopenia as a cardiovascular risk indicator in Chinese middle-aged and older adults

Background The triglyceride-glucose (TyG) index and the sarcopenia index (SI) reflect insulin resistance and skeletal muscle mass, respectively. We constructed the TyG-Sarcopenia Index (TyG-SI) to integrate metabolic and muscle-related information and examined whether it is associated with prevalent and incident cardiovascular disease (CVD) in Chinese middle-aged and older adults. Methods Using the China Health and Retirement Longitudinal Study (CHARLS) 2015 and 2020 waves, we analysed 10,266 adults aged 45 years or older (4,766 [46.4%] men; median age 61 years, IQR 53–67) cross-sectionally and followed 6,934 baseline CVD-free adults for approximately 5 years. TyG-SI was calculated as TyG divided by SI, multiplied by 1000. Associations were estimated with logistic regression and Cox proportional hazards models adjusted for age, sex, education, marital status, residence, smoking, alcohol consumption, BMI, hypertension, and cancer. Restricted cubic splines, ROC comparisons, exploratory subgroup analyses, and five-fold out-of-fold XGBoost Cox SHAP analyses supplemented these models. Results Overall, 2,092 of 10,266 participants (20.4%) had prevalent CVD; prevalence rose across TyG-SI quartiles from 14.4% to 26.3% ( P for trend < 0.001), with an odds ratio of 1.36 (95% CI 1.15–1.60) for Q4 versus Q1. During follow-up, 804 of 6,934 participants (11.6%) developed CVD; per SD higher TyG-SI, the adjusted hazard ratio was 1.14 (95% CI 1.06–1.22) for incident CVD. Single-indicator AUCs were 0.580, 0.579, and 0.577 for prevalent CVD, heart disease, and stroke, and out-of-fold incremental ΔAUC beyond conventional risk factors was small (0.00188 for prevalent CVD; 0.00226 for incident CVD, 95% CI including zero). In XGBoost-SHAP, TyG-SI accounted for 18.0% of total out-of-fold feature contribution, after age, hypertension, and BMI. Conclusions TyG-SI, derived from four routine biomarkers, was associated with prevalent and incident CVD in Chinese middle-aged and older adults after multivariable adjustment. Its single-indicator discriminative ability was weak and its incremental discrimination beyond conventional risk factors was small; these findings support further evaluation as a candidate complementary risk marker rather than an established clinical tool.