Abstract
<title>Abstract</title> <p> Background The “lean insulin-resistant paradox” challenges traditional body mass index (BMI)-based stroke risk stratification, leaving a significant high-risk population undetected. The triglyceride-glucose (TyG) index offers a practical measure of insulin resistance. We aimed to investigate the joint association of TyG and BMI with incident stroke in a national Chinese cohort, with emphasis on the normal-weight insulin-resistant phenotype. Methods This prospective cohort study included 8302 participants aged 45 years and older without a history of stroke at baseline from the China Health and Retirement Longitudinal Study (CHARLS), with a maximum follow-up of 9 years. Data on demographic characteristics, health status, functioning, and biomarkers were extracted for analysis. Multivariable-adjusted Cox proportional hazards models combined with restricted cubic splines were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) and mediation analyses. Mediation analyses were then performed to quantify the proportion of this association explained. Subgroup stratification was performed to assess effect modification. Sensitivity analyses were performed to evaluate the robustness of the primary findings. Receiver operating characteristic (ROC) curves were applied to evaluate the predictive performance of individual and combined indices. Results During the follow-up period, 554 participants developed incident stroke. The mean (SD) age of the participants was 59.64 (9.10) years, and 3872 (46.6%) were men. The multivariable-adjusted HR for stroke was 1.39 (95% CI: 1.16–1.65) per unit increase in TyG and 1.37 (95% CI: 1.04–1.79) per unit increase in BMI. Restricted cubic spline regression revealed significant dose–response associations of TyG and BMI with stroke risk. The combined TyG and BMI exhibited the highest predictive efficacy (AUC: 0.598, 95% CI: 0.574–0.621), superior to TyG alone (AUC: 0.577, 95% CI: 0.553–0.600) and BMI alone (AUC: 0.578, 95% CI: 0.553–0.602). Kaplan–Meier analysis showed that participants with both high TyG (≥ 8.59) and high BMI (≥ 25.63 kg/m²) had the highest cumulative incidence of stroke ( <italic>P</italic> < 0.001). The corresponding HR for incident stroke was 3.32 (95% CI: 2.39–4.61) in Model 2, which was substantially attenuated to 1.19 (95% CI: 0.75–1.87) after further adjustment in Model 3. BMI significantly mediated 10.6% of the association between the TyG index and stroke, while TyG simultaneously mediated 31.9% of the association between BMI and stroke. Conclusion Joint assessment of TyG and BMI unmasks a hidden yet clinically significant population: lean individuals with insulin resistance whose elevated stroke risk is overlooked by BMI-centric screening. Integrating metabolic phenotyping into routine risk assessment may reduce the population-level burden of stroke, particularly among those currently misclassified as "low risk" by body weight alone. </p>