Joint association of triglyceride-glucose (TyG) and atherogenic index of plasma (AIP) with stroke risk: findings from a nationwide prospective cohort

Abstract

Background: The triglyceride-glucose (TyG) index and the atherogenic index of plasma (AIP) are established surrogate markers for insulin resistance and dyslipidemia, respectively, each independently linked to stroke. However, their combined predictive value for stroke is underexplored. Consequently, this study aims to evaluate the predictive performance of the combined TyG and AIP indices for incident stroke in a middle-aged and elderly Chinese cohort.

Methods: We included 7652 adults aged ≥ 45 years without baseline stroke from the China Health and Retirement Longitudinal Study (CHARLS). TyG and AIP were calculated and dichotomized by median. The primary outcome was new-onset stroke. Cox proportional hazards models assessed associations, examining interactions on both additive (Relative Excess Risk due to Interaction, RERI) and multiplicative scales. Restricted cubic spline (RCS) analysis was used to examine nonlinear relationships. Predictive performance was assessed using area under the receiver operating characteristic curve (AUC), and model performance differences were evaluated using the DeLong test, net reclassification improvement (NRI), and integrated discrimination improvement (IDI).

Result: Over 9 years, 744 strokes (9.72%) occurred. High TyG (HR = 1.41, 95% CI 1.21–1.64) and high AIP (HR = 1.44, 1.24-1.67) independently increased risk. The combination showed highest risk (HR = 1.49, 1.27-1.75). AIP had a nonlinear association (inflection ~ 0.22); TyG was linear. The combined AUC (0.633) was not significantly better (DeLong test, P > 0.05), and both NRI and IDI were non-significant. Subgroup analysis revealed stronger effects of combined indicators in aged < 60 years old and those without diabetes or heart disease history. Sensitivity analysis supported robust results. Multiplicative interaction was significant (P = 0.021), while additive interaction was non-significant (RERI = 0.280, 95% CI -0.220-0.779).

Conclusion: Both the TyG index and AIP are independent risk factors for stroke. The primary value of combining these two indicators lies in identifying individuals at extremely high risk, rather than significantly enhancing overall predictive performance. This combined metric can serve as a comprehensive metabolic risk assessment tool, particularly suitable for primary stroke prevention in individuals under 60 years of age without a history of diabetes or heart disease. Future studies are needed to further validate its risk reclassification capability and intervention value.

Graphical Abstract:

Supplementary Information: The online version contains supplementary material available at 10.1186/s12933-026-03136-w.

Keywords: Atherogenic index of plasma; CHARLS; Glycerol triglyceride glucose index; Stroke.