Joint associations of the atherogenic index of plasma and surrogate indices of insulin resistance with incident cardiovascular disease in middle-aged and older Chinese adults

Background: The triglyceride-glucose (TyG) index, estimated glucose disposal rate (eGDR), and metabolic score for insulin resistance (METS-IR) are established surrogate markers of insulin resistance (IR), while the atherogenic index of plasma (AIP) reflects atherogenic dyslipidemia. Nevertheless, evidence regarding their combined value in cardiovascular disease (CVD) risk stratification remains limited. We therefore examined the joint associations of AIP and three IR surrogate indices with incident CVD. Methods: In this prospective research, 4,117 adults aged 45 years or more from the China Health and Retirement Longitudinal Study (CHARLS) were included, all of whom were free of cardiovascular disease at the beginning. The median was used to split AIP and each IR surrogate index into two categories. Cox models, restricted cubic spline analyses, 7-year time-dependent ROC analyses, integrated discrimination improvement, net reclassification improvement, mediation analyses, subgroup analyses, and sensitivity analyses were conducted. Results: During follow-up, 749 participants developed incident CVD. In fully adjusted models, the highest combined categories of TyG-AIP, eGDR-AIP, and METS-IR-AIP were associated with higher CVD risk:1.21, 95% CI: 1.03, 1.43; 1.82, 95% CI: 1.39, 2.37 and 1.43, 95% CI:1.19, 1.73. Among participants with high AIP, only eGDR showed a significant nonlinear association with incident CVD. AIP alone provided limited improvement in discrimination, whereas eGDR showed the greatest predictive gain, increasing the 7-year AUC from 0.578 to 0.628; METS-IR showed a smaller improvement, while TyG added little. Reclassification analyses showed a similar pattern. Mediation analyses indicated that the association between AIP and incident CVD was partly mediated by eGDR and, to a lesser extent, by METS-IR, whereas no significant mediation was observed for TyG. No significant multiplicative or additive interaction was observed between AIP and TyG or eGDR, whereas AIP and METS-IR showed a significant antagonistic interaction on the multiplicative scale. Conclusion Higher AIP and adverse IR profiles were jointly associated with incident CVD. Among the three surrogate indices, eGDR showed the most informative overall profile. Combined assessment of AIP and eGDR may improve cardiovascular risk stratification.