Associations between triglyceride-glucose-derived indices combined with the atherogenic index of plasma and the precise severity of newly diagnosed coronary artery disease: a quantile regressions analysis

Background Metabolic dysfunction, including insulin resistance, inflammation, and dyslipidaemia, plays a critical role in the development of coronary artery disease (CAD). The triglyceride-glucose (TyG) index, the C-reactive protein-triglyceride-glucose index (CTI), and the atherogenic index of plasma (AIP) are accessible biomarkers reflecting these disturbances. However, conventional methods of mean regression result in a loss of detailed information on disease severity. This study is aimed at evaluating the associations of the TyG index, CTI, and AIP with the quantitative severity of CAD across its entire distribution using quantile regression analysis. Methods This retrospective study included 2,885 patients with newly diagnosed CAD, as confirmed by coronary angiography. CAD severity was quantitatively assessed using the Gensini score. Metabolic indices, including the TyG index, CTI, and AIP, were calculated, and their composite indices—the TyG-AIP and CTI-AIP—were also derived. Correlation analysis and quantile regression analyses were performed to evaluate the associations between these indices and CAD severity across the distribution of the Gensini score. Results All the metabolic indices were positively but weakly linearly correlated with the Gensini score (all p < 0.001). The CTI demonstrated the strongest linear correlation with the Gensini score (partial r ≈ 0.29, adjusted for multiple covariates). Quantile regression analyses revealed significant heterogeneity in the associations between metabolic indices and CAD burden across the distribution of Gensini scores. All the metabolic indices were positively correlated with the Gensini score (all p < 0.001), with the CTI showing the strongest adjusted linear correlation. Quantile regression demonstrated progressively stronger associations across higher conditional quantiles, which remained significant after multivariable adjustment. Compared with the individual indices, the CTI-AIP and TyG-AIP showed stronger associations across most quantiles. Equality-of-slope testing confirmed significant heterogeneity across quantiles for all indices (all p < 0.01). Conclusion TyG-derived indices and AIP were significantly associated with Gensini scores. Quantile regression revealed marked heterogeneity in these associations, with CTI-AIP showing the strongest effect after multivariable adjustment in patients with moderate to high Gensini scores. Clinically, these indices may serve as potential biomarkers, with significant association across the distribution of coronary heart disease burden, especially in patients with newly diagnosed CAD.