Abstract / Summary
Diabetic kidney disease (DKD) is a common microvascular complication of type 2 diabetes mellitus (T2DM). The cholesterol-high-density lipoprotein-glucose (CHG) index is a newly proposed lipid-glycemic composite marker; however, evidence regarding its association with DKD remains limited. Therefore, this study aimed to evaluate the association between CHG and DKD in individuals with T2DM and to compare it with the well-established triglyceride-glucose (TyG) index. In this hospital-based cross-sectional study, 1,379 adults with T2DM were included. DKD was defined as an eGFR < 60 mL/min/1.73 m² and/or albuminuria (UACR ≥ 30 mg/g). The CHG and TyG indices were calculated and analyzed both continuously and in quartiles. Associations with DKD were examined using multivariable logistic regression with progressive covariate adjustment. Restricted cubic splines (RCS) assessed the shape of these associations and potential nonlinearity. ROC curves, decision curve analysis, and reclassification indices were used to compare discrimination and exploratory apparent model performance. Prespecified subgroup analyses and exploratory sensitivity analyses assessed robustness. Among 1,379 patients with T2DM, 554 (40.2%) had DKD. In the primary model (Model 2, adjusted for age, sex, BMI, smoking, drinking, and diabetes duration), both CHG (per 1-SD OR 1.46, 95% CI 1.29–1.65) and TyG (per 1-SD OR 1.44, 95% CI 1.27–1.63) were significantly associated with DKD (both P < 0.001). These associations remained robust in the extended model (Model 3, Model 2 + SBP + DBP + lipid-lowering medication) and after further adjustment for HbA1c. Restricted cubic spline analysis showed significant overall associations (P-overall < 0.001) without significant non-linearity for CHG (P for non-linearity = 0.348) or the TyG index ( P = 0.904). ROC analyses showed comparable discrimination between CHG and TyG for DKD (AUC 0.684 vs. 0.683; DeLong P = 0.715). In exploratory analyses, adding CHG to the base model yielded small improvements in reclassification (IDI 0.026, 95% CI 0.012–0.046; continuous NRI 0.277, 95% CI 0.160–0.392) and higher net benefit than the base model across threshold probabilities of 0.30–0.60 (nominal P values, not adjusted for multiple testing). A nominally significant interaction was observed for diabetes duration ( P = 0.006), with a stronger CHG–DKD association in patients with diabetes duration ≥ 10 years. The association was driven predominantly by albuminuria. Elevated CHG and TyG indices are independently associated with prevalent DKD in T2DM, with approximately linear dose–response patterns and comparable discriminative performance. Because of the cross-sectional design, these findings demonstrate association rather than prediction and require prospective validation. Not applicable.