Nonlinear associations and diagnostic performance of surrogate insulin resistance indices for diabetic retinopathy in type 2 diabetes

This study assessed the association between DR and surrogate insulin resistance (IR) indices, including TyG, VAI, LAP, METS-IR, and the TG/HDL-C ratio, and evaluated their ability to identify DR in patients with type 2 diabetes (T2D). In this cross-sectional, matched case-control study, patients with T2D were classified according to DR status. Associations between surrogate IR indices and DR and their diagnostic performance were examined using RCS models, Generalized linear mixed-effects models, Threshold effect analysis, and ROC analysis. A total of 1230 patients with T2D were included (615 with DR; 615 without). RCS models showed significant nonlinear associations between all indices and DR, with higher values associated with increased odds of DR. VAI showed the strongest association, with ORs (95% CI) of 1.69 (1.24–2.30) and 2.40 (1.72–3.35) for the second and third tertiles, respectively. LAP, METS-IR, and the TG/HDL-C ratio were also significantly associated with DR, whereas TyG was not significantly associated. All indices demonstrated good discriminative ability (AUC ≥ 0.758), with LAP showing the highest AUC (0.780, 95% CI: 0.722–0.837). Surrogate IR indices showed significant nonlinear associations with DR, supporting a link between IR and DR pathophysiology. These indices may serve as practical tools for identifying DR using routine clinical data.