Abstract
PURPOSE: To evaluate the predictive value of serum cystatin C and indirect bilirubin (IBIL), both individually and in combination, for sight-threatening diabetic retinopathy (STDR) in type 2 diabetes mellitus (T2DM). METHODS: Retrospective cross-sectional study. Comprehensive demographic, clinical, and laboratory data were collected. Statistical analyses included between-group comparisons, univariable and multivariable logistic regression, and receiver operating characteristic (ROC) curve analysis. Patients were stratified into cystatin C tertiles to examine dose-response relationships. RESULTS: 151 T2DM patients (79 non-STDR, 72 STDR) recruited from two clinical centers. STDR patients showed significantly higher cystatin C (1.10 [0.91, 1.31] vs. 0.88 [0.80, 1.01] mg/L, P < 0.001) and lower IBIL levels (6.90 ± 3.89 vs. 9.72 ± 4.98 μmol/L, P < 0.001). Multivariable analysis confirmed cystatin C (OR = 1.342 per 0.1 mg/L increase, 95% CI 1.111-1.621, P = 0.002) and IBIL (OR = 1.139 per μmol/L decrease, 95% CI 1.031-1.259, P = 0.011) as independent STDR predictors. The highest cystatin C tertile had 7.576-fold increased STDR odds (95% CI 2.560-22.419, P < 0.001). ROC analysis showed cystatin C (>1.025 mg/L) predicted STDR with area under the receiver operating characteristic curve (AUROC)=0.737, while IBIL (<6.09 μmol/L) had AUROC = 0.667. Combination strategies provided flexible performance: "OR" rule (either high cystatin C or low IBIL) achieved 80.6% sensitivity, while "AND" rule (both high cystatin C and low IBIL) reached 94.9% specificity. CONCLUSIONS: Serum cystatin C and IBIL are independent predictors of STDR in T2DM. Their combination offers a flexible screening approach-achieving either high sensitivity or high specificity-providing a practical tool for risk stratification and early detection of STDR.