Correlation between the triglyceride-glucose index and retinal vascular parameters in the fundus: a cross-sectional study

Background The triglyceride-glucose (TyG) index is a simple, cost-effective indicator of insulin resistance, strongly linked to cardiovascular/cerebrovascular risk. Advances in deep learning enable quantitative retinal microvascular analysis from non-invasive fundus photography. This study assessed correlations between retinal vascular characteristics and the TyG index to explore relationships between glucose-lipid metabolism and microvascular alterations. Methods This cross-sectional study included 712 inpatients from Beijing Tongren Hospital Cardiovascular Center. Patients were stratified by TyG index quartiles (Q1–Q4). Retinal vascular parameters were automatically quantified from color fundus photographs (CFPs) using a deep learning segmentation model. Linear regression and subgroup analyses evaluated associations with TyG. Results Significant intergroup differences existed in hypertension, diabetes, revascularization history, Gensini score, body mass index (BMI), fasting blood glucose (FBG), total cholesterol (TC), triglycerides (TG), low- density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and glycated hemoglobin (HbA1c) ( P < 0.01). Higher TyG levels correlated with increased BMI, TC, HbA1c, FBG, LDL-C, and decreased HDL-C. Retinal parameters fractal dimension (FD), vessel density (VD), mean blood vessel diameter (MBVD), mean venous diameter (MVD), and arteriovenous ratio (AVR) significantly differed between groups ( P < 0.01), particularly Q1 vs. Q4 and Q2 vs. Q4. After full adjustment (age, sex, medical/personal history, BMI, glucose/lipids), linear regression showed FD, VD, and AVR were negatively correlated with TyG, while MVD and MBVD were positively correlated. Restricted cubic spline (RCS) confirmed linearity ( P -nonlinear >0.05). Subgroup analysis further demonstrated that these linear associations (specifically for FD and VD) remained significant within the non-diabetic population. Significant interactions with age were observed ( P -interaction < 0.05), where elevated TyG was associated with decreased FD and VD, and increased MBVD and MVD specifically in participants aged ≥65 years. Conclusion TyG levels negatively correlate with retinal FD, VD, and AVR, and positively correlate with MBVD and MVD. Significant interactions exist in individuals ≥ 65 years, where TyG correlates strongly with FD, VD, MBVD, and MVD.