Macular microvascular parameters in type 2 diabetes without clinical retinopathy across strata of diabetes duration: a cross-sectional swept-source OCTA study

To determine which macular optical coherence tomography angiography (OCTA) parameters distinguish diabetic from non-diabetic eyes without clinical retinopathy, and which vary across strata of type 2 diabetes mellitus (T2DM) duration. In this single-centre cross-sectional study, 311 eyes of 183 participants (83 control eyes; 228 eyes with T2DM stratified by duration < 5, 5–<10, and ≥ 10 years) underwent 3 × 3 mm swept-source OCTA. Fourteen parameters were compared by generalized estimating equations (GEE) with an exchangeable working correlation clustered on participant ID, with GEE-based pairwise contrasts (Bonferroni) and Benjamini–Hochberg false-discovery-rate (FDR) correction across the 14 outcomes. Models were adjusted for age, sex, body mass index, hypertension, hyperlipidaemia and smoking. A separate analysis restricted to diabetic participants compared the three duration strata; discrimination used participant-level cluster bootstrap with optimism correction. Eight of 14 parameters differed among the four groups after FDR correction. Vessel diameter index (VDI) was higher in all three diabetes groups than in controls in both plexuses after full adjustment and FDR correction (VDI-SVC β = +0.100 to + 0.126; VDI-DVC β = +0.027 to + 0.037; all FDR P ≤ 0.001). No statistically significant differences were detected among the categorized duration strata for VDI-SVC or VDI-DVC (adjusted FDR P = 0.607 and 0.399, respectively). VLD-SVC was lower in the ≥ 10 y group than in controls (β = −1.493, FDR P < 0.001); that contrast is not a duration gradient. Unadjusted duration differences in VD-SVC, small-vessel density, VLD-SVC and choriocapillaris non-perfusion (FDR P = 0.007–0.013) did not survive both covariate adjustment and FDR correction (smallest adjusted FDR P = 0.128). The largest AUC was 0.734 (95% CI 0.663–0.798); none of the exploratory AUC comparisons survived FDR correction. Vessel caliber indices of both plexuses are higher in diabetic than in non-diabetic eyes independently of systemic covariates; no statistically significant VDI differences were detected among the categorized duration strata. Apparent duration-related density differences were confined to unadjusted analyses and did not survive adjustment and multiplicity correction. These modest associations require longitudinal confirmation and are not suitable for clinical use.