Stage-related differences in retinal sublayer thickness across the full course of type 2 diabetes

Retinal neurodegeneration is emerging as an early indicator of Type 2 diabetes (T2D). However, large-scale cohort studies evaluating differences in retinal structure across all stages of T2D and their predictive value for incident risk remain limited. In this prospective UK Biobank study, we explored stage-related differences in retinal sublayers' thickness across five categories of T2D severity (healthy, prediabetes, T2D without any complication, T2D with only microvascular complications, and T2D with both micro- and macrovascular complications). Through use of spectral-domain optical coherence tomography imaging and ordinal logistic regression analyses, the relationships between autosegmented retinal fovea thicknesses and disease stages were evaluated. Across increasingly advanced stages of T2D, significant thinning was observed in overall macular thickness (odds ratio [OR] = 0.983, 95% confidence interval [C]I: 0.981–0.986), the retinal nerve fiber layer (OR = 0.967, 95% CI: 0.960–0.974), and the photoreceptor layer (OR = 0.960, 95% CI: 0.952–0.968). Over a median 11.42-year follow-up, 435 of 21,626 initially healthy individuals developed T2D. Cox proportional hazards models, adjusted for sociodemographic factors, lifestyle, body mass index, and metabolic comorbidities, revealed that lower baseline overall macular thickness (hazard ratio [HR] = 0.988, 95% CI: 0.979–0.997), inner nuclear layer thickness (HR = 0.927, 95% CI: 0.880–0.977), and total macular volume (HR = 0.538, 95% CI: 0.336–0.860) were significantly associated with an increased risk of incident T2D. In conclusion, reduced overall macular thickness is a consistent marker of T2D severity and an independent risk factor for incident T2D, highlighting the value of noninvasive retinal measurements for risk prediction and disease monitoring.