Retinal microvascular geometry as a biomarker of cognitive function in patients living with type 1 diabetes

Aims: Retinal vessels may serve as biomarkers of cognitive function in type 1 diabetes (T1D). This cross-sectional study evaluated whether parameters from the Retina-based Microvascular Health Assessment System (RMHAS) are associated with Montreal Cognitive Assessment (MoCA) scores, beyond relevant covariates. Methods: 42 adults with T1D underwent MoCA, 5-minute Low-Frequency to High-Frequency (LF/HF) ratio and retrieval of 5-year historical clinical and biochemical data. Estimated Glucose Disposal rate (eGDR) was calculated. Fundus images were analyzed with RMHAS. Retinal variables were filtered by Pearson correlation and reduced via principal component analysis (PCA). Hierarchical linear and logistic regression models, plus AUROC analysis, were performed. Results: Median MoCA score was 25 [IQR 23–27]; 59.5% of participants had low MoCA (<26). A clinical model (age, sex, glucose coefficient of variation [CV], mean systolic blood pressure, eGDR and triglyceride CV) correlated with MoCA (R = 0.601, p = 0.013). Adding retinal PCA factors improved fit (R square change = 0.280, p = 0.006). For low MoCA, the clinical model (LF/HF ratio, age) showed R square = 0.208 (p = 0.018); retinal factors increased fit (R square = 0.250, p = 0.005). The combined model showed promising accuracy (AUROC 0.930, p < 0.001). Conclusion: Retinal parameters associated with cognitive performance by MoCA score, emerging as a promising biomarker for cognitive risk stratification in T1D.