Narrower retinal microvascular calibre is associated with reduced estimated glomerular filtration rate in a gender dependent manner in Type 2 diabetes

Background and aims: The retinal blood vessels provide a non-invasive “window” to study changes in the microcirculation, with similarities between the microvascular physiology in the eye and kidney. Previous studies have suggested change in retinal vessel calibre may be a useful predictor of estimated glomerular filtration rate (eGFR) in patients with diabetes but the reported direction of this association has not always been consistent. Our objective was to evaluate longitudinal changes in retinal vascular morphometry and eGFR in a population with type 2 diabetes mellitus (T2D). Materials and methods: GoDARTS is a bioresource with comprehensive longitudinal electronic medical records including hospital admissions, mortality, clinical biochemistry and medicine prescription records linked by patient-specific identifiers for a population of individuals with T2D. Through routine diabetic retinopathy (DR) screening, the earliest available fundus image of 1072 individuals were assessed with a follow-up (FuP) image captured 2 to 4 years later. Biochemical and blood pressure (BP) assessments during a period of 6 months either side of the date of each retinal photograph were obtained and the median value used to reflect the burden of these risk factors. Median serum creatinine was used to estimate GFR with the CKD-EPI equation. Retinal vascular measurements were captured using VAMPIRE 3.0 software to determine calibre-related coefficients, tortuosity and fractal dimension. Results: The mean time between baseline and follow-up was 3.01 years. At follow-up, eGFR and diastolic BP were significantly lower but HbA1c was higher, compared with baseline measures. At follow-up retinal arterioles and venules were significantly narrowed with associated diminished fractal dimension and increased venular tortuosity, compared with baseline (P<0.001). Following linear regression to adjust for potential confounders, retinal parameters were no longer associated with percentage change in eGFR. Using logistic regression, no retinal parameters were found to be associated with DR at baseline. Conclusion: While we were able to demonstrate significant temporal changes in retinal vascular calibre during the study period we were unable to identify any changes in retinal parameters significantly associated with declining eGFR in individuals with T2D over a 3-year time period. This study has used existing clinical data to evaluate potential early prognostic indicators. Failure to detect significant associations may be due to several reasons. (1) Variation in retinal parameters may not reflect ongoing changes in eGFR. (2) Our sample has insufficient power to detect associations of this effect size. (3) The 3-year time period between sampling is of insufficient duration to detect significant change in eGFR beyond normal age-related decline. (4) A specific eGFR threshold may be more relevant to detectable morphological changes. (5) It is possible that the progressive changes observed in the retina simply track temporal changes in the kidney.