Cerebral small vessel disease (cSVD), characterized by pathologic changes in the structure and function of small brain vessels, is detectable on brain MRI in the absence of clinical symptoms. However, imaging cerebral small vessels themselves in vivo remains costly and challenging. There is growing interest in investigating whether retinal microvascular imaging features could be proxies for changes in the brain microvasculature. Using a multipronged approach, we explored the relation of retinal microvascular characteristics with MRI markers of cSVD (MRI-cSVD).
First, we explored this relationship in older community persons from the population-based 3C-Dijon cohort. MRI-cSVD was assessed on a 1.5-Tesla MRI at baseline, comprising white matter hyperintensity volume (WMHV), lacunes, and a composite extreme cSVD phenotype (WMHV extreme distributions and presence/absence of lacunes). At 10-year follow-up, participants underwent measurements of retinal microvascular features on fundus using the Singapore “I” Vessel Assessment software. To support 3C-Dijon findings, we conducted a comprehensive literature review up to July 2024 from PubMed/EMBASE and used 2-sample Mendelian randomization (MR) leveraging large-scale genome-wide association studies, to assess causality and directionality.
In 670 3C-Dijon participants (median age 70.7, 65.7% women), multivariable analyses (adjusted for age, sex, axial length, and cardiovascular risk factors) showed a significant association of lower arteriolar fractal dimension (FDa) with extreme cSVD (odds ratio [OR] 1.68, 95% CI 1.20–2.34) after multiple testing correction (p < 0.0042), and at p < 0.05, associations of lower FDa and smaller arteriolar caliber with larger WMHV (β = 0.0534 [95% CI 0.0075–0.0569] and 0.0519 [95% CI 0.0045–0.0993]), and of greater venular tortuosity (TORTv) with lacunes (OR 1.45, 95% CI 1.05–2.00). Lower FDa was also associated with poorer executive function. The systematic review of the literature identified 12 studies (N = 7,796) that showed mostly consistent direction of effects for FDa (5/6 studies), TORTv (4/6), and arteriolar caliber (10/11), although statistical significance was observed in 5 individual studies only. Two-sample MR based on large genome-wide association studies (N = 5,292–52,798) showed evidence for a potentially causal association of greater TORTv with extreme cSVD and larger WMHV (p = 0.0017 and 0.049), with no evidence for reverse causation.
We provide multimodal evidence that geometric characteristics of the retinal microvasculature are associated with increased burden of MRI-cSVD and possibly worse executive function.