Retinal Microvascular Abnormalities as Surrogate Markers of Cerebrovascular Ischemic Disease: A Meta-Analysis.

BACKGROUND

To determine the predictive value of retinal microvascular abnormalities for cerebrovascular ischemic diseases (CVDs), we aimed to investigate the quantitative association between retinal microvascular changes and CVD subcategories: white matter hyperintensities (WMHIs), lacunar infarcts (LIs), and cerebral infarctions (CIs).

METHODS

Using Meta-analyses Of Observational Studies in Epidemiology guidelines, we searched 6 databases through September 2016 for studies evaluating the linkage between retinal microvascular abnormalities and WMHI, and LI and CI. Studies were included if they reported odds ratios (ORs) and 95% confidence intervals or raw patient level data (that were computed into ORs). Unadjusted and vascular risk-factor adjusted ORs were pooled into meta-analysis using DerSimonian Laird random effects model. Study quality and dissemination biases were assessed and integrated.

RESULTS

From 24,444 search-identified records, 28 prospective studies encompassing 56,379 patients were eligible for the meta-analysis. After vascular risk-factor adjustment, focal arteriolar narrowing was associated with WMHI (OR, 1.24 [1.01-1.79]), LI (OR, 1.77 [1.14-2.74]), and CI (OR, 1.75 [1.14-2.69]). Venular dilation was associated with LI (OR, 1.46 [1.10-1.93]), and retinal hemorrhages with WMHI (OR, 2.23 [1.34-3.70]). Any retinopathy exhibited significant association with CI (OR, 1.96 [1.65-2.50]). Heterogeneity was significant (I2>50%) for all syntheses except retinal hemorrhages and WMHI, and retinopathy and CI (I2=0 ⋅ 0%). Associations remained significant after adjustments for quality and publication bias.

CONCLUSIONS

We found the most significant association between retinal hemorrhages and WMHI. Focal arteriolar narrowing and retinopathy predicted CVD subtypes after risk-factor adjustment, suggesting that features different than traditional vascular risk factors, are involved in CVD pathophysiology.