Cerebral Small Vessel Disease Outperforms Brain Atrophy as an Imaging Biomarker in Diabetic Retinopathy

This study aimed to examine microvascular lesions and neurodegenerative changes in diabetic retinopathy (DR) compared to type 2 diabetes mellitus (T2DM) without DR (NDR) using structural MRI and to explore their associations with DR.

243 patients with NDR and 122 patients with DR were included. Participants underwent conventional brain MRI scans, clinical measurements, and fundus examinations. Cerebral small vessel disease (CSVD) imaging parameters were obtained using AI-based software, manually verified, and corrected for accuracy. Volumes of major cortical and subcortical regions representing neurodegeneration were assessed using automated brain segmentation and quantitative techniques. Statistical analysis included T-test, chi-square test, Mann-Whitney U test, multivariate analysis of variance (MANCOVA), multivariate logistic regression, area under the receiver operating characteristic curve (AUC), and Delong test.

DR group exhibited significant differences in 11 CSVD features. Meanwhile, DR showed an atrophy trend in the frontal cortex, occipital cortex, and subcortical gray matter (GM) compared to NDR. After adjustment, DR patients exhibited greater perivascular spaces (PVS) numbers in the parietal lobe (OR = 1.394) and deep brain regions (OR = 1.066), greater dilated perivascular spaces (DPVS) numbers in the left basal ganglia (OR = 2.006), greater small subcortical infarcts (SSI) numbers in the right hemisphere (OR = 3.104), and decreased left frontal PVS (OR = 0.824), total left DPVS (OR = 0.714), and frontal cortex volume (OR = 0.959) compared to NDR. Further, the CSVD model showed a larger AUC (0.823, 95% CI: 0.781-0.866) than the brain atrophy model (AUC = 0.757, 95% CI: 0.706-0.808).

Microvascular and neurodegeneration are significantly associated with DR. CSVD is a better imaging biomarker for DR than brain atrophy.