Diabetic Retinopathy May Be Predicted Using Microvascular Metrics

Fluorescence angiogram of retina showing severity 1-A of diabetic retinopathy, associated with diabetes mellitus. The angiogram shows distinct, multiple microaneurysms.
Risk prediction is advancing due to discoveries that highlight microvascular changes and their effect on disease progression.

Peripapillary metrics, as derived using optical coherence tomography angiography (OCT-A) images, could broaden prediction models for diabetic retinopathy (DR) and referable diabetic retinopathy (RDR) as low vessel densities of the superficial capillary plexus are associated with the incidence of DR and RDR among patients with type 2 diabetes, according to data posted in the American Journal of Ophthalmology.

The prospective investigation follows 1033 eyes of 1033 patients (age range 30-80 years) with type 2 diabetes from baseline to 2-year follow-up visits. Participants were all ocular treatment-naïve, presented with visual acuity (VA) better than 20/200 Snellen, and did not have other serious systemic and ocular disorders. Assessment included en face images with swept-source OCT-A (SS-OCT-A), and the retinal superficial capillary plexus layer was isolated. Peripapillary vessel density (pVD) and vessel length density (pVLD) served as primary metrics for this analysis.

After 2 years, 222 (25.1%) participants without DR developed it, 11 (7.4%) DR eyes progressed, and 43 (4.17%) in the total sample were diagnosed with referable DR.

“Lower pVD and pVLD at the [optic nerve head] region are associated with 2-year incident DR and RDR, and capable to optimize prediction models for the risk of incident DR and RDR after adjusted established risk factors,” according to the report. 

Multivariable models revealed that low baseline pVD and pVLD in scans — circular 1.5 mm to 2.25 mm diameter and whole-image 3×3 mm2 area surrounding the ONH — significantly raised risk for DR (P <.05). By quadrant, reduced pVD and pVLD in temporal and superior sectors displayed greater risk for DR (P <.05).

Superficial capillary plexus pVD and pVLD did not show significant associations with DR progression, likely due to a small sample size for this particular measure, patients with mostly mild disease, and a relatively short follow-up period. Notably, when adding the OCT-A metrics to a standard prediction model built with factors such as HbA1c, body mass index, diabetes duration, age, and DR severity at baseline, the statistical area under the curve (AUC) model increased its performance by up to approximately 19% to predict RDR onset by 2 years (all P =.009). 

Previous cross-sectional studies showed reduced peripapillary perfusion in individuals with preclinical DR, as well as that pVD can be inversely related to disease severity. Prospective trials demonstrated VD and other vascular metrics predicted DR progression or onset of diabetic macular edema, although these cohorts displayed more severe levels of disease.

The current investigators suggest 2 primary reasons for lower pVD and pVLD — hyperglycemia prompts pericyte and endothelial cell apoptosis resulting in smaller peripapillary vessel diameter, and the renin-angiotensin system activation occurring with diabetes leads to decreased perfusion density of small vessels near the ONH. These factors can amplify retinal ischemia, especially for pVD in the superior and temporal quadrants.

Limitations of the study included a regional population, relatively small OCT-A field of view, and no assessment included for deep capillary plexus layers. Alternately, strengths were a large cohort overall and systematic evaluation of often-used peripapillary submetrics.


Yuan M, Wang W, Kang S, et al. Peripapillary microvasculature predicts the incidence and development of diabetic retinopathy: an SS-OCTA studyAm J Ophthalmol. Published online July 16, 2022. doi:10.1016/j.ajo.2022.07.001