Both increased retinal venular tortuosity and decreased fractal dimension are associated with incident diabetic retinopathy (DR) development, independent from classic disease risk factors, according to research results published in Diabetologia. 

Although cross-sectional studies have evaluated the relationship between retinal vessel trait measurements, fundus camera images, and DR in adults ages 60 to 75 years, prospective studies are needed to evaluate whether retinal vascular measurements might be used as biomarkers to predict subsequent DR development. 

Using data from the Edinburgh Type 2 Diabetes Study (ET2DS), researchers sought to determine if retinal vessel trait measurements at baseline were associated with incident diabetic retinopathy spanning a 10-year follow-up period. 


Continue Reading

At baseline, data collection included fasting blood samples for total serum cholesterol, high-density lipoprotein cholesterol, and HbA1c, early morning urine samples for creatinine and albumin measurements, and height, weight, and systolic and diastolic blood pressure. Retinal vessel traits were measured via Vascular Assessment and Measurement Platform for Images of the Retina (VAMPIRE) software. 

The cohort included 1066 participants, 340 of whom had prevalent DR at baseline. Over the follow-up period, 11.4% of participants of the 718 with no prevalent disease at baseline developed retinopathy. The median number of screening visits was 8 (range, 1-12), and the majority of participants had mild disease (R1 n=77, R2 n=1, R3 n=2, R4 n=2). 

Participants who developed retinopathy generally had increased HbA1c (60 mmol/mol vs 55 mmol/mol; 7.6% vs 7.2%). Both central retinal arteriole equivalent (CRAE) and ventral retinal venular equivalent (CRVE) vessel measurements and fractal dimension were reduced in participants with incident retinopathy; venular tortuosity, however, was increased. 

In the unadjusted model, decreased CRAE was associated with incident retinopathy (odds ratio, 0.93; 95% CI, 0.87 to 0.99). After adjusting for age and sex, as well as adding cardiometabolic and diabetes-related risk factors, the relationship was maintained. However, the addition of vascular disease history and CRVE resulted in a loss of statistical significance (OR, 0.65; 95% CI, 0.87-1.03). 

Both unadjusted and multivariable-adjusted models showed that the evidence for an association between decreased CRVE and incident retinopathy was weak (unadjusted OR, 0.95; 95% CI, 0.91 to 1.00; aOR, 0.95; 0.91 to 1.00). This association was not evident after further adjustments were made for diabetes-related risk factors, vascular risk factors, and arterial width. 

Arterial tortuosity was not associated with incident retinopathy (unadjusted OR, 0.99; 95% CI, 0.91 to 1.22), although evidence was found for a “relatively strong” association between increased venular tortuosity and incident retinopathy (OR, 1.43; 95% CI, 1.11 to 1.84), maintained after each block of covariates were added (aOR, 1.51; 95% CI, 1.15 to 1.98). 

After adjusting for cardiometabolic and diabetes risk factors, a relationship between decreased fractal dimension and incident retinopathy was noted (OR, 0.76; 95% CI, 0.60 to 0.98). 

Both venular tortuosity and fractal dimensions were added separately to a model that included the most highly cited diabetic retinopathy risk factors — HbA1c, ACR, and systolic blood pressure. Discriminative ability for venular tortuosity improved from 0.624 to 0.640 based on C statistics and confirmed by the Likelihood ratio test. When fractal dimension was added, the C statistic decreased from 0.625 to 0.621, suggesting that this addition did not improve the model. 

In a 170-participant subgroup, researchers evaluated if there was evidence of a change in individual retinal traits over time. No difference was noted in terms of width or tortuosity, but a statistically significant decrease in fractal dimension was noted. 

Several study limitations were noted; the C statistic did not “meet the general convention” of a “good model” and the ET2DS cohort was “not large enough to appropriately develop a full prediction model.” Researchers also cite insufficient power to allow for subgroup analysis by retinopathy severity as a limitation. 

“There is gathering evidence from this analysis and other similar studies that retinal vessel traits have the potential to assist in the prediction of vascular outcomes such as diabetic retinopathy,” the study shows. “Findings such as ours could assist screening programs in determining who is most at risk of developing sight-threatening retinopathy in order to inform stratified screening intervals.” 

Future research should focus on overcoming the heterogeneity of findings from different software types, evaluating the change in retinal traits over time, and developing cut points within venular tortuosity to aid clinicians, researchers add. 

Reference

Forster RB, Garcia ES, Sluiman AJ, et al; on behalf of the Edinburgh Type 2 Diabetes Study (ET2DS) Investigators. Retinal venular tortuosity and fractal dimension predict incident retinopathy in adults with type 2 diabetes: the Edinburg Type 2 Diabetes Study. Diabetologia. 2021;64(5):1103-1112. doi:10.1007%2Fs00125-021-05388-5

This article originally appeared on Ophthalmology Advisor