New Baseline Models of DR Prediction Emerge at ARVO

Thomas Castles

Study helps modernize a 20-year-old testing model.

New baseline models of predicting how diabetic retinopathy severity will progress have emerged from the annual meeting of the Association of Research in Vision and Ophthalmology (ARVO) in Baltimore, Maryland.

According to Cecilia S. Lee, MD, of the University of Washington, more needed to be done to determine the risk factors for developing proliferative diabetic retinopathy (PDR) and vitreous hemorrhages — two important clinical endpoints when studying diabetic retinopathy.

“We understand that baseline diabetic retinopathy severity predicts the risk of progression, but the last ETDRS [Early Treatment Diabetic Retinopathy Testing] models were completed 20 years ago,” Lee said during a presentation at ARVO. “We wanted to see if this statement still holds true.”

To find out, Lee and a team of researchers sourced electronic medical records from 19 different clinical eye centers in the United Kingdom, where each location was the sole care provider of its diabetic retinopathy patients.

Clinicians at the eye centers evaluated the type of non-proliferative diabetic retinopathy (NPDR) of each patient, notated the presence or absence of intraretinal microvascular abnormalities (IRMAs), venous beading and venous loops, and quantitated hemorrhages.

The study included all patients at the eye centers with diabetes that had been forwarded for diabetic retinopathy screening, initial and follow-up data over a 7-year period, and relevant information for demographic factors, visual acuity, diabetic retinopathy severity, and the presence of the 3 key clinical features that define severe NPDR.

This left researchers with roughly 25,000 individuals and 50,000 eyes to study, predominantly those of Caucasian descent with an average age of 65 years. The mean presenting visual acuity was approximately 20/30, and about 70% of the patient cohort had diabetic retinopathy.

“50% of the eyes that started with severe NPDR at baseline progressed to PDR by year 5 of the study,” Lee said. “The higher level of NPDR they started with, the higher their risk of progressing to PDR.”

But that wasn’t the most interesting finding of the study, she said.

“What is more interesting was the answer to our question, ‘Did clinical features of severe NPDR differentially predict the risk of PDR?’” Lee explained.

According to the study, eyes with hemorrhages and IRMA progressed faster to PDR than compared to those with venous beading. Similarly, the eyes that had IRMA were 1.8 times more likely to develop PDR than the eyes with venous beading.

By year 5, 10% of eyes that started with severe NPDR progressed to vitreous hemorrhaging, and unsurprisingly, the eyes that started with high baseline diabetic retinopathy progressed more quickly to vitreous hemorrhaging.

Further, the eyes that had 4 quadrant hemorrhaging were approximately 4 times more likely develop vitreous hemorrhaging than the eyes that had venous beading. The eyes that had IRMA were more likely to develop vitreous hemorrhaging that the eyes with venous beading, but this data set was not significantly significant.

“We learned that baseline severity continues to predict progression in diabetic retinopathy, despite the 20-year span that we’ve had since the completion of ETDRS,” Lee said. “But what’s more significant, is that the most significant feature of proliferative diabetic retinopathy was IRMA.”

To conclude, Lee confirmed her hypothesis that clinical features are important variables in diabetic retinopathy screening, and that “this has important implications for future screening frequency of our individual patients while we approach personalized medicine.”

Limitations of Lee’s study included a lack of systemic data and an assumption of the reliability of the physicians who graded study subjects.