according to a new study using data from the UK Biobank Eye and Vision Consortium and the European Prospective Investigation into Cancer (EPIC)–Norfolk study.
The researchers, from St. George’s University of London, Cambridge University, Kingston University, Moorfields Eye Hospital, and University College London, developed a method of artificial intelligence (AI)–enabled imaging of the retina’s vascular network that could accurately predict CVD and death, without the need for blood tests or blood pressure measurement.
The system “paves the way for a highly effective, noninvasive screening test for people at medium to high risk of circulatory disease that doesn’t have to be done in a clinic,” they said. “In the general population it could be used as a noncontact form of systemic vascular health check, to triage those at medium-high risk of circulatory mortality for further clinical risk assessment and appropriate intervention.” Optometry specialists welcomed the prospect and hailed it as “an exciting development.”
Retinal vessels give an accurate early indicator of CVD
The study, published online in the British Journal of Ophthalmology, was based on previous research showing that the width of retinal arterioles and venules seen on retinal imaging may provide an accurate early indicator of CVD, whereas current risk prediction frameworks aren’t always reliable in identifying people who will go on to develop or die of circulatory diseases.
The researchers developed a fully automated AI-enabled algorithm, called Quantitative Analysis of Retinal vessels Topology and Size (QUARTZ), to assess the potential of retinal vasculature imaging plus known risk factors to predict vascular health and death. They applied QUARTZ to retinal images from 88,052 UK Biobank participants aged 40-69 years, looking specifically at the width, vessel area, and degree of tortuosity of the retinal microvasculature, to develop prediction models for stroke, heart attack, and death from circulatory disease.
They then applied these models to the retinal images of 7,411 participants, aged 48-92 years, in the EPIC-Norfolk study. They then compared the performance of QUARTZ with the widely used Framingham Risk Scores framework.
The participants in the two studies were tracked for an average of 7.7 and 9.1 years, respectively, during which time there were 327 circulatory disease deaths among 64,144 UK Biobank participants (average age, 56.8 years) and 201 circulatory deaths among 5,862 EPIC-Norfolk participants (average age, 67.6 years).
Vessel characteristics important predictors of CVD mortality
Results from the QUARTZ models showed that in all participants, arteriolar and venular width, venular tortuosity, and width variation were important predictors of circulatory disease death. In addition, in women, but not in men, arteriolar and venular area were separate factors that contributed to risk prediction.
Overall, the predictive models, based on age, smoking, and medical history (antihypertensive or cholesterol lowering medication, diabetes, and history of stroke or MI) as well as retinal vasculature, captured between half and two-thirds of circulatory disease deaths in those most at risk, the authors said.
Compared with Framingham Risk Scores (FRS), the retinal vasculature (RV) models captured about 5% more cases of stroke in UK Biobank men, 8% more cases in UK Biobank women, and 3% more cases among EPIC-Norfolk men most at risk, but nearly 2% fewer cases among EPIC-Norfolk women. However, the team said that, while adding RV to FRS resulted in only marginal changes in prediction of stroke or MI, a simpler noninvasive risk score based on age, sex, smoking status, medical history, and RV “yielded comparable performance to FRS, without the need for blood sampling or BP measurement.”