Circulatory Mortality, Stroke, MI Predicted With AI-Enabled Retinal Vasculometry

AI-enabled retinal vasculometry has been identified as a noninvasive biomarker for the prediction of myocardial infarction, stroke, and circulatory mortality.

The use of artificial intelligence (AI)-enabled retinal vasculometry offers a noninvasive biomarker for the prediction of circulatory mortality, myocardial infarction (MI), and stroke, according to the results of a study published in the British Journal of Ophthalmology.

Researchers sought to examine whether the inclusion of AI-enabled retinal vasculometry improves existing risk algorithms for MI, incident stroke, and circulatory mortality.

AI-enabled retinal vessel image analysis processed images from a total of 88,052 participants from the UK Biobank (UKB), who were 40 to 69 years of age at image capture. Additionally, a total of 7411 participants from European Prospective Investigation into Cancer (EPIC)-Norfolk, who were aged 48 to 92 years at image capture, were included. Data on retinal arteriolar and venular width, area, and vessel tortuosity were obtained.

The researchers developed a fully automated AI-enabled system named QUARTZ (Quantitative Analysis of Retinal Vessels Topology and Size) for examining the retinal tree, which permits detailed vasculometry quantification in large population studies. The primary outcome was circulatory mortality, defined using ICD-10 codes.

The study included a total of 64,144 UKB participants who experienced 327 circulatory deaths and 5862 EPIC-Norfolk participants who experienced 201 circulatory deaths. The mean participant age in the UKB group was 56.8 years and the median follow-up was 7.7 years. In the EPIC-Norfolk group, the mean age was 67.6 years and the median follow-up was 9.1 years.

RV offers an alternative predictive biomarker to traditional risk-scores for vascular health, without the need for blood sampling or blood pressure measurement.

Study results showed that prediction models for circulatory mortality among men and women had optimism-adjusted C-statistics between 0.75 and 0.77, as well as R2 statistics between 0.33 and 0.44. In terms of incident stroke and MI, the addition of retinal vasculometry to Framingham risk scores did not improve model performance in either cohort, although the simpler RV model did perform equally or better than the Framingham risk score.

Limitations include that UKB and EPIC-Norfolk are both “healthy” cohorts, with relatively low event rates compared with those from other geographically similar middle-aged cohorts. Additionally, because the prevalence of current smoking is very low in UKB, this limited the ability to examine interactions with retinal vasculometry. Further, the percentage of non-White participants in UKB is low.

 “RV [retinal vasculometry] offers an alternative predictive biomarker to traditional risk-scores for vascular health, without the need for blood sampling or blood pressure measurement,” the study authors wrote. “Further work is needed to examine RV in population screening to triage individuals at high-risk.”

This article originally appeared on The Cardiology Advisor

References:

Rudnicka AR, Welikala R, Barman S, et al. Artificial intelligence-enabled retinal vasculometry for prediction of circulatory mortality, myocardial infarction and stroke. Br J Ophthalmol. Published online October 4, 2022. doi:10.1136/bjo-2022-321842