A groundbreaking study published in Heart reveals that a vascular “fingerprint” in the retina—the light-sensitive tissue layer at the back of the eye—can predict stroke risk with the same accuracy as traditional methods, eliminating the need for invasive lab tests.
This “fingerprint,” consisting of 29 measurable indicators of vascular health, represents a practical, cost-effective tool for primary care and low-resource settings. Researchers emphasize its potential to transform stroke risk assessment globally, particularly as stroke impacts 100 million people annually and accounts for 6.7 million deaths, mostly linked to modifiable factors such as high blood pressure, poor diet, and smoking.
As reported by medicalxpress, the retina’s intricate vascular network closely mirrors that of the brain, making it an ideal candidate for detecting systemic health issues like diabetes. However, inconsistencies in prior studies and the specialized imaging technique—fundus photography—have limited its use in stroke risk prediction.
Leveraging artificial intelligence, the Retina-based Microvascular Health Assessment System (RMHAS) has paved the way for identifying retinal biomarkers that accurately predict stroke risk. The study analyzed retinal vascular indicators in 68,753 participants from the UK Biobank, accounting for demographic, socioeconomic, and health factors.
The final analysis included 45,161 individuals, aged 55 on average, over a 12.5-year follow-up. Of these, 749 experienced a first-time stroke. Researchers identified 29 retinal indicators significantly linked to stroke risk. These included measures of density, complexity, caliber, and twistedness of retinal vessels. Changes in density and caliber increased stroke risk by 10–19% and 10–14%, respectively, while reduced complexity and twistedness indicators increased risk by 10.5–19.5%.
Even when combined only with age and sex, the retinal fingerprint performed as well as traditional risk factors in predicting stroke.
While the study is observational and its findings may not apply to diverse populations due to the predominantly white UK Biobank cohort, researchers underscore the model’s promise.
“Given the accessibility of retinal imaging and the inclusion of basic factors like age and sex, this approach is a practical and easily implementable method for assessing stroke risk, especially in primary care and low-resource environments,” the researchers concluded.