Predicting the Degree of Coronary Artery Stenosis Through Retinal Vascular Characteristics and Minimal Clinical Information.

Purpose: To describe the correlative relationship between fundus vascular characteristics and the severity of coronary artery stenosis.

Patients And Methods: A total of 1447 patients scheduled for hospitalization at Beijing Anzhen Hospital Union Hospital for coronary angiography to assess coronary artery conditions between February 2022 and December 2022 were selected. Of these, 1418 cases were ultimately obtained, with 29 cases excluded due to failure to complete coronary angiography or fundus photography. Severity of coronary artery stenosis was assessed using Gensini score. Fundus images were obtained via non-dilated fundus photography, and retinal arteriovenous diameters, arteriovenous ratios, curvatures, and dimensions were subsequently measured automatically using a computer program.

Results: Pearson's correlation coefficients between the Gensini Score and fundus vascular characteristics were found to be as follows: The diameters of the superior temporal artery (-0.08), The diameters of the superior temporal vein (-0.03), the arteriovenous diameter ratio (-0.10), the retinal arterial dimension (-0.09), the retinal vein dimension (-0.06), the retinal arterial curvature (-0.06), the retinal vein curvature (-0.01). The area under the curve (AUC) for logistic regression modeling with the Gensini Score binarized (dichotomized into two categories based on a cutoff of 4 points) as the dependent variable and fundus vascular characteristics as the independent variable was 0.59 ± 0.05 (95% CI). The AUC for logistic regression modeling with the combination of fundus vascular characteristics and clinical information (sex, age, height, weight, smoking history, creatinine, total cholesterol, fasting blood glucose) yielded an AUC of 0.71 ± 0.07 (95% CI).

Conclusion: Coronary stenosis is inversely correlated with fundus vascular characteristics. The combination of fundus vascular characteristics with a limited amount of clinical data (sex, age, height, weight, smoking history, creatinine, total cholesterol, fasting blood glucose) may serve as a non-invasive tool for predicting the presence or absence of severe stenosis in coronary arteries.