Severity Stratification of Coronary Heart Disease using novel inner Ellipse based Foveal Avascular Zone Biomarkers

Abstract

Purpose : Given the similarities between the retinal and coronary microvasculature, the retina holds promising potential to serve as non-invasive screening tool for coronary heart disease (CHD). Particularly, irregularities of the Foveal avascular zone (FAZ) may have been undervalued in their diagnostic potential. In this prospective cross-sectional study, we developed and assessed novel FAZ metrics. We aimed to discern whether FAZ alterations can serve as indicators for CHD presence and severity.

Methods : 6x6 fovea-centered OCT-A (Plex Elite 9000, Carl Zeiss Meditec) images of both eyes were acquired in patients that underwent coronary angiography during the same visit. The Gensini Score (GS) was utilized to quantify CHD severity. Six shape metrics as circularity and solidity were assessed. Additionally, three novel biomarkers based on the inner ellipse(iE) were defined: absolute iE difference, Hausdorff Distance (HD) and Chamfer Distance (CD).

Results : 280 patients were included, of which 73 had no CHD. Eight out of nine metrics showed significant associations with the GS in the left eye. However, significant differences across three CHD severity groups were only demonstrated by our novel metrics. Utilizing the Chamfer Distance, age and sex, patients with and without CHD could be distinguished with an average AUC of 0.89 (CI95% 0.84; 0.95). Moreover, three CHD severity groups could be discerned with a macro average AUC of 0.77 (CI95% 0.72; 0.84).

Conclusions : A comprehensive assessment of FAZ shape descriptors, never examined in the context of CHD previously, was performed. Utilizing the novel biomarkers, CHD patients could be distinguished with high efficiency. Further, our results showed promising potential for CHD severity staging using retinal surrogates allowing a paradigm shift for non-invasive CHD screening.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.

 

A) FAZ and iE illustration B) AUC for the binary logistic regression (BLR) model utilizing the Chamfer Distance in addition to age and sex, achieving an AUC of 89% in discerning CHD and non-CHD patients. In comparison to that C) shows the BLR baseline model with an of AUC 74%, not including any FAZ metric but only age and sex. D) shows macro-AUCs for the ordinal logistic regression model (OLR), discerning three CHD severity stages with a mean AUC of 77%, in contrast to the OLR baseline model in E) reaching only 61%. All AUCs were averaged after 5-fold cross validation.