Background: The artificial intelligence (AI)-based cardiovascular disease (CVD) risk scoring applied to retinal imaging has recently emerged as a promising tool for medical check-up and CVD prevention. In the subclinical stage of CVD, identifying residual risk factors is crucial for early prevention. In addition, atherogenic lipid markers such as non-high-density lipoprotein cholesterol (HDL-C) cholesterol, triglyceride-to-HDL-C ratio, and remnant cholesterol, have been proposed as important factors of residual risk. But there is no data about association between AI-derived retinal image based cardiovascular (CV) risk scoring and atherogenic lipid markers.
Methods: We analyzed data from 58,905 participants who underwent routine health examinations and retinal imagine at two health screening centers. Retinal photographs were assessed using the AI-based CVD risk, which estimates 5-year risk of major adverse cardiovascular events. Logistic regression was performed to evaluate the association of lipidemia profiles and metabolic syndrome with moderate-to-high CVD risk group, adjusting for age, sex, and related clinical factors
Results: The mean age was 46.16±14.67 years, and 49.48% were men. According to the AI-based CVD risk score, 36.6% of participants were categorized as having a moderate-to-high risk (score ≥30). Remnant cholesterol (odds ratio [OR]=1.12, 95% confidence interval [CI]: 1.02-1.22) were independently associated with higher AI-based CVD risk in multivariate regression model. Metabolic syndrome (OR=1.29, 95% CI: 1.17-1.42), age (OR=1.30, 95% CI: 1.29-1.30) and male sex (OR=6.21, 95% CI: 5.60-6.88) were significant with atherogenic lipid markers and AI-based CVD risk score in multivariate regression model.
Conclusion: In this subclinical population, AI-based CVD risk scores were significantly associated with atherogenic lipid markers, especially remnant cholesterol. These findings highlight the clinical usefulness of AI-derived retinal image-based CV risk scoring in early stages of CVD risk detection.