Cardiovascular Disease Genetic Risk Prediction Models: A Systematic Review

This review provides a comprehensive evaluation of genetic cardiovascular disease (CVD) risk prediction models developed or validated for the general population. The study aims to assess these models’ methodological quality, predictive performance, and validation approaches. A systematic search of electronic databases was conducted to identify studies published between 1990 and 2024. Data on study design, participant characteristics, predictor variables, statistical modeling approaches, validation methods, and performance metrics were extracted and analyzed. A total of nine studies met the inclusion criteria. Among them, five studies (56%) focused on model development, two studies (22%) performed internal validation, and two studies (22%) were externally validated. The most frequently included predictors were age (n = 9; 100%), sex (n = 7; 78%), LDL cholesterol (n = 6; 67%), and diabetes (n = 5; 56%). The genetic variants most commonly integrated into the models included APOE (n = 4), LPA (n = 5), and PCSK9 (n = 4). The AUROC scores ranged from 0.72 to 0.83, with Hosmer–Lemeshow calibration reported in five studies (56%). Cross-validation was the predominant validation method utilized in 56% of studies, whereas external validation was applied in only 22%. Although polygenic risk scores (PRS) have improved cardiovascular risk prediction, methodological inconsistencies, limited external validation, and lack of standardization remain key challenges. Future research should prioritize large-scale external validation studies, incorporate broader population-based datasets, and establish standardized performance evaluation frameworks to enhance model reliability and clinical applicability.