Sex-stratified polygenic risk scores for coronary artery disease incidence: Insights from a 20-year cohort study

Polygenic risk scores (PRS) utilize genetic variants to evaluate susceptibility to diseases. While the performance of PRS has been extensively studied in European populations, its accuracy and validation in the Middle East, particularly in Iran, remain underexplored. We aimed to assess whether PRSs improve coronary artery disease (CAD) prediction beyond clinical risk factors using sex-stratified analyses.

This longitudinal study leveraged data from the Tehran cardiometabolic genetic study (TCGS) cohort (N = 16,226), a long-term population-based study in Iran, including participants free of baseline CAD. We evaluated three established CAD-PRSs (PRS241, PRS175, PRS161) derived from GWAS of European cohorts, alongside a meta-PRS and multi-ancestry data. Multivariable Cox proportional hazards regression models were employed to assess the association between each PRS and CAD incidence, adjusting for established clinical covariates. Sensitivity analyses were conducted by sequentially adjusting for clinical risk factors to evaluate the independent effect of the PRS. Incremental predictive value beyond clinical factors was evaluated using net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision curve analysis (DCA).The association of clinical risk factors associated with PRS was also investigated.

Over a median follow-up of 15 years in 4,594 individuals, 991 CAD events occured. PRS241 demonstrated consistent sex-stratified associations, with a 70% increased risk in females (HR = 1.70, 95% CI: 1.28–2.25) and 54% in males (HR = 1.54, 95% CI: 1.22–1.93) comparing highest versus lowest quartiles. In sensitivity analyses, adjustment for HDL, LDL, hypertension, and diabetes resulted in only modest attenuation of these associations, with PRS241 maintaining significant independent risk. PRS161 was significantly associated with CAD only in males (38% increased risk), while PRS175 showed no significant associations in either sex. Integrating PRS241 into clinical risk models yielded small improvements in CAD prediction, indicated by enhanced discrimination (female: ΔC = 0.004; male: ΔC = 0.004) and reclassification (males: relative IDI = 0.0255). DCA further confirmed small net benefit for PRS241. Also, PRS241 and PRS161 were linked to adverse HDL levels in both sexes (PRS161: β = − 0.0068, FDR P-value = 0.002 in females ; β = − 0.006, FDR P-value = 0.010 in males. PRS241: β = − 0.005, FDR P-value = 0.007 in females ; β = − 0.005, FDR P-value = 0.012 in males). Notably, PRS241 also showed sex-stratified associations with hypertension (OR = 1.11, FDR P-value = 0.04) and diabetes (OR = 1.66, FDR P-value = 0.008) exclusively in females.

These findings demonstrate that incorporation of PRS241 into established clinical risk factors provides small but complementary value for CAD risk prediction. This may facilitate improved risk stratification and support more targeted primary prevention strategies in both sexes.