SUN-631 Lipids and Hemoglobin A1c as Potential Mediators of the Relationship between Later Age at Menarche and Increased Risk of Coronary Artery Disease

Abstract A. Sonawalla: None. D.I. Chasman: None. Y. Chan: None. Background: Epidemiological studies have shown a U-shaped relationship between age at menarche (AAM) and risk of coronary artery disease (CAD) in women, with both earlier and later AAM associated with an increased risk of CAD. Earlier AAM is clearly associated with known CAD risk factors, but studies have reported varying associations of later AAM with known CAD risk factors; hence, mediators of increased CAD risk in later AAM are uncertain. In prior work, we showed that the relationship between AAM and CAD depends on the underlying source of variation in AAM: later AAM due to common genetic variants is associated with a decreased risk of CAD, whereas later AAM due to causes other than common genetic variants is associated with an increased risk of CAD. Objective: To determine if the associations of AAM with CAD risk factors depend on the underlying source of variation in AAM, and to identify potential mediators of the association of later AAM with increased risk of CAD. Methods: We primarily analyzed data from 201,037 women in the UK Biobank and conducted validation studies on data from 23,268 women in the Women’s Genome Health Study (WGHS). We used polygenic scores (PGS) for AAM to estimate the effects of one source of variation in AAM: common genetic variants. To construct a PGS for each individual, we used PRS-CS and PRSice-2 with summary statistics from the largest published GWAS for AAM. From linear regression models, we estimated for each individual both the change in AAM attributable to the PGS [ΔAAM(PGS)] and the change in AAM independent of the PGS [ΔAAM(non-PGS)]. We then used linear regression and linear splines (with covariates of age and age squared) to study the relationship of ΔAAM(PGS) and ΔAAM(non-PGS) with CAD risk factors: HDL-C, LDL-C, triglycerides (TG), HbA1c, systolic and diastolic blood pressure (SBP and DBP), BMI and waist-hip ratio (WHR). Results: Higher ΔAAM(PGS), reflecting common genetic variants driving later AAM, was linearly associated with higher HDL-C, and lower TG, HbA1c, SBP, DBP, BMI and WHR; no association was seen with LDL-C. In contrast, ΔAAM(non-PGS) showed non-linear associations that differed from those with ΔAAM(PGS), particularly when driving later AAM. For example, for HDL-C, when ΔAAM(PGS) and ΔAAM(non-PGS) were positive (i.e., driving later AAM), the slopes of the associations differed for ΔAAM(PGS) vs. ΔAAM(non-PGS) (0.74 [95% CI 0.54 to 0.95], p = 8 x 10-13 vs. -0.28 [-0.37 to -0.20], p = 2 x 10-10). Some CAD risk factors (HDL-C, TG and HbA1c), showed a U-shaped relationship very similar to that seen with CAD itself. Validation studies using WGHS data produced similar results. Conclusions: Similar to CAD itself, CAD risk factors show divergent associations with later AAM driven by common genetic variants vs. later AAM independent of common genetic variants. Changes in HDL-C, triglycerides and HbA1c may mediate the association of later AAM with increased risk of CAD. Sunday, June 2, 2024