Rare coding variants in DNA damage repair genes associated with timing of natural menopause

Abstract

The age of menopause is associated with fertility and disease risk, and its genetic control is of great interest. We use whole-exome sequences from 132,370 women in the UK Biobank to test for associations between rare damaging variants and age at natural menopause. Rare damaging variants in five genes are significantly associated with menopause: CHEK2 (p = 3.3 × 10^-51), DCLRE1A (p = 8.4 × 10^-13), and HELB (p = 5.7 × 10^-7) with later menopause and TOP3A (p = 7.6 × 10^-8) and CLPB (p = 8.1 × 10^-7) with earlier menopause. Two additional genes are suggestive: RAD54L (p = 2.4 × 10^-6) with later menopause and HROB (p = 2.9 × 10^-6) with earlier menopause. In a follow-up analysis of repeated questionnaires in women who were initially premenopausal, CHEK2, TOP3A, and RAD54L genotypes are associated with subsequent menopause. Consistent with previous genome-wide association studies (GWASs), six of the seven genes are involved in the DNA damage repair pathway. Phenome-wide scans across 398,569 men and women revealed that in addition to known associations with cancers and blood cell counts, rare variants in CHEK2 are also associated with increased risk for uterine fibroids, polycystic ovary syndrome, and prostate hypertrophy; these associations are not shared with higher-penetrance breast cancer genes. Causal mediation analysis suggests that approximately 8% of the breast cancer risk conferred by CHEK2 pathogenic variants after menopause is mediated through delayed menopause.

Keywords: CHEK2; CLPB; DCLRE1A; HELB; HROB; RAD54L; TOP3A; breast cancer; exome sequencing; menopause.

Figure 1

Flowchart of genome-wide analyses of menopause timing

Figure 2

Manhattan plot of genome-wide scan of gene-level tests of menopause timing p value is from TTE analysis using a Cox proportional hazards model.

Figure 3

Cumulative incidence of menopause among CHEK2 PTV carriers and non-carriers between the ages of 40 and 60 years. Shaded area is the 95% confidence interval from a survival model fitted to the data.

Figure 4

Flowchart of phenome-wide analyses of ANM-associated and breast-cancer-associated genes

Figure 5

Results of phenome-wide analyses of PTVs in CHEK2 and four other canonical breast cancer genes Volcano plots for each gene show beta and p value from REGENIE. Colored in red are hematological quantitative traits, colored in blue are cancer and neoplasms (ICD10 codes beginning with letter C or D), and colored in black are other traits.

Figure 6

Mediation analysis of the proportion of CHEK2 PTV’s breast cancer effect mediated by delaying ANM The bottom arrow shows the modeled total effect of CHEK2 genotype increasing risk for breast cancer after 60 years of age (from a logistic regression model). The upper left arrow shows the modeled amount by which CHEK2 delays ANM (using linear regression among postmenopausal women). The upper right arrow shows the modeled effect of delaying ANM by 1 year on risk for breast cancer after 60 years of age (using logistic regression).