Discovery of novel ancestry specific genes for androgens and hypogonadism in Million Veteran Program Men

Abstract

Given the various roles of testosterone in men's health, we conducted a multi-ancestral genetic analysis of total testosterone, free testosterone, SHBG, and hypogonadism in men within the Million Veteran Program (MVP). Here we identified 157 significant testosterone genetic variants, of which 8 have significant ancestry-specific associations. These variants implicate several genes, including SERPINF2, PRPF8, BAIAP2L1, SHBG, PRMT6, and PPIF, related to liver function. Genetic regulators of testosterone have cell type-specific effects in the testes, liver, and adrenal gland and are associated with disease risk. We conducted a meta-analysis amongst ancestry groups to identify 188 variants significantly associated with testosterone, of which 22 are novel associations. We constructed genetic scores for total testosterone, SHBG levels, and hypogonadism and find that men with higher testosterone genetic scores have lower odds of diabetes, hyperlipidemia, gout, and cardiac disorders. These findings provide insight into androgen regulation and identify novel variants for disease risk stratification.

Fig. 1. Testosterone variants affect gene expression in specific cell types.

Bar plot of colocalization probability (PP.H4) with GTEx gene expression in testes (orange), liver (yellow), adrenal gland (green) for (A) total testosterone, (B) free testosterone, (C) sex-hormone binding globulin (SHBG) levels and (D) hypogonadism. Full colocalization results are given in Supplementary Data 3.

Fig. 2. GWAS of Testosterone levels and hypogonadism in MVP men.

Circos plot of METAL p-values (-log base 10) of GWAS for total testosterone, free testosterone, SHBG, and hypogonadism in the MVP. The green line indicates a suggestive threshold (p < 10⁻⁶), and the red line indicates a genome-wide significance threshold (p < 5 × 10⁻⁸). The gray band correspond to Cochran's Q statistic to measure heterogeneity. Genes implicated through colocalization analysis (PP H4 > 0.8) are labeled on the outer ring and colored by cell type localization signal: liver (yellow), adrenal gland (green), testes (orange), multiple cell types (purple). A complete list of associations and nearby genes can be found in Supplementary Data 6.

Fig. 3. Genes implicated in total testosterone, free testosterone, SHBG and hypogonadism GWAS analysis.

Implication performed using variant effect predictor (VEP). Network plot of genes (coral), variants (gray), and phenotypes (teal). Scatter points are scaled by the number of edges.

Fig. 4. Total testosterone, SHBG GS and hypogonadism GRS construction and internal validation.

Decile plots of total testosterone GS (A), SHBG GS (B) and hypogonadism GRS (C) and corresponding levels of total testosterone (ng/dL), SHBG levels (nmol/dl) and odds of hypogonadism, respectively, all MVP (n = 658582) and for EUR (n = 454996), AFR (n = 122068), AMR (n = 56513) and EAS (n = 10518) ancestry groups. Data are presented as mean values +/− standard error of the mean (SEM).

Fig. 5. Association of Total Testosterone, SHBG GS and Hypogonadism GRS with Disease Risk in MVP.

CoxPH Hazards ratio of association of total testosterone GS, SHBG GS and hypogonadism GRS with diabetes (A), hyperlipidemia (B), obesity (C), gout (D), liver disorders (E) and cardiac disorders (F) in EUR (total testosterone n = 292983, SHBG n = 373815, hypogonadism n = 313728), AFR (total testosterone n = 69927, SHBG n = 94358, hypogonadism n = 75851), AMR (total testosterone n = 34304, SHBG n = 45567, hypogonadism n = 37089) and EAS (total testosterone n = 6985, SHBG n = 8319, hypogonadism n = 7405) ancestries in the MVP, excluding individuals from discovery analyses. Data are presented as mean values +/− SEM. Asterisk (*) indicates associations significant after Benjamini-Hochberg multiple test correction. Full association statistics are given in Supplementary Data 8.