Genetic Determinants of Circulating Estrogen Levels and Evidence of a Causal Effect of Estradiol on Bone Density in Men

Abstract

Context: Serum estradiol (E2) and estrone (E1) levels exhibit substantial heritability.

Objective: To investigate the genetic regulation of serum E2 and E1 in men.

Design, setting, and participants: Genome-wide association study in 11,097 men of European origin from nine epidemiological cohorts.

Main outcome measures: Genetic determinants of serum E2 and E1 levels.

Results: Variants in/near CYP19A1 demonstrated the strongest evidence for association with E2, resolving to three independent signals. Two additional independent signals were found on the X chromosome; FAMily with sequence similarity 9, member B (FAM9B), rs5934505 (P = 3.4 × 10-8) and Xq27.3, rs5951794 (P = 3.1 × 10-10). E1 signals were found in CYP19A1 (rs2899472, P = 5.5 × 10-23), in Tripartite motif containing 4 (TRIM4; rs17277546, P = 5.8 × 10-14), and CYP11B1/B2 (rs10093796, P = 1.2 × 10-8). E2 signals in CYP19A1 and FAM9B were associated with bone mineral density (BMD). Mendelian randomization analysis suggested a causal effect of serum E2 on BMD in men. A 1 pg/mL genetically increased E2 was associated with a 0.048 standard deviation increase in lumbar spine BMD (P = 2.8 × 10-12). In men and women combined, CYP19A1 alleles associated with higher E2 levels were associated with lower degrees of insulin resistance.

Conclusions: Our findings confirm that CYP19A1 is an important genetic regulator of E2 and E1 levels and strengthen the causal importance of E2 for bone health in men. We also report two independent loci on the X-chromosome for E2, and one locus each in TRIM4 and CYP11B1/B2, for E1.

Figure 1.

Manhattan plots for the genome-wide meta-analysis results. (a) E2 adjusted for age and BMI; (b) E2 adjusted for, age, BMI, testosterone, and SHBG; and (c) E1 adjusted for age and BMI. Red line indicates P = 5 × 10⁻⁸. Genome-wide significant loci are indicated by green. In the analysis of E1, one SNP on chromosome 1 reached the threshold for genome-wide significance (P < 5 × 10⁻⁸), but had a minor allele frequency of <0.01 in all but two cohorts; therefore, this SNP was discarded from further analyses.

Figure 2.

Proposed mechanisms underlying the associations between genome-wide significant SNPs and serum levels of E2 and T. SNPs associated with elevated levels of both E2 and T are expected to be located upstream of T. SNPs associated with elevated levels of E2 but no increase in T levels are expected to affect aromatase activity or E2 clearance. The allele associated with increased serum E2 is given for each SNP. Upwards arrow represents increase, downward arrow represents decrease, and arrow in parentheses represents nonsignificant decrease. The proposed effect of E2 on BMD is also indicated. T, testosterone.

Figure 3.

Forest plot of Mendelian randomization analyses showing the effect of E2 on BMD. Effect size of E2 on BMD expressed as SD increase in BMD per pg/mL E2. The horizontal lines represent confidence interval; the central vertical line represents precision. The values are based on a meta-analysis of all five E2-associated SNPs (rs727479, rs2899472, rs16964258, rs5934505, rs5951794). The horizontal axis shows the scale of the effects. FNBMD, femoral neck bone mineral density; LSBMD, lumbar spine bone mineral density.