No Association of Polymorphisms in CYP17, CYP19, and HSD17-B1 with Plasma Estradiol Concentrations in 1,090 British Women

The production of endogenous estradiol, and thus the risk of breast cancer, may be affected by functional polymorphisms in the genes coding for enzymes in the steroid biosynthesis pathway. Potentially important polymorphisms have been identified in three genes: CYP17, which encodes for a P450 enzyme that catalyzes 17α-hydroxylase and C_17-20 lyase activities in the biosynthesis of androgens; CYP19, which encodes for aromatase, the enzyme involved in the conversion of androgens to estrogens; and HSD17-B1, which encodes for the 17β-hydroxysteroid oxidoreductase type 1 enzyme that reduces estrone to estradiol. Several studies have observedan association between risk for breast cancer and these polymorphisms (1). However, other studies have not found such an association, and the biological relationship of these polymorphisms with circulating estrogens has yet to be established. We assessed whether polymorphisms in CYP17, CYP19, or HSD17-B1 are associated with circulating levels of estradiol, the most biologically potent estrogen. Specifically, we examined a T to C substitution 34 bp upstream of the translation initiation site in the 5′ untranslated region of CYP17 among 634 premenopausal and 455 postmenopausal British women. Among the postmenopausal women, we also examined a [TTTA]_n repeat allele beginning at bp 682 (intron 4) of CYP19, an associated TCT insertion/deletion polymorphism, a C to T substitution in the untranslated region of exon 10 of CYP19, and an A to G substitution in exon 6 of HSD17-B1 1,954 bp downstream of the first nucleotide in the translation initiation codon.

Study participants included 636 premenopausal and 456 postmenopausal British women recruited to the Oxford arm of the European Prospective Investigation into Cancer and Nutrition and previously studied in a cross-sectional analysis of sex hormones and risk factors for breast cancer (2, 3). Details of the estradiol assays are described elsewhere (2). One postmenopausal and two premenopausal women were excluded from the current study due to insufficient buffy coat being available for molecular analyses.

Genotyping on DNA extracted from buffy coat samples was conducted blind after the hormone assays had been completed. Eight hundred and ninety-eight samples were genotyped at the Cancer Research UK Genotyping Facility in Oxford. Blinded quality control samples (3.4%) were inserted to validate genotype identification procedures: concordance for the blinded samples was 100% for the CYP17 assay, 93.3% for the CYP19 intron 4 assay, 100% for the CYP19 exon 10 assay, and 100% for the HSD17-B1 assay. One hundred and ninety-eight samples were genotyped for the CYP17 single nucleotide polymorphism at the Biomedical Research Centre in Dundee as part of a previous unpublished study using similar methods. Genotype frequencies from the two laboratories were statistically similar: CYP17 TT, TC, and CC genotype frequencies were 40.0%, 45.9%, and 14.1%, respectively, for samples assayed in Oxford and 38.4%, 47.0%, and 14.7%, respectively, for samples assayed in Dundee (χ² = 0.18, P = 0.9). Seven samples were genotyped for CYP17 at both laboratories and there was 100% concordance between the two sets of results.

Estradiol values were logarithmically transformed to reduce the positive skewness of the distributions, and geometric mean hormone values and the corresponding95% confidence intervals are presented. ANOVA was used to evaluate the association between genotypes andestradiol concentrations. Estradiol concentrations in postmenopausal women, classified by their number of putative high-risk alleles, including single nucleotide polymorphisms in CYP17, CYP19 exon 10, and HSD17-B1 were also examined. Statistical analyses were repeated with adjustment for potential confounders including age (5-year age groups), body mass index [calculated as weight/height² (kg/m²) and grouped <20, 20<22.5, 22.5<25, 25<27.5, ≥27.5], vigorous exercise (0, 1-2, 3-4, ≥5 hours per week), and number of days the blood sample was in the post (0-1, 2, ≥3 days; among postmenopausal women only). In the analyses of data from premenopausal women, adjustment was also made for stage of menstrual cycle (early follicular, late follicular, mid-cycle, early luteal, and late luteal defined as ≥22, 16-21, 12-15, 3-11, and 0-2 days before next menstrual period), smoking status (never, former, current), alcohol consumption (<0.50, 0.50<8.0, 8.0<16.0, ≥16.0 g/d), and hours since last meal (0-1, 2, 3-4, ≥5 hours).

There was no statistically significant variation in plasma estradiol concentrations by CYP17 genotype in premenopausal or postmenopausal women, either before or after adjustment for potential confounders. Table 1 shows the results from the multivariable model. We also found no association between CYP19 or HSD17-B1 variants and estradiol concentrations among postmenopausal women. When subjects were classified according to the number of putative high-risk alleles in CYP17, CYP19 exon 10, and HSD17-B1, we did not observe a significant dose-response relationship between estradiol concentrations and number of high-risk alleles, nor did we observe significant heterogeneity in estradiol concentrations by the number of high-risk alleles either before or after adjustment for potential confounders.

The results of this study suggest that there is no association between plasma estradiol concentrations and a polymorphism in CYP17 in either premenopausal or postmenopausal women, in contrast with the findings of two early studies that found the same polymorphism to be significantly associated with estradiol concentrations (4, 5) but in agreement with several more recent investigations (6-10).

Aromatase, encoded by CYP19, catalyzes the last stage in estrogen production; therefore, the effects of variants of this gene are of particular interest in relation to plasma estradiol concentrations. Previous studies have found elevated estradiol concentrations in carriers of the CYP19 8-repeat allele (10, 11) and the longer repeat alleles (12 and 13 repeats; ref. 12) compared with noncarriers, and reduced estradiol levels in carriers of the 7-repeat allele (11) and the 3-bp deletion polymorphism (10), although associations reached statistical significance in only one study (10). Our results, however, suggest no association between these polymorphisms in CYP19 and plasma estradiol concentrations in postmenopausal women. Our null findings for the CYP19 exon 10 polymorphism in relation to estradiol concentrations are consistent with results from the only other study of this variant and estradiol (13).

Polymorphisms in the gene HSD17-B1 may also be important given the role of 17β-hydroxysteroid oxidoreductase type 1 in the reduction of estrone to the more biological potent estradiol, yet to the authors' knowledge, no previously published studies have investigated polymorphisms in this gene in relation to estradiol concentrations. We found no association between a polymorphism in this gene and estradiol concentrations in postmenopausal women. We also found no evidence that the effects of individual polymorphisms in genes involved in steroid biosynthesis were cumulative as has been suggested previously (14).

This study had >80% power at a 0.05 significance level to detect differences of 25% in plasma concentrations between women with different genotypes. The results suggest that there is no large effect of the polymorphisms studied in CYP17, CYP19, and HSD17-B1 on endogenous estradiol concentrations; however, more data would be needed to establish whether or not there is a small effect.