Abstract
Context: Few candidate genes for polycystic ovary syndrome (PCOS) are widely agreed upon largely due to lack of replication. Type 6 17β-hydroxysteroid dehydrogenase (HSD17B6) gene expression is increased in PCOS ovarian theca. Previous genetic study of HSD17B6 reported significant association of rs898611 with PCOS risk and metabolic phenotypes.
Objective: Our objective was to replicate association between polymorphisms in HSD17B6 and PCOS in a well-characterized replication cohort.
Design: We conducted a case-control association study.
Setting: Subjects were recruited from reproductive endocrinology clinics; controls were recruited from the surrounding communities of the University of Alabama at Birmingham and Cedars-Sinai Medical Center in Los Angeles. Genotyping occurred at Cedars-Sinai Medical Center.
Participants: Participants included 335 White women with PCOS and 198 White controls.
Main Measurements: We assessed HSD17B6 genotype, PCOS status, and metabolic traits.
Results: The minor allele of rs898611 was not associated with PCOS; however, it was associated with increased body mass index (P = 0.031), increased fasting insulin (P = 0.008), decreased fasting glucose/insulin ratio (P = 0.038), and increased homeostasis model assessment of insulin resistance (HOMA-IR) (P = 0.021). rs10459247 and rs10876920 were associated with increased fasting insulin (P = 0.031 and 0.019, respectively), and rs10876920 was also associated with increased HOMA-IR (P = 0.046). Haplotype T-A-T-C was associated with reduced fasting insulin (P = 0.046), and haplotype C-A-C-T was associated with increased body mass index (P = 0.032).
Conclusions: Although we did not replicate association between PCOS and rs898611, we replicated associations of this variant and others in HSD17B6 with metabolic traits. These replication data suggest a role for HSD17B6 in PCOS. How HSD17B6, an enzyme involved in steroid metabolism, may influence BMI and insulin resistance in PCOS remains to be determined.
Variants in the HSD17B6 gene are associated with body mass index and indices of insulin resistance in women with PCOS in an independent replication study.
Expression data allow identification of novel candidate genes. Ovarian theca expression data found increased levels of type 6 17β-hydroxysteroid dehydrogenase (HSD17B6) mRNA in polycystic ovary syndrome (PCOS) (1). The HSD17B6 enzyme has both epimerase and oxidative activities and converts 3α-androstanediol to dihydrotestosterone and androsterone to epiandrosterone (2,3,4).
Previous genetic study reported significant association between the minor allele of rs898611 and PCOS; this and other HSD17B6 single-nucleotide polymorphisms (SNPs) were associated with body mass index (BMI), fasting insulin, homeostasis model assessment of insulin resistance (HOMA-IR), and fasting glucose to insulin ratio (GIR) (5). Based on increased expression of HSD17B6 in PCOS ovaries and the above reported associations, we performed a replication study to confirm association between HSD17B6 and PCOS traits in a larger, well-characterized independent cohort.
Subjects and Methods
Subjects and phenotyping
We studied 335 unrelated White PCOS patients and 198 White control women recruited at two centers, the University of Alabama at Birmingham (287 PCOS and 187 controls) and Cedars-Sinai Medical Center (48 PCOS and 11 controls). Study subjects were premenopausal, nonpregnant, on no hormonal therapy, including oral contraceptives, for at least 3 months, and met 1990 National Institutes of Health criteria for PCOS (6). Parameters for defining hirsutism, hyperandrogenemia, ovulatory dysfunction, and exclusion of related disorders were previously reported (7).
Controls were healthy women, with regular menstrual cycles and no evidence of hirsutism, acne, alopecia, or endocrine dysfunction and had not taken hormonal therapy (including oral contraceptives) for at least 3 months. Controls were recruited by word of mouth and advertisements calling for healthy women.
Subjects were evaluated according to a previously described protocol (7). Fasting glucose and insulin were also obtained in a subset of the cases (75%). Homoeostasis model assessments (HOMA-IR and HOMA for β-cell function) were calculated using standard formulas (8). Table 1 presents clinical characteristics.
Table 1.
Clinical characteristics of PCOS and control subjects
| Control (n = 198) | PCOS (n = 335) | |
|---|---|---|
| Age (yr) | 32.5 (17.0) | 27.0 (11.1)a |
| BMI (kg/m2) | 24.7 (5.9) | 33.4 (14.7)a |
| mFG score | 0 (0) | 7.0 (5.0)a |
| Hirsute (%) | 0 | 71.9% |
| Total testosterone (ng/dl) | 40.5 (27.0) | 76.0 (33.0)a |
| Insulin (μIU/ml) | 6.9 (6.5) | 16.0 (18.0)a |
| Glucose (mg/dl) | 86.0 (10.0) | 86.0 (13.0) |
| HOMA-IR | 1.50 (1.51) | 3.37 (4.21)a |
| HOMA-%B | 124.6 (108.2) | 238.0 (262.2)a |
| GIR | 11.97 (10.24) | 5.47 (7.44)a |
Data are median (interquartile range). To convert total testosterone from ng/dl to nmol/liter, multiply by 0.03467; to convert insulin from μIU/ml to pmol/liter, multiply by 7.175; to convert glucose from mg/dl to mmol/liter, multiply by 0.05551. HOMA-%B, β-Cell function estimated by the homeostatic model assessment; mFG, modified Ferriman-Gallwey hirsutism score.
P < 0.001 compared with control group.
All subjects gave written informed consent, and the study was performed according to the guidelines of the Institutional Review Boards of University of Alabama at Birmingham and Cedars-Sinai Medical Center.
Genotyping and haplotype determination
Five SNPs, rs898611, rs12809466, rs11171968, rs10459247, and rs10876920, were selected because they capture (r2 > 0.8) all of the variation across the HSD17B6 gene, plus 10 kb upstream, in the Caucasian population of HapMap (release 24) (9). The initial HSD17B6 study (5) was performed before HapMap and consequently genotyped different SNPs, except for rs898611. SNPs were genotyped using Applied Biosystems (Foster City, CA) TaqMan Assays-On-Demand according to manufacturer’s instructions. The overall genotyping success rate was 96.8% (ranging from 94.1–97.3% for individual SNPs). Duplicate genotyping of 96 samples for one SNP yielded 100% concordance.
The program Haploview (version 4.1) was used to calculate linkage disequilibrium (LD, the D′ statistic) between each pair of SNPs and determine haplotypes and their frequencies (10). The solid spine of LD algorithm in Haploview was used to determine haplotype blocks. Only subjects whose haplotype assignment was higher than 95% certain were analyzed.
Statistical analysis
Unpaired t tests and χ2 tests were used to compare clinical characteristics between cases and controls; quantitative traits were log- or square-root-transformed as appropriate to reduce nonnormality. Data are presented as median (interquartile range).
Genotypic association with presence/absence of PCOS was evaluated using logistic regression, adjusting for recruitment site, BMI, and age. Association between genotype and BMI, fasting insulin, HOMA-IR, GIR, and total testosterone was performed using analysis of covariance adjusting for site, age, and BMI in all analyses except those in which BMI was the dependent variable, wherein analyses were adjusted for site and age. We restricted association analyses to these traits because they were associated with HSD17B6 variants in the original study (5).
To confirm any significant (taken as P < 0.05) associations, we estimated empirical P values by permutation. For significant associations, the samples were permuted by shuffling phenotypic data 1000 times, and subsequent association analyses were carried out to obtain the distribution of the test statistic under the null hypothesis of no association. Empiric P values were obtained as the proportion of the 1000 replicates with a P value less than or equal to the nominal ones obtained from the unshuffled data. These empiric P values are reported in Results.
In the original report, minor allele homozygotes of rs898611 manifested an odds ratio for PCOS of 3.0 (5). We used the Genetic Power Calculator (11), inputting this odds ratio, PCOS prevalence of 7%, and our sample size of 335 cases and 198 controls. This revealed that our sample had higher than 99% power (α = 0.05) to detect the original effect.
Results
We genotyped five SNPs (Table 2). SNP rs11171968 was extremely rare (minor allele frequency 0.004) and was thus dropped from analysis. LD among these markers (D′) ranged from 0.86–1.0 (average pairwise D′ of 0.97). Haploview generated a single haplotype block in HSD17B6 (supplemental Fig. 1, published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org).
Table 2.
SNP information for the HSD17B6gene region
| Variant | Location | Alleles (major/minor) | PCOS MAF | Control MAF | Overall MAF |
|---|---|---|---|---|---|
| rs898611 | Intron 1 | T/C | 0.391 | 0.376 | 0.385 |
| rs12809466 | Intron 1 | A/G | 0.071 | 0.061 | 0.067 |
| rs10459247 | Intron 2 | T/C | 0.262 | 0.244 | 0.255 |
| rs10876920 | Intron 3 | C/T | 0.482 | 0.500 | 0.489 |
MAF, Minor allele frequency.
No SNPs were associated with PCOS status. PCOS carriers of the minor allele at rs898611 had increased BMI [major allele homozygotes, 32.7 (14.2) kg/m2; minor allele carriers, 34.7 (13.7) kg/m2; P = 0.031]. Three SNPs in HSD17B6 were associated with fasting insulin (Table 3). Major allele carriers at rs898611, rs10459247, and rs10876920 had significantly reduced fasting insulin compared with minor allele homozygotes. Carriers of the major allele at rs898611 and rs10876920 also had reduced HOMA-IR compared with minor allele homozygotes, and carriers of the major allele at rs898611 also had increased GIR (Table 3).
Table 3.
Association of variants in HSD17B6 with metabolic traits
| Major allele carriers | Minor allele homozygotes | P value | |
|---|---|---|---|
| Insulin (μIU/ml) | |||
| rs898611 | 15.0 (17.0) | 21.0 (23.2) | 0.008 |
| rs12809466 | 16.0 (17.7) | 11.0 (NA) | 0.50 |
| rs10459247 | 15.0 (17.2) | 21.0 (20.5) | 0.031 |
| rs10876920 | 15.0 (17.0) | 20.0 (20.5) | 0.019 |
| HOMA-IR | |||
| rs898611 | 3.18 (3.79) | 4.62 (4.82) | 0.021 |
| rs12809466 | 3.44 (4.16) | 2.77 (NA) | 0.58 |
| rs10459247 | 3.20 (3.86) | 4.50 (3.27) | 0.16 |
| rs10876920 | 3.18 (3.78) | 4.37 (4.08) | 0.046 |
| GIR | |||
| rs898611 | 5.73 (7.33) | 4.38 (4.16) | 0.038 |
| rs12809466 | 5.44 (6.96) | 9.27 (NA) | 0.45 |
| rs10459247 | 5.73 (7.46) | 4.70 (3.05) | 0.11 |
| rs10876920 | 5.73 (7.45) | 4.99 (5.80) | 0.093 |
Results are median (interquartile range). NA, Interquartile range not available because there is only one subject homozygous for the minor allele of rs12809466.
None of the common (frequency >3%) haplotypes (Table 4) were associated with PCOS. Haplotype 1 (T-A-T-C) carriers had reduced fasting insulin compared with noncarriers [carriers, 15.0 (16.5) μIU/ml; noncarriers, 18.0 (18.5); P = 0.046]. Haplotype 2 (C-A-C-T) was associated with increased BMI [carriers, 35.3 (14.5) kg/m2; noncarriers, 33.1 (14.1); P = 0.032].
Table 4.
Haplotype information for the HSD17B6gene region
| Haplotype | Alleles | PCOS frequency | Control frequency | Overall frequency |
|---|---|---|---|---|
| 1 | T-A-T-C | 0.484 | 0.481 | 0.483 |
| 2 | C-A-C-T | 0.252 | 0.240 | 0.248 |
| 3 | T-A-T-T | 0.122 | 0.145 | 0.130 |
| 4 | C-G-T-T | 0.071 | 0.060 | 0.067 |
| 5 | C-A-T-T | 0.029 | 0.053 | 0.038 |
Haplotypes are composed of SNPs rs898611, rs12809466, rs10459247, and rs10876920.
None of the above associations were observed in the controls. We detected no association between genotype and total testosterone level in cases or controls.
Discussion
Variants in HSD17B6 were previously associated with BMI and metabolic traits (5). We have replicated the specific association between the minor allele of rs898611 and increased BMI, increased HOMA-IR, increased fasting insulin, and decreased GIR and HSD17B6 haplotypes with BMI and insulin. We did not replicate association between rs898611 and PCOS susceptibility.
PCOS candidate gene studies have not been well replicated. Challenges include interstudy heterogeneity, overestimation of effect size, and insufficient statistical power to replicate the effect (12). There have been few adequately powered replication studies in replication cohorts larger than the discovery cohort. Well designed replication studies may clarify the field; examples include replication studies of HSD17B5 (13), CYP11A1 (14), and the insulin gene (15).
Our study includes several features that facilitated replication of associations between HSD17B6 and metabolic traits. Our cohort was larger than the discovery cohort, providing sufficient power to detect an effect, and was of the same race/ethnicity. Although the discovery cohort is Australian and the replication cohort American, both are Caucasian, and their comparability is demonstrated by the similar allele frequencies for rs898611 (0.36 and 0.39, respectively).
Heterogeneity of diagnostic criteria complicates PCOS research. For this report, both the Australian (discovery) and American (replication) cohorts were each diagnosed by a single clinician, although not the same clinician, strictly adhering to 1990 National Institutes of Health criteria. Ascertainment of both cohorts was also similar, with PCOS subjects recruited consecutively from reproductive endocrinology clinics and controls recruited from the surrounding population. Thus, the effect of diagnostic heterogeneity in these analyses should be minimal.
HSD17B6 may regulate the equilibrium of active androgen available for the androgen receptor, catalyzing the production of potent androgens such as 5α-dihydrotestosterone from 3α-androstanediol (16). This enzyme favors production of strong androgens from weaker androgenic substrates over the production of weaker androgens.
The associations in the discovery investigation (5) and this replication are primarily with metabolic traits. Association of a gene involved in androgen metabolism with metabolic traits, particularly insulin-related traits, deserves comment. HSD17B6 is expressed in the brain, kidney, spleen, lung, testis, adrenal, prostate, and ovary as well as insulin-responsive tissues such as liver (3,4,17) and adipose (G. Chazenbalk, personal communication), where it might influence insulin sensitivity. Its substrate and product in the ovary remain unknown. The mechanism by which this steroidogenic enzyme may influence circulating insulin levels remains unclear; however, HSD17B6 expression is highest in the liver (3,4,17), the main site of insulin clearance. Functional studies are required to elucidate how the androgen and insulin pathways interact in PCOS.
We did not replicate association between rs898611 and PCOS. This may be the result of the winner’s curse, the phenomenon of overestimation of genetic effect size in initial studies (18). If the odds ratio in the discovery cohort was actually substantially lower than 3.0, our cohort may have been underpowered to detect the effect. Other possible explanations for the lack of association with PCOS in our cohort include statistical chance, unaccounted genetic, lifestyle and/or environmental differences between the discovery and replication cohorts, and that the initial association with PCOS was a false positive. HSD17B6 might not affect PCOS susceptibility but may act as a modifier gene influencing insulin resistance and obesity.
Hyperandrogenemia is central to PCOS; however, accurate measurement of circulating androgens is challenging (19). Inconsistencies in this trait may have reduced our ability to detect association with total testosterone levels. Circulating steroid levels may be independent of levels seen in specific tissues (20). Thus, SNPs in HSD17B6 might be associated with unmeasured tissue-specific levels of androgens such as dihydrotestosterone as well as metabolic traits.
Replication in candidate gene studies is also relevant to the issue of correction for multiple testing. Replication of results is the optimal solution to multiple testing. Because our goal was to test the prior hypothesis that HSD17B6 variants are associated with PCOS and metabolic traits in affected women, we did not use the conservative Bonferroni approach that is appropriate for initial exploratory analyses. To avoid false positives arising from multiple testing, we tested for association only with the traits that were associated with HSD17B6 in the initial report (5) and confirmed all significant results with permutation analysis.
The functional significance of the associated variants is unknown. Perhaps the associated variants influence steroidogenesis in adipose tissue, accounting for the associations with obesity and insulin resistance. These variants may affect promoter activity, transcription factor binding or mRNA splicing. rs898611 is in strong LD with rs2277339, located in the promoter (5). Studies are needed to elucidate the role of HSD17B6 in relation to insulin resistance. Characterization of variation in the gene by sequencing may identify coding variants in LD with the associated variants. The expression patterns in ovarian tissues and insulin-responsive tissues and transcriptional regulation also need to be determined.
Variation in HSD17B6 is associated with several metabolic phenotypes. We have successfully replicated the previously reported (5) associations between the minor allele of rs898611 and increased BMI, increased fasting insulin, increased HOMA-IR, and reduced GIR in PCOS subjects. This indicates that replication of candidate gene results is possible in confirming the role of a gene in the etiology of PCOS. Functional work on the role of this gene in PCOS is warranted. HSD17B6 may be an interface between hyperandrogenemia and hyperinsulinemia.
Supplementary Material
Footnotes
This work was supported by National Institutes of Health Grants R01-HD29364 and K24-HD01346 (to R.A.), R01-DK79888 (to M.O.G.), and M01-RR00425 (General Clinical Research Center Grant from the National Center for Research Resources), the Winnick Clinical Scholars Award (to M.O.G.), and an endowment from the Helping Hand of Los Angeles, Inc.
Disclosure Summary: M.R.J., R.M., J.C., X.G., and M.O.G. have nothing to declare. R.A. has received consulting fees from Procter, Gamble, Merck, Co., and Organon.
First Published Online October 16, 2009
Abbreviations: BMI, Body mass index; GIR, glucose to insulin ratio; HOMA-IR, homeostasis model assessment of insulin resistance; HSD17B6, type 6 17β-hydroxysteroid dehydrogenase; LD, linkage disequilibrium; PCOS, polycystic ovary syndrome; SNP, single-nucleotide polymorphism.
References
- Wood JR, Nelson VL, Ho C, Jansen E, Wang CY, Urbanek M, McAllister JM, Mosselman S, Strauss 3rd JF 2003 The molecular phenotype of polycystic ovary syndrome (PCOS) theca cells and new candidate PCOS genes defined by microarray analysis. J Biol Chem 278:26380–26390 [DOI] [PubMed] [Google Scholar]
- Duester G 2000 Families of retinoid dehydrogenases regulating vitamin A function: production of visual pigment and retinoic acid. Eur J Biochem 267:4315–4324 [DOI] [PubMed] [Google Scholar]
- Moeller G, Adamski J 2006 Multifunctionality of human 17β-hydroxysteroid dehydrogenases. Mol Cell Endocrinol 248:47–55 [DOI] [PubMed] [Google Scholar]
- Huang XF, Luu-The V 2000 Molecular characterization of a first human 3(α→β)-hydroxysteroid epimerase. J Biol Chem 275:29452–29457 [DOI] [PubMed] [Google Scholar]
- Jones MR, Italiano L, Wilson SG, Mullin BH, Mead R, Dudbridge F, Watts GF, Stuckey BG 2006 Polymorphism in HSD17B6 is associated with key features of polycystic ovary syndrome. Fertil Steril 86:1438–1446 [DOI] [PubMed] [Google Scholar]
- Zawadzki J, Dunaif A 1992 Diagnostic criteria for polycystic ovary syndrome: towards a rational approach. In: Dunaif A, Givens JR, Haseltine FP, Merriam GR, eds. Polycystic ovary syndrome. Boston: Blackwell Scientific; 377–384 [Google Scholar]
- Azziz R, Woods KS, Reyna R, Key TJ, Knochenhauer ES, Yildiz BO 2004 The prevalence and features of the polycystic ovary syndrome in an unselected population. J Clin Endocrinol Metab 89:2745–2749 [DOI] [PubMed] [Google Scholar]
- Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC 1985 Homeostasis model assessment: insulin resistance and β-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28:412–419 [DOI] [PubMed] [Google Scholar]
- The International HapMap Consortium 2003 The International HapMap Project. Nature 426:789–796 [DOI] [PubMed] [Google Scholar]
- Barrett JC Fry B, Maller J, Daly MJ 2005 Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21:263–265 [DOI] [PubMed] [Google Scholar]
- Purcell S, Cherny SS, Sham PC 2003 Genetic Power Calculator: design of linkage and association genetic mapping studies of complex traits. Bioinformatics 19:149–150 [DOI] [PubMed] [Google Scholar]
- Goodarzi MO 2008 Looking for polycystic ovary syndrome genes: rational and best strategy. Semin Reprod Med 26:5–13 [DOI] [PubMed] [Google Scholar]
- Goodarzi MO, Jones MR, Antoine HJ, Pall M, Chen YD, Azziz R 2008 Nonreplication of the type 5 17β-hydroxysteroid dehydrogenase gene association with polycystic ovary syndrome. J Clin Endocrinol Metab 93:300–303 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gaasenbeek M, Powell BL, Sovio U, Haddad L, Gharani N, Bennett A, Groves CJ, Rush K, Goh MJ, Conway GS, Ruokonen A, Martikainen H, Pouta A, Taponen S, Hartikainen AL, Halford S, Järvelin MR, Franks S, McCarthy MI 2004 Large-scale analysis of the relationship between CYP11A promoter variation, polycystic ovarian syndrome, and serum testosterone. J Clin Endocrinol Metab 89:2408–2413 [DOI] [PubMed] [Google Scholar]
- Powell BL, Haddad L, Bennett A, Gharani N, Sovio U, Groves CJ, Rush K, Goh MJ, Conway GS, Ruokonen A, Martikainen H, Pouta A, Taponen S, Hartikainen AL, Halford S, Zeggini E, Järvelin MR, Franks S, McCarthy MI 2005 Analysis of multiple data sets reveals no association between the insulin gene variable number tandem repeat element and polycystic ovary syndrome or related traits. J Clin Endocrinol Metab 90:2988–2993 [DOI] [PubMed] [Google Scholar]
- Biswas MG, Russell DW 1997 Expression cloning and characterization of oxidative 17β- and 3α-hydroxysteroid dehydrogenases from rat and human prostate. J Biol Chem 272:15959–15966 [DOI] [PubMed] [Google Scholar]
- Belyaeva OV, Chetyrkin SV, Clark AL, Kostereva NV, SantaCruz KS, Chronwall BM, Kedishvili NY 2007 Role of microsomal retinol/sterol dehydrogenase-like short-chain dehydrogenases/reductases in the oxidation and epimerization of 3α-hydroxysteroids in human tissues. Endocrinology 148:2148–2156 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kraft P 2008 Curses–winner’s and otherwise–in genetic epidemiology. Epidemiology 19:649–651; discussion 657–658 [DOI] [PubMed] [Google Scholar]
- Rosner W, Auchus RJ, Azziz R, Sluss PM, Raff H 2007 Utility, limitations, and pitfalls in measuring testosterone: an Endocrine Society position statement. J Clin Endocrinol Metab 92:405–413 [DOI] [PubMed] [Google Scholar]
- Leon L, Bacallao K, Gabler F, Romero C, Valladares L, Vega M 2008 Activities of steroid metabolic enzymes in secretory endometria from untreated women with polycystic ovary syndrome. Steroids 73:88–95 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.