Association study of CYP17 and HSD11B1 in polycystic ovary syndrome utilizing comprehensive gene coverage

Abstract

Cytochrome P450-C17 enzyme (CYP17) is an important component of the androgen synthesis pathway, a pathway that is dysfunctional in polycystic ovary syndrome (PCOS). Variation in 11-beta hydroxysteroid dehydrogenase (HSD11B1) is associated with cortisone reductase deficiency, a condition with a phenotype similar to PCOS. Both CYP17 and HSD11B1 genes have been previously studied for their possible relationship with PCOS, yielding inconsistent results. In this study, we evaluated the association between variation in these genes and PCOS. Two-hundred and eighty-seven Caucasian PCOS women and 187 Caucasian controls were genotyped for single nucleotide polymorphisms (SNPs) that were specifically chosen to allow full coverage of CYP17 and HSD11B1, including four SNPs in CYP17 and eight SNPs in HSD11B1. SNP and haplotype association analyses were conducted. Our results indicate that variants in the two genes are not associated with PCOS, or with the quantitative traits characteristic of PCOS, suggesting that these genes are not major risk factors for the syndrome.

Introduction

Polycystic ovary syndrome (PCOS), characterized by hyperandrogenism, oligo-ovulation and polycystic ovarian morphology, is a leading cause of infertility and affects 7% of women (Goodarzi and Azziz, 2006). Evidence for a genetic basis for the hyperandrogenemia of PCOS is found in the increased rates of hyperandrogenemia in the family members of PCOS probands, both with and without PCOS (Legro et al., 1998). In an effort to identify the genetic basis of PCOS, a number of genes encoding major enzymes of the androgen biosynthetic pathway have been examined and associations reported, although the support for these associations in replication studies has not been unanimous (Goodarzi, 2008). One such candidate gene is CYP17, which codes for the cytochrome P450 enzyme that catalyzes the addition of a hydroxyl group at carbon 17 of the steroid D ring of pregnenolone and progesterone to produce 17-hydroxypregnenolone and 17-hydroxyprogesterone, respectively (Gilep et al., 2011). It also has a lyase activity, acting to convert 17-hydroxyprogesterone and 17-hydroxypregnenolone to androstenedione and dehydroepiandrosterone (DHEA), respectively.

Investigators searching for PCOS genes have also looked to cortisone reductase deficiency (CRD), a condition that is partly attributed to a decrease in 11β hydroxysteroid dehydrogenase (11β-HSD1) activity. HSD11B1 is a keto reductase, whose function is to reduce cortisone to cortisol in the liver; a decrease in its activity leads to adrenal hyperandrogenism via the compensatory elevation of adrenocorticotropic hormone (Draper et al., 2003). The phenotype of CRD is in some ways similar to that of PCOS and includes traits such as hirsutism, oligo-amenorrhea and infertility in women, raising interest in the possible involvement of HSD11B1 in PCOS.

In this study, we investigated the association of CYP17 and HSD11B1 SNPs and haplotypes with PCOS susceptibility and with biochemical features of subjects with PCOS in a well-characterized cohort of Caucasian PCOS and control subjects.

Materials and Methods

Subjects and phenotyping

A total of 287 consecutive unrelated non-Hispanic White women with PCOS and 187 unrelated White control women were recruited from the Birmingham, AL, USA area as previously described (Goodarzi et al., 2006). Clinical characteristics are given in Table I. PCOS was diagnosed using the 1990 National Institutes of Health criteria (Zawadzki and Dunaif, 1992). The comprehensive physical examination and hormonal evaluation of these subjects have been previously described in detail (Azziz et al., 2004). All subjects gave written informed consent, and the study was performed according to the guidelines of the Institutional Review Boards of University of Alabama and Cedars-Sinai Medical Center.

Table I.

Clinical characteristics of the study cohort.

	Control (n= 187)	PCOS (n= 287)
Age (years)	33.0 (17.0)	27.5 (11.5)a
BMI (kg/m2)	24.1 (6.4)	34.7 (13.5)a
WHR	0.78 (0.08)	0.83 (0.10)a
mFG score	0 (0)	7.0 (5.0)a
Hirsute (%)	0	73.9a
Total testosterone (nmol/l)	1.42 (0.92)	2.77 (1.07)a
Free testosterone (pmol/l)b	12.1 (9.0)	29.1 (16.3)a
DHEAS (μmol/l)	2.58 (2.03)	5.66 (4.61)a
SHBG (nmol/l)b	220.0 (120.0)	150.0 (70.0)a
Insulin (pmol/l)	49.5 (45.9)	129.2 (129.2)a
Glucose (mmol/l)	4.77 (0.56)	4.77 (0.72)
HOMA-IR	0.92 (0.83)	2.29 (1.93)a
HOMA-%B	103.9 (59.5)	175.3 (99.3)a

BMI, body mass index, WHR, waist-to-hip ratio; mFG, modified Ferriman-Gallwey hirsutism score; DHEAS, dehydroepiandrosterone sulfate; SHBG, sex hormone-binding globulin; HOMA-IR, insulin resistance estimated by the homeostatic model assessment; HOMA-%B, beta-cell function estimated by the homeostatic model assessment.

Data are presented as median (interquartile range). Fasting insulin, glucose and HOMA variables were available in a subset (∼70%) of the cohort, all of whom were non-diabetic.

^aP < 0.001 compared with control group.

^bSHBG activity was measured by competitive-binding analysis, using Sephadex G-25 and [³H]T as the ligands; free testosterone was calculated as previously described (Boots et al., 1998). This assay gives SHBG values of ∼100–300 nmol/l in normal adult women.

Single nucleotide polymorphisms genotyping and haplotype determination

Single nucleotide polymorphisms (SNPs) were selected using frequency and validation data from the CEU (Utah residents with ancestry from northern and western Europe) population of the International HapMap database (release 24; http://hapmap.ncbi.nlm.nih.gov/) (The International HapMap Consortium, 2003) with the aim of exploiting linkage disequilibrium (LD) to capture the common haplotypes in these genes. CYP17 maps to chromosome 10q24.3 and spans 6.9 kb. HSD11B1 maps to chromosome 1q32–q41 and spans 30 kb. Genotyping of SNPs was by the 5′-exonuclease assay (Taqman MGB, Applied Biosystems, Foster City, CA, USA) as previously described (Livak, 1999; Goodarzi et al., 2003). SNPs rs10883783, rs1004467, rs6162 (Ser65Ser) and rs743572 were selected to represent the common haplotypes of the region for CYP17. SNP rs743572 is the –34 C/T promoter SNP examined in several prior studies (Diamanti-Kandarakis et al., 1999; Marszalek et al., 2001; Kahsar-Miller et al., 2004; Echiburu et al., 2008; Park et al., 2008; Pusalkar et al., 2009; Unsal et al., 2009). SNPs rs12086634, rs2236902, rs2282739, rs11119328, rs846906, rs11808690, rs6672256 and rs12060922 were selected to represent the common haplotypes of the region for HSD11B1. SNP rs12086634 (1971T/G) has been implicated in CRD (Draper et al., 2003). These 12 SNPs capture 34 of 37 (92%) of the CEU HapMap SNPs at r² > 0.8 for the two genes.

Statistical analysis

Haploview 4.2 (Barrett et al., 2005) was used to determine haplotypes as well as haplotype blocks, using the solid spine of LD algorithm. SNPs and haplotypes were tested for association with PCOS diagnosis and component phenotypes of PCOS. Associations of genotype with PCOS were evaluated using logistic regression. Associations with androgens, the modified Ferriman-Gallwey (mFG) score and insulin-related traits were evaluated using analysis of covariance in the PCOS women. All analyses were adjusted for age and BMI. Significance was taken at P < 0.008 to account for the effects of multiple testing, considering that we analyzed one LD group of SNPs from each of two genes (CYP17 and HSD11B1) against three families of traits (PCOS diagnosis, androgenic traits and metabolic traits), yielding a correction factor of six (0.05/6 = 0.008).

The sample size of 287 cases and 187 controls has an excellent power (≥90%) to detect the association of risk alleles of frequency ≥0.2 with PCOS at an odds ratio ≥2.0 and a fair-to-good power (40–90%) to detect the association at odds ratios 1.5 and 1.75. Detailed power calculations given in the Supplementary data, Table SI reveal a lower power to detect the association of rare risk alleles (frequency ≤0.1) with PCOS at an odds ratio <1.75.

Results

CYP17

We genotyped four SNPs in CYP17 (Fig. 1). SNP frequencies are shown in Table II. A strong LD (D′= 1) was observed between each pair of CYP17 SNPs. None of the CYP17 SNPs or haplotypes were associated with PCOS diagnosis or with component quantitative traits of PCOS [free or total testosterone, DHEA sulfate, sex hormone-binding globulin, mFG score, fasting glucose, fasting insulin or homeostatic model assessments of insulin resistance (HOMA-IR) and beta-cell function (HOMA-%B)]. The common haplotypes for CYP17 are displayed in Table III.

Figure 1.

Gene structure and LD plot for the cytochrome P450-C17 (CYP17) gene. The gene structure is shown at top, with exons shown as filled boxes and introns as connecting lines. The locations of the genotyped SNPs relative to the exons are indicated. The LD plot at the bottom displays D′ values (%) for each pair of SNPs in the box at the intersection of the diagonals from each SNP. The solid blocks indicate D′ = 1 (100%) for the corresponding pair of variants [all logarithm of the odds (LOD) scores ≥2 for the corresponding pair of variants; the LOD score compares the likelihood of obtaining the D′ value if the two SNPs are linked, to the likelihood of observing the same D′ purely by chance; a LOD score of 2 indicates a 100-fold likelihood]. Within the gene, SNPs were considered together in one haplotype block as indicated.

Table II.

MAF and position information on CYP17 and HSD11B1 SNPs.

SNP designation	Variation	Location in gene	Overall frequency	PCOS frequency	Control frequency
CYP17
rs10883783	T/A	Intron 7	0.323	0.331	0.310
rs1004467	T/C	Intron 3	0.096	0.096	0.094
rs6162	G/A	Exon 1	0.448	0.460	0.428
rs743572	A/G	5′ region	0.420	0.429	0.407
HSD11B1
rs12086634	T/G	Intron 3	0.202	0.197	0.211
rs2236902	A/G	Intron 4	0.017	0.014	0.020
rs2282739	C/T	Intron 4	0.237	0.238	0.237
rs11119328	C/A	Intron 4	0.170	0.165	0.177
rs846906	C/T	Intron 4	0.179	0.178	0.180
rs11808690	T/G	Intron 4	0.200	0.196	0.208
rs6672256	T/A	Intron 4	0.191	0.189	0.193
rs12060922	G/A	Intron 4	0.211	0.210	0.213

Note: MAF, minor allele frequency. Allele frequency data are from genotyping of 474 subjects. CYP17, cytochrome P450-C17 gene. HSD11B1, 11-beta hydroxysteroid dehydrogenase gene.

Table III.

CYP17 and HSD11B1 haplotypes and haplotype frequencies.

Designation	Haplotype	Overall frequency	PCOS frequency	PCOS counta	Control frequency	Control counta
CYP17
1	TTGA	0.557	0.547	307	0.572	199
2	ATAG	0.322	0.332	186	0.307	107
3	TCAG	0.097	0.096	54	0.097	34
4	TTAA	0.024	0.025	14	0.024	8
HSD11B1
1	TACCCTTG	0.574	0.575	317	0.574	198
2	TACCTTTG	0.178	0.176	97	0.181	62
3	GATACGAA	0.168	0.163	90	0.177	61
4	TATCCTTG	0.019	0.024	13	0.010	4
5	TGTCCTTG	0.017	0.014	8	0.020	7
6	GATCCGTA	0.016	0.015	9	0.016	6
7	GATCCGAA	0.015	0.017	10	0.012	4

Note: CYP17 haplotypes are composed of SNPs rs10883783, rs1004467, rs6162 and rs743572. HSD11B1 haplotypes are composed of SNPs rs12086634, rs2236902, rs2282739, rs11119328, rs846906, rs11808690, rs6672256 and rs12060922.

^aCount represents the number of chromosomes assigned a particular haplotype by the expectation maximization algorithm.

HSD11B1

We genotyped eight SNPs spanning HSD11B1 (Fig. 2). SNP frequencies are displayed in Table II. LD among markers (D′) in HSD11B1 ranged from 0.18 to 1.0 (an average pairwise D′ of 0.93). None of the HSD11B1 SNPs or haplotypes were associated with PCOS susceptibility or with component quantitative traits of PCOS. The common haplotypes are displayed in Table III.

Figure 2.

Gene structure and LD plot for the 11-beta hydroxysteroid dehydrogenase (HSD11B1) gene. The gene structure is shown at top, with exons shown as filled boxes and introns as connecting lines. The locations of the genotyped SNPs relative to the exons are indicated. The LD plot at the bottom displays D′ values (%) for each pair of SNPs in the box at the intersection of the diagonals from each SNP. The solid blocks indicate D′ = 1 (100%) for the corresponding pair of variants. The red blocks indicate a logarithm of the odds (LOD) score ≥2 for the corresponding pair of variants; blue or white solid blocks indicate a LOD score <2. The LOD score compares the likelihood of obtaining the D′ value if the two SNPs are linked, to the likelihood of observing the same D′ purely by chance; a LOD score of 2 indicates a 100-fold likelihood. Within the gene, SNPs were considered together in one haplotype block as indicated.

Discussion

In this study we did not observe association between polymorphisms in the HSD11B1 and CYP17 gene regions and PCOS risk or quantitative traits of PCOS, despite the predicted effect of these genes on androgen levels.

Although hyperandrogenism is a distinct feature of PCOS, studies of CYP17, which encodes the rate-limiting step in androgen synthesis, have yielded mixed results, which could have resulted from smaller sample sizes in those studies and the fact that most prior studies looked at only one or two variants per gene. Most studies focused almost exclusively on a variant in the 5′ promoter region (−34 T/C, rs743572). While a few studies reported an association of this variant with PCOS (Diamanti-Kandarakis et al., 1999; Pusalkar et al., 2009) or polycystic ovarian morphology (Carey et al., 1994), several others found no such association (Gharani et al., 1996; Liovic et al., 1997; Techatraisak et al., 1997; Marszalek et al., 2001; Park et al., 2008; Unsal et al., 2009).

Regarding quantitative traits, the CYP17 promoter variant has been associated with androgen levels (Diamanti-Kandarakis et al., 1999; Perez et al., 2008; Pusalkar et al., 2009), adiposity and insulin levels (Echiburu et al., 2008) in some studies; alternatively, other investigators found no association with androgen levels (Gharani et al., 1996; Techatraisak et al., 1997; Marszalek et al., 2001; Kahsar-Miller et al., 2004; Unsal et al., 2009). Only one study employed a haplotype approach, genotyping seven CYP17 SNPs including rs743572 in Korean subjects; while in this study no SNP associations were observed, one haplotype was found to be associated with PCOS (Park et al., 2008). Studies with negative association results tended to have larger sample sizes than those that had positive association results, decreasing the chances of false positives. For example, an early positive study (Carey et al., 1994) was found to have negative results when the sample size was expanded (Gharani et al., 1996). Our current negative results appear to be consistent with the balance of prior reports.

Most association studies of HSD11B1 in PCOS arrived at the same conclusion as the present report. HSD11B1 was selected as a candidate gene given the resemblance of the CRD phenotype to PCOS. One study found that the CRD-associated rs12086634 polymorphism in HSD11B1, which we included in our study, was associated with hyperandrogenism in lean PCOS patients and may also prevent obesity in these patients (Gambineri et al., 2006). Limitations of that study were the abundance of lean PCOS subjects and the lack of power in identifying the same association in the controls. A subsequent study by the same group found that HSD11B1 variants were not associated with PCOS; rather HSD11B1 variation was associated with a higher risk of metabolic syndrome in the general population, regardless of the diagnosis of PCOS (Gambineri et al., 2011). Additional HSD11B1 studies support our negative association result, such as one study examining the HSD11B1 83557insA mutation and its possible association with PCOS (San Millan et al., 2005) and another study seeking to find a possible connection between CRD-associated HSD11B1 variants and PCOS susceptibility (Draper et al., 2006).

We believe that our results bring clarity to the conflicting literature concerning the role of CYP17 and HSD11B1 in PCOS. Our comprehensive approach using haplotype techniques, allowing us to cover the entire gene region, increases confidence that these genes do not play an important role in PCOS. Also, our larger sample size compared with several prior studies reduced the chance of false positive results, as did our use of a P value cutoff corrected for multiple testing. Given these advantages of our study, we conclude that genetic variation in the CYP17 and HSD11B1 loci are not major risk factors for PCOS.

Supplementary data

Supplementary data are available at http://molehr.oxfordjournals.org/.

Authors’ roles

A.K.C. contributed to writing the manuscript. A.K.C. and M.O.G. contributed to the genetic association analysis. R.A. and M.O.G. contributed to design of the study and editing of the manuscript. R.A. contributed to recruitment of study subjects.

Funding

This work was supported by the National Institutes of Health [HD029364 to R.A. and DK079888 to M.O.G., CTSI Grant UL1RR033176]; the Helping Hand of Los Angeles, Inc. and the Winnick Clinical Scholars Award [to M.O.G].

Conflict of interest

None declared.