Lack of association between common polymorphisms in the 17β-hydroxysteroid dehydrogenase type V gene (HSD17B5) and precocious pubarche

Journal of Steroid Biochemistry & Molecular Biology 105 (2007) 176–180

Lack of association between common polymorphisms in the 17␤-hydroxysteroid dehydrogenase type V gene (HSD17B5) and precocious pubarche Clive J. Petry a,∗ , Ken K. Ong a,b , Dianne L. Wingate a , Francis de Zegher c , Lourdes Ib´an˜ ez d , David B. Dunger a b

a Department of Paediatrics, University of Cambridge, Cambridge CB2 2QQ, UK M.R.C. Epidemiology Unit, Strangeways Research Laboratory, Wort’s Causeway, Cambridge CB1 8RN, UK c Department of Woman & Child, University of Leuven, 3000 Leuven, Belgium d Endocrine Unit, Hospital Sant Joan de D´ eu, University of Barcelona, 08950 Barcelona, Spain

Received 15 December 2006; accepted 29 January 2007

Abstract Background: 17␤-Hydroxysteroid dehydrogenase (type V; HSD17B5) is a key enzyme involved in testosterone production in females. A single nucleotide polymorphism (SNP) in the promoter region of its gene was recently found to be associated with polycystic ovary syndrome (PCOS) and its related hyperandrogenaemia. Precocious pubarche (PP) is a clinical entity pointing to adrenal androgen excess from mid-childhood onward and is associated with ovarian androgen excess from puberty onward. It is therefore a strong risk factor for PCOS. Methods: To investigate associations between this promoter SNP along with three exonic SNPs (one non-synonymous and two synonymous) from the same gene, and PP, a case–control study was performed in 190 girls with PP (84 of which were also tested for functional ovarian hyperandrogenism) from Barcelona, Spain and 71 healthy controls. Clinical features and hormone concentrations relevant to hyperandrogenism were compared by HSD17B5 genotype and haplotype. Results: Neither HSD17B5 genotypes nor haplotype were associated with PP, or subsequent androgen excess in girls from Barcelona (all P > 0.05). Conclusions: HSD17B5 SNPs predicted to have functional effects do not appear to be a risk factor for PP in girls from Barcelona, despite these girls being at high risk of developing androgen excess in adulthood. © 2007 Elsevier Ltd. All rights reserved. Keywords: Aldoketo reductase; 3␣-Hydroxysteroid dehydrogenase; Hyperandrogenism; Polycystic ovary syndrome

1. Introduction 17␤-Hydroxysteroid dehydrogenase (EC 1.1.1.62) catalyses a number of oxidation/reduction reactions including the conversion of androstenedione to testosterone and dehydroepiandrosterone to androstenediol. Twelve human isoforms of the enzyme have been described which vary both in where they are expressed and in their substrate specificities [1]. Two of these isoforms are involved in testosterone pro∗ Corresponding author at: Department of Paediatrics, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 2QQ, UK. E-mail address: [email protected] (C.J. Petry).

0960-0760/$ – see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jsbmb.2007.01.004

duction: type 3 being expressed mainly in the testes and type 5 (HSD17B5) being more ubiquitously expressed including in the ovaries [2] and the adrenal gland (this isoform also being known as aldo-keto reductase family 1, member C3 (AKR1C3), dihydrodiol dehydrogenase 3, prostaglandin F synthase and 3␣-hydroxysteroid dehydrogenase—type II, due to the multifunctionality of the enzyme). HSD17B5 is therefore crucial for androgen production in women [3,4]. Polycystic ovary syndrome (PCOS) is the commonest cause of infertility in young women. Its diagnosis requires two of the following symptoms and signs: (i) oligo- or anovulation, (ii) clinical and/or biochemical signs of hyperandrogenism and/or (iii) polycystic ovaries (along with the

C.J. Petry et al. / Journal of Steroid Biochemistry & Molecular Biology 105 (2007) 176–180

exclusion of other aetiologies including congenital adrenal hyperplasia, androgen-secreting tumours and Cushing’s syndrome) [5]. In certain populations women who develop signs of ovarian androgen excess post-menarche present before puberty with exaggerated adrenarche and/or precocious pubarche (PP), defined as the appearance of pubic hair before age 8 [6,7]. Girls with PP become insulin resistant and dyslipidaemic just like women with PCOS, especially if the PP was preceded by fetal growth restriction. PP may therefore be a forerunner of PCOS [7,8]. Given the potential importance of HSD17 in androgen production and the hyperandrogenaemia in both PCOS and PP, HSD17B5 is a candidate gene for the development of these conditions. Whilst isoforms 1–3 are not linked with PCOS [9], a common and possibly functional single nucleotide polymorphism (SNP) in the HSD17B5 5 -promoter was recently found to be associated with PCOS [10]. Having previously found PP and its features to be associated with common polymorphisms in the insulin variable number of tandem repeats [11], the androgen receptor [12] and aromatase [13,14] genes, in the present study we therefore sought association between common variations in the HSD17B5 gene and PP in girls.

177

Healthy controls (n = 71) were either short-normal Barcelona–Spanish girls (heights between the 10th and 25th percentiles) attending a clinic for concerns about growth (although all these girls were therefore in the familial range for height and none were subsequently found to have a pathological cause for their relative shortness) or were subjects undergoing minor non-endocrine-related surgery such as those on the ear, nose or throat. None of the controls had a history of PP or PCOS or showed any symptoms or signs of androgen excess. The study protocol was approved by the Institutional Review Board of Barcelona University Hospital of Sant Joan de D´eu. Informed consent was obtained from parents and/or study subjects, including permission for the collection and genotyping of DNA samples. 2.2. Assays Serum testosterone, androstenedione, DHEAS, and sex hormone binding globulin were measured as described previously [7]. The intra- and inter-assay coefficients of variation were 9.2% and 10.4%, 6% and 11.9%, 5.3% and 3.9%, and 3.0% and 4.4%, respectively. 2.3. HSD17B5 Genotyping

2. Materials and methods 2.1. Subjects All girls with PP were recruited over a 5-year period at the Hospital Sant Joan de D´eu (Barcelona, Spain) as described previously [6–8,13]. Girls were considered eligible for the study if they had PP due to premature adrenarche, as suggested by elevated plasma dehydroepiandrosterone sulphate (DHEAS) and/or androstenedione concentrations at clinical diagnosis [6]. Exclusion criteria, as published previously, include those exclusions from the diagnosis of PCOS [5–8]. One hundred and ninety girls with PP fulfilled these criteria. Eighty four of these subjects were subsequently also studied postmenarche in the follicular phase (days 3–8) of the menstrual cycle or after 2 months of amenorrhea, with assessment of biochemical and clinical features including hirsutism scores [15] and baseline androgen levels as described [6]. Serum 17-hydroxyprogesterone levels in response to leuprolide acetate, a gonadotrophin releasing hormone agonist (Procrin, 500 ␮g sc; Abbott, Madrid, Spain) were measured, and biochemical ovarian hyperandrogenism was defined as a peak greater than 4.8 nM [6].

Four HSD17B5 SNPs were genotyped (Table 1). SNP1 was previously associated with PCOS [10] and the remaining three polymorphisms were coding SNPs (one nonsynonymous and two synonymous) chosen for the relatively high likelihood of them being functional significant [16], as predicted by SNPselector [17]. All SNPs were genotyped by mismatch PCR amplification [18] and restriction fragment length polymorphism analyses. 2.3.1. SNP 1 SNP 1 (rs3763676) is an A → G SNP in the 5 -promoter region of HSD17B5 [10]. Ten nanograms DNA was amplified (10 ␮l total volume) along with 1× reaction buffer, 200 ␮M each dNTP (Promega Ltd., Poole, UK), 2.0 mM magnesium, 12 pmol each primer (see Table 1 for primer sequences), 10% (v/v) glycerol and 0.5 units Biotaq DNA Polymerase (Bioline, London, UK). The mix was heated to 94 ◦ C for 5 min, followed by 20 cycles of 94 ◦ C (45 s), 55 ◦ C (45 s, dropping 0.5 ◦ C per cycle) and 72 ◦ C (45 s). After this the samples underwent 15 cycles of 94 ◦ C (45 s), 45 ◦ C (45 s) and 72 ◦ C (45 s), prior to a final incubation at 72 ◦ C for 10 min. Ten microliters of the resulting PCR product was incubated at 37 ◦ C for 16 h along with 1 unit of DraIII and appropriate

Table 1 Oligonucleotide primer sequences used for amplifying single nucleotide polymorphisms in the 17␤-hydroxysteroid dehydrogenase gene Polymorphism

Forward primer sequence (5 → 3 )

Reverse primer sequence (5 → 3 )

SNP1 (rs3763676) SNP2 (rs12529) SNP3 (rs7741) SNP4 (rs12387)

ctccacagaccatataagacaccct gacaagtgacagggaatggattccaaact tactacctttggttgctcctccaggtccc ggtccgaccagccttggaaaactcacttaa

Tcccaatacaggcatgaagtg cccaatacgggtttcacttctactagg cttcggatggccagtccaacctgctc aggtgttccacttacggttcttcttctgtg

178

C.J. Petry et al. / Journal of Steroid Biochemistry & Molecular Biology 105 (2007) 176–180

buffers (New England Biolabs (N.E.B.), Hitchin, UK). Separating the resulting products by agarose gel electrophoresis produced a 223 bp band for the ‘A’ allele and 198 and 25 bp bands for the ‘G’ allele. 2.3.2. SNP 2 SNP 2 (rs12529) is a non-synonymous coding SNP located in exon 1 of HSD17B5. It codes for a glutamine to histidine change at amino acid position 5 of the enzyme, which is predicted to involve a buried charge change [17,19]. The cycling conditions were similar to those for SNP 1, except that the annealing temperatures were 7 ◦ C higher throughout. The reaction mix was the same as for SNP 1 with the exception of the primer sequences (Table 1). Ten microliters of the resulting PCR product was incubated at 37 ◦ C for 16 h along with 1 unit of PstI and appropriate buffers (N.E.B.). This produced a band of 170 bp for the ‘C’ allele and 142 and 28 bp bands for the ‘G’ allele. 2.3.3. SNP 3 SNP3 (rs7741) is a synonymous coding SNP located in exon 2 of HSD17B5. The cycling conditions were the same as those for SNP 1, except that the annealing temperatures were 6 ◦ C higher throughout. The reaction mix was the similar to that for SNP 1 except that the primer sequences were different (Table 1), the concentration of magnesium was 1.0 mM and there was no glycerol added. Ten microliters of the resulting PCR product was incubated at 37 ◦ C for 16 h along with 1 unit of AvaI and appropriate buffers (N.E.B.). This produced a band of 141 bp for the ‘A’ allele and 114 and 27 bp bands for the ‘G’ allele. 2.3.4. SNP 4 SNP 4 (rs12387) is a synonymous coding SNP located in exon 3 of HSD17B5. The cycling conditions were similar to those for SNP 1, except that the annealing temperatures were 12 ◦ C higher throughout. The reaction mix was the same as for SNP 1 with the exception of the primer sequences (Table 1) and the fact that the magnesium concentration was 2.5 mM. Ten microliters of the resulting PCR product was incubated at 37 ◦ C for 16 h along with 1 unit of AfIII and appropriate buffers (N.E.B.). This produced a band of 226 bp for the ‘A’ allele and 196 and 30 bp bands for the ‘G’ allele.

initially entering all four SNPs, again using circulating testosterone concentrations as the dependent variable. Hormone concentrations and quantitative phenotypic data were compared across individual HSD17B5 genotypes by one-way ANOVA; where appropriate adjustment was made for puberty stage. Frequencies of HSD17B5 genotypes (and associations with qualitative clinical outcomes) were compared using the χ2 - or Fisher’s exact tests. HSD17B5 haplotypes were reconstructed from the four SNP genotypes using Phase (Version II) [20]. Hormone and quantitative phenotypic data were compared across HSD17B5 haplotypes by one-way ANOVA. Frequencies of HSD17B5 haplotypes were compared using the χ2 goodness-of-fit test. P < 0.05 was considered statistically significant throughout.

3. Results 3.1. Linkage disequilibrium between different HSD17B5 SNPs Genotype frequencies for the different SNPs are shown in Table 2. The strongest linkage disequilibrium (LD) in the dataset was between SNPs 1 and 2 (r2 = 0.39, D = 0.96). There was also evidence of weaker LD between SNPs 1 and 3 (Fig. 1). 3.2. Clinical outcomes None of the HSD17B5 SNP genotypes or haplotype was associated with PP (SNP1 χ2 = 0.96, P = 0.6; SNP2 χ2 = 1.09, P = 0.6; SNP3 χ2 = 0.47, P = 0.8; SNP4 χ2 = 2.09, P = 0.4; haplotype P = 0.9). Neither were any of the SNPs associated with clinical features of ovarian androgen excess such as acne (SNP1 P = 0.6; SNP2 P = 0.7; SNP3 P = 0.5; SNP4 P = 0.6; Table 2 Genotype numbers and frequencies for the four HSD17B5 single nucleotide polymorphisms that were genotyped in this study HSD17B5 SNP

Genotype

SNP1 (rs3763676)

A/A A/G G/G

74 (39.2 %) 88 (46.6 %) 27 (14.3 %)

23 (32.9 %) 37 (52.9 %) 10 (14.3 %)

SNP2 (rs12529)

C/C C/G G/G

70 (36.8 %) 85 (44.7 %) 35 (18.4 %)

24 (34.3 %) 36 (51.4 %) 10 (14.3 %)

SNP3 (rs7741)

A/A A/G G/G

123 (64.7 %) 62 (32.6 %) 5 (2.6 %)

47 (68.1 %) 21 (30.4 %) 1 (1.4 %)

SNP4 (rs12387)

A/A A/G G/G

157 (83.5 %) 28 (14.9 %) 3 (1.6 %)

55 (79.7 %) 14 (20.3 %) 0 (0 %)

2.4. Statistical analyses All genotype frequencies were consistent with Hardy– Weinberg equilibrium. Statistical analyses were performed using SPSS for windows (Version 11.5.0) (SPSS Inc., Chicago, USA). There was 80% statistical power to detect a difference between minor allele frequencies of 0.5 and 0.3. Initial HSD17B5 SNP analyses were performed by entering all four SNP genotypes as co-variates into an ANOVA model with circulating testosterone concentrations as the dependent variable. Backward linear regression was then performed (with probability of F of 0.05 for entry and 0.1 for removal)

Precocious pubarche

Controls

Up to four samples per SNP (1.5 %) failed repeated genotyping attempts. All distributions of those samples that were successfully genotyped were consistent with Hardy Weinberg equilibrium. There was no significant difference in these genotype distributions between the PP and control girls from Barcelona for any of the SNPs.

C.J. Petry et al. / Journal of Steroid Biochemistry & Molecular Biology 105 (2007) 176–180

Fig. 1. A linkage disequilibrium plot (expressed as r2 and D ) between the four HSD17B5 SNPs genotyped in this study (SNP1 rs3763676; SNP2 rs12529; SNP3 rs7741; SNP4 rs12387), plotted using JLIN software [26].

haplotype P = 0.1), or hirsutism score (overall P = 0.3; SNP1 P = 0.7; SNP2 P = 0.7; SNP3 P = 0.7; SNP4 P = 0.6; haplotype P = 0.9). 3.3. Associations with plasma androgen concentrations The four HSD17B5 SNPs were not associated with serum testosterone concentrations (overall model r2 = 1.4%, P = 0.6), unless puberty stage was adjusted for (overall model r2 = 44.8%, P < 0.0001). Even then none of the individual SNPs were associated with them (SNP1 P = 0.6; SNP2 P = 0.8; SNP3 P = 0.2; SNP4 P = 0.7). Likewise the HSD17B5 haplotype was not associated with serum testosterone concentrations (r2 = 7.1%, P = 0.9), even when puberty stage (P < 0.001) was adjusted for (overall model r2 = 42.1%, P = 0.8). Backward linear regression also removed all the HSD17B5 SNPs from the model (data not shown). There were also no detectable associations with serum androstenedione (overall P = 0.6), DHEAS (overall P = 0.8), 17-hydroxyprogesterone response to gonadotrophin releasing hormone agonist stimulation (overall P = 0.6), or sex hormone binding globulin concentrations (overall P = 0.4).

4. Discussion In this study we failed to find an association between four common polymorphisms in the HSD17B5 gene and either PP, its resulting hyperandrogenaemia or postmenarcheal functional ovarian hyperandrogenism in girls from Barcelona. Three of the polymorphisms that we genotyped were coding SNPs: SNP2 (non-synonymous) and SNPs 3 and 4 (both syn-

179

onymous). The remaining 5 -promoter polymorphism, SNP1, was previously shown to be associated with PCOS [10]. As evidence suggests that PP may be a precedent of PCOS [7,8], we sought association between the HSD17B5 SNP1 and either PP or the subsequent development of ovarian androgen excess. This SNP is located 71 bp upstream of the HSD17B5 transcription initiation site and a few base pairs upstream from the Sp1/Sp3 binding site of the promoter [21] suggesting that the functionality of this SNP relates to an alteration of the binding affinity of Sp1/Sp3 to the promoter [19]. There are a number of reasons why we may not have been able to corroborate or extend the association between SNP1 and androgen excess. Firstly the sample size of our cohort may have lacked the statistical power to detect such an association. This is unlikely though unless the associated risk is very small since associations with phenotype have been found previously with polymorphisms in different genes where SNP allele frequencies were similar in our population to those found for HSD17B5 SNP1 [11,13,14]. In these studies there would have been similar and clearly sufficient statistical power to detect such associations. Secondly any association may be population specific, or due to ethnic differences between our Barcelona population and the mixed race American population used in the original study [10]. Instead it may merely reflect the polygenic nature of PCOS, suggesting that polymorphisms in different genes alter responses differently in diverse populations. Alternatively, it may be that the original association relates to a chance finding in the modestly sized population that was studied, especially since the risk allele for PCOS was only associated with a slight trend for increased circulating testosterone concentrations [10]. In contrast with this suggestion, however, are results from the activity assays testing the SNPs functionality which indicated that the polymorphism moderately increases HSD17B5 gene promoter activity and increases its affinity for the transcriptional factors Sp1/Sp3 [10]. Apart from SNP1, the only HSD17B5 SNP that we genotyped for which association has previously been sought with clinical phenotype is SNP2, which was associated with risk of lung cancer in a Chinese population exposed to high levels of polycyclic aromatic hydrocarbons [22]. This suggests that either this SNP, or one in strong LD with it, is functional. Due to the multiple specificities of HSD17B5 and the complexities of polycyclic aromatic hydrocarbon metabolism though, this is not necessarily the same enzymatic activity that converts androstenedione to testosterone that interests us. In conclusion we could find no association between PP and HSD17B5, despite the fact that the SNPs were predicted to have a relatively high likelihood of functionality [10,17]. Hapmap data (release #21/phase II Jul06, NCBI B35 assembly, dbSNP b125) gives only one haplotype block (defined by having a strong LD backbone of D > 0.8) covering all of the length of the HSD17B5 gene, plus 4 kb of its 5 promoter region and 1 kb of its 3 -untranslated region in its CEPH population of Utah residents with northern and western European ancestry [23]. Haplotype block structure

180

C.J. Petry et al. / Journal of Steroid Biochemistry & Molecular Biology 105 (2007) 176–180

and their resultant SNP tags appear somewhat robust across different European populations [24], so the Hapmap CEPH population may have similar haplotype block structure to the Barcelona population investigated in our study. The SNPs selected for genotyping in our study were selected on the basis of having relatively high probabilities of being functionally significant, a genotyping approach that can more easily elucidate functional SNPs than is possible with tagging SNP approaches that rely on LD [25]. In the Hapmap CEPH population using these four SNPs as tags captures all 64 common alleles within the HSD17B5 haplotype block with a mean r2 of 0.606. However an r2 of greater than 0.8 is found between only 38% of these captured alleles. Whilst the SNPs genotyped in our study therefore do not capture all the common genetic diversity in HSD17B5 and therefore we can not state with certainty that common polymorphisms in HSD17B5 are definitely not major risk factors for PP in our population, what is clear is that any polymorphisms that are risk factors are certainly not predictable using current means.

Acknowledgements The authors are extremely grateful to all the girls who agreed to participate in this study. DBD is supported by the Wellcome Trust and the Medical Research Council. FdZ is a Clinical Investigator of the Fund for Scientific Research, Flanders, Belgium.

References [1] R. Mindnich, G. Moller, J. Adamski, The role of 17 beta-hydroxysteroid dehydrogenases, Mol. Cell. Endocrinol. 218 (1–2) (2004) 7–20. [2] K.N. Qin, R.L. Rosenfield, Expression of 17 beta-hydroxysteroid dehydrogenase type 5 in human ovary: a pilot study, J. Soc. Gynecol. Invest. 7 (1) (2000) 61–64. [3] V. Luu-The, Analysis and characteristics of multiple types of human 17beta-hydroxysteroid dehydrogenase, J. Steroid Biochem. Mol. Biol. 76 (1–5) (2001) 143–151. [4] V. Luu-The, I. Dufort, G. Pelletier, F. Labrie, Type 5 17betahydroxysteroid dehydrogenase: its role in the formation of androgens in women, Mol. Cell. Endocrinol. 171 (1–2) (2001) 77–82. [5] Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group, Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome, Fertil. Steril. 81 (1) (2004) 19–25. [6] L. Ib´an˜ ez, N. Potau, R. Virdis, M. Zampolli, C. Terzi, M. Gussinye, A. Carrascosa, E.E. Vicens-Calvet, Postpubertal outcome in girls diagnosed of premature pubarche during childhood: increased frequency of functional ovarian hyperandrogenism, J. Clin. Endocrinol. Metab. 76 (6) (1993) 1599–1603. [7] L. Ib´an˜ ez, C. Valls, N. Potau, M.V. Marcos, F. de Zegher, Polycystic ovary syndrome after precocious pubarche: ontogenty of the low birthweight effect, Clin. Endocrinol. 55 (5) (2001) 667–672. [8] L. Ib´an˜ ez, N. Potau, I. Francois, F.F. de Zegher, Precocious pubarche, hyperinsulinism and ovarian hyperandrogenism in girls: relation to reduced fetal growth, J. Clin. Endocrinol. Metab. 83 (10) (1998) 3558–3562.

[9] M. Urbanek, R.S. Legro, D.A. Driscoll, R. Azziz, D.A. Ehrmann, R.J. Norman, J.F. Strauss III, R.S. Spielman, A. Dunaif, Thirty-seven candidate genes for polycystic ovary syndrome: strongest evidence for linkage is with follistatin, Proc. Natl. Acad. Sci. USA 96 (15) (1999) 8573–8578. [10] K. Qin, D.A. Ehrmann, N. Cox, S. Refetoff, R.L. Rosenfield, Identification of a functional polymorphism of the human type 5 17␤hydroxysteroid dehydrogenase gene associated with polycystic ovary syndrome, J. Clin. Endocrinol. Metab. 91 (1) (2006) 270–276. [11] L. Ib´an˜ ez, K. Ong, N. Potau, M.V. Marcos, F. de Zegher, D. Dunger, Insulin gene variable number of tandem repeat genotype and the low birth weight, precocious pubarche, and hyperinsulinism sequence, J. Clin. Endocrinol. Metab. 86 (12) (2001) 5788–5793. [12] L. Ib´an˜ ez, K.K. Ong, N. Mongan, J. Jaaskelainen, M.V. Marcos, I.A. Hughes, F. de Zegher, D.B. Dunger, Androgen receptor gene CAG repeat polymorphism in the development of ovarian hyperandrogenism, J. Clin. Endocrinol. Metab. 88 (7) (2003) 3333–3338. [13] C.J. Petry, K.K. Ong, K.F. Michelmore, S. Artigas, D.L. Wingate, A.H. Balen, F. de Zegher, L. Ib´an˜ ez, D.B. Dunger, Association of aromatase (CYP 19) gene variation with features of hyperandrogenism in two populations of young women, Hum. Reprod. 20 (7) (2005) 1837–1843. [14] C.J. Petry, K.K. Ong, K.F. Michelmore, S. Artigas, D.L. Wingate, A.H. Balen, F. de Zegher, L. Ib´an˜ ez, D.B. Dunger, Associations between common variation in the aromatase gene promoter region and testosterone concentrations in two young female populations, J. Steroid Biochem. Mol. Biol. 98 (4–5) (2006) 199–206. [15] D. Ferriman, J.D. Gallwey, Clinical assessment of body hair growth in women, J. Clin. Endocrinol. Metab. 21 (11) (1961) 1440–1447. [16] N. Risch, K. Merikangas, The future of genetic studies of complex human diseases, Science 273 (5281) (1996) 1516–1517. [17] H. Xu, S.G. Gregory, E.R. Hauser, J.E. Stenger, M.A. Pericak-Vance, J.M. Vance, S. Zuchner, M.A. Hauser, SNPselector: a web tool for selecting SNPs for genetic association studies, Bioinformatics 21 (22) (2005) 4181–4186. [18] K. Hao, T. Niu, C. Sangokoya, J. Li, X. Xu, SNPkit: an efficient approach to systematic evaluation of candidate single nucleotide polymorphisms in public databases, Biotechniques 33 (4) (2002) 822–828. [19] W. Qiu, M. Zhou, F. Labrie, S.X. Lin, Crystal structures of the multispecific 17beta-hydroxysteroid dehydrogenase type 5: critical androgen regulation in human peripheral tissues, Mol. Endocrinol. 18 (7) (2004) 1798–1807. [20] M. Stephens, N.J. Smith, P. Donnelly, A new statistical method for haplotype reconstruction from population data, Am. J. Hum. Genet. 68 (4) (2001) 978–989. [21] K. Qin, R.L. Rosenfield, Characterization of the basal promoter element of the human type 5 17beta-hydroxysteroid dehydrogenase gene, Biochim. Biophys. Acta 1728 (3) (2005) 115–125. [22] Q. Lan, J.L. Mumford, M. Shen, D.M. Demarini, M.R. Bonner, X. He, M. Yeager, R. Welch, S. Chanock, L. Tian, R.S. Chapman, T. Zheng, P. Keohavong, N. Caporaso, N. Rothman, Oxidative damage-related genes AKR1C3 and OGG1 modulate risks for lung cancer due to exposure to PAH-rich coal combustion emissions, Carcinogenesis 25 (11) (2004) 2177–2181. [23] International HapMap Consortium, A haplotype map of the human genome, Nature 437 (7063) (2005) 1299–1320. [24] J.C. Mueller, E. Lohmussaar, R. Magi, M. Remm, T. Bettecken, P. Lichtner, S. Biskup, T. Illig, A. Pfeufer, J. Luedemann, S. Schreiber, P. Pramstaller, I. Pichler, G. Romeo, A. Gaddi, A. Testa, H.E. Wichmann, A. Metspala, T. Meitinger, Linkage disequilibrium patterns and tagSNP transferability among European populations, Am. J. Hum. Genet. 76 (3) (2005) 387–398. [25] H.K. Tabor, N.J. Risch, R.M. Myers, Candidate-gene approaches for studying complex genetic traits: practical considerations, Nature Rev. Genet. 3 (5) (2002) 391–397. [26] K.W. Carter, P.A. McCaskie, L.J. Palmer, JLIN: a java based linkage disequilibrium plotter, BMC Bioinformatics 7 (2006) 60.