|Home | About | Journals | Submit | Contact Us | Français|
Hydroxysteroid-dehydrogenase-17b (HSD17b) genes control the last step in estrogen biosynthesis. The isoenzymes HSD17b2 and HSD17b4 in the uterus preferentially catalyze the conversion of estradiol, the most potent and active form of estrogen, to estrone, the inactive form of estrogen. Endometrial adenocarcinoma is linked to excessive exposure to estrogens. We hypothesized that single nucleotide polymorphisms (SNPs) in genes HSD17b2 and HSD17b4 may alter the enzyme activity, estradiol levels and risk of disease.
Pairwise tag SNPs were selected from the HapMap Caucasian database to capture all known common (minor allele frequency>0.05) genetic variation with a correlation of at least 0.80. Forty-eight SNPs were genotyped in the case-control studies nested within the Nurses’ Health Study (NHS) (cases=544, controls=1296) and the Women’s Health Study (WHS) (cases=130, controls=389). The associations with endometrial cancer were examined using conditional logistic regression to estimate odds ratio and 95% confidence intervals adjusted for known risk factors. Results from the two studies were pooled using fixed effects models. We additionally investigated whether SNPs are predictive of plasma estradiol and estrone levels in the NHS using linear regression.
Four intronic SNPs were significantly associated with endometrial cancer risk (p-value <0.05). After adjustment for multiple testing, we did not observe any significant associations between SNPs and endometrial cancer risk or plasma hormone levels.
This is the first study to comprehensively evaluate variation in HSD17b2 and HSD17b4 in relation to endometrial cancer risk. Our findings suggest that variation in HSD17b2 and HSD17b4 does not substantially influence the risk of endometrial cancer in Caucasians.
Hydroxysteroid dehydrogenase 17b (HSD17b) genes are involved in the synthesis and metabolism of sex steroid hormones. There are at least eleven human HSD17b isoenzymes expressed in a variety of tissues such as the ovary, placenta, uterus, liver, adipose tissue, prostate and testis(1). HSD17b Type 2 (HSD17b2) and HSD17b Type 4 (HSD17b4) are expressed in the uterus(2, 3) and control the last step in the formation of estrogens. These enzymes preferentially catalyze the conversion of estradiol (E2), the most potent and active form of estrogen, to estrone (E1), the inactive form of estrogen (4). HSD17b2 and HSD17b4 are located on two different chromosomes, chromosome 16 and 5 respectively, span 63 Kb and 90 Kb each, and are composed of 5 and 24 exons respectively. HSD17b2 and HSD17b4 function as intracrine regulators modulating local estrogen levels (5).
Endometrial cancer is a disease linked to prolonged exposure to estrogen unopposed by progesterone. This mechanism is supported by studies demonstrating an association of reproductive factors and exogenous hormone use (oral contraceptives, postmenopausal hormones) with endometrial cancer (6). Family history of endometrial cancer has also been associated with sporadic endometrial cancer (7), suggesting genetic variability may play a role in the development of endometrial cancer . We hypothesized that genetic variation in the form of single nucleotide polymorphisms (SNPs) in estrogen metabolism genes HSD17b2 and HSD17b4 may influence enzyme activity, estradiol levels and risk of disease. A number of genes along the estrogen biosynthesis pathway have been previously examined in relation to endometrial cancer, however, no previous studies have examined HSD17b2 and HSD17b4 (8). In addition, only a small number of SNPs in these two genes have been investigated in relation to other health outcomes (9–15). We used the Nurses’ Health Study (NHS)and Women’s Health Study (WHS) nested case-control studies to comprehensively evaluate SNP variation in genes HSD17b2 and HSD17b4 and their association with endometrial cancer risk and circulating hormone levels.
The NHS case-control study of endometrial cancer is nested within the NHS prospective cohort study established in 1976 when 121,700 married female nurses aged between 30–55 years and residing in 11 US states, agreed to participate in the study. The nurses were followed every two years by completing a self-administered mailed questionnaire with detailed information on lifestyle factors and disease status (6). In 1989–90, 32,826 women completed a blood questionnaire and provided a blood sample. In 2000–2002, 33,040 women who had not provided a blood sample completed a buccal cell questionnaire and provided a buccal cell sample. Cases in this study consisted of women who provided a blood or buccal cell sample, were diagnosed with invasive endometrial cancer between 1976 through June 1, 2004 and were confirmed by medical records. Controls were randomly selected participants from the cohort who provided a blood or buccal cell sample, and had no previous report of hysterectomy and cancer through the questionnaire cycle in which the case was diagnosed. Controls were matched to cases, using a 2:1 or 3:1 ratio, on age, menopausal status at specimen collection and prior to diagnosis, postmenopausal hormone use at specimen collection, date of specimen collection, type of biospecimen, and fasting status at blood draw.
The WHS case-control study of endometrial cancer is nested within the completed WHS randomized clinical trial that examined the use of low-dose aspirin and vitamin E for the primary prevention of cancer and cardiovascular disease. The trial consists of 39,876 female health professionals across the US, aged 45 years or older when randomization began in April 1993 (16, 17). Upon randomization, every six months for the first year, and annually thereafter, participants completed a detailed questionnaire that provided information on known or potential risk factors for endometrial cancer (18). From 1993–95, 28,345 women in the WHS provided blood samples. The cases in this analysis included women who provided a blood sample and were diagnosed with endometrial cancer from blood collection through June 1, 2002. Only cases confirmed by a pathologist were considered. Controls were randomly selected from the rest of the participants who provided a blood sample, and had no previous report of cancer and hysterectomy. Controls were matched to cases, using a 3:1 ratio, on age, menopausal status and postmenopausal hormone use at specimen collection, date of specimen collection and fasting status at blood draw.
SNPs were selected using data from the HapMap population sample with European ancestry (CEU) (phase I+II release 24). SNP data were downloaded from a region including the gene and 20Kb upstream and 10Kb downstream of the gene, in order to capture 5’ and 3’ regulatory regions. We excluded SNPs with Minor Allele Frequency (MAF) <0.05. Following exclusions, a total of 71 SNPs were listed in HapMap in gene HSD17b2 and a total of 87 SNPs in gene HSD17b4. All SNPs within the genes were in introns except 3 non-synonymous SNPs in HSD17b4. One SNP was in the 5’ untranslated region (UTR) of HSD17b2. Using the software Haploview and taking advantage of the linkage disequilibrium structure between SNPs, we randomly selected pairwise tag SNPs correlated with a minimum r2 of 0.80 with all remaining SNPs in the regions of interest. We forced certain SNPs to be chosen as tag SNPs, for comparability of results between studies and because of possible functional significance. These included tag SNPs identified by the Breast and Prostate Cohort Consortium (personal communication), non-synonymous SNPs (n=3), SNPs in regulatory regions (n=1), and SNPs previously reported to be associated with other diseases (n=2 of which one SNP overlapped with an exonic SNP) (14, 15).
Power calculations were estimated using the program Quanto. The sample size of ~700 case-control pairs in this study provided more that 82% power to detect a minimum log-additive odds ratio of 1.3 for an allele with 20% or higher frequency at the 0.05 level.
DNA was extracted from leukocyte cell and buccal cell samples with the QIAGEN-QIAamp 96 DNA Blood Kit. The SNPs were genotyped using Taqman assay on Biotrove OpenArray® Real-Time qPCR system (Woburn, MA). Laboratory personnel were blinded to case status, and a random 5% of the samples were repeated for genotyping quality control. The concordance for the duplicate samples was 100%. Missing data was less than 10%. Both DNA extraction and genotyping were performed at the Harvard Partners Genotyping Facility at the Dana–Farber/Harvard Cancer Center High Throughput Genotyping Core.
In a previous study of breast cancer risk in the NHS, estradiol and estrone plasma levels were measured in 643 postmenopausal control women with no history of cancer (except non-melanoma skin cancer) and no prior PMH use in the last three months (19). A subset of our tag SNPs (n=23/48; RS2042429, RS4445895, RS2955163, RS996752, RS2955162 in HSD17b2 and RS25640, RS32646, RS382719, RS442923, RS463513, RS2455466, RS2457221, RS2636961, RS2678070, RS3797371, RS3850201, RS6897978, RS10064000, RS10478424, RS11748477, RS11749784, RS12653702, RS17388769 in HSD17b4) was previously genotyped in these controls (unpublished data). For our analysis we restricted to women at risk for endometrial cancer (intact uterus) (n=471) and who had genotype data available for HSD17b2 and HSD17b4 SNPs.
Genotype frequencies among controls were tested for Hardy-Weinberg equilibrium using a chi-square test. The success rate (SR) for each SNP was calculated as the percent of successfully genotyped samples for each SNP. Conditional logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between HSD17b2 and HSD17b4 SNP genotypes and endometrial cancer risk. We used the additive model modeling the number of minor alleles as a continuous variable. Individuals homozygous for the more common allele were coded as having zero copies of the minor allele, individuals heterozygous as having one copy, and individuals homozygous for the rare allele as having two copies. We conducted analysis adjusted for the matching factors only and separate analysis additionally adjusted for age at menarche, parity and age at first birth, body mass index at diagnosis (BMI; kg/m2), smoking status at diagnosis, postmenopausal hormone use at diagnosis, age at menopause at diagnosis, and ever oral contraceptive use. We used the fixed effects model to combine results from the two cohorts after testing for heterogeneity. We adjusted for multiple testing by controlling the false discovery rate (20). All analyses were restricted to Caucasians (98%). We used linear regression adjusted for age and laboratory batch to evaluate the association between the number of copies of minor allele and log-transformed hormone levels. SAS Version 9.1 software (SAS Institute, Cary, NC) was used for all analyses.
The NHS comprised of 544 cases (300 blood, 244 buccal cell) and 1,296 controls (817 blood, 479 buccal cell), and the WHS study of 130 cases and 389 matched controls. The complete nested case-control study was composed of 686 endometrial cancer cases and 1,729 matched controls. The basic characteristics of the two study populations are described in previous studies (21).
For gene HSD17b2 pairwise tagging yielded 24 tag SNPs that captured all 71 SNPs in the region of interest with a mean r2 of 0.95. For gene HSD17b4 pairwise tagging yielded 24 tag SNPs that captured all 87 SNPs in the region of interest with a mean r2 of 0.96. One SNP in HSD17b4 failed genotyping, however, this SNP only tagged itself, therefore coverage of the region remained high. We successfully genotyped a total of 47 SNPs; 24 SNPs in HSD17b2 and 23 SNPs in HSD17b4.
Genotype frequencies in controls were in accordance with Hardy-Weinberg equilibrium. Success rates were above 90% except for seven SNPs (SR > 85%), and one SNP (rs 2911422) failed genotyping. MAFs in our study were similar to MAFs listed in HapMap. All of the SNPs but 10 (see Table 2 footnote) had P-values greater than 0.05 for the test of heterogeneity comparing the NHS and WHS findings. Results adjusted for the matching factors only were similar to results additionally adjusted for other factors (data not shown). In the pooled analysis, we observed a total of four intronic SNPs in genes HSD17b2 and HSD17b4 with a p-value <0.05 (Table 1, ,2).2). However, after adjustment for multiple testing (47 tests) we did not observe any statistically significant associations with the polymorphisms in HSD17b2 and HSD17b4 and endometrial cancer risk. In the individual studies, a small number of SNPs in HSD17b4 including a non-synonymous SNP were found to be significant in the WHS, however, this is likely attributed to chance given the number of tests conducted, the smaller sample size in the WHS, and the fact that these SNPs had a low MAF (0.08–0.09). Among a group of NHS controls (n=471) with plasma hormone measurements, SNPs in HSD17b2 and HSD17b4 were not associated with estradiol or estrone levels (data not shown).
This is the first study to examine genetic variation in HSD17b2 and HSD17b4 genes in relation to endometrial cancer risk. We evaluated the association between 47 tag SNPs that efficiently covered the genes HSD17b2 and HSD174 and endometrial cancer risk, in two case-control studies nested within the NHS and WHS. After combining results from the two studies, we observed no significant associations between polymorphisms in HSD17b2 and HSD17b4 and endometrial cancer risk among Caucasians.
Previous studies have examined a small number of SNPs (n<3) in HSD17b2 in relation to mammographic density (12), prostate cancer (10), human spermatogenic defect (11) and circulating sex hormones (9, 12). None of these studies observed any significant associations. Similarly, a handful of studies have examined a select number of SNPs (n<13) in HSD17b4 in relation to testicular germ cell tumors (13), ovarian cancer (14), and androgen deprivation prostate cancer therapy (15). The study by Beesly et al. (14) observed a borderline association of non-synonymous SNP rs17145454 with the clear cell subtype of ovarian cancer but not with overall ovarian cancer suggesting the association maybe due to chance. The study by Ross et al. (15) observed a significant association of intronic SNP rs24543 with androgen deprivation prostate cancer therapy. We genotyped these two SNPs as well as newly discovered non-synonymous SNPs in our genes of interest and SNPs in known regulatory regions. We also included SNPs genotyped in the Breast and Prostate Cancer Consortium where no association was observed with breast and prostate cancer (personal communication). We did not observe an association with any of the above selected SNPs nor with the remaining tag SNPs we genotyped.
Mice knock-out studies on HSD17b2 and HSD17b4 highlight the importance of these genes for survival to birth and adulthood. Mice lacking the gene HSD17b2 exhibit placenta structural abnormalities and die at the embryo stage (22). Mice overexpressing HSD17b2 show growth retardation, ovarian dysfunction and mammary gland hyperplasia (23). Mice lacking the gene HSD17b4 do not die at the embryo stage but exhibit growth retardation and structural abnormalities in several organs such as the eye (atrophic retina), brain, and testis (infertility) (24). HSD17b4 in addition to sex steroid metabolism plays a central role in fatty acid metabolism. Humans deficient in HSD17b4, also named multifunctional protein or D-bi-functional protein (DBP), are diagnosed with a developmental syndrome known as Zellweger syndrome otherwise called DBP deficiency. Patients die within their first to third year of life, and have brain abnormalities and seizures (24). The deficiency is caused by point mutations, deletions or truncations that severely affect the structure and activity of HSD17b4 (25). We observed that all known SNPs in these genes as listed in the HapMap database were located in introns, except 3 non-synonymous SNPs in HSD17b4. Taken together all of the above observations underscore the important role these two genes play in biological function and support that these proteins are more likely to be functionally conserved (26).
We also did not observe any association between a subset of our tag SNPs and plasma hormone levels suggesting that variation in these genes does not influence levels of circulating hormones. It is possible that circulating hormone levels may not accurately reflect local hormone levels which may be influenced by variation in HSD17b2 and HSD17b4.
This is the first report to investigate genetic variation in HSD17b2 and HSD17b4 in relation to endometrial cancer risk. The main strengths of this study include the comprehensive evaluation of SNPs in HSD17b2 and HSD17b4 in two prospective case-control studies. We selected 47 tagging SNPs that provided complete coverage across the genes and regulatory regions. In addition, we investigated SNPs with possible functional significance and we replicated findings from previous studies on other health outcomes. The use of two prospective case-control studies in which to validate our findings and the homogeneous population are among other strengths of the study. However, our study was limited to Caucasians therefore the results may not be generalized to other ethnicities. We also did not examine rare variants and may not have had sufficient power to detect weak associations. The lack of significant associations does not rule out the possibility that polymorphisms in these genes may have a larger impact when they interact with other genes or lifestyle factors. Our findings suggest that common genetic variation in HSD17b2 and HSD17b4 does not substantially influence the risk of endometrial cancer in Caucasian women.
This work was supported by the National Institute of Health (grant numbers: CA87969, CA49449, CA82838, CA047988, HL043851, NICHD K12 HD051959-01); the American Cancer Society (grant number: RSG-00-061-04-CCE); and the HSPH-Cyprus Initiative for the Environment and Public Health funded by the Republic of Cyprus (to S.K).
We thank J. Prescott, K. Terry, H. Ranu, and P. Soule, for assistance, and we thank the participants in the Nurses’ Health Study and the Women’s Health Study for their dedication and commitment.
Conflict of interest statement
The authors declare that there are no conflicts of interest.
Author contributionsAll authors contributed to this study and approved the enclosed final manuscript. S.K. performed all analyses, assisted with study design, and prepared the manuscript; M.McG. assisted with study design, data analysis, and manuscript editing; and I.L., J.B, P.K, and I.DV. were involved in all stages of this project, including obtaining funding, study design, data collection, statistical support, and manuscript editing.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.