Figures
Abstract
Endometrial cancer (EC) contributes substantially to total burden of cancer morbidity and mortality in the United States. Family history is a known risk factor for EC, thus genetic factors may play a role in EC pathogenesis. Three previous genome-wide association studies (GWAS) have found only one locus associated with EC, suggesting that common variants with large effects may not contribute greatly to EC risk. Alternatively, we hypothesize that rare variants may contribute to EC risk. We conducted an exome-wide association study (EXWAS) of EC using the Infinium HumanExome BeadChip in order to identify rare variants associated with EC risk. We successfully genotyped 177,139 variants in a multiethnic population of 1,055 cases and 1,778 controls from four studies that were part of the Epidemiology of Endometrial Cancer Consortium (E2C2). No variants reached global significance in the study, suggesting that more power is needed to detect modest associations between rare genetic variants and risk of EC.
Citation: Chen MM, Crous-Bou M, Setiawan VW, Prescott J, Olson SH, Wentzensen N, et al. (2014) Exome-Wide Association Study of Endometrial Cancer in a Multiethnic Population. PLoS ONE 9(5): e97045. https://doi.org/10.1371/journal.pone.0097045
Editor: Paolo Peterlongo, IFOM, Fondazione Istituto FIRC di Oncologia Molecolare, Italy
Received: February 24, 2014; Accepted: April 15, 2014; Published: May 8, 2014
Copyright: © 2014 Chen et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The Nurses' Health Study (NHS) investigators are supported by the NCI, NIH Grants Number 1R01 CA134958, 2R01 CA082838, P01 CA087969, and R01 CA49449. MMC is supported by training grant 5T32CA009001-38 from the NCI. MCB is also supported by a Sara Borrell postdoctoral fellowship from the Spanish Ministry of Health, Carlos III Health Institute. The Multiethnic Cohort Study (MEC) is supported by the NCI, NHI Grants Number CA54281, CA128008, and 2R01 CA082838. The Fred Hutchinson Cancer Research Center (FHCRC) is supported by NCI, NIH Grant Number 2R01 CA082838, NIH RO1 CA105212, RO1 CA 87538, RO1 CA75977, RO3 CA80636, NO1 HD23166, R35 CA39779, KO5 CA92002 and funds from the Fred Hutchinson Cancer Research Center. The EDGE Study (Estrogen, Diet, Genetics, and Endometrial Cancer) is supported by NCI, NIH Grants Number 2R01 CA082838, R01CA83918 (Olson PI) and P30-CA008748 (Cancer Center Support Grant). The Alberta Health Services Study (AHS) was supported by operating grants obtained from the National Cancer Institute of Canada with funds from the Canadian Cancer Society and the Canadian Institute for Health Research. The AHS is also supported by NCI, NIH Grant Number 2R01 CA082838. Dr Christine Friedenreich is supported by career awards from Alberta Innovates-Health Solutions and the Alberta Cancer Foundation through the Weekend to End Women's Cancers Breast Cancer Chair. Dr Linda Cook was supported through the Canada Research Chairs program. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Endometrial cancer (EC), a cancer of the uterine epithelial lining that typically occurs near or after menopause, is the most common cancer of the female reproductive organs and the 10th leading cause of cancer death in women in the developed world [1]–[3]. EC is strongly associated with estrogen-only post-menopausal hormone therapy [4], [5] and excess body weight [6] due to increased aromatization of C-19 steroids by excess adipose tissue [7]. These risk factors support the “unopposed estrogen” hypothesis in which EC may develop because of the unchecked mitogenic effects of estrogen in the absence of sufficient progesterone [8]. Some studies have shown that family history increases risk two to three-fold in younger women who have a first-degree female relative with EC [9], [10], while among older women the association is less strong. In addition, there is an increased risk of EC in women with Lynch syndrome [11], a hereditary autosomal dominant condition that confers a high risk of colorectal cancer as well. These observations suggest that germline genetics may contribute to EC susceptibility.
Genome-wide association studies (GWAS) have successfully identified more than a hundred susceptibility loci for a variety of cancer types [12]. Three GWAS studies of EC have been conducted to date with only one identifying a novel genome-wide significant locus, rs4430796, (p = 7.1 × 10−10) associated with EC [13] at the HNF1B gene region on chromosome 17q12. Two independent studies subsequently replicated the association with rs4450796 [14], [15]. However, two other GWAS studies of EC [14], [16] were not able to identify additional genome-wide significant loci, suggesting that common variants with large effects may not highly contribute to the familial risk of EC.
Most risk alleles discovered through GWAS have modest effect sizes that do not account for much heritability of common diseases [17]. Moreover, GWAS studies have focused on common variants (>5%) in the general population. Low frequency variants make up a large fraction of genetic variation in humans and may explain a substantial portion of the heritability in cancer etiology. Recent exome-sequencing studies have found rare variants in candidate susceptibility genes for familial colorectal cancer [18], breast cancer [19], and prostate cancer [20], suggesting that analysis of rare variants may also provide insight into the etiology of EC. However, exome-sequencing studies require samples sizes that are not amenable to large epidemiological studies due to the high cost currently needed to achieve sufficient statistical power.
There has been a push to develop statistically powerful, yet relatively inexpensive, methods to detect associations for rare variants with larger effect sizes. Illumina has recently developed the Infinium HumanExome BeadChip (exome array) from non-synonymous variants found at least 3 times on more than 2 data sets from the whole-exome sequencing of more than 12,000 individuals. This array provides a platform from which we can begin to survey the landscape of rare variation in a large number of samples.
We genotyped rare variants in a multiethnic population of 3,067 women (1,169 EC cases and 1,898 controls) from the Epidemiology of Endometrial Cancer Consortium (E2C2) [21] in order to test the hypothesis that rare variants in coding regions may be associated with EC risk.
Methods
Ethics committee from each participating study (Alberta Health Services; Estrogen, Diet, Genetics and Endometrial Cancer Study; Multiethnic Cohort Study) obtained written informed consent from all study participants. All written consent was approved from the Institutional Review Board (IRB) from each institution (Alberta Health Services, Canada; Memorial Sloan Kettering, USA; University of Hawaii Cancer Center, USA; Keck School of Medicine-University of Southern California, USA).
Alberta Health Services, Memorial Sloan Kettering, University of Hawaii Cancer Center, and University of Southern California institutional review boards specifically approved the present study (Exome-Wide Association Study of Endometrial Cancer), as well as the written consent obtained from participants.
Participating studies also obtained IRB certification, permitting data sharing according to the NIH Policy for Sharing of Data Obtained in NIH Supported or Conducted Genome- Wide Association studies (GWAS).
Study Population
Exome array genotyping was performed on 3,067 samples from 3 retrospective case-control studies: the Alberta Health Services Study (AHS) [22], the Estrogen, Diet, Genetics and Endometrial Cancer study (EDGE) [23], and the Fred Hutchinson Cancer Research Center (FHCRC) study and 1 case-control study nested within the prospective Multiethnic Cohort Study (MEC) [24]. Studies participating in this analysis are described in Table 1 and in our previous GWAS[14]. Of the women included in the study, 1,169 were EC cases and 1,898 were controls. Cases were restricted to those diagnosed with the most common subtype of EC (type I) while controls were cancer free and had an intact uterus. Controls were matched to cases by age and study site.
Genotyping and Quality Control
DNA was extracted at each study site from buffy coat or cheek-cell samples following the manufacturer's protocol and genotyped at the University of Southern California using the Infinium Human Exome BeadChip (Illumina Inc., San Diego, CA) as part of the Stage II replication of the E2C2 GWAS. The BeadChip included 9,232 custom markers, 2,211 of which are specifically relevant to EC, in addition to the 247,870 markers coding primarily for protein-altering variants already included in the BeadChip's default design.
Genotype calling was performed with Illumina GenCall on all samples (n = 3,067) using the MEC cluster file (16,000 multiethnic samples) for the non-custom markers and autoclustering for the custom markers. Variants were excluded from analyses if call rates were < 90% (n = 115), the variant was monomorphic (n = 77,521), the loci had no observed founders and missing all genotypes (n = 1,962), the variant was an insertion or deletion allele (n = 117), or the variant deviated from Hardy-Weinberg equilibrium at p-value < 0.0001 in any ethnic group (n = 248). The final disease trait analysis data set contained 177,139 successfully genotyped variants.
In total, 3,031 out of 3,067 samples were successfully genotyped with call rates ≥ 90%. Of these, we removed 40 duplicate samples (genotype concordance rate > 99.9%) used for assay quality control and 15 samples for other quality control reasons. We conducted principal components analysis (PCA) to identify self-reported ethnicity outliers and infer ancestry with EIGENSOFT v 4.2 [25] using 47,097 custom and non-custom SNPs with genotyping rates > 90% and MAF > 1%. The HapMap phase II (build 37) CEU, YRI, and JPT-CHB samples were used as reference populations. Using the first 5 principal components, we determined 7 individuals that were ethnicity outliers and excluded them from analyses. After further removal of 136 outliers (more than 3.5 standard deviations from the mean) of sample heterozygosity by ethnicity, 2,833 women (1,055 EC cases and 1,778 controls) remained for disease trait analysis.
Statistical Analysis
Single variant association analysis.
Single variant analyses were performed overall and stratified by self-reported ethnic group. For each SNP, we estimated odds ratios (OR) and 95% confidence intervals (CI) using unconditional logistic regression, assuming an additive genetic model (0, 1, 2 copies of the minor allele) and adjusting for body mass index (BMI in kg/m2), age, study site, plate, and the first 4 principal components to account for population stratification. All single variant analyses were performed using PLINK v 1.07 [26].
Gene-based analysis.
As an additional method to discover rare variants associated with EC, gene-based testing was performed using SKAT-O [27] over all ethnicities. SKAT-O combines gene-burden tests and SKAT, a SNPset level test for association using kernel machine methods, in special cases for an optimized approach that maximizes power. These analyses were also adjusted for BMI, age, study site, plate and the first 4 principal components. In total, 16,245 genes with at least one variant were tested.
Statistical significance.
We determined single variant association to reach global significance if the unadjusted p-value was <2.82 × 10−7, corresponding to a Bonferroni correction for 177,139 tests. Gene-based associations were considered significant for unadjusted p-values <3.08 × 10−6, corresponding to a Bonferroni correction for 16,245 tests.
In accordance to NIH/NCI policy all data will be submitted to the database of Genotypes and Phenotypes (dbGaP, http://www.ncbi.nlm.nih.gov/gap).
Results
Association analyses included 177,139 successfully genotyped variants with MAF > 0 from a total of 257,102 variants included in the array. Population characteristics of the four participating studies (AHS, EDGE, FHCRC, and MEC) are described in Table 1. Mean age at diagnosis for cases ranged from 58.5 years in AHS to 65.5 years in MEC and mean BMI at diagnosis for cases ranged from 28.8 kg/m2 in MEC to 32.3 kg/m2 in AHS and EDGE. Of the 3,067 samples genotyped, 2,833 were included in the analysis. There were no differences in age, BMI, and ethnicity between excluded cases and those included in the analysis (results not shown). Of these 2,833 individuals, there were 254 self-reported African-Americans, 347 self-reported Asians, 1,686 self-reported Caucasians, 79 self-reported Hawaiians, 360 self-reported Latinas, and 107 who did not report a specific ethnicity (Table 2).
Variant Distribution among Reported Ethnicities
In this study population, 77,521 variants (30.4%) were found to be monomorphic across all reported ethnicities and 177,139 variants (69.6%) were polymorphic in at least one ethnic population with 74.0% of polymorphic alleles having MAF ≤ 1% (Figure 1). Of the variants that were polymorphic in at least one ethnic population, 42.0% in African Americans, 71.7% in Asians, 34.9% in Caucasians, 69.7% in Hawaiians, 49.5% in Latinas, and 60.0% in those of unknown ethnicity were monomorphic (Figure 2). The MAF distributions were fairly similar among Asians, Hawaiians, and those who did not report a specific ethnicity while African Americans, Caucasians, and Latinas shared more similarities in MAF with each other than with Asians, Hawaiians, and those of unknown ethnicity. About 20.2% (n = 35,912) of variants were shared by all 5 reported ethnicities while Caucasians and Latinas had the most variants in common at 41.1% (n = 72,878) (Figure 3). Caucasians had the most unique polymorphic variants (18.7%), followed by African-Americans (14.0%), Latinas (3.2%), Asians (2.7%), those who did not report ethnicity (1.0%), and Hawaiians (0.4%).
The number of variants is plotted by the minor allele frequency over all ethnicities. These variants include those that are monomorphic in all ethnicities.
The number of variants is plotted by the minor allele frequency for each ethnicity. All these variants are polymorphic in at least one reported ethnicity.
Numbers of shared variants are shown at intersections. The total numbers of polymorphic variants by ethnicity are listed in the upper-left hand corner.
Single Variant Association for Endometrial Cancer
No variants reached global significance in single variant association of EC for all ethnicities combined (Figure 4a, Table 3) when correcting for multiple comparisons using the Bonferroni adjustment (p <2.82 × 10−7). The strongest associations were for variants with >0.05 MAF (Table 3) located within 50 kb of the long non-protein coding intergenic RNA, LINC00520 (rs1953358, OR = 1.36, p = 4.76 × 10−7) and in the intron region of PROS1 (rs8178648, OR = 1.71, p = 1.53 × 10−6), which codes for protein S, a cofactor to protein C in the anti-coagulation pathway. In Caucasians, who make up the majority of the overall analysis, only rs8178648 remained suggestively associated with OR = 1.98 and p = 3.35 × 10−6 (Figure 4b, Table 3). There were no globally significant or suggestive variants in African Americans, Asians, Hawaiians, Latinas, and those who did not report ethnicity (Table S1).
Results of single variant analyses (−log10p) are plotted against chromosome position (NCBI build 37) for association over all ethnicities (A) and for associations within Caucasians (B). Suggestive variants are labeled above. Results were adjusted for age at diagnosis, BMI, study site, plate, and the first four principal components.
Gene-based Analysis of Endometrial Cancer
None of the gene-based tests of association were globally significant (p < 3.08 × 10−6) after adjusting for multiple comparisons (Table S2). Of the 16,245 genes tested, the most significant EC association was with KRT81 (p = 2.21 × 10−5), a member of the keratin gene family located on 12q13. PROS1, where rs8178648 is located, was not significantly associated with EC (p = 0.6789) when testing over all ethnicities neither when testing only in Causasians (results not shown).
Discussion
We present an initial exploration into whether rare variants are associated with EC risk in a multiethnic population from the E2C2. No variants reached global significance (p < 2.82 × 10−7) in the single variant association analyses of EC in all ethnicities combined or when stratified by reported ethnicity. Additionally, no gene-based test of association reached global significance (p < 3.08 × 10−6).
Among all ethnicities, rs8178648 on chromosome 3 maintained a suggestive association with EC (OR = 1.707, 95% CI: 1.363–2.123, p = 1.53 × 10−6). The variant lies within the intron region of PROS1, a gene coding for protein S, a cofactor in the anticoagulant pathway that causes autosomal dominant hereditary thrombophilia when mutated [28]. PROS1 expression has been reported to be elevated in aggressive prostate cancer tissue [29] and thyroid cancer tissue [30], suggesting it may have a role in cancer etiology or progression. PROS1 has been found to be directly upregulated by progestins [31] and downregulated by 17β-Estradiol, an estrogen that regulates gene expression via the estrogen receptor [32], making it susceptible to imbalances in the sex hormone metabolic pathway, which is implicated in EC etiology. However, PROS1 was not significantly associated with EC (p = 0.6789) when using SKAT-O and no other GWAS have found significant or suggestive variants in this gene.
One weakness of this study is our limited sample size, which was not sufficiently powered to detect rare variants with modest effects associated with EC. Additionally, the exome array content is predominantly based on European ancestry whereas our study included a substantial number of samples with other ancestries. Incomplete exome array coverage of all functional variants and indels that may impact EC risk may also have limited the scope of our study. However, our analysis is one of only two studies [33] using the exome array to examine associations between rare variants and complex diseases in large multiethnic populations. Our study is also the first to utilize the exome array with EC and serves as an extension to our previous examination of common variants on EC risk.
A previous GWAS [13] identified one novel locus near HNF1B, rs4430796, inversely associated with EC risk. We replicated the findings in our GWAS [14], but no other common variants associated with EC have been determined. Exome arrays that focus on rare variants, which are hypothesized to have larger effect sizes than common variants, have been used to successfully identify new loci influencing insulin processing and secretion in type 2 diabetics [34]. To date, analyses of cancer sites using exome arrays have failed to find strong evidence that rare variants are highly associated with cancer, revealing only one variant significantly associated with breast cancer and none with prostate cancer [33]. Similarly, we have not identified any loci significantly associated with EC. Due to our limited sample size, our study was estimated to be sufficiently powered to detect ORs > 2.53 for low frequency variants (MAF = 0.02). An OR of 2.00 (MAF = 0.01) would also need around 4,250 cases and 7,250 controls to be sufficiently powered. Even for variants with higher MAFs similar to what was observed for rs8178648, a study detecting a per-allele OR of 1.70 would require at least 1,107 cases and 1,871 controls to be considered sufficiently powered (β = 0.80). Therefore, larger studies need to be conducted in order to detect novel associations with rare variants.
In conclusion, our study found no evidence that rare variants with large effect sizes are associated with EC risk. Though we were able to identify a few suggestive associations, as with rs8178648, much larger studies would be needed to identify a more modest influence of rare variants on the risk of EC.
Supporting Information
Table S1.
1–5. Single variant association results. Top 100 most significant single variant associations with endometrial cancer by ethnic groups.
https://doi.org/10.1371/journal.pone.0097045.s001
(XLSX)
Table S2.
SKAT-O gene based association results. SKAT-O gene based associations with endometrial cancer for all ethnicities combined.
https://doi.org/10.1371/journal.pone.0097045.s002
(XLSX)
Author Contributions
Conceived and designed the experiments: IDV PK. Analyzed the data: MMC MCB. Wrote the paper: MMC MCB IDV. Reviewed the analysis and results and provided insightful comments on the manuscript: VWS JP SHO NW AB LB CC CC LSC JD CMF SEH PH BEH DJH LLM XL JL LL IO SP SP LP TRR HR CS FS XS X-oS NSW LX DVDB HPY HY SC CH.
References
- 1. Jemal A, Bray F (2011) Center MM, Ferlay J, Ward E, et al (2011) Global cancer statistics. CA Cancer J Clin 61: 69–90
- 2. Bray F, Ren JS, Masuyer E, Ferlay J (2013) Global estimates of cancer prevalence for 27 sites in the adult population in 2008. Int J Cancer 132: 1133–1145
- 3.
American Cancer Society. (2013) Cancer Facts & Figures 2013. American Cancer Society, Atlanta.
- 4. Persson I, Adami HO, Bergkvist L, Lindgren A, Pettersson B, et al. (1989) Risk of endometrial cancer after treatment with oestrogens alone or in conjunction with progestogens: results of a prospective study. BMJ 298: 147–151.
- 5. Grady D, Gebretsadik T, Kerlikowske K, Ernster V, Petitti D (1995) Hormone replacement therapy and endometrial cancer risk: a meta-analysis. Obstet Gynecol 85: 304–313
- 6. Bergstrom A, Pisani P, Tenet V, Wolk A, Adami HO (2001) Overweight as an avoidable cause of cancer in Europe. Int J Cancer 91: 421–430.
- 7. Schmandt RE, Iglesias DA, Co NN, Lu KH (2011) Understanding obesity and endometrial cancer risk: opportunities for prevention. Am J Obstet Gynecol 205: 518–525
- 8. Kaaks R, Lukanova A, Kurzer MS (2002) Obesity, endogenous hormones, and endometrial cancer risk: a synthetic review. Cancer Epidemiol Biomarkers Prev 11: 1531–43.
- 9. Gruber SB, Thompson WD (1996) A population-based study of endometrial cancer and familial risk in younger women. Cancer and Steroid Hormone Study Group. Cancer Epidemiol Biomarkers Prev 5: 411–7.
- 10. Lucenteforte E, Talamini R, Montella M, Dal Maso L, Pelucchi C, et al. (2009) Family history of cancer and the risk of endometrial cancer. Eur J Cancer Prev 18: 95–99
- 11. Vasen HF, Watson P, Mecklin JP, Jass JR, Green JS, et al. (1994) The epidemiology of endometrial cancer in hereditary nonpolyposis colorectal cancer. Anticancer Res 14: 1675–1678.
- 12.
Hindorff LA, MacArthur J, Morales J, Junkins HA, Hall PN, et al.. (2009) A Catalog of Published Genome-Wide Association Studies. NHGRI. http://www.genome.gov/gwastudies. Accessed 14 May 2013.
- 13. Spurdle AB, Thompson DJ, Ahmed S, Ferguson K, Healey CS, et al. (2011) Genome-wide association study identifies a common variant associated with risk of endometrial cancer. Nat Genet 43: 451–454
- 14. De Vivo I, Prescott J, Setiawan VW, Olson SH, Wentzensen N, et al. (2014) Genome-wide association study of endometrial cancer in E2C2. Hum Genet 133: 211–224
- 15. Setiawan VW, Haessler J, Schumacher F, Cote ML, Deelman E, et al. (2012) HNF1B and endometrial cancer risk: results from the PAGE study. PLoS One 7: e30390
- 16. Long J, Zheng W, Xiang YB, Lose F, Thompson D, et al. (2012) Genome-wide association study identifies a possible susceptibility locus for endometrial cancer. Cancer Epidemiol Biomarkers Prev 21: 980–987
- 17. Zuk O, Hechter E, Sunyaev SR, Lander ES (2012) The mystery of missing heritability: Genetic interactions create phantom heritability. Proc Natl Acad Sci U S A 109: 1193–1198
- 18. Gylfe AE, Katainen R, Kondelin J, Tanskanen T, Cajuso T, et al. (2013) Eleven candidate susceptibility genes for common familial colorectal cancer. PLoS Genet 9: e1003876
- 19. Thompson ER, Doyle MA, Ryland GL, Rowley SM, Choong DY, et al. (2012) Exome sequencing identifies rare deleterious mutations in DNA repair genes FANCC and BLM as potential breast cancer susceptibility alleles. PLoS Genet 8: e1002894
- 20. Fitzgerald LM, Kumar A, Boyle EA, Zhang Y, McIntosh LM, et al. (2013) Germline missense variants in the BTNL2 gene are associated with prostate cancer susceptibility. Cancer Epidemiol Biomarkers Prev 22: 1520–1528
- 21. Olson SH, Chen C, De Vivo I, Doherty JA, Hartmuller V, et al. (2009) Maximizing resources to study an uncommon cancer: E2C2—Epidemiology of Endometrial Cancer Consortium. Cancer Causes Control 20: 491–496
- 22. Friedenreich CM, Cook LS, Magliocco AM, Duggan MA, Courneya KS (2010) Case-control study of lifetime total physical activity and endometrial cancer risk. Cancer Causes Control 21: 1105–1116
- 23. Fortuny J, Sima C, Bayuga S, Wilcox H, Pulick K, et al. (2009) Risk of endometrial cancer in relation to medical conditions and medication use. Cancer Epidemiol Biomarkers Prev 18: 1448–1456
- 24. Kolonel LN, Henderson BE, Hankin JH, Nomura AM, Wilkens LR, et al. (2000) A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol 151: 346–357.
- 25. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, et al. (2006) Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet 38: 904–909
- 26. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, et al. (2007) PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81: 559–575
- 27. Lee S, Wu MC, Lin X (2012) Optimal tests for rare variant effects in sequencing association studies. Biostatistics 13: 762–775
- 28. Andersen BD, Bisgaard ML, Lind B, Philips M, Villoutreix B, et al. (2001) Characterization and structural impact of five novel PROS1 mutations in eleven protein S-deficient families. Thromb Haemost 86: 1392–1399.
- 29. Saraon P, Musrap N, Cretu D, Karagiannis GS, Batruch I, et al. (2012) Proteomic profiling of androgen-independent prostate cancer cell lines reveals a role for protein S during the development of high grade and castration-resistant prostate cancer. J Biol Chem 287: 34019–34031
- 30. Griffith OL, Melck A, Jones SJ, Wiseman SM (2006) Meta-analysis and meta-review of thyroid cancer gene expression profiling studies identifies important diagnostic biomarkers. J Clin Oncol 24: 5043–5051
- 31. Hughes Q, Watson M, Cole V, Sayer M, Baker R, et al. (2007) Upregulation of protein S by progestins. J Thromb Haemost 5: 2243–2249
- 32. Suzuki A, Sanda N, Miyawaki Y, Fujimori Y, Yamada T, et al. (2010) Down-regulation of PROS1 gene expression by 17beta-estradiol via estrogen receptor alpha (ERalpha)-Sp1 interaction recruiting receptor-interacting protein 140 and the corepressor-HDAC3 complex. J Biol Chem 285: 13444–13453
- 33. Haiman CA, Han Y, Feng Y, Xia L, Hsu C, et al. (2013) Genome-wide testing of putative functional exonic variants in relationship with breast and prostate cancer risk in a multiethnic population. PLoS Genet 9: e1003419
- 34. Huyghe JR, Jackson AU, Fogarty MP, Buchkovich ML, Stancakova A, et al. (2013) Exome array analysis identifies new loci and low-frequency variants influencing insulin processing and secretion. Nat Genet 45: 197–201