Genome-wide association studies have found type 2 diabetes-associated variants in the HNF1B gene to exhibit reciprocal associations with prostate cancer risk. We aimed to identify whether these variants may have an effect on cancer risk in general versus a specific effect on prostate cancer only.
In a collaborative analysis, we collected data from GWAS of cancer phenotypes for the frequently reported variants of HNF1B, rs4430796 and rs7501939, which are in linkage disequilibrium (r2 = 0.76, HapMap CEU). Overall, the analysis included 16 datasets on rs4430796 with 19,640 cancer cases and 21,929 controls; and 21 datasets on rs7501939 with 26,923 cases and 49,085 controls. Malignancies other than prostate cancer included colorectal, breast, lung and pancreatic cancers, and melanoma. Meta-analysis showed large between-dataset heterogeneity that was driven by different effects in prostate cancer and other cancers. The per-T2D-risk-allele odds ratios (95% confidence intervals) for rs4430796 were 0.79 (0.76, 0.83)] per G allele for prostate cancer (p<10−15 for both); and 1.03 (0.99, 1.07) for all other cancers. Similarly for rs7501939 the per-T2D-risk-allele odds ratios (95% confidence intervals) were 0.80 (0.77, 0.83) per T allele for prostate cancer (p<10−15 for both); and 1.00 (0.97, 1.04) for all other cancers. No malignancy other than prostate cancer had a nominally statistically significant association.
The examined HNF1B variants have a highly specific effect on prostate cancer risk with no apparent association with any of the other studied cancer types.
Citation: Elliott KS, Zeggini E, McCarthy MI, Gudmundsson J, Sulem P, et al. (2010) Evaluation of Association of HNF1B Variants with Diverse Cancers: Collaborative Analysis of Data from 19 Genome-Wide Association Studies. PLoS ONE 5(5): e10858. doi:10.1371/journal.pone.0010858
Editor: Florian Kronenberg, Innsbruck Medical University, Austria
Received: April 8, 2010; Accepted: April 28, 2010; Published: May 28, 2010
Copyright: © 2010 Elliott 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: JPAI is supported by funding for the Tufts Clinical and Translational Science Institute and the Center for Genetic Epidemiology and Modeling is supported by grant UL1 RR025752 from the National Institutes of Health. KSE is supported by the Wellcome Trust (WT075491/Z/04). EZ is supported by the Wellcome Trust (WT088885/Z/09/Z). NH and SM were funded by the Australian National Health and Medical Research Council (NHMRC) and the National Institutes of Health (grant R01-CA88363). HG is supported by Swedish Cancer Society (Cancerfonden), Swedish Research Council, and Linneus grant “Predication and prevention of breast cancer and prostate cancer”. RE and KM are supported by Cancer Research UK Grant C5047/A8385; C5047/A7357, The Institute of Cancer Research and The Everyman Campaign and The Prostate Cancer Research Foundation. The authors acknowledge support from the NIHR to the Biomedical Research Centre at The Institute of Cancer Research and The Royal Marsden NHS Foundation Trust. DEN, JLD and FCH would like to acknowledge the support of the National Cancer Research Institute (NCRI) formed by the Department of Health, the Medical Research Council and Cancer Research UK. The Prostate Testing for Cancer and Treatment (ProtecT) study is funded by the UK National Institute for Health Research Health Technology Assessment Programme (projects 96/20/06, 96/20/99). The NCRI provided funding through ProMPT (Prostate Mechanisms of Progression and Treatment) and this support is gratefully acknowledged. The Cambridge prostate biorepository also received funding from the NIHR Comprehensive Biomedical Research Centre Grant. This support is gratefully acknowledged. The views and opinions expressed therein are those of the authors and are not necessarily those of the funding bodies. deCODE acknowledges grant support from the 6th Framework Program of the EU: contract 018827 (Polygene). The Australian Melanoma Family Study (AMFS) was supported by grants from the NHMRC, Cancer Councils of Queensland, NSW and Victoria, and by National Institutes of Health grant RO1-CA83115 to the GenoMEL consortium. The Framingham Heart Study (FHS) The Framingham Heart Study of the National Heart Lung and Blood Institute of the National Institutes of Health and Boston University School of Medicine were supported by the National Heart, Lung and Blood Institute's Framingham Heart Study Contract No. N01-HC-25195 and its contract with Affymetrix, Inc for genotyping services (Contract No. N02-HL-6-4278). Analyses reflect intellectual input and resource development from the Framingham Heart Study investigators participating in the SNP Health Association Resource (SHARe) project. A portion of this research was conducted using the Linux Cluster for Genetic Analysis (LinGA-II) funded by the Robert Dawson Evans Endowment of the Department of Medicine at Boston University School of Medicine and Boston Medical Center.
Competing interests: The authors have declared that no competing interests exist.
See acknowledgments for author list.
A large number of epidemiological studies have suggested correlations between type 2 diabetes (T2D) and various cancers, , . Most evidence suggests an inverse correlation between T2D and prostate cancer, ,  although not all studies agree on this. Several studies also suggest positive correlations between other cancers and T2D, , . It is unclear whether these correlations, if true, represent casual relationships and whether they may also reflect some shared genetic background. Recently, with the advent of genome-wide association studies (GWAS), a large number of genetic variants have been identified that confer susceptibility to T2D or specific types of cancer. An interesting observation has been that specific variants in the HNF1B gene (formerly TCF2) have been demonstrated to be associated both with the risk of prostate cancer, ,  and the risk of T2D,  with the effects being in the opposite direction for these two phenotypes.
HNF1B was previously known to be mutated in individuals with maturity-onset diabetes of the young type 5 (MODY 5), but a biological explanation of the impact of the identified common variation on T2D and prostate cancer risk remains elusive. The identified genetic effects are small in magnitude even for prostate cancer and T2D, representing odds ratios [ORs] per allele in the range of 1.2 ,  and 0.9 , , respectively. Therefore, small effects for other cancer types would not be readily detectable, unless very large studies were performed or data were combined from several studies.
A definitive answer on whether HNF1B variants modulate also the risk of other malignancies, or show specificity for prostate cancer, requires large sample sizes. Here we present the results of a large collaborative meta-analysis of HNF1B, rs4430796 and rs7501939, which have the most consistent associations with both prostate cancer and T2D. Relevant data were collected on the two variants from GWAS on cancer phenotypes in Caucasian populations in order to examine whether they have an effect on cancer risk in general, on few specific cancers, or only on prostate cancer.
Database of contributed information
All the originally contacted investigators of cancer-related GWA studies agreed to participate in this collaborative analysis, with the exception of the investigators of 3 GWA studies , ,  (1 on breast cancer, 1 on colorectal cancer and 1 on neuroblastoma), 1 of which had no data on the requested variants, as they had used a Affymetrix platform. Investigators who agreed to participate in the collaborative analysis contributed data on 13 datasets for rs4430796 and 19 datasets for rs7501939 , , , , , , , , , , , , , , , , , . For 5 datasets, data were available only for the latter polymorphism either because the polymorphism was not available on the platform used or the SNP failed quality control criteria.
The contributing teams and datasets are shown in Table 1 with data on the number of cases and controls for each polymorphism and for each type of cancer. Datasets from the Framingham cohort contained imputed data for both polymorphisms since an Affymetrix platform had been used, rs4430796 data from the M.D. Anderson Cancer Center was imputed since this SNP had not been directly genotyped, and melanoma data from AMFS and Q-MEGA contained counts from pooling experiments, otherwise all other datasets had direct genotyping on individual participants. Detailed demographic and other characteristics of the study populations can be found in the respective primary publications of these GWA studies , , , , , , , , , , , , , , , , , , .
Table 1. Characteristics of datasets included in the collaborative meta-analysis.doi:10.1371/journal.pone.0010858.t001
Overall, the collaborative analysis included data on rs4430796 for 19,640 cancer cases and 21,929 controls; for prostate cancer there were 11,145 cases and 9,650 controls, while for all other cancers there were 8,495 cases and 12,279 controls. The collected data on rs7501939 included 26,923 cases and 49,085 controls; for prostate cancer there were 12,898 cases and 40,371 controls, while for the other cancers there were 14,025 cases and 43,893 controls. Malignancies other than prostate cancer in these datasets included colorectal, breast, lung and pancreatic cancers, and melanoma (Table 1). deCODE contributed data on 4 different cancers and had a common population control group for all 4 of them. The Framingham Heart Study (FHS) contributed data on 4 different cancers and had a common population control group (subjects ≥65 years at the last contact who are not nuclear family member of the cancer cases) for all 4 studies with the exception of prostate and breast cancer which used male and female only controls respectively. The common control groups for deCODE and FHS are only counted once in the total sample sizes above.
The meta-analysis of all datasets (Table 2, Figure 1) showed a per T2D risk allele association with both rs4430796 (G allele OR 0.91 [95% CI: 0.88, 0.94] p = 3×10−10) and rs7501939 (T allele OR 0.91 [95% CI: 0.88, 0.94] p = 5v10−10) according to fixed effects calculations, while by random effects calculations there was nominal significance (OR 0.94 [95% CI: 0.88, 1.00], p = 0.033 for rs4430796 and 0.93 [95% CI: 0.86, 1.01], p = 0.07 for rs7501939). The reason for this diversity is that there was very large between-study heterogeneity in the effect sizes (I2 of 82% [95% CI: 73-89%] and 80% [95% CI: 70-86%], respectively, for the two polymorphisms; Q-test p-value <0.001 for both polymorphisms), and this makes the fixed effects calculations less reliable. Results were qualitatively similar when we increased the variance in deCODE, FHS, and IARC estimates to account for the overlapping control group (not shown).
Figure 1. Association of rs4430796 and rs7501939 with diverse cancer types.
Panel A shows results for rs4430796 and panel B shows results for rs7501939. Each study is shown by its odds ratio and 95% confidence intervals). Prostate cancer studies appear on the top and other cancer studies follow in alphabetical order. For the abbreviations of the names of the studies see Table 1. The summary diamond at the bottom corresponds to the fixed effects summary. Weight indicates the relative proportion of the total evidence found in each study (the weight is inversely proportional to the variance).doi:10.1371/journal.pone.0010858.g001
Table 2. Summary of results for association between rs4430796 and rs7501939 and diverse cancer types.doi:10.1371/journal.pone.0010858.t002
The heterogeneity was largely driven by the diversity in the effect sizes between prostate cancer and all other cancers. A meta-analysis limited to prostate cancer datasets gave consistent associations with both rs4430796 (OR per copy of T2D risk allele (A) 0.79 [95% CI: 0.76, 0.83], p<10−15 by fixed effects and 0.79 [95% CI: 0.74, 0.84] p = 10−13 by random effects), and rs7501939 (OR per copy of T2D risk allele (T) 0.80 [95% CI: 0.77, 0.83] p<10−15 by fixed effects and 0.79 [95% CI: 0.74, 0.85], p = 2×10−11 by random effects) (Table 2). There was some residual between-study heterogeneity even within the prostate cancer datasets (I2 of 42% [95%CI: 0–79%] and 56% [95% CI: 0–82%], respectively, for the two polymorphisms; Q-test p-value 0.037 and 0.14, respectively), although the heterogeneity pertained only to the exact magnitude of the genetic effects and a nominally statistically significant association was seen in each of the datasets except for the Framingham study where the number of prostate cancer cases was more limited.
Conversely, the results for all other cancers suggested no significant association and results were consistent across studies. The summary OR was 1.03 and 1.00 for the two polymorphisms respectively (p = 0.14 and 0.81 by fixed effects) and the 95% CIs excluded ORs deviating more than 7% from the null (OR = 1.00) for rs4430796 and more than 4% from the null for rs7501939 (Table 2). The Q-test p-value was 0.99 and 0.45 for the two polymorphisms respectively and random effects estimates were thus identical to fixed effects estimates.
There was also no convincing evidence for an association between either of the two polymorphisms and any of the other cancers (besides prostate cancer), when each cancer type was evaluated separately. Point estimates were in the opposite direction (odds ratio 1.03–1.05) for pancreatic and lung cancer, but were not nominally statistically significant (Table 2). The difference between the prostate cancer and other cancers' effect estimates was beyond chance (p<0.05) for both polymorphisms.
The current collaborative analysis documents that both rs4430796 and rs7501939 have robust support for association with prostate cancer, while we did not observe any convincing evidence for an association of any of the other cancers examined with either polymorphism. When data from all other cancers, excluding prostate cancer, were combined the summary effects had 95% CIs that excluded even subtle associations. Apart from prostate cancer, when other datasets for each individual cancer type was combined, the 95% CIs consistently excluded associations with modest effects. This would suggest that the effects mediated by these polymorphisms are specific to T2D and prostate cancer and they do not involve any other cancer types.
The HNF1B gene encodes a transcription factor and it was initially identified as a MODY gene. Subsequent studies have suggested that mutations in this gene may also be related to renal disease and chromophobe renal cell carcinoma. No GWAS evaluating kidney cancer were included in our analysis, and no kidney cancer GWAS has been published to-date. The expression profile of the gene shows a tissue-specific pattern. It is essential for embryonic survival and is expressed in the gut, kidney, liver, lung, pancreas, prostate, thymus and genital tract , . It could be speculated that the lack of association with some cancers studied here may be due to the low or absent expression of this gene in those tissues (for example breast cancer). We did not have data on liver cancer, thymoma or genital tract cancer, but data on lung, pancreatic, and colorectal cancer showed no association, with point estimates very near to the null.
The two variants that we assessed are not necessarily the functional culprits. GWA studies typically derive markers of phenotypes that are probably linked with the functional genetic variation. However, identifying the functional variants is difficult. Even if they could be identified, it is unlikely that substantially large genetic effects for other cancers would exist, if the tagging variants have so consistently null effects. Another caveat is that we only examined populations of Caucasian descent. This reduces the heterogeneity that could be due to different LD patterns in populations of different ancestry. However, it would be worthwhile to investigate the associations of the HNF1B variants for T2D, prostate cancer, and other cancers, also in non-Caucasian populations. Preliminary data suggest that both of the examined variants had consistent associations with T2D in Caucasian, Asian (Hong Kong), and West African ancestry participants, while the association of rs4430796 with prostate cancer risk was found to be even stronger in the Japanese than in Caucasian populations. Moreover, it would be useful to dissect associations with specific disease subsets. Even within the analyzed Caucasian-descent populations, we observed some modest between-study heterogeneity in the strength of the association between the HNF1B variants and prostate cancer. This may be due to different associations in different sub-phenotypes. For example, some data suggest that the rs4430796 A allele may primarily increase the risk for early-onset (<50 years) prostate cancer rather than later-onset disease.
In conclusion, while the two examined HNF1B variants conclusively have pleiotropic effects on both T2D and prostate cancer, the pleiotropy apparently does not extend to other cancer types. Genetic effects may offer a way to dissect comorbidity between specific cancers and metabolic phenotypes. Besides HNF1B, other gene loci have started appearing where variants are identified that modulate susceptibility to both T2D and some malignancy, e.g. prostate cancer for the JAZF1 locus gene ,  and melanoma for the CDKN2A locus , although different, unlinked variants are implicated in the susceptibility to the malignancy and T2D, respectively. The elucidation of correlated pleiotropic effects on diverse phenotypes will require very large studies, given the generally subtle effects involved. Collaborative efforts between multiple teams, as in the current study, may offer a suitable approach to answer such questions.
Eligible GWA investigations and data
We used the NHGRI catalogue of published GWA studies, a comprehensive database of GWA investigations to identify GWA studies on cancer phenotypes published as of May 20, 2008. We also performed additional PubMed searches to identify whether any additional GWA studies on cancer phenotypes had been published until then. We focused on solid cancers, excluding hematologic malignancies. Given that these GWAS did not include any studies on pancreatic cancer (of special interest, given the association with T2D), we also identified GWAS on pancreatic cancer that had not been published by that time, so as to ensure their inclusion.
We communicated with the corresponding and principal investigators of all of these studies to request their participation in the collaborative meta-analysis. The investigators of these studies were asked to contribute relevant data on genotype frequencies in cancer cases and non-cancer controls for the HNF1B variants, rs4430796 and rs7501939. The risk alleles for prostate cancer are A and C for rs4430796 and rs7501939 respectively. The risk alleles for T2D are G and T for rs4430796 and rs7501939 respectively. The two SNPs have modestly high LD in Caucasians, but low LD in Africans (r2 = 0.77 and 0.22 in CEU and YRI, respectively). Investigators were requested to provide all GWA data that they had obtained for evaluation of any cancer phenotype, including any additional unpublished datasets. Additional genotyping for the two specific variants was encouraged, when a GWA platform had been used that did not directly genotype these polymorphisms (e.g. Affymetrix or Perlegen rather than Illumina). When a study had data on more than one cancer type, data were requested to be provided separately for each cancer type. Investigators were asked to provide also information and clarifications about the design of their studies, and to ensure that population stratification and cryptic relatedness had been appropriately addressed and appropriate quality controls were available for the genotyping. All GWAS investigations that contributed data on these SNPs used stringent QC standards (as described in detail in their original publications) and the two SNPs fulfilled these standards. Approval from local institutional review boards and steering committees was obtained, as deemed necessary for each study by its investigators. The contributed data were checked for completeness and with logical queries and any missing or unclear information was clarified through communication with the contributing investigators.
For each SNP, we performed meta-analyses including the data from all eligible cancer studies (regardless of the specific cancer phenotype addressed) and also subgroup meta-analyses, with each subgroup limited to studies addressing a specific cancer phenotype. A separate analysis compared the results of the association for prostate cancer versus the association for all other cancers combined.
All analyses followed the per allele (log-additive) model of inheritance with effect sizes expressed in the odds ratio (OR) scale using both fixed and random effects models. Heterogeneity testing used the Q statistic (considered statistically significant at p<0.10), and the I2 metric and its 95% CIs . Analyses excluding data from studies with pooled genotyping gave similar results (not shown).
Based on the accumulated total sample size and given the minor allele frequencies of these two polymorphisms in HapMap CEU (47% for rs4430796 A allele and 47% for rs7501939 T allele), the meta-analysis had 95% or higher power to detect an association of OR = 1.10 at alpha = 0.05 with each of the two polymorphisms for overall cancer risk, prostate cancer risk, or other cancer risk. Reported p-values are two-tailed. Analyses were performed in STATA 10.0 (College Station, Texas).
Australian Melanoma Family Study investigators: Graham J Mann1, John L Hopper2, Joanne F Aitken3, Richard F Kefford1, Graham G Giles4, Bruce K Armstrong5.
1Westmead Institute of Cancer Research, University of Sydney at Westmead Millennium Institute and Melanoma Institute Australia, Westmead, New South Wales, Australia. 2Centre for Molecular, Environmental, Genetic and Analytic Epidemiology, School of Population Health, University of Melbourne, Melbourne, Victoria, Australia. 3Viertel Centre for Research in Cancer Control, The Cancer Council Queensland, Brisbane, Queensland, Australia. 4Cancer Epidemiology Centre, The Cancer Council Victoria, Carlton, Victoria, Australia. 5School of Public Health, University of Sydney, New South Wales, Australia.
PanScan Consortium: Gloria M. Petersen*1, Laufey Amundadottir*2,3, Charles S. Fuchs*4,5, Peter Kraft*6,7, Rachael Z. Stolzenberg-Solomon*3, Kevin B. Jacobs 3,8,9, Alan A. Arslan10,11,12, H. Bas Bueno-de-Mesquita13, Steven Gallinger14, Myron Gross15, Kathy Helzlsouer16, Elizabeth A. Holly17, Eric J. Jacobs18, Alison P. Klein19,20, Andrea LaCroix21, Donghui Li22, Margaret T. Mandelson21,23, Sara H. Olson24, Harvey A. Risch25, Wei Zheng26, Demetrius Albanes3, William R. Bamlet1, Christine D. Berg 27, Marie-Christine Boutron-Ruault28, Julie E. Buring29,30, Paige M. Bracci17, Federico Canzian31, Sandra Clipp32, Michelle Cotterchio33, Mariza de Andrade1, Eric J. Duell34, J. Michael Gaziano35,36, Edward L. Giovannucci5,6,37, Michael Goggins38, Göran Hallmans39, Susan E. Hankinson5,6, Manal Hassan22, Barbara Howard40, David J. Hunter5,6, Amy Hutchinson3,8, Mazda Jenab34, Rudolf Kaaks31, Charles Kooperberg21, Vittorio Krogh41, Robert C. Kurtz42, Shannon M. Lynch3, Robert R. McWilliams1, Julie B. Mendelsohn3, Dominique S. Michaud6,43, Hemang Parikh2,3, Alpa V. Patel18, Petra H.M. Peeters43,44, Aleksandar Rajkovic45, Elio Riboli43, Laudina Rodriguez46, Daniela Seminara47, Xiao-Ou Shu26, Gilles Thomas3,48, Anne Tjønneland49, Geoffrey S. Tobias3, Dimitrios Trichopoulos6,50, Stephen K. Van Den Eeden51, Jarmo Virtamo52, Jean Wactawski-Wende53, Zhaoming Wang3,8, Brian M. Wolpin4,5, Herbert Yu25, Kai Yu3, Anne Zeleniuch-Jacquotte11,12, Joseph F. Fraumeni, Jr.3, Robert N. Hoover*3, Patricia Hartge*3, Stephen J. Chanock*2,3.
1Department of Health Sciences Research, College of Medicine, Mayo Clinic, Rochester, MN, USA. 2Laboratory of Translational Genomics, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, USA. 3Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department Health and Human Services, Bethesda, MD, USA. 4Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, USA. 5Channing Laboratory, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA. 6Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA. 7Department of Biostatistics, Harvard School of Public Health, Boston, MA, USA. 8Core Genotyping Facility, SAIC-Frederick Inc., NCI-Frederick, Frederick, MD, USA. 9Bioinformed Consulting Services, Gaithersburg, MD, USA. 10Department of Obstetrics and Gynecology, New York University School of Medicine, New York, NY, USA. 11Department of Environmental Medicine, New York University School of Medicine, New York, NY, USA. 12New York University Cancer Institute, New York, NY, USA. 13National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands and Department of Gastroenterology and Hepatology, University Medical Centre Utrecht, Utrecht, The Netherlands. 14Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada. 15Department of Laboratory Medicine/Pathology, School of Medicine, University of Minnesota, Minneapolis, MN, USA. 16Prevention and Research Center, Mercy Medical Center, Baltimore, MD, USA. 17Department of Epidemiology & Biostatistics, University of California San Francisco, San Francisco, CA, USA. 18Department of Epidemiology, American Cancer Society, Atlanta, GA, USA. 19Department of Oncology, the Johns Hopkins University School of Medicine, Baltimore, MD, USA. 20Department of Epidemiology, the Bloomberg School of Public Health, The Sol Goldman Pancreatic Research Center, The Johns Hopkins Medical Institutions, Baltimore, MD, USA. 21Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA, USA. 22Department of Gastrointestinal Medical Oncology, University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA. 23Group Health Center for Health Studies, Seattle, WA, USA. 24Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, NY, USA. 25Yale University School of Public Health, New Haven, CT, USA. 26Department of Medicine and Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, TN, USA. 27Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA. 28Inserm, Paris-Sud University, Institut Gustave-Roussy, Villejuif, France. 29Divisions of Preventive Medicine and Aging, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA. 30Department of Ambulatory Care and Prevention, Harvard Medical School, Boston, MA, USA. 31German Cancer Research Center (DKFZ), Heidelberg, Germany. 32Johns Hopkins Bloomberg School of Public Health, George W. Comstock Center for Public Health Research and Prevention, Hagerstown, MD, USA. 33Cancer Care Ontario and Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada. 34Catalan Institute of Oncology (ICO), Barcelona, Spain. 35Physicians' Health Study, Divisions of Aging, Cardiovascular Disease, and Preventive Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA. 36Massachusetts Veterans Epidemiology Research and Information Center, Veterans Affairs Boston Healthcare System, Boston, MA, USA. 37Department of Nutrition, Harvard School of Public Health, Boston, MA, USA. 38Departments of Oncology, Pathology and Medicine, The Johns Hopkins University School of Medicine, Baltimore, MD, USA. 39Department of Public Health and Clinical Medicine, Nutritional Research, Umeå University, Umeå, Sweden. 40MedStar Research Institute, Georgetown University, Hyattsville, MD 41Nutritional Epidemiology Unit, Fondazione IRCCS Istituto Nazionale dei Tumoridi Milano, Milan, Italy. 42Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, NY, USA. 43Divisionof Epidemiology, Public Health and Primary Care, Imperial College London, London, UK. 44Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, The Netherlands. 45Departments of Obstetrics and Gynecology and Reproductive Sciences, University of Pittsburgh, Pittsburgh, PA, USA 46Public Health and Participation Directorate, Health and Health Care Services Council, Asturias, Spain. 47Division of Cancer Control and Population Sciences, National Cancer Institute, Bethesda, MD, USA. 48Synergie-Lyon-Cancer, INSERM U590, Centre Leon Berard, 69373 Lyon, Cedex 08, France 49Institute of Cancer Epidemiology, Danish Cancer Society, Copenhagen, Denmark. 50Bureau of Epidemiologic Research, Academy of Athens, Greece 51Division of Research, Kaiser Permanente, Northern California Region, Oakland, CA, USA. 52Department of Chronic Disease Prevention, National Institute for Health and Welfare, Helsinki, Finland. 53Department of Social and Preventive Medicine, University at Buffalo, SUNY Buffalo, NY, USA.
Conceived and designed the experiments: KSE EZ MIM JPAI. Analyzed the data: KSE JPAI. Contributed reagents/materials/analysis tools: KSE JG PS SNS ST LTA HG JX VG RAE DEN JLD FCH KM SJH MRS BZ LCC KMB AMFSI NKH SM IPT ML CIA JMM WBI DFE PB PC RB DFG TR DJH SJC KS. Wrote the paper: KSE EZ MIM JG JPAI.
- 1. Inoue M, Iwasaki M, Otani T, Sasazuki S, Noda M, et al. (2006) Diabetes mellitus and the risk of cancer: results from a large-scale population-based cohort study in Japan. Arch Intern Med 166: 1871–1877.
- 2. Stattin P, Bjor O, Ferrari P, Lukanova A, Lenner P, et al. (2007) Prospective study of hyperglycemia and cancer risk. Diabetes Care 30: 561–567.
- 3. Rousseau MC, Parent ME, Pollak MN, Siemiatycki J (2006) Diabetes mellitus and cancer risk in a population-based case-control study among men from Montreal, Canada. Int J Cancer 118: 2105–2109.
- 4. Kasper JS, Giovannucci E (2006) A meta-analysis of diabetes mellitus and the risk of prostate cancer. Cancer Epidemiol Biomarkers Prev 15: 2056–2062.
- 5. Gong Z, Neuhouser ML, Goodman PJ, Albanes D, Chi C, et al. (2006) Obesity, diabetes, and risk of prostate cancer: results from the prostate cancer prevention trial. Cancer Epidemiol Biomarkers Prev 15: 1977–1983.
- 6. Calton BA, Chang SC, Wright ME, Kipnis V, Lawson K, et al. (2007) History of diabetes mellitus and subsequent prostate cancer risk in the NIH-AARP Diet and Health Study. Cancer Causes Control 18: 493–503.
- 7. Will JC, Vinicor F, Calle EE (1999) Is diabetes mellitus associated with prostate cancer incidence and survival? Epidemiology 10: 313–318.
- 8. Manolio TA, Brooks LD, Collins FS (2008) A HapMap harvest of insights into the genetics of common disease. J Clin Invest 118: 1590–1605.
- 9. Gudmundsson J, Sulem P, Steinthorsdottir V, Bergthorsson JT, Thorleifsson G, et al. (2007) Two variants on chromosome 17 confer prostate cancer risk, and the one in TCF2 protects against type 2 diabetes. Nat Genet 39: 977–983.
- 10. Sun J, Zheng SL, Wiklund F, Isaacs SD, Purcell LD, et al. (2008) Evidence for two independent prostate cancer risk-associated loci in the HNF1B gene at 17q12. Nat Genet 40: 1153–1155.
- 11. Thomas G, Jacobs KB, Yeager M, Kraft P, Wacholder S, et al. (2008) Multiple loci identified in a genome-wide association study of prostate cancer. Nat Genet.
- 12. Winckler W, Weedon MN, Graham RR, McCarroll SA, Purcell S, et al. (2007) Evaluation of common variants in the six known maturity-onset diabetes of the young (MODY) genes for association with type 2 diabetes. Diabetes 56: 685–693.
- 13. Horikawa Y, Iwasaki N, Hara M, Furuta H, Hinokio Y, et al. (1997) Mutation in hepatocyte nuclear factor-1 beta gene (TCF2) associated with MODY. Nat Genet 17: 384–385.
- 14. Tenesa A, Farrington SM, Prendergast JG, Porteous ME, Walker M, et al. (2008) Genome-wide association scan identifies a colorectal cancer susceptibility locus on 11q23 and replicates risk loci at 8q24 and 18q21. Nat Genet 40: 631–637.
- 15. Gold B, Kirchhoff T, Stefanov S, Lautenberger J, Viale A, et al. (2008) Genome-wide association study provides evidence for a breast cancer risk locus at 6q22.33. Proc Natl Acad Sci U S A 105: 4340–4345.
- 16. Maris JM, Mosse YP, Bradfield JP, Hou C, Monni S, et al. (2008) Chromosome 6p22 locus associated with clinically aggressive neuroblastoma. N Engl J Med 358: 2585–2593.
- 17. Cust AE, Schmid H, Maskiell JA, Jetann J, Ferguson M, et al. (2009) Population-based, case-control-family design to investigate genetic and environmental influences on melanoma risk: Australian Melanoma Family Study. Am J Epidemiol 170: 1541–1554.
- 18. Eeles RA, Kote-Jarai Z, Giles GG, Olama AA, Guy M, et al. (2008) Multiple newly identified loci associated with prostate cancer susceptibility. Nat Genet 40: 316–321.
- 19. Petersen GM, Amundadottir L, Fuchs CS, Kraft P, Stolzenberg-Solomon RZ, et al. (2010) A genome-wide association study identifies pancreatic cancer susceptibility loci on chromosomes 13q22.1, 1q32.1 and 5p15.33. Nat Genet 42: 224–228.
- 20. Brennan P, McKay J, Moore L, Zaridze D, Mukeria A, et al. (2007) Uncommon CHEK2 mis-sense variant and reduced risk of tobacco-related cancers: case control study. Hum Mol Genet 16: 1794–1801.
- 21. Hung RJ, McKay JD, Gaborieau V, Boffetta P, Hashibe M, et al. (2008) A susceptibility locus for lung cancer maps to nicotinic acetylcholine receptor subunit genes on 15q25. Nature 452: 633–637.
- 22. Amos CI, Wu X, Broderick P, Gorlov IP, Gu J, et al. (2008) Genome-wide association scan of tag SNPs identifies a susceptibility locus for lung cancer at 15q25.1. Nat Genet 40: 616–622.
- 23. Zanke BW, Greenwood CM, Rangrej J, Kustra R, Tenesa A, et al. (2007) Genome-wide association scan identifies a colorectal cancer susceptibility locus on chromosome 8q24. Nat Genet 39: 989–994.
- 24. Brown KM, Macgregor S, Montgomery GW, Craig DW, Zhao ZZ, et al. (2008) Common sequence variants on 20q11.22 confer melanoma susceptibility. Nat Genet 40: 838–840.
- 25. Yeager M, Orr N, Hayes RB, Jacobs KB, Kraft P, et al. (2007) Genome-wide association study of prostate cancer identifies a second risk locus at 8q24. Nat Genet 39: 645–649.
- 26. Duggan D, Zheng SL, Knowlton M, Benitez D, Dimitrov L, et al. (2007) Two genome-wide association studies of aggressive prostate cancer implicate putative prostate tumor suppressor gene DAB2IP. J Natl Cancer Inst 99: 1836–1844.
- 27. Tomlinson IP, Webb E, Carvajal-Carmona L, Broderick P, Howarth K, et al. (2008) A genome-wide association study identifies colorectal cancer susceptibility loci on chromosomes 10p14 and 8q23.3. Nat Genet 40: 623–630.
- 28. Stacey SN, Manolescu A, Sulem P, Thorlacius S, Gudjonsson SA, et al. (2008) Common variants on chromosome 5p12 confer susceptibility to estrogen receptor-positive breast cancer. Nat Genet 40: 703–706.
- 29. Rafnar T, Sulem P, Stacey SN, Geller F, Gudmundsson J, et al. (2009) Sequence variants at the TERT-CLPTM1L locus associate with many cancer types. Nat Genet 41: 221–227.
- 30. Thorgeirsson TE, Geller F, Sulem P, Rafnar T, Wiste A, et al. (2008) A variant associated with nicotine dependence, lung cancer and peripheral arterial disease. Nature 452: 638–642.
- 31. Gudmundsson J, Sulem P, Manolescu A, Amundadottir LT, Gudbjartsson D, et al. (2007) Genome-wide association study identifies a second prostate cancer susceptibility variant at 8q24. Nat Genet 39: 631–637.
- 32. Gudmundsson J, Sulem P, Rafnar T, Bergthorsson JT, Manolescu A, et al. (2008) Common sequence variants on 2p15 and Xp11.22 confer susceptibility to prostate cancer. Nat Genet.
- 33. Easton DF, Pooley KA, Dunning AM, Pharoah PD, Thompson D, et al. (2007) Genome-wide association study identifies novel breast cancer susceptibility loci. Nature 447: 1087–1093.
- 34. http://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000007.v9.p4.
- 35. Edghill EL, Bingham C, Ellard S, Hattersley AT (2006) Mutations in hepatocyte nuclear factor-1beta and their related phenotypes. J Med Genet 43: 84–90.
- 36. Rebouissou S, Vasiliu V, Thomas C, Bellanne-Chantelot C, Bui H, et al. (2005) Germline hepatocyte nuclear factor 1alpha and 1beta mutations in renal cell carcinomas. Hum Mol Genet 14: 603–614.
- 37. Reber M, Cereghini S (2001) Variant hepatocyte nuclear factor 1 expression in the mouse genital tract. Mech Dev 100: 75–78.
- 38. Coffinier C, Barra J, Babinet C, Yaniv M (1999) Expression of the vHNF1/HNF1beta homeoprotein gene during mouse organogenesis. Mech Dev 89: 211–213.
- 39. Ioannidis JP, Thomas G, Daly MJ (2009) Validating, augmenting and refining genome-wide association signals. Nat Rev Genet 10: 318–329.
- 40. Yamada H, Penney KL, Takahashi H, Katoh T, Yamano Y, et al. (2009) Replication of prostate cancer risk loci in a Japanese case-control association study. J Natl Cancer Inst 101: 1330–1336.
- 41. Levin AM, Machiela MJ, Zuhlke KA, Ray AM, Cooney KA, et al. (2008) Chromosome 17q12 variants contribute to risk of early-onset prostate cancer. Cancer Res 68: 6492–6495.
- 42. Zeggini E, Scott LJ, Saxena R, Voight BF, Marchini JL, et al. (2008) Meta-analysis of genome-wide association data and large-scale replication identifies additional susceptibility loci for type 2 diabetes. Nat Genet 40: 638–645.
- 43. Zeggini E, Weedon MN, Lindgren CM, Frayling TM, Elliott KS, et al. (2007) Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science 316: 1336–1341.
- 44. Hindorff LA, Sethupathy P, Junkins HA, Ramos EM, Mehta JP, et al. (2009) Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc Natl Acad Sci U S A 106: 9362–9367.
- 45. Kavvoura FK, Ioannidis JP (2008) Methods for meta-analysis in genetic association studies: a review of their potential and pitfalls. Hum Genet 123: 1–14.
- 46. Higgins JP, Thompson SG, Deeks JJ, Altman DG (2003) Measuring inconsistency in meta-analyses. Bmj 327: 557–560.
- 47. Ioannidis JP, Patsopoulos NA, Evangelou E (2007) Uncertainty in heterogeneity estimates in meta-analyses. Bmj 335: 914–916.
- 48. Petersen GM, Amundadottir L, Fuchs CS, Kraft P, Stolzenberg-Solomon RZ, et al. A genome-wide association study identifies pancreatic cancer susceptibility loci on chromosomes 13q22.1, 1q32.1 and 5p15.33. Nat Genet.