Genetic Predictors of Circulating 25-Hydroxyvitamin D and Risk of Colorectal Cancer

Colorectal cancer is the second leading cause of cancer death in men and women in the United States. It is estimated that in 2013, a total of 142,820 U.S. men and women will be diagnosed with cancer of the colon and rectum (1). Family history is a strong risk factor for colorectal cancer (2, 3), which is consistent with the existence of shared etiologic and genetic determinants among relatives. Known genetic mutations account for about 30% to 50% of the familial risk (4); much of the remaining familial aggregation is unexplained. Genome-wide association studies (GWAS) of sporadic colorectal cancer have identified at least 20 independent loci statistically significantly associated with risk (5–12). However, these variants cumulatively explain only a very small fraction of colorectal cancer risk (13).

Beyond inherited risk, there is a large body of evidence supporting the role of nongenetic factors, including vitamin D status, in the etiology of colorectal cancer. The first human evidence to suggest an association between vitamin D and colorectal cancer was the ecologic study by Garland and Garland based on data collected in the 1950′s and 60′s which showed a strong inverse association between colon cancer mortality and solar UVB radiation exposure in the United States. (14). Since then, most but not all case–control and cohort studies have found an inverse association between intake of vitamin D (both diet and supplements) and colorectal cancer risk (15–18), with even stronger associations observed using directly measured circulating 25(OH)D (19–22), an integrated biomarker of vitamin D status (23). Randomized clinical trials of vitamin D supplementation, including the Women's Health Initiative (24) and the British Oxford Trial (25), have not showed reductions in colorectal cancer incidence. However, these trials have generally tested low doses of vitamin D and each included less than 7 years of follow-up, which is likely insufficient to show an effect on cancer incidence, particularly in light of the long latency of disease. Similarly, large meta-analyses of clinical trials have not shown robust evidence for a protective role of vitamin D in the development of colorectal cancer (26) despite some indication of a preventative role in the development of adenomas (27).

Circulating 25(OH)D levels are a function of dietary sources and exposure of the skin to sunlight, specifically UVB rays. In addition to environmental determinants, twin and family studies suggest that genetic factors contribute substantially to circulating vitamin D levels, with heritability estimates ranging from 43% to 80% (28–31).

Two published GWAS of 25(OH)D have uncovered single-nucleotide polymorphisms (SNPs) significantly associated with lower circulating 25(OH)D levels in four gene regions that seem to have functional relevance: GC (group-specific component vitamin D–binding protein); CYP2R1 (cytochrome P450, family 2, subfamily R, polypeptide 1), encoding C-25 hydroxylase that converts vitamin D to the active ligand for the vitamin D receptor; DHCR7/NADSYN1 [7-dehydrocholesterol (7-DHC) reductase/nicotinamide adenine dinucleotide synthetase] (32) with roles in the synthetic vitamin D pathway (33, 34); and CYP24A1 (encoding 24-hydroxylase involved in the degradation of both 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D; refs. 33, 34). Both the GC SNP rs2282679 and the DHCR7/NADSYN1 SNPs are located in intronic regions, with the GC SNP showing the largest magnitude of association with 25(OH)D < 75 nmol/L [odds ratios (OR), 1.63; 95% confidence intervals (CI), 1.53–1.73; ref. 34]. The SNP rs10741657 is proximal to the CYP2R1 gene and rs6013897 is proximal to CYP24A1 yet the precise associations with gene expression are yet to be determined.

We investigated the association between these SNPs previously identified as associated with 25(OH)D and risk of colorectal cancer in 13 cohorts that are part of the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO) and the Colon Cancer Family Registry (CCFR). Within a subset of participants from the Nurses' Health Study (NHS), Health Professionals Follow-up Study (HPFS), and Physician's Health Study (PHS) with measured prediagnostic plasma 25(OH)D levels, we also examined the joint effects of plasma 25(OH)D and 25(OH)D-associated SNPs on colorectal cancer risk.

The analysis included a total of 10,061 cases and 12,768 controls of European ancestry drawn from 13 studies within GECCO and CCFR. Details on the studies are provided in Table 1, and are described in detail in the Supplementary Data and Supplementary Table S1. In brief, each study defined colorectal cancer cases as colorectal adenocarcinoma, confirmed by medical records, pathologic reports, or death certificates. All participants provided informed consent and studies were approved by their respective Institutional Review Boards. None of the studies included in GECCO or CCFR contributed subjects to the any previous GWAS of 25(OH)D with the exception of a subset of the NHS subjects (n = 1,342) who participated in the validation stage of one study (33).

We used genotype data from GECCO and CCFR. GECCO consisted of participants within the French Association Study Evaluating RISK for sporadic colorectal cancer (ASTERISK); Hawaii Colorectal Cancer Studies 2 and 3 (Colo2&3); Darmkrebs: Chancen der Verhütung durch Screening (DACHS); Diet, Activity, and Lifestyle Study (DALS); HPFS; Multiethnic Cohort (MEC); NHS; Ontario Familial Colorectal Cancer Registry (OFCCR); PHS; Prostate, Lung, Colorectal Cancer, and Ovarian Cancer Screening Trial (PLCO); VITamins And Lifestyle (VITAL); and the Women's Health Initiative (WHI). Phase I genotyping on a total of 1,709 colon cancer cases and 4,214 controls from PLCO, WHI, and DALS (PLCO Set 1, WHI Set 1, and DALS Set 1) was conducted using Illumina HumanHap 550 K, 610 K, or combined Illumina 300 K and 240 K, and has been described previously (12). A total of 650 colorectal cancer cases and 522 controls from OFCCR are included in GECCO from previous genotyping using Affymetrix platforms (35). A total of 5,540 colorectal cancer cases and 5,425 controls from ASTERISK, Colo2&3, DACHS, DALS Set 2, MEC, PMH, PLCO Set 2, VITAL, and WHI Set 2 were successfully genotyped using Illumina HumanCytoSNP. A total of 2,004 colorectal cancer colorectal cancer cases and 2,244 controls from HPFS (2 sets), NHS (2 sets), and PHS (2 sets) were successfully genotyped using Illumina HumanOmniExpress. The CCFR included a population-based case–control set of participants from sites in the United States, Canada, and Australia successfully genotyped using Illumina Human1M or Human1M-Duo (36).

DNA was extracted from samples of white blood cells or, in the case of a subset of NHS, HPFS, DACHS, MEC, and PLCO samples, and all VITAL samples from buccal cells using conventional methods (37). All studies included 1% to 6% blinded duplicates to monitor quality of the genotyping. All individual-level genotype data were managed centrally at University of Southern California (Los Angeles, CA; CCFR), the Ontario Institute for Cancer Research (Toronto, Ontario, Canada; OFCCR), the University of Washington (Seattle, WA; HPFS, NHS, and PHS), or the GECCO and CCFR Coordinating Center (CC) at the Fred Hutchinson Cancer Research Center (all other studies) to ensure consistent quality assurance and quality control and statistical analysis. Details on the quality assurance and quality control can be found in Supplementary Table S2. In brief, samples were excluded on the basis of call rate, heterozygosity, unexpected duplicates, gender discrepancy, and unexpectedly high identity-by-descent or unexpected concordance (>65%) with another individual. For missing SNP data, all GECCO studies were imputed to HapMap II release 24, with the exception of OFCCR, which was imputed to HapMap II release 22. CCFR was imputed using IMPUTE (38), OFCCR was imputed using BEAGLE (39), and all other studies were imputed using MACH (40). All SNPs met quality-control measures for Hardy–Weinberg equilibrium in controls (HWE; P ≥ 10⁻⁴) and minor allele frequency (MAF ≥ 1%) or imputation R² > 0.3.

Dietary information, including calcium, folate, fiber, and alcohol intake, was available for Colo2&3, DALS, HPFS, MEC, NHS, PLCO I, PLCO II, VITAL, WHI; calcium, folate, and alcohol was available in PHS; and calcium and alcohol in ASTERISK and DACHS. Regular use of nonsteroidal anti-inflammatory drugs (NSAID) was available for CCFR, Colo2&3, DACHS, HPFS, MEC, NHS, and VITAL. All studies collected data on smoking status, red meat consumption, physical activity, body mass index (BMI), and hormone replacement therapy in postmenopausal women with the exception of ASTERISK. ASTERISK was restricted to cases with no family history of colorectal cancer. We adopted a flexible approach to retrospective covariate harmonization as previously described (41, 42).

In previous studies, we measured plasma levels of 25(OH)D in a subset of the cases and controls with genetic data that were nested within the NHS, HPFS, and PHS (total cases n = 672 and total controls n = 909) using a radioimmunosorbent assay in the laboratory of Dr. Bruce W. Hollis (Medical University of South Carolina, Charleston, SC). The median intra-assay coefficient of variation from blinded quality-control samples was 11.8% in NHS, 10.1% in HPFS, and 13.8% in PHS. Cases and their controls were analyzed in the same batch and laboratory personnel were blinded to case, control, and quality-control status (21, 43, 44).

The statistical analyses of the GECCO and CCFR samples were conducted at a central data analysis center on individual-level data to ensure a consistent analytic approach. For each study, we estimated the association between each SNP and risk for colorectal cancer by calculating βs, ORs, standard errors (SEs), 95% CIs, and P values using log-additive genetic models relating the genotype dose (0, 1, or 2 copies of the allele) to risk of colorectal cancer. For imputed SNPs, we used the dosage (expected number of copies of the minor allele) when testing associations, which has been shown to give unbiased estimates (45). We also created a genetic risk score (GRS), composed of four SNPs from four distinct gene regions to ensure no single gene was overrepresented in the score using an allelic scoring system based on summing the number of risk alleles [previously associated with lower 25(OH)D], yielding a possible range of 0 to 8 alleles to derive estimates of allelic OR.

Minimally adjusted models included covariates for age, sex (when appropriate), center (when appropriate), batch effects (ASTERISK only), and the first three principal components from EIGENSTRAT to account for population substructure. Multivariate models were additionally adjusted for family history of colorectal cancer, BMI, NSAID use, smoking status, alcohol use, dietary calcium, folate and red meat intake, sedentary status, and hormone replacement therapy based on covariate availability. We repeated the minimally adjusted model analyses stratified by anatomic site (colon and rectum).

We conducted inverse-variance weighted, fixed-effects meta-analysis to combine β estimates and SEs from log-additive models across individual studies. We chose to focus on fixed-effects to improve power and assessed heterogeneity across studies utilizing random effects models (46).

For analyses of the joint effect of plasma 25(OH)D and our GRS composed of 25(OH)D-associated SNPs, we included the 672 cases and 909 controls in NHS, HPFS, and PHS among whom we had previously measured prediagnostic levels of 25(OH)D and also had genotype data (21, 44). We calculated ORs and 95% CI for colorectal cancer comparing extreme quartiles of 25(OH)D defined according to cohort-specific cutoff points determined by the distribution in controls (44). We compared the GRS-associated risk for colorectal cancer across categories of high versus low vitamin D levels and quartiles of vitamin D, as well as tested for multiplicative interactions between GRS and a 1 ng/mL increase in 25(OH)D and high/low vitamin D using a product term in the model and assessing its significance by the Wald method.

We used PLINK, R (47, 48), and SAS 9.2 (SAS Institute, Inc.) to conduct the statistical analyses. We estimated our power to detect an association between a GRS and colorectal cancer using the method of Tosteson and colleagues. (49). These calculations account for the strength of association between the vitamin D SNPs and circulating 25(OH)D.

Our study included 10,061 colorectal cancer cases and 12,768 controls. Overall 53% were female and the mean age at colorectal cancer diagnosis was 64 years (±9.6 SD), 55.0% past/current smokers and with risk allele frequencies ranging from 16% to 61%. Table 1 summarizes the characteristics of the studies included in the analyses. Analyses of each individual SNP in models first minimally and then fully adjusted, did not show a statistically significant association with colorectal cancer risk (Table 2). In analyses stratified by anatomic site, each of the five SNP associations remained nonsignificant (Table 3), in tests of associations with cancers of the colon and rectum.

We considered the possibility that a combination of SNPs associated with circulating 25(OH)D in prior GWAS may be associated with risk of colorectal cancer. However, an analysis of a GRS composed of the risk alleles from the four SNPs associated with plasma 25(OH)D and colorectal cancer risk did not show any significant association (Table 4). The Fig. 1 forest plot depicts the ORs and 95% CIs of the GRS and colorectal cancer association of the individual GECCO and CCFR studies showing a balanced distribution of study-specific ORs around the null value of 1.0.

Among the subset of NHS, HPFS, and PHS subjects with measured prediagnostic plasma 25(OH)D levels (N = 672 cases and N = 909 controls), we observed that a one-allele change in our GRS was associated with a mean 1.5 ng/mL decrease in 25(OH)D (P < 0.0001). Compared with the lowest quartile of 25(OH)D, the highest quartile of 25(OH)D was associated with an OR of 0.66 (95% CI, 0.47–0.92) for colorectal cancer. We did not observe a differential association of GRS with risk of colorectal cancer according to strata of quartile levels of plasma 25(OH)D or according to high or low levels of 25(OH)D (P_{heterogeneity} > 0.05).

Consistent evidence from epidemiologic studies supports an inverse association between circulating 25(OH)D, the best integrated biomarker of vitamin D status, and risk of colorectal cancer. However, contrary to expectation, we did not observe a statistically significant association between SNPs associated with circulating 25(OH)D and colorectal cancer, marginally or in an additive GRS.

A number of epidemiologic studies have reported inverse associations between 25(OH)D and colorectal cancer. A meta-analysis of five nested case–control studies reported a pooled OR of 0.49 (95% CI, 0.35–0.68) for colorectal cancer comparing the highest quintile (median 37 ng/mL) of 25(OH)D with the lowest (6 ng/mL) (19). Another recent systematic review of nine studies observed a pooled relative risk (RR) for colorectal cancer of 0.67 (95% CI, 0.54–0.80) comparing extreme quintiles of 25(OH)D (22). Overall, the estimated OR of colorectal cancer for a 10 ng/mL increase in circulating 25(OH)D was 0.74 (95% CI, 0.63–0.89) with the relationship appearing approximately linear (22). Several mechanisms could explain an anticancer benefit for vitamin D: reduction of cell proliferation; inhibition of angiogenesis; promotion of cell differentiation; and stimulation of apoptosis (23, 50–57). Vitamin D also has an anti-inflammatory effect, reducing PTGS-2 (COX-2) expression and decreasing levels of the inflammatory marker C-reactive protein (58–60).

Prior clinical trials testing the association between vitamin D and cancer have been null. In a 5-year British placebo-controlled trial with cancer assessed as a secondary outcome, 100,000 IU of vitamin D₃ every 4 months over 5 years was not associated with colorectal cancer incidence (RR = 1.02; 95% CI, 0.60–1.74; 25). Similarly, among 36,000 women in the WHI calcium–vitamin D trial, a combination of calcium (1,000 mg/day) plus low-dose vitamin D₃ (400 IU/day) for a mean of 7 years did not reduce colorectal cancer incidence (RR = 1.08; 95% CI, 0.86–1.34; 24). However, the interpretation of these null results is tempered by several important limitations. First, the relatively low doses of vitamin D used were probably inadequate to yield a substantial contrast between the treatment and placebo groups. Second, the duration of follow-up was probably too short to observe an influence on incidence of cancer. Observational data suggest that any influence of calcium and vitamin D intake on colorectal cancer risk could require at least 10 years to emerge, consistent with our understanding of the prolonged dwell time of the adenoma–carcinoma pathway (18). On the other hand, a Nebraska population-based placebo-controlled trial of calcium alone or calcium plus vitamin D₃ (1,100 IU/day) observed a significantly lower cancer incidence among those supplemented with calcium and vitamin D over just 4 years of treatment (61). However, follow-up of total cancers was a secondary outcome and there were only a small number of colorectal cancer cases, limiting the interpretation of these results.

More than 90% of circulating 25(OH)D is protein-bound with the GC encoded vitamin D–binding protein being the major carrier of 25(OH)D. Less than 1% of vitamin D circulates in its unbound form (62). Vitamin D–binding protein is a multifunctional protein that also binds fatty acids and may have immune functions independent of its role as a carrier of vitamin D (63). Prior studies have observed that unbound 25(OH)D was more strongly related to bone mineral density (64) and parathyroid hormone levels among hemodialized patients (65), than total 25(OH)D, thereby implicating a role for vitamin D–binding protein in modifying the biologic activity of circulating vitamin D. The available estimates of the association between 25(OH)D and colorectal cancer, as well as genetic markers of 25(OH)D, are based solely on total circulating 25(OH)D levels (19, 22, 33, 34). It is unclear how these estimates might change when accounting for vitamin D–binding protein levels or by individually examining free and protein-bound 25(OH)D.

Prior studies have examined individual SNPs in CYP24A1 or GC (66, 67) in association with risk of colorectal cancer. A prior DALS multicenter population-based case–control study of 1,600 colorectal cancer cases found a statistically significant association between one CYP24A1 polymorphism and overall risk of colon cancer, particularly for proximal colon cancer, as well as an association between three CYP24A1 polymorphisms and distal colon cancer (67). However, the correlation between these CYP24A1 genetic markers investigated by the DALS study and the CYP24A1 marker examined in this full combined analysis study is very low (r² < 0.1). Because our a priori hypothesis was that SNPs most strongly associated with 25(OH)D levels would be associated with colorectal cancer, we did not consider total genetic variation in CYP24A1 with colorectal cancer risk. It is possible that alternative CYP24A1 SNPs may be associated with colorectal cancer through mechanisms independent of 25(OH) D levels. A prior study of the CCFR cohort of 1,750 sibships found no evidence for associations between GC and the risk of colorectal cancer and no evidence for modification of the association by calcium and/or vitamin D intake (66). Studies of additional genes that are not significantly associated with circulating 25(OH)D but are implicated in the vitamin D pathway, including the vitamin D receptor (VDR) SNPs Fok1 and BsmI, have yielded inconsistent results (68–71).

The lack of association that we observed between genetic markers associated with circulating 25(OH)D and colorectal cancer is consistent with prior clinical trials of vitamin D and colorectal cancer and would, at least initially, seem to argue against a causal association between vitamin D and colorectal cancer (72). However, prior work has showed that these 4 SNPs, although correlated with circulating 25(OH)D, explain only a small fraction (5%) of the variance in circulating 25(OH) (73). Recently, a Scottish case–control study observed a significant association between direct plasma measurements of 25(OH)D and colorectal cancer risk, yet failed to replicate the association using an instrumental-variable method of Mendelian randomization with the same four genome-wide significant risk loci examined in our analysis (74). The investigators attributed these inconsistent results to a presumed weak correlation between these SNPs and 25(OH)D, as well as a limited sample size of 2,001 cases of colorectal cancer and 2,237 controls. Given our significantly larger sample size of approximately 10,000 cases and 12,500 controls and assuming a correlation between our GRS and 25(OH)D of r = 0.17, with a 10 ng/mL increase in 25(OH)D associated with OR = 0.74 for colorectal cancer (22), our power to detect a one-allele change in our GRS is 96% (significance level of 0.05). However, if the true magnitude of association with a 10 ng/mL increase in 25(OH)D is in fact an OR of 0.85 for colorectal cancer, we would have only had 56% power to detect a one-allele change in our GRS.

We are not certain of the precise pathway or biologic mediators by which 25(OH)D influences colorectal cancer risk. Our GRS assumes that each included SNP would be associated with increased colorectal cancer risk according to their observed association with lower 25(OH)D. If this assumption is invalid, combining the alleles into this GRS would reduce our power to detect associations with colorectal cancer. If we remove the GC SNP and repeat our power calculation, our observed correlation between our GRS and 25(OH)D becomes r = 0.11, resulting in 68% power to detect a one-allele change in our proxy score.

We acknowledge some limitations. First, our study includes only populations of European descent, which limits the generalizability of our findings. However, the circulating 25(OH)D SNPs that we examined were identified in GWAS of populations of European descent and so the underlying genetic associations should hold in our study population. Moreover, limiting our analysis to European descent populations minimizes the potential for confounding by population structure. Second, if these SNPs are correlated with another locus that influences the risk of colorectal cancer, this could confound our results (72). Third, despite our large sample size, we had limited power to detect associations between individual SNPs and risk of colorectal cancer.

In conclusion, our findings do not support an association between SNPs associated with circulating 25(OH)D and risk of colorectal cancer. This may be due to the fact that these SNPs account for only a small portion of the variance observed in circulating 25(OH)D levels and that those alleles associated with low-circulating levels of 25(OH)D may not affect colorectal cancer risk in the same direction. Future studies are needed to examine the role of unbound and protein-bound 25(OH)D, along with other biomarkers of the vitamin D pathway, in the development of colorectal cancer.

No potential conflicts of interest were disclosed.

Conception and design: L.T. Hiraki, C.M. Hutter, B.J. Caan, C.S. Fuchs, E. Giovannucci, J.L. Hopper, D. Seminara, P. Kraft, U. Peters, A.T. Chan

Development of methodology: S.J. Chanock, D. Duggan, C.S. Fuchs, A. Hazra, J.L. Hopper, P. Kraft, U. Peters, A.T. Chan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.M. Hutter, S.I. Berndt, S. Bézieau, H. Brenner, B.J. Caan, G. Casey, J. Chang-Claude, S.J. Chanock, D.V. Conti, D. Duggan, C.S. Fuchs, S. Gallinger, R. Hayes, M. Hoffmeister, T.J. Hudson, M.A. Jenkins, S. Küry, L. Le Marchand, J. Ma, J.E. Manson, P.A. Newcomb, K. Ng, J.D. Potter, R.E. Schoen, M.L. Slattery, J. Wactawski-Wende, E. White, B.W. Zanke, U. Peters, A.T. Chan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L.T. Hiraki, C. Qu, C.M. Hutter, D.V. Conti, C.S. Fuchs, E. Giovannucci, A. Hazra, J.L. Hopper, M Lemire, J.E. Manson, H. Nan, P.A. Newcomb, K. Ng, F. Schumacher, K. Wu, P. Kraft, U. Peters, A.T. Chan

Writing, review, and/or revision of the manuscript: L.T. Hiraki, C. Qu, C.M. Hutter, J.A. Baron, S.I. Berndt, S. Bézieau, H. Brenner, G. Casey, J. Chang-Claude, C.S. Fuchs, E. Giovannucci, R. Hayes, A. Hazra, B. Henderson, M. Hoffmeister, J.L. Hopper, T.J. Hudson, M.A. Jenkins, S. Küry, L. Le Marchand, J. Ma, J.E. Manson, P.A. Newcomb, K. Ng, J.D. Potter, R.E. Schoen, F. Schumacher, D. Seminara, M.L. Slattery, J. Wactawski-Wende, E. White, K. Wu, P. Kraft, U. Peters, A.T. Chan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.T. Hiraki, C. Qu, T.A. Harrison, M. Hoffmeister, S. Küry, L. Le Marchand, J.E. Manson, F. Schumacher, P. Kraft, U. Peters, A.T. Chan

Study supervision: C.S. Fuchs, M. Hoffmeister, J. Wactawski-Wende, P. Kraft, U. Peters, A.T. Chan

S. Bézieau and S. Kury are affiliated with ASTERISK which was funded by a regional Programme Hospitalier de Recherche Clinique (PHRC) and supported by the Regional Council of Pays de la Loire, the Groupement des Entreprises Françaises dans la Lutte contre le Cancer (GEFLUC), the Association Anne de Bretagne Génétique, and the Ligue Régionale Contre le Cancer (LRCC).

J.A. Baron, G. Casey, D.V. Conti, J.L. Hopper, M.A. Jenkins, P.A. Newcomb, and F.R. Schumacher are affiliated with CCFR which is supported by the National Cancer Institute, NIH under RFA # CA-95-011 and through cooperative agreements with members of the CCFR and principal investigators (P.I.s). This genome wide scan was supported by the National Cancer Institute, NIH by U01 CA122839. The content of this article does not necessarily reflect the views or policies of the National Cancer Institute or any of the collaborating centers in the CFRs, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government or the CFR. The following CCFR centers contributed data to this article and were supported by the following sources: Australasian Colorectal Cancer Family Registry (U01 CA097735), Seattle Colorectal Cancer Family Registry (U01 CA074794), and Ontario Registry for Studies of Familial Colorectal Cancer (U01 CA074783).

S. Gallinger, T.J. Hudson, M. Lemire, and B.W. Zanke are affiliated with OFCCR which is supported by the NIH, through funding allocated to the Ontario Registry for Studies of Familial Colorectal Cancer (U01 CA074783); see CCFR section above. As subset of ARCTIC, OFCCR is supported by a GL2 grant from the Ontario Research Fund, the Canadian Institutes of Health Research, and the Cancer Risk Evaluation (CaRE) Program grant from the Canadian Cancer Society Research Institute. T.J. Hudson and B.W. Zanke are recipients of Senior Investigator Awards from the Ontario Institute for Cancer Research, through generous support from the Ontario Ministry of Research and Innovation.

B. Henderson and L. Le Marchand are affiliated with COLO2&3/MEC. COLO2&3 is supported by the National Cancer Institute (R01 CA60987). MEC is supported by R37 CA54281, P01 CA033619, and R01 CA63464.

H. Brenner, J. Chang-Claude, and M. Hoffmeister are affiliated with DACHS which was supported by grants from the German Research Council (Deutsche Forschungsgemeinschaft, BR 1704/6-1, BR 1704/6-3, BR 1704/6-4, and CH 117/1-1), and the German Federal Ministry of Education and Research (01KH0404 and 01ER0814).

B.J. Caan, J.D. Potter, and M.L. Slattery are affiliated with DALS which was supported by the National Cancer Institute, NIH, U.S. Department of Health and Human Services (R01 CA48998 to M.L. Slattery).

T.A. Harrison, C.M. Hutter, U. Peters, and C. Qu are affiliated with GECCO which is supported by National Cancer Institute, NIH, U.S. Department of Health and Human Services (U01 CA137088). Funding for the genome-wide scan of DALS, PLCO, and WHI was provided by the National Cancer Institute, NIH, U.S. Department of Health and Human Services (R01 CA059045).

A.T. Chan, C.S. Fuchs, E.L. Giovannucci, L.T. Hiraki, A. Hazra, P. Kraft, H. Nan, K. Ng, and K. Wu are affiliated with HPFS, NHS, and PHS. HPFS was supported by the NIH (P01 CA 055075, UM1 CA167552, R01 137178, and P50 CA 127003), NHS by the NIH (R01 137178, P50 CA 127003, and P01 CA 087969) and PHS by the NIH (CA42182). A.T. Chan is a Damon Runyon Clinical Investigator and also supported by K24 DK098311. L. T. Hiraki was supported by a Canadian Institute of Health Research (CIHR) Fellowship Award.

S.I. Berndt, S.J. Chanock, R.B. Hayes, and R.E. Schoen are affiliated with PLCO which was supported by the Intramural Research Program of the Division of Cancer Epidemiology and Genetics and supported by contracts from the Division of Cancer Prevention, National Cancer Institute, NIH, DHHS. Control samples were genotyped as part of the Cancer Genetic Markers of Susceptibility (CGEMS) prostate cancer scan, supported by the Intramural Research Program of the National Cancer Institute. The datasets used in this analysis were accessed with appropriate approval through the dbGaP online resource (http://www.cgems.cancer.gov/data_acess.html) through dbGaP accession number 000207v.1p1.c1. [National Cancer Institute (2009) CGEMS data website. http://cgems.cancer.gov/data_access.html; ref. 75]. Control samples were also genotyped as part of the GWAS of Lung Cancer and Smoking [76]. Funding for this work was provided through the NIH, Genes, Environment and Health Initiative [NIH GEI] (Z01 CP 010200). The human subjects participating in the GWAS are derived from the Prostate, Lung, Colon and Ovarian Screening Trial and the study is supported by intramural resources of the National Cancer Institute. Assistance with genotype cleaning, as well as with general study coordination, was provided by the Gene Environment Association Studies, GENEVA Coordinating Center (U01 HG004446). Assistance with data cleaning was provided by the National Center for Biotechnology Information. Funding support for genotyping, which was conducted at the Center for Inherited Disease Research, Johns Hopkins University (Baltimore, MD) was provided by the NIH GEI (U01 HG 004438). The datasets used for the analyses described in this article were obtained from dbGaP at http://www.ncbi.nlm.nih.gov/gap through dbGaP accession number phs000093.

E. White is affiliated with VITAL which is supported in part by the NIH (K05 CA154337) from the National Cancer Institute and Office of Dietary Supplements.

J. Manson and J. Wactawski-Wende are affiliated with WHI. The WHI program is funded by the National Heart, Lung, and Blood Institute, NIH, U.S. Department of Health and Human Services through contracts HHSN268201100046C, HHSN268201100001C, HHSN268201100002C, HHSN268201100003C, HHSN268201100004C, and HHSN271201100004C.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.