|Home | About | Journals | Submit | Contact Us | Français|
The etiology of prostate cancer remains elusive, although steroid hormones probably play a role. Considering the carcinogenic potential of estrogen metabolites as well as altered intraprostatic estrogen biosynthesis during the development of prostate cancer, we investigated associations between repeat polymorphisms of three key estrogen-related genes (CYP11A1, CYP19A1, UGT1A1) and risk of prostate cancer in the Prostate Cancer Prevention Trial (PCPT), designed to test finasteride versus placebo as a chemoprevention agent. Using data and specimens from 1154 cases and 1351 controls who were frequency matched on age, family history of prostate cancer and PCPT treatment arm, we used logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (95% CIs) separately in the placebo and finasteride arms. Among men in the placebo arm, CYP19A1 7/8 genotype carriers had a significantly higher risk of prostate cancer compared with those with the 7/7 genotype (OR = 1.70, 95% CI = 1.16–2.5), regardless of Gleason grade. This genotype was also associated with elevated serum estrogen levels. For the (TA)n repeat polymorphism in UGT1A1, the heterozygous short (<7 repeats)/long (≥7 repeats) genotype was significantly associated with the risk of low-grade prostate cancer (OR = 1.34, 95% CI = 1.05–1.70) compared with the short/short genotype. No significant association was found with CYP11A1. These associations were not observed among men in the finasteride arm. The results indicate that repeat polymorphisms in genes involved in estrogen biosynthesis and metabolism may influence risk of prostate cancer but that their effects may be modified by factors altering hormone metabolism, such as finasteride treatment.
Prostate cancer is the most common cancer in men, other than non-melanoma skin cancer, and is the second leading cause of cancer death in USA men (1). Prostate cancer is expected to contribute to 28% of newly diagnosed cancers and 11% of cancer deaths in 2011 (1). Although the etiology of prostate cancer has not been fully elucidated, established risk factors such as family history of prostate cancer indicate that genetic factors probably play a role in the development of the disease (2). Given the fact that the prostate is an androgen-dependent organ and that prostate cancer almost always responds initially to androgen-deprivation therapy (3,4), genetic variants in androgen-related genes have been the primary focus in the search for genetic risk factors for prostate cancer (5). However, genes involved in estrogen biosynthesis and metabolism could also be potential candidates.
It has long been hypothesized that estrogen may participate in prostate carcinogenesis due to the mutagenic effects of estrogen metabolites (6,7). Animal studies support this hypothesis, evidenced by an increased risk of prostatic dysplasia and carcinoma with estrogen treatment in a Noble rat model (8). The role of estrogen was further documented in aromatase-deficient mice, in which no prostate cancer developed, although blockage of the conversion of testosterone to estrogen led to enlargement of the prostate gland (9). However, results from pooled analyses in human observational studies do not support significant associations between circulating estrogen levels and prostate cancer risk (10–12), and no associations were found in our recent evaluation of serum estrogens in relation to risk in the Prostate Cancer Prevention Trial (PCPT) (13). However, local metabolism of testosterone to estrogens in the prostate may be more relevant than circulating levels. In fact, using biopsy tissues from human prostate, Ellem et al. (14) found that prostate cancer cells had high levels of transcription and expression of aromatase, in contrast to the absence of the enzyme in normal prostate epithelial cells, indicating enhanced local estrogen biosynthesis during the prostatic carcinogenesis process. Our hypothesis is that polymorphisms in estrogen-related genes alter estrogen metabolism, in particular local estrogen metabolism, which may in turn modify prostate cancer risk. Therefore, it is expected that polymorphisms in estrogen-related genes may be associated with an increase of prostate cancer risk, however, such an association may not be modified by polymorphism-associated changes in circulating estrogen levels.
In the current study, we focused on repeat polymorphisms of CYP11A1, CYP19A1 and UGT1A1, three critical genes playing central roles in estrogen biosynthesis and metabolism. As shown in Figure 1, the cholesterol P450 side-chain cleavage enzyme CYP11A1 catalyzes the conversion of cholesterol to pregnenolone, which is the first and rate-limiting step of biosynthesis for both testosterone and estrogen (15). The production of estrogen requires aromatase CYP19A1, which converts testosterone to estrogen (16). Aromatase deficiency inhibits estrogen biosynthesis and increases androgen levels in animal models (9). UGT1A1 catalyzes the glucuronidation of estrogen metabolites, some of which are potential carcinogens (7,17). Glucuronidation inactivates these metabolites and facilitates their excretion, representing one of the main detoxification pathways of estrogens (18). Thus, in a case–control study nested in the PCPT, we investigated repeat polymorphisms of these genes in relation to prostate cancer risk as well as to levels of circulating estrogens.
The study was conducted within a subset of men who participated in the PCPT, a randomized phase III trial testing whether finasteride, a 5α-reductase inhibitor, could reduce the 7 years period prevalence of prostate cancer. The PCPT design and participant characteristics have been described in detail previously (19). Briefly, a total of 18882 men aged ≥55 years with a normal digital rectal examination, a prostate-specific antigen level of ≤3 ng/ml and no history of prostate cancer, severe benign prostate hyperplasia symptoms or other clinically significant coexisting conditions were randomized to receive finasteride (5 mg/day) or placebo. Participants underwent annual screening, and for-cause biopsies were recommended for participants with abnormal digital rectal examination and/or an increase in prostate-specific antigen. After 7 years on study, all men who were not previously diagnosed with prostate cancer were offered an end-of-study biopsy to determine presence or absence of cancer. The Gleason scoring system was used to classify tumors as low grade (Gleason score < 7) or high grade (Gleason score ≥ 7). The diagnoses were confirmed by at least two different pathologists, and discordant results were resolved by including an additional referee pathologist. Institutional review boards at all participating institutions approved the study protocols and all participants provided informed consent.
The primary case definition in the PCPT is biopsy-proven presence of prostate cancer. From an initial pool of 2401 potential cases, men were excluded if: the detection of prostate cancer was after the trial was unblinded (n = 173); detection was outside the established timeframe of the 7 years end-of-study biopsy (n = 91) or an adequate baseline serum sample was not available (n = 328). This resulted in a total of 1809 cases of prostate cancer with available baseline serum included in this nested case–control study. Controls were defined as men with negative end-of-study biopsies, and they were frequency matched to cases on age in 5 years increments, treatment arm (finasteride versus placebo) and family history of prostate cancer (first degree relatives). Controls were oversampled on race to include all non-white subjects, so as to maximize power for subgroup analyses.
Data on age, race/ethnicity, education, physical activity, alcohol consumption and history of smoking were collected at recruitment using self-administered questionnaires. Height and weight were measured to calculate body mass index (BMI; kg/m2). Non-fasting blood samples were drawn at the baseline visit (~3 months prior to randomization) and kept at room temperature for 30–60 min before centrifugation. Serum was removed and frozen immediately after centrifugation. Serum samples were generally shipped by overnight express to Esoterix (Calabasas, CA) within 24 h of collection, but some were stored for up to 7 days at −20°C before shipment. Upon receipt, the sample was thawed to remove an aliquot for analysis of prostate-specific antigen. An additional tube of blood was collected with anticoagulant for isolation of white blood cells. The whole blood samples were shipped overnight chilled and processed at NCI Frederick. Plasma and white blood cells were aliquoted and stored at −70°C. A total of 1154 cases and 1351 controls with adequate DNA for genotyping were included in the final analysis. Participants who were excluded due to lack of adequate DNA were comparable with participants with adequate DNA in terms of demographic characteristics such as age, BMI, race, family history and treatment (results not shown).
DNA was extracted from white blood cells using Qiagen M48 robot (Valencia, CA) at NCI Frederick and then shipped to the Roswell Park Cancer Institute Genomics Core Facility for genotyping by polymerase chain reaction (PCR) amplification followed by Capillary Electrophoresis fragment analysis. The primers for CYP11A1 were forward: CTCTGAGTCAGCTGTACTG and reverse: GAGCTATCTTGCCAGCTTG; for CYP19A1 were forward: GTCTATGAATGTGCCTTTTT and reverse: GTTTGACTCCGTGAGTTTGA and for UGT1A1 were forward: CGTGACACAGTCAAACATTAACTT and reverse: CAGCAGTGGCTGCCATCCAC. All primers were acquired from Applied Biosystems (Foster City, CA). Each 25 μl of PCR mixture included 2.5 U Qiagen HotStar Taq Polymerase, 3 mM MgCl2, 0.5 mM deoxynucleoside triphosphates, 100 nM primers and 10 ng template DNA. All PCR amplifications were performed in 96-well plates using a Bio-Rad DNA Engine thermal cycler (Hercules, CA). Thermal cycling conditions were as follows: 94°C × 15 min followed by 44 cycles of 94°C × 20 s, 56°C × 30 s and 72°C × 60 s with a final incubation at 72°C for 3 min. The PCR products were subjected to capillary electrophoresis on an Applied Biosystems 3130xl Genetic Analyzer. The number of microsatellite repeats was determined using GeneMapper v3.7 software (Applied Biosystems) and the local Southern method sizing algorithm. MapMarker 1000 labeled with X-Rhodamine was used as the size standard (BioVentures, Murfreesboro, TN). For quality control purposes, 5% blind duplicate samples were distributed through the plates. The call rates for the three assays were >95% and the concordance rate between duplicate pairs was >95%.
Serum samples were shipped to the laboratory of Dr Frank Stanczyk at the University of Southern California (Los Angeles, CA) for measurement of sex hormone concentrations, as described previously (20). Briefly, levels of serum testosterone, estrone and estradiol were determined by radioimmunoassay after organic solvent extraction and chromatography on Celite columns (Celite Corporation, Santa Barbara, CA). Sex hormone-binding globulin was measured by chemiluminescent immunoassay on the Immulite analyzer (Siemens Medical Solutions Diagnostics, Malvern, PA).
Standard chi-square tests were used to compare genotype frequencies between case and control groups. Because of the skewed distributions of sex hormone levels, geometric means were computed and compared among genotype groups in controls only using generalized linear models adjusted for age (continuous), BMI (continuous), race (white or non-white), first-degree family history of prostate cancer (yes or no) and sex hormone-binding globulin (continuous). To accommodate the frequency-matched study design, unconditional logistic regression models were used to estimate odds ratios (ORs) and 95% confidence intervals (95% CIs) for the association between gene polymorphisms and prostate cancer risk (21). Polychotomous logistic regression was used to test for associations with low-grade and high-grade prostate cancer compared with controls. In addition to matching variables, we also included other covariates in the models, including age (continuous), race (white or non-white), BMI (continuous) and first-degree family history of prostate cancer (yes or no). All analyses were performed separately by treatment arm. To test whether the associations of genotypes with prostate cancer risk were mediated through circulating sex hormones, we controlled for serum levels of sex hormones in the models in addition to other covariates. To test for potential effect modification by circulating testosterone and estrogens, which are substrates for CYP19A1 and UGT1A1, respectively, levels were dichotomized into low and high categories based on the distributions of baseline concentrations among controls, and separate logistic regression models were then fit after stratification. Interaction was tested by including a multiplicative term between genotypes and sex hormone levels in the models. All analyses were two sided and statistical significance was defined as P < 0.05; analyses were performed using SAS 9.2 (SAS Institute, Cary, NC).
Allele distributions of CYP11A1, CYP19A1 and UGT1A1 repeat polymorphisms among controls are presented in Figure 2. Each gene had a unique distribution pattern, which is consistent with the findings from other studies (22–24). For CYP11A1 (TAAAA)n polymorphism, six alleles containing, respectively, 4, 6, 7, 8, 9 and 10 repeats were observed, with 4 and 6 being most common. For CYP19A1 (TTTA)n polymorphism, seven alleles with the length of repeat ranging from 7 to 13 followed a bimodal distribution, with a peak at 7 repeats and another at 11 repeats. For UGT1A1 (TA)n polymorphism, four alleles with, respectively, 5, 6, 7 and 8 repeats were observed, with 6 and 7 repeats being most common. Allele frequencies were compared between cases and controls, and significant differences in distribution were found for CYP11A1 and for CYP19A1 (Supplementary Table 1 is available at Carcinogenesis Online), with the 9-repeat allele of CYP11A1 and 8-repeat allele of CYP19A1 more common in cases than in controls (8.3 versus 5.8% and 11.5 versus 8.8%, respectively, χ2 P < 0.01 for both).
Taking advantage of our relatively large sample size, genotypes with allele frequency ≥1% were examined individually in relation to prostate cancer risk. As shown in Table I, among men randomized to the placebo arm, there were no associations between CYP11A1 repeat polymorphism and prostate cancer risk. For CYP19A1, compared with its common genotype (7/7), 7/8 genotype was significantly associated with an increased risk of prostate cancer (OR = 1.70, 95% CI = 1.16–2.50). The association was observed for both low-grade (OR = 1.57, 95% CI = 1.03–2.38) and high-grade (OR = 1.98, 95% CI = 1.07–3.65) prostate cancer. Although we were unable to test the gene-dosage effect due to rare number of participants with the 8/8 genotype in the study, at least one copy of the 8-repeat allele was associated with prostate cancer risk when compared with the 7/7 genotype (OR = 1.46, 95% CI = 1.06–2.01). When the UGT1A1 repeat polymorphism was dichotomized into short (≤6) and long (>6) repeats based on allele distribution, the short/long genotype was associated with an increased risk of prostate cancer, but the association was limited to low-grade prostate cancer (OR = 1.34, 95% CI = 1.05–1.70). No significant associations were observed for the homozygous long/long genotypes with either low- or high-grade prostate cancer.
Associations between genotypes of CYP11A1, CYP19A1, UGT1A1 and prostate cancer risk among men in the finasteride arm are shown in Table II. In contrast to the findings among men in the placebo arm, there were no significant associations between CYP19A1 and UGT1A1 genotypes and prostate cancer risk, regardless of Gleason grade. However, unlike the null CYP11A1 findings in the placebo arm, the 6/9 genotype was significantly associated with an increased risk of high-grade prostate cancer (OR = 2.46, 95% CI = 1.11–5.47); no association was found with low-grade prostate cancer (OR = 1.00, 95% CI = 0.46–2.17).
General characteristics of cases and controls were compared, and only race significantly differed between cases and controls as well as across genotypes (Supplementary Tables 2 and 3 are available at Carcinogenesis Online). However, when the analyses were limited to white men only, the results for both placebo arm and finasteride arm were unchanged (data not shown).
The relationships between sex hormones and repeat polymorphisms were examined among controls (Table III). There was no overall significant difference in serum sex hormone levels by CYP11A1 repeat genotypes. For CYP19A1, serum estrone levels significantly differed by genotypes and individuals with at least one copy of the 8-repeat allele had slightly higher levels of serum estrogen. Men with the 8/8 genotype had the highest levels of serum estradiol (37.8 pg/ml) and estrone (55.2 pg/ml) and lowest level of serum testosterone (314.4 ng/dl) (result not shown). For UGT1A1, the heterozygous short/long genotype was associated with lower levels of serum estradiol. No gene-dosage effect was observed for UGT1A1; the homozygous long/long genotype was not associated with a further decrease of serum estradiol levels.
To test whether the associations between repeat polymorphisms and prostate cancer risk were mediated by polymorphism-associated changes in circulating sex hormone levels, we included serum levels of estrone, estradiol or testosterone in the models. The results for the genotypes remained unchanged after control for these sex hormones (data not shown). However, when stratified by serum testosterone and estrogens, substrates for enzyme CYP19A1 and UGT1A1, the associations between prostate cancer risk and CYP19A1 and UGT1A1 genotypes varied across above-median and below-median groups (Table IV). In the placebo arm, a significant association was observed between 7/8 genotype of CYP19A1 and an increased risk of prostate cancer (OR = 2.11, 95% CI = 1.22–3.67) in men with baseline serum testosterone above the median level, although the interaction test was not statistically significant. Similarly, an increased risk of prostate cancer with UGT1A1 short/long genotype was only observed among men with an above-median level of serum estradiol (OR = 1.66, 95% CI = 1.19–2.31), with a test for interaction of borderline significance (P = 0.05). Mutual adjustment for other sex hormones in the interaction models did not noticeably change the ORs (<5%) (results not shown). There was no effect modification by circulating levels of sex hormones among men in the finasteride arm.
In this nested case–control study from the PCPT, we found that prostate cancer risk was significantly associated with repeat polymorphisms in CYP19A1 (encoding aromatase) and UGT1A1 (encoding uridine diphosphate glucuronyltransferase), but not with CYP11A1 (encoding the cholesterol P450 side-chain cleavage enzyme) in the placebo arm. In contrast, in the finasteride arm, we observed no association with either CYP19A1 or UGT1A1, whereas a significant association was found between CYP11A1 repeat polymorphism and high-grade prostate cancer. Although we need to be cautious in interpretation of findings due to lack of gene-dosage effect and limited sample size in certain categories, the different relationship in the two treatment arms may indicate that, although some genetic variants may affect susceptibility to prostate cancer, such effects may not be notable when factors altering sex hormone metabolism pathways, such as finasteride treatment, are considered. Finasteride inhibits the conversion of testosterone to dihydrotestosterone, which may increase estrogen levels and alter the effect of genetic factors on estrogen metabolism. Indeed, both estrone and estradiol levels were modestly increased after finasteride treatment in the PCPT (13).
The polymorphic (TAAAA)n repeat in CYP11A1 is located in the promoter region of the CYP11A1 gene. Although similar (TAAAA)n repeat polymorphism in other genes have been shown to be associated with increased transcription and protein expression (25), the function of this repeat polymorphism in the CYP11A1 gene has not been characterized. In our study, the only significant association with CYP11A1 repeat polymorphism was between the 6/9 genotype and high-grade prostate cancer in the finasteride treatment arm. However, this is likely to be a chance finding due to small sample size in this group (12 cases and 21 controls) as well as lack of gene-dosage effect. The lack of association between CYP11A1 (TAAAA)n repeat polymorphism and prostate cancer risk is consistent with the findings that serum sex hormone levels did not differ by CYP11A1 genotypes. Several previous studies have investigated the association between CYP11A1 (TAAAA)n repeat polymorphism and prostate cancer risk (26–30), with only one study reporting a positive association with advanced prostate cancer in a Japanese population (30). Interestingly, the allelic distribution of the CYP11A1 (TAAAA)n repeat polymorphism differs considerably between Japanese and European populations, with a higher prevalence of 6-repeat allele in Japanese populations compared with the higher prevalence of 4-repeat allele in European populations (24,30). Therefore, ethnicity may at least partly explain why the positive association was not replicated in European populations.
The CYP19A1 gene encodes aromatase, which catalyzes the conversion of androstenedione and testosterone to estrogen. The (TTTA)n repeat polymorphism in CYP19A1 is located in intron 4 and is not close to any intronic splice sites, therefore, it is unlikely that this polymorphism directly affects aromatase activity. However, a strong degree of linkage disequilibrium between the TTTA repeat polymorphism and an exonic C/T polymorphism was reported (31). Long length of TTTA repeats was found to be associated with the T allele, which showed a high-activity phenotype with increased messenger RNA levels as well as activity of aromatase. Several studies have investigated the CYP19A1 repeat polymorphism in relation to prostate cancer with inconsistent findings (32–34). When dichotomized, Cussenot et al. (32) found that the long length allele (>7 repeats) was associated with a higher risk of prostate cancer. Using same classification, we found that the short/long genotype was associated with increased risk of prostate cancer (OR = 1.34, 95% CI = 1.03–1.75), but not the long/long genotype (results not shown). Further examining each individual genotype, only the 7/8 genotype was significantly associated with both low-grade and high-grade prostate cancer when compared with the 7/7 genotype. An increased risk of prostate cancer was also observed with the homozygous 8/8 genotype (OR = 1.53), although the association was not statistically significant, probably due to small sample size (seven cases and six controls). When all participants with at least one copy of the 8-repeat allele were combined, a significant association was also observed with increased prostate cancer risk compared with men with the 7/7 genotype (OR = 1.46, 95% CI = 1.06–2.01). Although it is possible that the 8-repeat allele of CYP19A1 may be associated with an increased risk of prostate cancer, there is a lack of biological evidence to explain why only the 8-repeat alleles but not other longer alleles (such as 9-, 10-, 11-, 12-, 13-repeat) are associated with increased prostate cancer risk.
In a group of postmenopausal women, Haiman et al. (23) found a modest association between 8-repeat allele and plasma estrogen levels, showing 8-repeat allele carriers had higher estrone and estradiol level than non-carriers. Consistent with these findings in women, in the PCPT, higher levels of serum estrogen were observed in men with at least one copy of 8-repeat allele; men with the 8/8 genotype had the highest level of serum estrogen and lowest level of serum testosterone in our study (results not shown). These results indicate that the 8-repeat allele of CYP19A1 may be associated with a high-aromatase-activity phenotype, which may enhance local estrogen metabolism in the prostate and contribute to prostate carcinogenesis. The increase of aromatase transcription and translation has been shown in prostate cancer tissues (14). The significant association between the CYP19A1 repeat polymorphism and prostate cancer risk was only observed among men with high (above-median) baseline level of serum testosterone, the substrate of aromatase, further supporting the hypothesis that genetic variants in CYP19A1 may modify prostate cancer risk by altering local estrogen metabolism. With high levels of substrate, difference in aromatase activity may appreciably affect the conversion of testosterone to estrogen, thus contributing significantly to local estrogen levels.
The UGT1A1 TA repeat polymorphism is located in the TATAA box of the UGT1A1 promoter, and the number of repeats (5–8) was found to be inversely correlated with transcription and expression of UGT1A1 as well as with enzymatic activity (35,36). As a consequence, the long allele of UGT1A1 (>6 repeats) is considered to be associated with decreased enzyme activity, which may theoretically result in elevated estrogen levels. However, in our data, low estrogen levels were observed among men carrying the long allele of UGT1A1. Compared with men with the short/short genotype, men with the short/long genotype had the lowest estrogen levels but had an increased risk of prostate cancer. No gene–dosage effect was observed, showing no significant association among men with the long/long genotype. Therefore, interpretation of the association between UGT1A1 genotypes and prostate cancer needs to be made with caution, and the possibility of chance cannot be ruled out. A recent study reported no association between UGT1A1 repeat polymorphism and prostate cancer risk (37).
This study benefits from the prospective collection of blood samples before disease diagnosis, which allowed us to evaluate the relationships between baseline sex hormone levels, genotypes and prostate cancer risk. We found that serum estrogen levels significantly differed by CYP19A1 and UGT1A1 repeat genotypes, however, controlling for sex hormone levels in the model did not change the associations between gene polymorphisms and prostate cancer risk, indicating that the effects of repeat polymorphisms in key estrogen metabolism genes on prostate cancer susceptibility are probably not mediated by circulating estrogen levels. Circulating sex hormone levels are controlled by a large number of genes involved in estrogen biosynthesis and metabolism, which were not examined in this study. Repeat polymorphisms in CYP19A1 and UGT1A1 only slightly altered circulating estrogen levels, showing <10% of changes in their level between differed genotypes as shown in Table III, which may partly explain the limited impact of controlling circulating hormone levels on gene–disease associations. However, when circulating sex hormone levels were dichotomized into high/low groups, the significant associations between CYP19A1 and UGT1A1 repeat polymorphisms and prostate cancer risk were restricted in men with high (above-median) levels of serum testosterone and estradiol, the specific substrates of CYP19A1 and UGT1A1 in estrogen metabolism pathway. With high level of circulating substrates, changes in local estrogen metabolic enzyme activities by genetic polymorphisms may have a large impact on intraprostatic estrogen levels. This finding supports our hypothesis that genetic variants in estrogen-related genes may modify prostate cancer risk by altering local estrogen metabolism; although as yet, there is lack of direct biological evidence for this hypothesis.
Although the sample size in this study was large, allowing for examination of individual genotypes, the study is still hampered by small sample size in certain categories. Taking the 8-repeat allele of CYP19A1 as an example, we found that men with the 7/8 genotype had a significantly higher risk of prostate cancer than men with the common 7/7 genotype. However, with a total of 13 men with the 8/8 genotype in the placebo arm, we could not reliably assess the additive effect of this allele on prostate cancer risk. Secondly, circulating estrogen levels may not reflect intraprostatic levels (38), and we were not able to examine the effects of estrogen-related gene polymorphisms on local estrogen metabolism. Lastly, functional characterization of potential genetic variants in the future is warranted.
In summary, we examined three critical genes involved in estrogen biosynthesis and metabolism in relation to prostate cancer. We found that repeat polymorphisms in CYP19A1 and UGT1A1 may be associated with altered estrogen levels as well as an increased risk of prostate cancer, although we could not rule out the possibility that these were chance findings. However, the different associations between estrogen-related genes and prostate cancer risk in the placebo and finasteride arms indicate that factors interfering with sex hormone metabolism pathways may modify the effect of genetic variants on prostate cancer susceptibility. Considering the low frequency of risk alleles and the potential interactions with environmental factors, a much larger study would be required to confirm these findings.
National Cancer Institute (P01 CA108964).
Conflict of Interest Statement: None declared.