Variation in the cytochrome P450 oxidoreductase (POR) gene, a key regulator of type II cytochrome P450 enzymes, may affect exposure to endogenous steroid hormones and breast cancer risk. We sequenced the POR locus and tested candidate polymorphisms G5G and A503V for association with breast cancer risk among women in the Multiethnic Cohort Study (1,615 cases and 1,962 controls). The single nucleotide polymorphism (SNP) A503V was common in all racial/ethnic populations (minor allele frequency, ≥0.05) but was not associated with risk. SNP G5G (A → G nucleotide change), which lies in a suggestive exonic splicing enhancer motif in exon 1, was common only in African Americans (minor allele frequency, 0.21) and the homozygous state was modestly associated with increased breast risk among all cases [345 cases and 426 controls; odds ratio (OR), 1.64; 95% confidence interval (CI), 0.89–3.04; P = 0.12] and among cases with advanced disease (95 cases: OR, 3.08; 95% CI, 1.42–6.70; P = 0.005). In an attempt to replicate this association, we genotyped SNP G5G in additional African American case-control studies (747 cases and 468 controls). Nonsignificant positive associations were noted with the GG genotype class in all studies. In the pooled analysis (1,038 cases and 877 controls with genotype data), the association was statistically significant among all cases (OR, 1.58; 95% CI, 1.04–2.41; P = 0.03) and stronger in those with advanced disease (411 cases and 877 controls; OR, 2.60; 95% CI, 1.56–4.34; P = 0.0002). These data suggest that African Americans harbor an allele at the POR locus that may increase breast cancer risk. [Cancer Res 2007;67(8):3565–8]
- genetic polymorphism
- breast cancer
Collaborative efforts are under way to examine variation in genes involved in steroid hormone biosynthesis as predictors of breast and prostate cancer risk ( 1). Many of these genes encode cytochrome P450 enzymes that function at key steps in gonadal and adrenal steroidogenesis (e.g., CYP17A1 and CYP19A1; refs. 2, 3). For catalytic activity, P450 enzymes require an interaction with P450 oxidoreductase (POR), a flavoprotein that shuttles electrons from NADPH to the iron atom of the P450 heme group. Rare mutations in POR and in the fibroblast growth factor receptor 2 lead to Antley-Bixler syndrome, a congenital condition that is characterized by craniosynostosis and other skeletal malformations ( 4). Mutations in POR have also been found in patients with and without Antley-Bixler syndrome presenting with ambiguous genitalia and/or disordered steroidogenesis (decreased 17α-hydroxylase, 17,20-lyase, 21-hydroxylase and aromatase activities; refs. 4– 8). Based on the critical role of this gene in P450 activity and steroid hormone production, we tested the hypothesis that common genetic variation at the POR locus may contribute to breast cancer risk.
Materials and Methods
This work was carried out in established population-based breast cancer case-control studies. The first study was nested within the Multiethnic Cohort Study (MEC; ref. 9), a cohort that includes >215,000 individuals from Hawaii and Los Angeles which was assembled between 1993 and 1996. The cohort is comprised predominantly of African Americans, Native Hawaiians, Japanese, Latinos, and Whites. Beginning in 1994, blood samples were collected from incident breast cancer cases identified by cohort linkage to Surveillance, Epidemiology, and End Results (SEER) registries, as well as a random sample of participants in the MEC to serve as controls for genetic analyses. Participation rates for providing blood samples were ≥65% for cases and controls. This analysis was based on 1,615 invasive breast cancer cases diagnosed up to April 2002 and 1,962 controls free of cancer up to April 2002 who were ages 44 to 86 years. Controls were frequency-matched to cases on 5-year age group and race/ethnicity (African Americans, 345/426; Native Hawaiians, 109/290; Japanese, 425/420; Latinos, 335/386; Whites, 401/440).
Associations observed in the MEC with candidate single nucleotide polymorphisms (SNPs) in the POR gene were examined in three additional studies. The first was a San Francisco Bay area–based case-control study (SF) initiated in 1995 that included invasive African American breast cancer cases ages 35 to 79 years identified through the regional SEER cancer registry and controls identified through random digit dialing ( 10, 11). Beginning in 1999, blood samples were collected for >70% of cases (n = 211) diagnosed between 1997 and 1999 and controls (n = 232). The second study was comprised of invasive African American breast cancer cases and controls ages 35 to 64 years from the Los Angeles component of the Women's Contraceptive and Reproductive Experiences Study (CARE; ref. 12). Incident cases diagnosed between 1994 and 1998 were identified through the Los Angeles SEER registry, and random digit dialing was used to identify controls from the same geographic area. Participation rates for the CARE Study were >70% for cases and controls. Blood specimens were collected from 82% of cases (n = 397) and 80% of controls (n = 236). The third study included invasive African American breast cancer cases from a Los Angeles–based case-control study among young women ages 20 to 49 years (the LIFE study; ref. 13). Incident cases diagnosed between 2000 and 2003 were identified from the Los Angeles SEER registry. In the LIFE study, the participation rate for African American cases was 60%, and in the current study, we used blood samples obtained from the first 69% of interviewed cases (n = 139). The overall participation rates for cases and controls in these studies ranged from 41% to 70%. Although the overall participation rates were not exceptional, it is unlikely that participation in any of these studies was associated with POR genotype, and that nonparticipation would have resulted in any systematic bias in this study.
To search for putative functional variants, we sequenced the 15 coding regions of the POR gene (and adjacent splice site regions) in 95 women with advanced breast cancer (SEER stage ≥2) from the MEC (19 of each racial/ethnic group). This sample was selected to (a) increase the probability of discovering variation that is biologically relevant, and (b) to achieve ≥85% power to detect a SNP with an ethnic-specific minor allele frequency (MAF) ≥0.05. Bidirectional sequencing was done on the ABI 3730 DNA Analyzer (Applied Biosystems, Foster City, CA) using published primers ( 6). The NCBI sequence, NM_000941.1 was used as the reference sequence for all identified variants.
We identified four synonymous SNPs (MAF ≥0.05 in any racial/ethnic group): G5G (rs10262966), P129P (rs12669302), A485A (rs6950661), and S572S (rs1057870); one nonsynonymous variant, A503V (rs1057868); and two rare nonsynonymous variants (V191L and L447M), observed in only one White or Latino case, respectively. No common variants were found at splice site junctions. The frequencies in each racial/ethnic population are shown in Supplementary Table S1.
Common SNPs with a greater likelihood of altering enzyme function were selected for genotyping in the MEC case-control samples. These included the nonsynonymous variant, A503V and G5G, which had been reported in a Japanese male with Antley-Bixler syndrome as a potentially causal mutation located within an exonic splicing enhancer motif in exon 1 ( 5). SNP genotyping was done by the TaqMan assay using the ABI 7900 (Applied Biosystems) in the USC Genomics Core. Concordance with blinded quality control samples in each study (∼5%) was >99%.
Odds ratios (OR) and 95% confidence intervals (CI) were estimated for each SNP genotype using unconditional logistic regression. Associations were examined in ethnic-stratified analyses, and pooled ORs (ORpooled) were estimated adjusting for race/ethnicity and age, and study when data were combined from all studies. Tests of homogeneity of ORs for SNP genotypes were also done across racial/ethnic group and study. The early onset cases in the LIFE study were analyzed with those from CARE as they were selected from the same geographic region, and we found no evidence of an association between POR variants and age at diagnosis in these studies (P = 0.68); removing the LIFE cases from the analysis did not affect the results. Analyses were also conducted by stage at diagnosis and hormone receptor (estrogen receptor and progesterone receptor) status, which was collected from the cancer registries. Adjustment for established breast cancer risk factors (in MEC analyses; ref. 14) had little effect on the results. The study protocol was approved by the Institutional Review Boards at all participating institutions.
Results and Discussion
The V allele (valine) at A503V was common in all racial/ethnic populations in the MEC (MAF ≥0.17 among controls) and genotype frequencies were not found to deviate from Hardy-Weinberg Equilibrium in any group. The V allele was not significantly associated with breast cancer risk in any racial/ethnic group [ORpooled, ethnicity versus AA genotype, AV: 0.97; 95% CI (0.84–1.12); VV: 0.92; 95% CI (0.74–1.19); Ptrend = 0.54; Table 1 ] and thus, was not investigated further. The G allele of the G5G variant (A → G nucleotide change) was relatively rare in all populations (MAF <0.03 in controls) except in African Americans (MAF, 0.21, r2 with A503V = 0.02). In this population, the homozygous genotype class GG was associated with a nonsignificant increase in breast cancer risk (OR, 1.64; 95% CI, 0.89–3.04; P = 0.12) compared with noncarriers of the G allele. This association was statistically significant among African Americans in the MEC with advanced disease (OR, 3.08; 95% CI, 1.42–6.70; P = 0.005; Table 2 ). The effect did not differ by estrogen receptor or progesterone receptor status (data not shown).
In an attempt to replicate the observed association among African Americans, we genotyped the G5G variant in the SF and CARE/LIFE case-control studies. Among controls, the MAF of the G allele was 0.25 in the SF study and 0.23 in the CARE/LIFE study; genotype frequencies were not found to deviate from Hardy-Weinberg Equilibrium in cases or controls in any of the study populations. In all studies, women with the GG genotype class were at a nonsignificantly increased relative risk compared with noncarriers of the G allele (SF: OR, 1.63; 95% CI, 0.70–3.80; P = 0.26; CARE/LIFE: OR, 1.47; 95% CI, 0.68–3.20; P = 0.33; Table 2). When the genotype data from all studies were combined (1,038 cases and 877 controls) the association was statistically significant (ORpooled, study = 1.58; 95% CI, 1.04–2.41; P = 0.03; Ptrend = 0.15). As observed in the MEC, among advanced cases, this association was statistically significant in the SF study (P = 0.002), but not in the CARE/LIFE study (P = 0.57), although a positive relationship was observed ( Table 2). When pooling the data across studies, a highly statistically significant association was observed with the GG genotype class among women with advanced disease (411 cases and 877 controls; ORpooled, study = 2.60; 95% CI, 1.56–4.34; P = 0.0002; Ptrend = 0.01). No evidence of heterogeneity for genotype class effects were noted across studies ( Table 2). Although the association among advanced cases was substantially weaker in the CARE/LIFE studies which were comprised of younger women, we found no evidence of modification of the effect by age within any of the four studies or overall (Supplementary Table S2).
Together, these data suggest that the G5G polymorphism contributes to breast cancer risk in African Americans, the only population that carried this variant at an appreciable frequency among the populations examined in the MEC. This association, which was limited to advanced breast cancer, proposes a role for this allele in disease progression. It has been speculated that this SNP disrupts an exonic splicing enhancer motif in exon 1 of the POR gene, which could have an effect on the efficiency of pre-mRNA splicing ( 5) and on regulating the activity of cytochrome P450 enzymes involved in steroid hormone biosynthesis. Additional work, however, will be needed to understand the functionality of this SNP and whether it is an underlying risk allele or is in linkage disequilibrium with an unknown causal allele in African Americans.
One explanation for these findings may be population stratification because race was based on self-report. We assessed and corrected for this potential confounding effect in the MEC samples using data for ∼1,000 SNPs from 60 genes that had been genotyped previously in these same subjects. Adjustment for stratification using the methods of Price et al. ( 15) had little effect on the results (GG versus AA genotype: all cases, OR, 1.65; advanced cases, OR, 3.34), and thus, we feel confident that population substructure was unlikely to explain these results.
Based on the frequency of the GG genotype for G5G in African Americans (∼0.05) and an associated relative risk of 1.6, we project the attributable risk in this population to be ∼3%. These findings support the need for genetic studies in minority populations to identify risk alleles with pan-ethnic effects but with varying allele frequencies across populations, as well as risk alleles that may be ethnic-specific, both of which may be missed in studies limited to non–Hispanic Whites who have been the focus of most studies to date.
In summary, our results suggest that the POR genotype contributes to breast cancer risk among African American women. Replication of these findings in even larger studies is warranted.
Grant support: National Cancer Institute (NCI) grants CA63464 and CA54281. The San Francisco–based study was supported by NCI grant CA77305 and grant 17-96-1-6071 by the U.S. Department of Defense. The Women's CARE Study was supported by the National Institute of Child Health and Human Development, with additional support from the NCI, through a contract with University of Southern California (N01 HD 3-3175), through an intra-agency agreement with the Centers for Disease Control and Prevention (Y01 HD 7022), and with support from SEER contract N01-CN-67010. Women's CARE Study control blood collection was supported by grant 7PB-0051 from the California Breast Cancer Research Program. The LIFE study was supported by NCI grants CA17054 and CA74847 and grant 4PB-0092 from the California Breast Cancer Research Program. The collection of cancer incidence data for the San Francisco Bay area and Los Angeles County used in this publication was supported by the California Department of Health Services as part of the statewide cancer reporting program mandated by the California Health and Safety Code Section 103885.
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.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
The ideas and opinions expressed herein are those of the authors, and no endorsement by the State of California, Department of Health Services, is intended or should be inferred.
- Received January 2, 2007.
- Revision received February 20, 2007.
- Accepted March 9, 2007.
- ©2007 American Association for Cancer Research.