Common variants in genes coding for chemotherapy metabolizing enzymes, transporters, and targets: a case–control study of contralateral breast cancer risk in the WECARE Study

Participants were identified through five population-based cancer registries: Los Angeles County Cancer Surveillance Program; Cancer Surveillance System of the Fred Hutchinson Cancer Research Center (Seattle); State Health Registry of Iowa; and the Cancer Surveillance Program of Orange County/San Diego-Imperial Organization for Cancer Control (Orange County/San Diego). These cancer registries contribute to the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) program. The fifth registry from which subjects were recruited was the Danish Breast Cancer Cooperative Group Registry, supplemented by data from the Danish Cancer Registry [16].

Details of CBC case and UBC control eligibility have been described previously [16]. Briefly, cases were women diagnosed prior to age 55 years, from 1985 to 2000, with invasive breast cancer that had not spread beyond regional lymph nodes. This had to be followed by a second in situ or invasive breast cancer diagnosed in the contralateral breast at least 1 year later. The ‘at-risk’ interval was defined as starting at the time of first breast cancer diagnosis and ending at reference date, that is, date of the second breast cancer diagnosis in cases (reference date) or the corresponding date in matched controls. Two controls were individually matched to each case on year of birth (in 5-year strata), year of diagnosis (in 4-year strata), registry region, and race/ethnicity. All women had to be alive at the time of contact and able to complete a telephone interview and donate a blood sample. Counter-matching based on registry-reported radiation treatment status was used to improve the statistical efficiency of the study design. Thus, for each radiation exposed case, one radiation exposed control and one unexposed control were selected from the relevant stratum; and for each unexposed case, two radiation exposed controls were selected [16].

Across the five cancer registries, 708 cases and 1,399 controls completed the study interview and provided a blood sample. Four individuals were excluded from the current analysis because they did not consent to genotyping beyond the initial ATM, BRCA1, and BRCA2 mutation screening. To minimize the potential influence of ancestral differences in genotype frequencies, all analyses were restricted to Caucasian women (n = 1,933) as recorded by the cancer registry. Further exclusions were made after genotyping (see below).

The data collection protocol was approved by the institutional review board at each of the participating centers and by the Ethical Committee System in Denmark. Each woman provided written informed consent. Details of the study questionnaire have been published previously and included questions about known breast cancer risk factors [16]. Medical records, pathology reports, and hospital charts, in addition to self-reported data (collected during the telephone interview), were used to collect detailed treatment information (surgery, chemotherapy, hormonal therapy, radiation therapy) on the first primary breast cancer as well as during the at-risk period. Information collected on chemotherapy and hormonal therapy included dates of administration, reason for treatment (e.g., primary disease, recurrence), and type of drug. The most common chemotherapeutic regimens received by women in the WECARE Study population were cyclophosphamide (CTX), methotrexate (MTX), 5-fluorouracil (5FU) (CMF) (63 % of women treated with chemotherapy were treated with CMF) and 5FU, doxorubicin (Adriamycin^®), CTX (FAC) or 5FU, epirubicin, CTX (FEC) (19 % of women treated with chemotherapy were treated with FAC/FEC) (Table 1). All other drug combinations were coded as ‘other’ chemotherapy. For the current analyses, a woman was classified as having received CMF or FAC/FEC if she received these combinations of drugs any time during her treatment for a first primary breast cancer and prior to the reference date.

Genes were selected for their known involvement in the metabolism, action, and transport of chemotherapeutic agents commonly used to treat breast cancer. A list of genes and their associated drugs can be found in Table 2.

DNA was prepared from blood samples by red cell lysis and standard methods of phenol/chloroform extraction. Samples were genotyped with Illumina’s HumanOmni1-Quad BeadChip (Illumina Inc., San Diego, CA, USA) as part of the WECARE Study’s GWAS effort. Default Omni1-Quad cluster definitions supplied by Illumina were used to call genotypes, and single nucleotide polymorphisms (SNPs) with GenTrain scores <0.36 were considered ‘no calls,’ and samples with call rates <95 % were excluded in addition to other exclusion criteria described below. Each 96 well plate included one inter-plate positive quality control sample (NA06990—Coriell Cell Repositories). In addition, 38 blinded and 46 unblinded quality controls replicates from the study sample were genotyped. Concordance rates for both the Coriell and study sample replicates were high: >99.99 %.

Additional genotyping in these genes was performed to broaden gene coverage. SNP lists from the HapMap project (http://​hapmap.​ncbi.​nlm.​nih.​gov/​) were imported into Tagger (in Haploview) [17], and haplotype tagging SNPs (tagSNPs) were selected based on patterns of linkage disequilibrium (LD) with boundaries suggested by Gabriel et al. [18]. tagSNPs were selected based on pairwise tagging with a minimum r ² of 0.90. Multiplex SNP genotyping was carried out using the Illumina Golden Gate™ assay on custom BeadChips (Illumina Inc., San Diego, CA, USA). Laboratory methods and sample control measures have been described previously [19].

The CYP2D6*4 (rs3892097) variant was genotyped by a modified MGB Eclipse probe assay (Epoch Biosciences, ELITech Group, Paris, France). The outer primers designed to exclude pseudogenes were 5′ AGCCTGCCCCAGCCAAGGGAGC 3′ and 5′ CTCGGTCTCTCGCTCCGCAC 3′. The internal primers were designed by Epoch Biosciences to encompass the SNP: 5′ AATAAATCATAACCCCTTACCCGCATCTC 3′ and 5′ GATCACGTTGCTCACGGCTTTGTCCAAGAG 3′. DNA was amplified using the standard Eclipse protocol except that in the first 15 of 50 cycles, and the annealing temperature was increased by ten degrees to 68 °C. This method resulted in 100 % concordance of genotypes among the 24 % blinded, re-sampled DNAs. A subset of samples (17 %) was confirmed by a second method, allele-specific tetra-primer PCR, and separation of the allele-specific fragment sizes on 1 % agarose [20].

Quality control steps applied to the genome-wide association study (GWAS) data lead to further subject exclusions: (a) Women with SNP call rates <95 % were excluded (n = 22); (b) Population stratification was investigated using EIGENSTRAT [21]; using the first two principal components, 9 outliers with significant African or Chinese ancestry were identified for exclusion; and (c) 14 additional participants were excluded due to incomplete matched sets. Identity by descent was examined using PLINK [22] identifying 3 pairs of sisters, including one pair of identical twins. These women were not excluded from the analysis. An additional 28 subjects were excluded because they had >5 % missing genotypes on the SNP BeadChips. Analyses are based on the remaining 1,860 participants (636 CBC cases and 1,224 UBC controls) with genotype data from both the Omni1-Quad and custom bead chips platforms.

Within the selected genes of interest, 260 SNPs were genotyped on the OMNI platform, 27 SNPs on the SNP BeadChip, and rs3892097 (CYP2D6*4) on a modified MGB Eclipse probe assay (for a total of 287 genotyped SNPs). SNPs with >10 % missing (n = 16) and those that were monomorphic (n = 87) were excluded. Although Hardy–Weinberg equilibrium may not strictly apply since all participants in the study were affected with breast cancer, 4 SNPs deviating from Hardy–Weinberg equilibrium (p < 0.001) were also excluded. This left 180 SNPs in or near 14 genes to be included in the analyses: 1 SNP in CYP1A1, 4 in CYP1B1, 2 in CYP2A6, 6 in CYP2B6, 4 in CYP2C9, 9 in CYP2D6, 26 in CYP3A4, 33 in CYP3A5, 13 in DHFR, 3 in GSTM1, 1 in GSTM2, 22 in GSTP1, 54 in MTHFR, and 2 in ABCB1 (Online Resource 1).

In analyses examining the impact of genotype on the association between CMF and risk of CBC, the chemotherapy regimen variable was coded as CMF, other chemotherapy regimens, and no chemotherapy. Similar coding was used for FAC/FEC analyses. In all instances, the comparison group was women who did not receive chemotherapy. In the CMF analyses, FAC/FEC was coded as ‘other chemotherapy’ and vice versa. SNPs in genes that code for enzymes involved in the metabolism or action of cyclophosphamide, methotrexate, 5-fluorouracil, or doxorubicin/epirubicin were included in the analyses (Table 2). Based on the combination of drugs used in either regimen, the same SNPs were included in the CMF and FAC/FEC analyses.

Rate ratios (RR) and 95 % confidence intervals (CI) were estimated using conditional logistic regression to examine the association between chemotherapeutic regimen (CMF or FAC/FEC) and risk of CBC, stratified by genotype for each SNP using the dominant model [0 (homozygous wild-type), 1 (heterozygous and homozygous variant)]. Models were run adjusting for age at first breast cancer diagnosis and included an ‘offset term’ (i.e., log weight ‘covariate’ in the model where the coefficient of this log weight is fixed at one [16]), taking into account the sampling probabilities of the counter-matching. Multi-variable adjusted models were also run including adjustment for age at first diagnosis, family history, stage and histology of first primary breast cancer, and other treatments (hormonal therapy and radiation therapy). The likelihood ratio test was used to test for heterogeneity of treatment effect across genotypes.

Age and multivariable-adjusted [as described above] analyses were also conducted to confirm the association between chemotherapy and CBC risk in the subgroup of women included in the current analyses [1,860 (88 %) of the 2,107 total number of women in the WECARE Study].

A conservative Bonferroni correction was used to determine the multiple comparison cut-point (α = 0.0003, obtained from (0.05/180 SNPs) at which results were considered statistically significant. All analyses were conducted using SAS 9.2 (SAS Institute Inc., Cary, NC, USA). Figures were generated using Microsoft^® Excel 2007.