Association of MTHFR gene polymorphisms with breast cancer survival

Breast cancer cases (n = 248) were women, who were recruited between February of 1993 and August of 2003 in the greater Baltimore area, as described previously [29]. All patients were identified through surgery lists and enrolled into the study prior to surgery. They were selected from a group of 361 breast cancer patients who were recruited at the University of Maryland Medical Center, the Baltimore Veterans Affairs Medical Center, Union Memorial Hospital, Mercy Medical Center, and the Sinai Hospital in Baltimore, Maryland. Selection was based on eligibility criteria and the availability of survival information. Cases were eligible if they had pathologically confirmed breast cancer, had residency in the greater Baltimore area, were of African-American or Caucasian descent by self-report, were diagnosed with breast cancer within the last six months prior to recruitment, and had, by self-report, no previous history of the disease. Patients were excluded if they were HIV, HCV, or HBV carriers, were IV-drug users, were institutionalized, or were physically or mentally unable to sign consent and complete the questionnaire. All participants signed a consent form and completed an interviewer-administered questionnaire. The blood collection was performed at the subjects' convenience but mostly at the conclusion of the interview. Of the patients that were eligible under the original protocol and could be contacted by phone, 83% agreed to participate in the study. One eligible patient died prior to surgery and is not included in the study population. Additional information to determine the ER status, disease stage, treatment, and survival was obtained from medical records and pathology reports, the Social Security Death Index, and the National Death Index. The Institutional Review Boards at the participating institutions approved the study.

A pathologist reviewed each tumor specimen to confirm the presence of tumor and the histology. Disease staging was performed according to the tumor-node-metastasis (TNM) system of the American Joint Committee on Cancer/the Union Internationale Contre le Cancer (AJCC/UICC). The two germline MTHFR SNPs, C677T and A1298C, were genotyped at the NCI Genotyping Core Facility (GCF), using the assay conditions described in the SNP500Cancer database [30]. We genotyped genomic DNA from fresh-frozen breast tissue (170 non-tumor tissues and 28 tumors) and buffy coat (n = 50) of the 248 patients. All genotype assays contained negative and positive controls, and 10% blinded duplicates consisting of patient samples. We successfully genotyped 98% of the cases for both genotypes (n = 243) and had a 100% concordance among blinded duplicates for the two assays. In addition, we analyzed the MTHFR C677T genotype of microdissected normal tissue from 21 of those 28 bulk tumor samples that were genotyped at the NCI GCF. The MTHFR genotype of these samples was determined by direct sequencing. We had 100% concordance between the C677T genotypes of the bulk tumor samples and the non-tumor tissue from these samples. The observed genotype frequencies for both SNPs were in Hardy-Weinberg equilibrium in our case population, and for a matched control population (n = 317). Genotype frequencies among African-Americans and Caucasians were in agreement with the frequency data for these race/ethnic groups in the SNP500Cancer database [30].

Stata 7.0 (Stata Corp, College Station, TX) and SAS 8.0 (SAS Institute, Cary, NC) statistical softwares were used for data analysis. All statistical tests were two-sided and an association was considered statistically significant with p < 0.05. The Kaplan-Meier method and the log-rank test were used for univariate survival analysis. The Cox Proportional-Hazards regression was used for multivariate survival analysis to calculate adjusted hazard ratios. A statistical test for interaction was performed in Stata to determine if the effect of the two MTHFR genotypes on breast cancer survival is modified by other factors. In the analysis, an interaction on breast cancer survival was coded as "xi: stcox i.Var1*Var2". The interaction term was coded with MTHFR genotypes as Var1 and other factors (e.g., race, alcohol, and smoking) as Var2. Var1 and Var2 were coded as binary variables. For this test, the result was coded as zero (or referent) for the common MTHFR genotype (C/C for SNP677 and A/A for SNP1298) and one for the variant genotype (C/T or T/T for SNP677; A/C or C/C for SNP1298). Alcohol consumption was coded as "no/yes" for this test with "no" as the referent indicating a consumption of 12 or less alcoholic drinks in life at the time of recruitment. Survival was determined for the period from the date of hospital admission to the date of the last completed search for death entries in the Social Security Index (August 18, 2004). The median and mean follow-up times for overall survival were 47 and 55 months, respectively, with a range of 12 to 140 months for the follow up time. A total of 59 out of the 248 patients in the study (24%) died during this period. We obtained death certificates of the deceased cases and censored all causes of death that were not related to breast cancer, such as accidents, violent crimes, stroke, heart attack, and liver cirrhosis, in our analysis. The cause of death was not related to breast cancer for 5 patients.

We studied the relationship between the MTHFR C677T and A1298C genotypes and disease outcome of 248 unselected breast cancer patients. The clinicopathological features of the patients are shown in Table 1. Most women were of African-American descent, and 132 patients received chemotherapy. Among those, 15 patients received a 5-fluorouracil/methotrexate-based therapy, 71 patients received a doxorubicin-based combination therapy, and the remaining 46 patients received either other agents or the exact treatment modality was not reported. The genotype distribution of the patients is shown in Table 2. The genotype frequencies for C677T and A1298C were significantly different between African-American and Caucasian women. We also observed a linkage between the two genotypes in our population (Table 3) that was consistent with published data [31, 32]. None of the patients was homozygous for both variant alleles, and only two patients were homozygous for one variant allele and heterozygous for the other variant allele.

We first examined the individual effects of the two genotypes on breast cancer survival assuming a dominant effect of the variant allele. C677T was not significantly associated with breast cancer survival in the unstratified Cox regression analysis with adjustments for age at diagnosis, race, A1298C, body mass index (BMI), ER status, TNM stage, and chemotherapy (Table 4). Stratification by ER-status and race revealed that the T allele (C/T or T/T) was protective in patients with an ER-negative tumor and in African-American patients. The association with ER-negative tumors was marginally statistically significant. The relationship between C677T and ER-negative but not ER-positive breast cancer was also seen in the Kaplan-Meier survival analysis (Figure 1a,b). To examine the independent effect of C677T, we repeated our analysis but included only those patients who were homozygous A/A at codon 1298 (n = 135, see Table 3). In this stratified analysis, the variant T allele at codon 677 (C/T or T/T) remained protective, albeit non-significantly, in patients with an ER-negative tumor (HR = 0.3; 95% CI, 0.06–1.47; n = 51) and in African-American patients (HR = 0.14; 95% CI, 0.02–1.09; n = 87).

The second MTHFR genotype, A1298C, was significantly associated with survival in the unstratified analysis (Table 5). In contrast to C677T, the variant allele (A/C or C/C) increased the risk of poor outcome. A stratification by the ER status showed that the variant C allele lead to an increased risk of dying only among patients with an ER-negative tumor, as shown in the Kaplan-Meier (Fig. 1c,d) and multivariate Cox regression analyses (Table 5). The variant C allele was also associated with a significantly increased risk of poor outcome among Caucasians and among patients, who did not receive chemotherapy (Table 5). To examine the independent effect of A1298C, we repeated our analysis but included only those patients who were homozygous C/C at codon 677 (n = 163, see Table 3). We found that the variant C allele at codon 1298 (A/C or C/C) remained a significant predictor of poor survival among ER-negative patients matched on the C677T genotype (HR = 3.27; 95% CI, 1.27–8.46; n = 62).

We also modeled the survival analysis assuming an additive effect (homozygous wild-type, heterozygous, homozygous variant) by C677T and A1298C. Again, we found that 677T was associated with an increased survival of patients who either had an ER-negative tumor (HR = 0.36; 95% CI, 0.13–1.02) or were African-Americans (HR = 0.12; 95% CI, 0.02–0.92). 1298C was associated with a decreased survival of patients with an ER-negative tumor (HR = 2.26; 95% CI, 1.17–4.37), Caucasians (HR = 2.53; 95% CI, 0.95–6.71), and patients who did not receive chemotherapy (HR = 2.74; 95% CI, 0.79–9.5). The results are very similar to those obtained with the assumption of a dominant effect by 1298C (Table 5). Thus, the additive and dominant models support the same conclusions.

Next, we studied the combined effect of the two SNPs after categorizing the genotypes into low and high MTHFR enzyme activity diplotypes based on the previously reported effect of the two variants on MTHFR activity [6]. Patients were classified as carriers of a high activity diplotype (n = 143) if they either had the CC (677) and AA (1298) or the CC (677) and AC (1298) genotype combinations (see Table 3). The CC(677)AC(1298) diplotype has 80–90% of the enzyme activity encoded by the CC(677)AA(1298) diplotype. All other genotype combinations were considered low activity diplotypes (n = 100) with a predicted MTHFR activity of 30–70% of the CC(677)AA(1298) diplotype. Carriers of a low activity diplotype had a non-significantly increased survival in the unstratified analysis and had a significantly increased survival if they were African-Americans (Table 6). Overall, the observed hazard ratios for the diplotypes were similar to the hazard ratios obtained for C677T but divergent from the hazard ratios obtained for A1298C.

Lastly, we examined interactions between the SNPs and other risk factors on breast cancer survival. We did not observe an interaction with the ER status and chemotherapy. We found a significant protective interaction between the variant C allele at codon 1298 (A/C or C/C) and alcohol consumption on breast cancer survival (A1298C, p _interaction = 0.017). Patients who reported to consume alcohol at recruitment had an improved survival, compared to nondrinkers, if they carried the 1298C allele because of the interaction. There was no such interaction with C677T or the MTHFR diplotype (C677T, p _interaction = 0.43; MTHFR diplotype, p _interaction = 0.64). We also observed interactions between the two genotypes and race/ethnicity on breast cancer survival (C677T, p _interaction = 0.026; A1298C, p _interaction = 0.088; MTHFR diplotype, p _interaction = 0.054). Because of the interaction, African-American women had improved survival, compared to Caucasian women, if they were carriers of either the variant T allele at codon 677 (C/T or T/T), the variant C allele at codon 1298 (A/C or C/C), or the low activity MTHFR diplotype. The data are consistent with the results of the Cox regression analyses for C677T, A1298C, and the MTHFR diplotypes that showed an increased survival hazard of Caucasian women when compared with African-American women in the stratified analyses (Tables 4 to 6).