Associations between null mutations in GSTT1 and GSTM1, the GSTP1 Ile¹⁰⁵Val polymorphism, and mortality in breast cancer survivors

The Glutathione-S Transferases (GST) are a phase II superfamily of cytosolic, mitochondrial and microsomal enzymes that catalyze the conjugation of reduced glutathione to electrophilic centers on a variety of substrates (Strange et al. 2001). This activity is acts a detoxification step for a variety of endogenous molecules and xenobiotics, including chemotherapeutic drugs. The mammalian cytosolic GSTs comprise 6 classes of dimeric isoenzymes alpha (α), mu (μ), pi (π), theta (τ), zeta (ζ) and omega (ω). GST-μ, GST-τ, and GST-π are encoded by the GSTM1, GSTT1, and GSTP1 genes, respectively; and these 3 genes have been studied in association with genetic susceptibility to cancer (Strange & Fryer 1999; Spurdle et al. 2010).

Homozygous deletion of the GSTM1 and GSTT1 genes (null genotype), are associated with a lack of enzyme function and increased vulnerability to cytogenetic damage (Seidegard et al. 1988). Individuals who have deletions in GSTM1 or GSTT1 may therefore be at increased cancer risk (Strange & Fryer 1999; Rebbeck 1997).

The GST π (P1) polymorphism (rs1695; an A→G transition at position 313) results in an Ile→Val change at codon 105 (Ile¹⁰⁵Val). The variant allele is associated with lower substrate-specific catalytic activity, including towards the alkylating anticancer agent chlorambucil (Hayes & Strange 2000; Pandya et al. 2000; Srivastava et al. 1999). A limited number of studies with conflicting results have investigated the association between polymorphisms in GST genes and mortality in breast cancer patients. The majority of these studied patients diagnosed prior to 1999. Five of six studies have samples of women undergoing chemotherapy and/or radiotherapy, and most examined only one GST gene (usually GSTP1). Four of the six (Ambrosone et al. 2001; Bewick et al. 2008; Lizard-Nacol et al. 1999; Sweeney et al. 2000) were based on small samples of patients (N<100; (Bewick et al. 2008; Lizard-Nacol et al. 1999) N=240-250 (Ambrosone et al. 2001; Sweeney et al. 2000)). One large study of 2430 breast cancer patients was comprised of women with early stage disease (94%) who were unlikely to have undergone chemotherapy, (Goode et al. 2002) and found no association with the only GST examined (GSTP1) and survival. One other large study of 1034 women from Shanghai, China, all treated with adjuvant chemotherapy, found a reduction in risk with the variant GSTP1 Val allele but no association with either GSTT1 or GSTM1 and risk of death (Yang et al. 2005). In two reports based on the same sample, women with breast cancer with null mutations for GSTM1 and GSTT1 had reduced risk of death compared to women with alleles present, (Ambrosone et al. 2001) and a reduction in mortality risk for women homozygous for the variant GSTP1 Val allele compared to those homozygous for the Ile allele (Sweeney et al. 2000; Yang et al. 2005). Finally, 2 other small studies examined associations between one GST polymorphism among women treated with high dose chemotherapy, one reported no association between survival and GSTM1 null; (Lizard-Nacol et al. 1999) another, that the GSTP1 Val/Val polymorphism was non-significantly associated with worse overall survival (Bewick et al. 2008).

Two studies (one in smokers, and the other in patients with diabetes), reported an association between GSTT1 and GSTM1 null mutations and lower levels of the inflammatory biomarker CRP, (Hayek et al. 2006; Miller et al. 2003) itself associated with poor survival (Pierce et al. 2009a). We thus examined this association in the Health, Eating, Activity and Lifestyle (HEAL) study.

We extend prior research by examining the association between three different GST isoenzymes (null mutations in GSTM1 and GSTT1, the Ile¹⁰⁵Val polymorphism in GSTP1), and all-cause and breast-cancer specific mortality in a multi-ethnic cohort of breast cancer survivors drawn from population-based cancer registries. This sample of breast cancer patients diagnosed from 1995–1999 includes a larger number of women undergoing chemotherapy and/or radiotherapy than most prior studies (Ambrosone et al. 2001; Bewick et al. 2008; Lizard-Nacol et al. 1999; Sweeney et al. 2000) and reflects more contemporary therapy regimens than those based on women treated in the mid-1980s-mid-1990s (Ambrosone et al. 2001; Bewick et al. 2008; Lizard-Nacol et al. 1999; Sweeney et al. 2000).

Differences in distribution of continuous variables between genotypes were estimated using analysis of variance (ANOVA). Differences in distributions of categorical variables were compared using the Chi-square test. As the numbers of patients homozygous for the GSTP1 variant allele were few, heterozygous and homozygous variant allele groups were combined (recessive model). Hazard ratios (HR) and 95% confidence intervals (CI) for breast cancer-specific or all-cause mortality were based on the partial likelihood for Cox's proportional hazards model (Cox 1972). The proportional hazard assumption was tested using Schoenfeld residuals, and no violation of the proportionality assumption was found. Age was used as the underlying time variable, with entry and exit time defined as the participant’s age at the baseline interview, and age at death from either breast cancer or any cause, or end of follow-up, respectively.

We based variable inclusion on a likelihood ratio test, with the following covariates included in models: race/ethnicity/study-site (to adjust for different distributions of race/ethnicity by study site); BMI (categorical <18.5 kg/m²; ≥18.5 and <25 kg/m²; ≥25 and <40 kg/m²; ≥40 kg/m²); SEER summary tumor stage (local vs. regional) and treatment received at diagnosis (surgery; surgery+radiotherapy; chemotherapy). Covariates considered but not included in the final model (they did not significantly change the likelihood ratio score): menopausal status, education, smoking status, tamoxifen use, and ER status. The Wald statistic was used to test for trend across levels.

We determined whether the association of GST variants with outcome was the same across subgroup categories, using a test of homogeneity of trends across groups; specifically stage, ER status; and treatment received. Due to small numbers of events in premenopausal participants, we did not compare pre- and postmenopausal subgroups.

All p-values are two-sided. Analyses were performed using STATA 11 (Statacorp, TX USA).

Carriers of the GSTT1 null mutation had significantly higher levels of CRP compared to non-carriers (mean: 4.54 vs. 3.01 mg/L; P=0.01). Levels were also significantly higher in participants with diabetes (mean: 10.23 vs. 2.92 mg/L; P=0.02), but there were no differences in participants without diabetes. In contrast the GSTM1 null mutation was associated with significantly lower levels of CRP among diabetics only (mean: 4.83 vs. 14.61 mg/dL). There were no associations between GSTP1 and CRP (data not shown).

We performed a 3-way gene analysis, examining combinations of GSTM1 null/present; GSTT1 null/present and GSTP1¹⁰⁵Ile/Ile vs. Ile/Val+Val/Val. An increased risk of all-cause mortality was observed for each group relative to the referent (GSTM1 present/GSTT1 present/GSTP1¹⁰⁵Ile/Ile), but none reached statistical significance (data not shown). Breast cancer-specific mortality demonstrated a different pattern, with carriers of GSTT1 null mutation/GSTM1 present/GSTP1¹⁰⁵Ile/Ile genotypes associated with a reduced risk (HR=0.12; 95% CI 0.01-1.16) compared to participants with GSTM1 present/GSTT1 present/homozygous wild-type for GSTP1. However this association was not significant (P=0.06).

We next analyzed the same endpoints for GSTP1, GSTM1 and GSTT1 polymorphisms in patient subgroups, using fully adjusted models. The GSTP1 Val allele was associated with an approximate 2-fold increased risk of all-cause mortality in patients with ER-positive tumors, compared to those with ER-negative, though this was not significant (P=0.08). There was no evidence of effect modification for other subgroups examined. Due to limited power we did not examine associations for breast cancer-specific mortality in these subgroups.

There are few studies on the role of GST isoenzymes on mortality in breast-cancer survivors drawn from community practice. As described earlier, the majority of these studies had small sample sizes, were based on participants diagnosed prior to 1999, and on women undergoing chemotherapy and/or radiotherapy. In addition, most examined only one GST gene (usually GSTP1).

Here we report that the variant Val allele of GSTP1 (rs1695) was associated with increased risk of all-cause mortality, but not breast cancer-specific mortality, in a cohort of 533 breast-cancer survivors. The amino acid substitution in GSTP1 Ile¹⁰⁵Val, results in an enzyme with altered activity, (Ali-Osman et al. 1997) including towards alkylating anticancer agents, (Hayes & Strange 2000; Pandya et al. 2000; Srivastava et al. 1999) and the decreased risk of mortality in carriers of the variant allele who receive chemotherapy may be attributable to longer exposure to the active agent in therapy. Patients homozygous for the variant allele had a lower risk of chemo-resistance when treated with doxorubicin (OR= 0.11; 95% CI 0.01-0.90; P=0.04) (Romero et al. 2012). However, when we examined the association of GSTP1 polymorphisms in subgroups of patients, we found no association between the polymorphism and treatment received.

In contrast, deletions in the GSTM1 or GSTT1 genes were not associated with mortality confirming results from one other study, (Yang et al. 2005) but not in another, (Ambrosone et al. 2001) though the latter studied women recruited between mid 1980s-1990s.

We also found an association between breast cancer-specific mortality with carriers of GSTT1 null mutation/GSTM1 present/GSTP1 Ile/Ile genotypes associated with a reduced risk of breast cancer-specific mortality compared to participants with GSTM1 present/GSTT1 present/homozygous wild-type for GSTP1. While this was not significant, we cannot discount inadequate power.

When we examined associations in patients who received any treatment (chemotherapy/radiotherapy; excluding those who received surgery only), homozygous carriers of the GSTP1 Val allele had a decreased risk of all-cause mortality, although this was not significant, compared to all participants. This is similar to another report in Chinese women who all received adjuvant chemotherapy; (Yang et al. 2005) another reported that the GSTP1 Val/Val polymorphism was non-significantly associated with worse overall survival in women treated with high-dose chemotherapy (Bewick et al. 2008).

We found significantly lower levels of CRP in carriers of the GSTM1 null mutation in patients with diabetes only; and higher levels of CRP in carriers of the GSTT1 null mutation; the latter differed from results in patients with diabetes and smokers (lower CRP among carriers of the GSTT1 null mutation); (Hayek et al. 2006; Miller et al. 2003) however these participants were otherwise healthy.

Limitations of our study include relatively small numbers of events, thus we were not able to evaluate the association with outcome by subgroups such as menopausal status; and three-way gene analyses are underpowered. We also had limited power to examine breast cancer-specific mortality. The cohort was established before some current treatments such as aromatase inhibitors and Her2/neu targeted therapies were available, and therefore we cannot estimate what associations GST isoenzymes might have with survival in women using these treatments. There was also a possible selection bias in this study. As blood for genotyping was obtained approximately 30-months post-diagnosis, and we excluded participants who were under treatment for recurrence, associations with early breast-cancer mortality would not be observed, and it is possible that genotype may have stronger associations closer to the time of diagnosis.

However, our study has a larger sample size than most prior studies examining the association between GST polymorphisms and survival, and participants also underwent more contemporary treatment protocols. Given the heterogeneity of published studies (different therapies, stages of disease, and recruitment periods), suggestions for further study include examining larger studies of pooled data of women diagnosed at similar time-periods who underwent similar treatment regimens, thus enhancing power to detect associations between GST isoenzymes and longer-term survival.