Results

Clinicopathological data
Patients (total 92) were characterized by two variables: age and menopausal status at diagnosis. The median age of these patients was 57 years, and status was premenopausal for 51% (n = 47) and postmenopausal for 48.9% (n = 45). All samples were composed of infiltrating ductal carcinoma with an area of intraductual carcinoma in 35 cases. Of the tumours, 33.7% (n = 31) were 2-5cm in size (T2), 27.2% (n = 25) were > 5cm (T3) and 39.1% (n = 36/92) were > 5cm with skin involvement (T4). Of tumours, 60.9% (n = 56) contained significant hormone receptors (HR+), and 39.1% (n = 36) did not (HR-). Clinical complete response was found in 19.6% (n = 18) of cases. A high histological grade (SBR3) was found in 34.8% (n = 32) of tumours. An intermediate level of pathological tumour size (pT2) was found in 42.4% (n = 39) of samples. Negative lymph nodes were found in 20.7% (n = 19) of the tumours, whereas 23.9% (n = 22) of tumours were associated with >1 involved nodes and 55.4% (n = 51/92) were associated with > 3 involved nodes. Mutations in p53 were detected in 30% (n = 28) of the cases studied.

Glutathione S-transferase M1 genotype determination
The PCR method described above allowed an internal standard controlled classification of GSTM1-deficient (GSTM1-null genotype) individuals. Of the 92 patients, 57.6% (n = 53) were classified as heritably GSTM1 deficient, and 42.4% (n = 39) were GSTM1-positive genotype. Paired samples of blood and breast tissue were analysed before treatment with primary chemotherapy from the same individual, and GSTM1 genotype was identical for the two samples (Fig. 1). Among the 39 patients with GSTM1-positive genotype, tissue samples obtained before and after treatment were available from 28 cases, allowing RNA extraction and GSTM1 expression using the RT-PCR method. Two of these patients had bilateral lesions, and measurement was determined in the two tumour localizations. Thus, total GSTM1 expression, as measured by the ratio of GSTM1 to β2- microglobulin values, were performed on 30 tumour specimens. GSTM1 RNA signal was detected in all of the tumours analysed before and after treatment. The median GSTM1 expression was 1.38 (range 0.02-23.27) in the untreated tumours, and 1.16 (range 0.01-6.56) in samples obtained after chemotherapy administration.

Glutathione S-transferase M1 and clinicopathological characteristics of the patients
Distribution of GSTM1 genotype and its relation with clinicopathological data of the patients are shown in Table 1. There were no statistically significant associations between GSTM1-null genotype and the parameters analysed: age, menopausal and hormonal status, clinical and pathological tumour size, grade, involved nodes and p53 gene mutations. GSTM1 expression measured by RT-PCR in 30 samples (corresponding to 28 cases) before and after treatment with primary chemotherapy was also compared with the clinicopathological characteristics of the patients. None of the parameters tested were related to GSTM1 expression determined before or after treatment (data not shown).
Relationship to clinical response to chemotherapy (Table 2) demonstrated that objective response (complete and partial responses) rate of the group with GSTM1-null genotype (75.5%) did not differ from that in those with GSTM1-positive genotype (76.9%). Thus, no significant relation was found between GSTM1 polymorphism and clinical response to chemotherapy (P = 0.8719). Also, no relation was obverved between GSTM1 RNA expression and clinical response to chemotherapy (P = 0.9524 and P = 0.5192 for before and after treatment, respectively). In contrast, clinical tumour size (P = 0.0177) and intraductal carcinoma (P = 0.0013) are strongly associated with clinical response. In multivariate analysis, the clinical tumour size (P = 0.0070, OR = 4.83, 95% CI = 1.45-16.10) and the absence of intraductal carcinoma (P = 0.0002, OR = 14.1, 95% CI = 2.52-78.50) remained the only factors linked to the clinical response.

Impact on survival of the patients
For disease-free survival, no differences were found between individuals with GSTM1-null genotype and those with positive-GSTM1 genotype (P =0.8094). Accordingly, no impact for RNA GSTM1 expression on disease-free survival (P = 0.8991 and P = 0.9096 for before and after treatment, respectively) was observed (Table 2). In contrast, the absence of hormone receptors (P = 0.0020) and the presence of p53 gene mutations (P = 0.0098) had an impact on disease-free survival. With multivariate analysis, hormone receptors status (P = 0.0002, OR=3.99, 95% CI = 1.92-8.29) and p53 gene mutations (P = 0.0138, OR = 2.36, 95% CI=1.22-4.59) remained significantly associated with metastasis recurrence risk.
No impact was also found (P = 0.9729) for GSTM1-null genotype on overall survival (Table 2), or for RT-PCR RNA expression (P = 0.1667 and P = 0.9637 for before and after treatment, respectively). Only the absence of hormone receptors (P = 0.0018), the presence of p53 gene mutations (P = 0.0071) and no response to primary chemotherapy (P = 0.0086) were associated with reduced overall survival of the patients. In multivariate analysis, hormone receptors status (P = 0.0003, OR = 5.23, 95% CI = 2.03-13.49) and p53 gene mutations (P = 0.0037, OR = 3.62, 95% CI = 1.53-8.53) were strongly related to the risk for death. Absence of clinical response to chemotherapy was less related to the overall survival (P = 0.0530, OR = 2.31, 95% CI = 1.02-5.26).