Significant survival improvement of patients with recurrent breast cancer in the periods 2001-2008 vs. 1992-2000

A total of 569 patients who were diagnosed and treated for recurrent breast cancer at the National Kyushu Cancer Center between 1992 and 2008 were eligible for this study. All patients had undergone primary surgery for breast cancer without any evidence of distant metastasis and then were clinically determined to have recurrent breast cancer. The diagnosis of recurrent disease was essentially confirmed by physical examination, X-rays, computer tomography (CT), magnetic resonance imaging (MRI), bone scintigraphy and/or other imaging modalities. A biopsy of metastasis was not essential for the diagnosis of recurrent disease. We excluded 133 (23.3%) cases that lacked either HR or HER-2 expression data, and 29 (5.1%) cases that were not followed-up, or who did not receive treatment after first diagnosis of recurrence. The remaining 407 patients were included in the study. Isolated ipsilateral breast cancer recurrence was excluded from the study because it was difficult to distinguish true recurrence from a new primary lesion. The clinicopathological factors and treatments for primary breast cancer and recurrent disease pertaining to these patients were entered prospectively into the hospital database. Patients were assigned to two cohorts based on the time of their first diagnosis of recurrence. Cohort A included patients diagnosed between 1992 and 2000; Cohort B included patients diagnosed between 2001 and 2008. As outlined above, patients were also assigned to four subgroups according to HR and HER-2 expression. The prognosis of the patients within each group was compared between the two time periods. Survival time was defined as the time from the date of first recurrence to the last follow-up examination, or death. Follow-up was ended in December 2009. The primary aims of this study were to evaluate the association between period of diagnosis of recurrent disease and overall survival (OS). As an exploratory analysis, we also determined the relationship between the survival improvement over time and the breast cancer subtype as defined by HR and HER-2 status.

Because prognostic variables were unevenly divided between the two main cohorts, a multivariate model was created to determine any association between these cohorts and survival rate, after accounting for other prognostic factors. Information regarding the year of first recurrent diagnosis, patient age, primary tumor size, number of positive axillary lymph nodes, HR status, HER-2 status, relapse free interval (RFI), sites of metastases, brain metastasis, adjuvant therapy, and medical treatments for recurrent disease was obtained directly from the database.

Associations between the two cohorts and clinicopathological parameters were analyzed using a Chi-square test for categorical variables, and the Wilcoxon rank sum test for continuous variables. Kaplan-Meier survival curves were plotted, and compared using the log-rank test. Cox proportional hazard models were used for both univariate and multivariate analyses. We analyzed new systemic treatment (AI and/or trastuzumab) for recurrent disease as a confounder in the Cox regression model in order to clarify whether introduction of these new agents was the main contributor to the survival improvement over time. The hazard ratios (95% CI) and P values were reported. All tests were two-sided, and P < 0.05 was considered statistically significant. Statistical analyses were carried out using SPSS II software (Version 11 for Windows; SAS Institute, Tokyo).

Cohort A contained 170 patients and cohort B contained 237 patients. The median observation time was 2.7 years for cohort A and 3.1 years for cohort B. There have been 286 deaths among the 407 patients. The distribution of patient characteristics within cohorts A and B is shown in Table 1. In all patients, HER-2 status was assessed by IHC and 102 patients (25%) were diagnosed with HER-2/c-erb-B2 overexpressing tumors. Two patients were diagnosed as having HER-2-amplified tumors by fluorescence in situ hybridization. The two cohorts were similar in terms of age at diagnosis of recurrence, primary tumor size, HER-2 over-expression, adjuvant treatments, and the site of first recurrence. There were differences in the number of axillary lymph nodes involved (≤ or >3 nodes), HR status, and RFI between the two cohorts. Patients with recurrent disease in cohort B tended to have less lymph node metastases, were more likely to have HR-positive disease, and had a longer disease-free interval compared with those in cohort A. In this study, 3 of the 60 patients with HER-2-positive disease (5%) received trastuzumab and 22 of the 167 patients with HR-positive disease (13%) received aromatase inhibitors as adjuvant therapy in cohort B. In these populations, the effect of these new agents may be attenuated. However, it has been shown that the use of trastuzumab is still effective when continued beyond disease progression [21] and that a change to a different AI is effective after the failure of another AI [22]; therefore, we decided to include these patients in this analysis.

The chemotherapy drugs, endocrine therapy agents, and trastuzumab prescribed for the treatment of recurrent disease within the two cohorts are shown in Table 2. Only the agents that were used commonly and approved in Japan were reviewed. Because AI, trastuzumab, capecitabine and vinorelbine were approved after 2001, and taxanes were approved in the late 1990s, there are significant differences in the frequency of use of these agents between the two cohorts. Twenty-two percent of cohort B, and 2.9% of cohort A, were treated with trastuzumab (P < 0.01). In this analysis, most patients with HER-2-positive tumors received trastuzumab-based treatment beyond the time of disease progression and the median number of trastuzumab treatment was 2.1 (range, 1-5). Fifty-six patients with HER-2-positive tumors received trastuzumab treatment. AIs were given to 55% of cohort B and 11% of cohort A (P < 0.01). One hundred forty-four patients with HR-positive tumors received AI treatment. The taxanes (46% vs. 25%, P < 0.01), capecitabine (42% vs. 7.6%, P < 0.01), and vinorelbine (25% vs. 2.4%, P < 0.01) were used more commonly in cohort B. Another trend was the reduced use of anthracyclines (22% vs. 48%, P < 0.01), tamoxifen (26% vs. 38%, P = 0.01), medroxyprogesterone acetate (15% vs. 47%, P < 0.01) and mitomycin (0.8% vs. 49%, P < 0.01) in cohort B.

In this unadjusted analysis, there was a significant improvement in survival rate for women diagnosed with recurrent breast cancer after 2001. The median survival time increased from 1.7 years in cohort A to 4.2 years in cohort B (Figure 1; P < 0.001). The 3-year survival rate increased from 28% in cohort A to 61% in cohort B, and the 5-year survival rate increased from 12% in cohort A to 41% in cohort B.

Univariate analysis showed that the site of first recurrence, brain metastasis, relapse free interval, hormone status, time period of recurrent breast cancer (pre- or post-2001), T stage, number of involved lymph nodes, adjuvant endocrine therapy, and used of AI and/or trastuzumab for treatment of recurrent disease were all associated with overall survival rates after breast cancer recurrence (Table 3). Because the distribution of patient characteristics differed between the two cohorts, a multivariate analysis was also performed. Both the time period and AI/trastuzumab treatment remained significant prognostic factors (cohort B vs. cohort A: HR = 0.70, P = 0.01; AIs and/or trastuzumab for recurrent disease: yes vs. no: HR = 0.46, P < 0.001). The presence of visceral metastases, brain metastasis, HR status, and T stage and the number of axillary metastases were also found to be significant adverse prognostic factors. HER-2 status was associated with prognosis in the multivariate analysis and adjuvant endocrine therapy was not associated with survival after disease recurrence.

We also performed a subgroup analysis in order to explore the association between the recent improvement in the survival of recurrent breast cancer patients and the breast cancer subtype as determined by HR and HER-2 status. For this purpose, patients were categorized into four subgroups based on their HR and HER-2 expression patterns. There have been 133 deaths among patients of the HR-positive/HER-2-negative subgroup; 34 deaths in the HR-positive/HER-2-positive subgroup; 32 deaths in the HR-negative/HER-2-positive subgroup; and 87 deaths in the HR-negative/HER-2-negative subgroup. The Kaplan-Meier curves illustrating the overall survival rates for each subgroup within the two cohorts are shown in Figure 2. For the HR-positive/HER-2-negative subgroup, the median survival in cohort A was significantly lower than that in cohort B (2.2 vs. 4.5 years, P < 0.001; Figure 2A; n = 205). For the HR-positive/HER-2-positive subgroup, the survival rate within cohort A was significantly lower than in cohort B (2.4 vs. 5.1 years, P = 0.001; Figure 2B; n = 57). For the HR-negative/HER-2-positive subgroup, the median survival rate within cohort A was significantly lower than that in cohort B (1.6 vs. 5.0 years, P < 0.001; Figure 2C; n = 47). In contrast, for the HR-negative/HER-2-negative subgroup, the prognosis remained poor over time and no significant survival improvement was seen between the two cohorts (1.0 vs. 1.4 years, P = 0.18, Figure 2D; n = 98). To clarify the effect of time period and AI/trastuzumab treatment on survival in each subtype, multivariate analyses in the 4 subgroups were performed (Table 4). When AI/trastuzumab and the time period were included in the same model, AI and/or trastuzumab was a significant favorable prognostic factor in the HR-positive/HER-2-negative subgroup (AIs and/or trastuzumab for recurrent disease: yes vs. no: HR = 0.37, P < 0.001) and the recent time period was significantly associated with prolonged survival in the HR-positive/HER-2-positive (cohort B vs. cohort A: HR = 0.39, P = 0.009) and HR-negative/HER-2-positive subgroups (cohort B vs. cohort A: HR = 0.12, P < 0.001). When AI/trastuzumab and the time period were separately analyzed, each factor was a strong prognostic factor in the HR-positive and/or HER-2-positive subtype but not in the HR-negative and HER-2-negative subtype in the presence of other confounders.