Presented in part at the 39th Annual Meeting of the American Society of Clinical Oncology, Chicago, Illinois, May 31 to June 3, 2003.
Over the past decade, a number of new therapeutic agents have become available in the treatment of metastatic breast cancer (MBC). This study characterized the use and assessed the impact on survival of population-based access to new agents for the treatment of MBC.
The dates of release in British Columbia of 7 new systemic agents for MBC during the 1990s were used to construct 4 time cohorts. All patients with a first diagnosis of distant metastases in each of the time cohorts were identified and characterized, and their survival was compared. Cox proportional regression modeling was used to assess for predictors of survival.
In total, 2150 patients with a first distant metastases diagnosed during 1 of the 4 cohort intervals were identified. Baseline characteristics between cohorts were similar, except a greater proportion of the later cohorts received adjuvant chemotherapy (P < .001), had positive estrogen receptor status (P = .01), and had a longer median time from initial diagnosis to MBC (P < .001). Survival in Cohort 1 (1991–1992) and Cohort 2 (1994–1995; median, 438 days and 450 days, respectively) was similar. Survival was longer in Cohort 3 (1997–1998; median, 564 days; P = .002) and improved further in Cohort 4 (1999–2001; median, 667 days; P = .05). In multivariate analysis, the later cohorts were associated independently with improved survival (P = .01 and P < .001, respectively).
Breast cancer is a major public health issue in industrialized nations: It is the most commonly diagnosed cancer and the second leading cause of cancer deaths in women. In 2006, across North America, close to 50,000 women have died from metastatic breast cancer (MBC).1, 2 Worldwide, it is estimated that >400,000 women will die of this disease.3 Patients with MBC have a median survival of 18 to 24 months.4 Surprisingly, to our knowledge, there have been no randomized controlled trials comparing a palliative, systemic regimen with best supportive care to demonstrate whether systemic treatment improves survival in patients with MBC. Today, it is unlikely that such a trial ever will be performed in this treatable disease.
There is recent epidemiological data to support a significant fall in breast cancer mortality over the past 15 years in both the U.K. and in the U.S.5, 6 Much of this improvement in outcome has been attributed to earlier detection strategies and to the greater use of adjuvant hormone and cytotoxic treatments.6–8 Over the past decade, a number of trials have demonstrated improved survival in MBC when newer hormone agents or chemotherapeutic regimen(s) were compared with a previous standard.9–14 However, it is recognized that patients in clinical trials are a highly selected group and are not necessarily representative of the general population of patients with MBC.
In the current study, from a population-based perspective, we examined temporal trends in the use of hormone and chemotherapeutic agents for the treatment of MBC and the corresponding survival rates. We sought to establish whether population-based access to newer systemic agents for the treatment of MBC improved survival.
MATERIALS AND METHODS
The province of British Columbia (BC) had a population of approximately 4 million with close to 2700 new cases of breast cancer and 630 deaths from breast cancer in 2006. Approximately 75% of all patients with breast cancer in the province are referred to a British Columbia Cancer Agency (BCCA) center. The BCCA has the mandate for cancer control for the entire province. This includes the operation of 4 regional cancer centers, which deliver all radiation therapy in the province, and the management of the provincial budget for all systemic cancer therapies. The BCCA also is responsible for the establishment of provincial, evidence-based guidelines regarding access to systemic agents in the treatment of all solid and hematologic malignancies. A central pharmacy database within the BCCA records the date, drug, and dose of all systemic agents and their indication in the treatment of a specific cancer in every patient. The BCCA manages a Breast Cancer Outcomes Database (BCOD), which contains detailed demographic, pathologic, staging, treatment, and outcome data for women who were diagnosed with breast cancer since January 1, 1989 and were referred to the BCCA. Information regarding the date and site of first local, regional, and distant recurrence is collected prospectively. Dates and causes of death are collected from the provincial death registry.
Based on the timing of the development and dissemination of breast cancer systemic therapy guidelines, we identified 4 time cohorts prior to and after the introduction of various systemic therapy agents used in the treatment of MBC in BC. Any patient who was diagnosed with MBC (either primary presentation or first recurrence of distant metastases) was included in that time cohort. For each individual patient, the diagnosis of a distant metastasis is categorized prospectively in the BCOD as either clinical, radiologic, or pathologic. Other criteria for study inclusion were age ≤75 years at diagnosis of MBC and the exclusion of patients who had another primary malignancy, except nonmelanoma skin cancer or a new primary, contralateral breast cancer. The first cohort, from January 1991 to December 1992, was selected as the baseline comparator. The second cohort, from January 1994 to December 1995, was the period during which paclitaxel and vinorelbine were released in BC for the treatment of MBC. The third cohort, from January 1997 to December 1998, was after the approval of the aromatase inhibitors and docetaxel for MBC. The fourth cohort, from July 1999 to June 2001, identified the period of access to trastuzumab and capecitabine. Patients in any of the cohort intervals identified by the BCOD were linked through the BCCA pharmacy database to collate the chemotherapeutic and/or hormone therapies that were dispensed for each individual. Ethical approval for the study was obtained from the governing research ethics board.
Baseline demographic and pathologic prognostic factors and the individual treatments dispensed for patients in the 4 cohorts were identified and compared between cohorts. An analysis of variance test was used to assess for statistical significance for continuous variables, whereas chi-square analysis was performed for categorical variables. Overall survival was defined as the interval from the date of diagnosis of distant metastasis to death from any cause. All patients who remained alive at the date of last analysis were censored. Kaplan-Meier survival curves within each cohort were calculated and compared with a Tarone-Ware test. A Cox regression model was used for the multivariable analysis. Assumptions of proportionality of hazards were verified by inspection of the log-hazard plots versus time and Schoenfeld residuals of the Cox model for all variables. Statistical analyses were performed with both SPSS software (version 10.1) and SPlus software (version 5).
The data were analyzed by 5 of the authors (S.K.C., C.H.S., Y.D., A.C. and I.A.O.). The article was drafted by S.K.C. and was modified after review by the other coauthors. S.K.C. had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Role of the Funding Source
This study was funded through the operating budget of the British Columbia Breast Cancer Outcomes Unit. There was no involvement of the funding source on study design, data collection, analysis, interpretation, or writing of the article.
In total, 2150 patients who fulfilled the study criteria were identified from the BCOD; 423 patients were identified for Cohort 1 (1991–1992), 561 patients were identified for Cohort 2 (1994–1995), 641 patients were identified for Cohort 3 (1997–1998), and 525 patients were identified for Cohort 4 (1999–2001). The proportion of patients within each cohort who presented with distant metastasis as their initial diagnosis of breast cancer was 27%, 18%, 20%, and 22% for Cohorts 1 through 4, respectively. The proportion of patients who developed distant metastasis within 5 years of their original primary breast cancer diagnosis was 70% for Cohort 1, 68% for Cohort 2, 57% for Cohort 3, and 50% for Cohort 4. At the time of analysis, 93%, 93%, 83%, and 58% of patients had died in Cohorts 1 through 4, respectively. Of the 2150 patients who were included in this study, the method of diagnosis of distant metastasis was clinical in 7.5% of patients, radiologic in 69.2% of patients, pathologic in 22% of patients, and unknown in the remaining 1.3% of patients. The baseline demographic and pathologic factors and the adjuvant therapies received between the cohorts are shown in Table 1. The 4 cohorts were similar in terms of age at diagnosis of distant metastasis, distribution of pathologic prognostic factors of the primary tumor (tumor size, grade, axillary lymph node involvement, and lymphatic or vascular invasion), and sites of first distant metastases (categorized as visceral vs soft tissue and/or bone). There was a difference in the proportions of estrogen receptor (ER) status between the 4 cohorts: Fifty-nine percent of patients in Cohorts 1 and 2 had positive ER status, 63% of patients in Cohort 3 had positive ER status, and 69% of patients in Cohort 4 had positive ER status (P = .01). However, this did not translate into a difference in the proportion of patients who received adjuvant tamoxifen between the cohorts. The use of adjuvant chemotherapy varied between the 4 cohorts. Only 43% of patients in Cohort 1 received any form of adjuvant chemotherapy versus close to 60% of patients in Cohorts 3 and 4 (P < .001). The interval from diagnosis of the primary breast cancer to the time of distant recurrence was significantly longer in the later cohorts (P < .001). This was associated with greater use of adjuvant chemotherapy and a lower percentage of patients presenting with de novo metastases in the later cohorts (20%–22% in Cohorts 3 and 4 vs 27% in Cohort 1; P = .004). Another contributing factor for this difference may have been the shorter interval between the earlier cohorts and inception of the BCOD (1989).
Table 1. Baseline Demographics of the 4 Time Cohorts*
Cohort 1: 1991–1992 (n = 423)
Cohort 2: 1994–1995 (n = 561)
Cohort 3: 1997–1998 (n = 641)
Cohort 4: 1999–2001 (n = 525)
MBC indicates metastatic breast cancer; ER, estrogen receptor; LV, lymphatic and/or vascular invasion.
Note that pathologic prognostic factors are of the primary tumor/lymph node and for patients with known data only.
Age at diagnosis of MBC, y
Age group: No. of patients (%), y
Interval from initial diagnosis to diagnosis of MBC, mo
Size of initial tumor, cm
Adjuvant therapy, %
Lymph node status at initial diagnosis, %
ER status at initial diagnosis, %
Grade at initial diagnosis, %
LV status at initial diagnosis, %
Site of first distant metastases, %
Soft tissue and/or bone
Systemic Therapies Received for MBC
The various chemotherapeutic and hormone agents that were received for the treatment of MBC by patients in each of the 4 time cohorts are outlined in Table 2. Only the agents that were used commonly and within the BCCA guidelines were reviewed. Over time, a greater proportion of patients received some form of palliative systemic therapy for the treatment of MBC. In the time cohort of from 1991 to 1992, 25% of patients did not receive either a palliative hormone or chemotherapeutic agent compared with only 10% in the later cohorts (P < .001). There was no significant difference in either the mean or the median number of treatments delivered between the cohorts. The utilization data demonstrated that, once a new agent was released for use (ie, funded by the BCCA), the newer agent began to be used. For example, just prior to formal funding approval of the nonsteroidal aromatase inhibitors (Cohort 2), this class of hormones was used in only 16% of patients compared with 44% (Cohort 3) after it was approved for use in the BCCA. Another trend was the reduced use of anthracyclines and tamoxifen for MBC in the later cohorts. This was correlated with an increased use of anthracyclines and tamoxifen in the adjuvant setting. The taxane docetaxel, the monoclonal antibody to HER-2 (trastuzumab), and the 5-fluororacil (5-FU) prodrug capecitabine all were used more commonly in the later time cohorts.
Table 2. Chemotherapies and Hormone Therapies Prescribed for the Treatment of Metastatic Breast Cancer for the 4 Time Cohorts
Cohort 1: 1991–1992 (n = 423)
Cohort 2: 1994–1995 (n = 561)
Cohort 3: 1997–1998 (n = 641)
Cohort 4: 1999–2001 (n = 525)
MBC indicates metastatic breast cancer; AI, aromatase inhibitor.
No. of treatments
Hormone therapy, %
Overall Survival Across the Cohorts
There was a significant improvement in survival for women with MBC over time. Although there was no difference between the first 2 cohorts, the later 2 cohorts had significant improvements in overall survival. The median survival for patients in Cohorts 1 through 4 was 436 days, 450 days, 564 days, and 661 days respectively. Tarone-Ware tests showed significantly improved survival for Cohort 3 versus Cohorts 1 and 2 (P = .002) and for Cohort 4 versus cohort 3 (P = .05). Kaplan-Meier survival curves for all 4 cohorts are shown in Figure 1. The 1-year survival rate was 55% for both Cohort 1 and Cohort 2, 64% for Cohort 3, and 71% for Cohort 4. The 2-year survival rate in the first 2 cohorts (1991–1995) was 33% to 34% compared with the corresponding 2-year survival rate of 44% to 45% for the later 2 cohorts (1997–2001). When the analysis was limited to patients who either presented with de novo distant metastases or developed a distant recurrence within 5 years of their initial primary breast cancer diagnosis, the results were similar (data not presented).
Multivariate Analysis of Predictors of Survival
The following variables were included in the prediction model jointly: age at diagnosis of metastasis, grade of the primary tumor (grade 1 and 2 vs grade 3), sites of first metastases (bone/soft tissue vs visceral), ER status (positive, negative, or unknown), use of adjuvant hormone therapy (yes or no), use of adjuvant chemotherapy (yes or no) and the 4 time cohorts. The results are summarized in Table 3. All factors except for adjuvant hormone or chemotherapy were highly significant, independent predictors of survival in patients with MBC. High grade, negative ER status, young age (<35 years), and older age (≥65 years) all were associated with a worse outcome, as expected. The site of distant metastases was significant for survival both as a sole predictor and in a model with the other variables; however, because this factor did not meet the assumption of proportionality, it was used only as a stratum in the final model. In the joint model, time cohort was identified as an independent predictor of survival. There was no significant difference in outcome between Cohort 1 and Cohort 2; however, both Cohort 3 and cohort 4 were associated with improved survival with a hazards ratio of 0.84 (P = .01) and 0.72 (P < .001), respectively.
Table 3. Cox Regression Analysis for Survival in Metastatic Breast Cancer*
Our current population-based study of a large cohort of women with a recent diagnosis of MBC is the first to our knowledge that demonstrates a significant improvement in survival over time. After adjustment for several covariates, the survival of women who were diagnosed with MBC in BC in the later part of the 1990s improved by approximately 30% compared with a similar cohort of women who were diagnosed in the early and middle parts of the 1990s. This improvement in survival was observed even though a greater proportion of women in the later cohorts already had been exposed to prior chemotherapy (in the adjuvant setting). The primary objectives of systemic therapy for women with MBC are to improve/maintain quality of life and, hopefully, to prolong survival. However, we believe there has never been a randomized controlled trial comparing a systemic agent with best supportive care to prove that survival can be improved by the treatment of MBC. In the current era of multiple, active agents available for the treatment of MBC, it is improbable and likely unethical that this type of study ever will be performed. In the early part of the 1990s, 25% of patients in BC did not receive any palliative systemic therapy, compared with only 10% in the later part of the decade. Thus, the current study appears to serve as a surrogate to suggest that systemic therapy for MBC is associated with improved survival.
A recent publication from a single institution also suggested an improvement in survival in recent cohorts of patients who were diagnosed with recurrent breast cancer.15 The authors analyzed the survival of 834 women who previously had received an adjuvant anthracycline regimen at the University of Texas M. D. Anderson Cancer Center and who subsequently developed recurrent breast cancer (local, regional, or distant) between 1974 and 2000. By dividing the period of their study into 5 time cohorts, they demonstrated in univariate analysis that the median survival improved from 15 months to 58 months in a comparison of their earliest cohort (1974–1979) with their later cohort (1995–2000). In multivariate analysis, the year of recurrence was associated with a trend toward improved survival. There are several important differences between the M. D. Anderson Cancer Center study and ours. Our study was larger, it was population-based, it included only patients who had distant metastases, and we linked the systemic treatments received by each patient with the cohort's outcome. A study from the French comprehensive cancer centers demonstrated a virtual doubling of the 5-year survival rate for women with MBC who were diagnosed in the late 1990s compared with the early 1980s, but those investigators also did not link individual treatment data for MBC and outcome.16
Over the past decade, there have been 6 large randomized trials that demonstrated an improvement in overall survival of 1 palliative agent/regimen compared with another systemic agent in the treatment of MBC.9–14 Two of those studies involved the use of the newer class of aromatase inhibitors versus a progestational agent in postmenopausal women with advanced breast cancer.9, 10 Two of the studies compared the taxane docetaxel either with another chemotherapeutic regimen (mitomycin C and vinblastine)11 or with another taxane (paclitaxel).14 The other 2 studies were in patients who received combinations of chemotherapy either with a monoclonal antibody to HER-2 (trastuzumab)12 or with combination docetaxel and an oral 5-FU agent (capecitabine)13 versus the standard chemotherapy alone.
From the results from our current study, we cannot attribute the improved survival conclusively to a single agent or regimen. However, we can state that there were differences in the use of newer agents/regimens over time, and these differences appear to be associated with improved survival. It is noteworthy that the greatest differences in survival were associated with the introduction of the aromatase inhibitors, docetaxel and trastuzumab, in the later 2 cohorts. These data appear to support the notion that widespread access to newer therapies that have demonstrated improved survival in highly selected patients on clinical trials do appear to translate into population-based survival benefits. In addition to potentially improving survival, some of the newer agents also may have less toxicity (eg, aromatase inhibitors compared with progesterone agents). Furthermore, if some of the older agents had greater morbidity that could lead to mortality, then the decreased use of these agents also may have contributed to the improved survival.
One potential confounding factor of this study may be lead-time bias if improved diagnostic testing led to an earlier diagnosis of metastatic disease. We do not believe that such an ascertainment bias would have had any significant impact on the improved survival seen. The follow-up guidelines within the province of BC over the entire period of the study were consistent with the American Society of Clinical Oncology recommendations of clinical follow-up without routine laboratory or radiologic investigations in the absence of clinical suspicion of recurrent disease.17 Our large number of patients, all studied within a recent decade (1990s), also makes it less likely that changes in imaging technology could explain the observed improved survival after 1995. Furthermore, there even was a survival difference over the 4 years between the 1997 to 1998 cohort and the 1999 to 2001 cohort. Other limitations of this study include the unequal duration of follow-up in later cohorts and the increased use of adjuvant chemotherapy in the later cohorts. In attempts to address the issue of unequal duration of follow-up between cohorts, we performed additional analyses in which the cohorts were limited to patients who either presented with de novo distant metastases or developed distant recurrence within 5 years of their initial primary breast cancer diagnosis, and the results were similar to those for the entire original cohort. The observation of increased use of adjuvant chemotherapy in the later cohorts, in fact, could have reduced the effectiveness of palliative systemic therapy in these groups, because a greater proportion were no longer “chemotherapy naive.”
In conclusion, we believe that this is the first study to demonstrate a population-based improvement in survival for women with a recent diagnosis of MBC. This improvement in survival appeared to be caused by the availability and use of newer and more effective systemic agents for the treatment of MBC. Although MBC still is an incurable disease, this study provides optimism for those women who are diagnosed with MBC today and should provide enthusiasm and evidence that continued research to discover better therapies may translate into further improvements in outcome in the future.