|
|
| |||||||||||||||||||||||||||||||||||||||
Background and Methods We examined the effect of adjuvant systemic therapy on survival after breast cancer among the residents of the Canadian province of British Columbia. Data on survival were collected for all women in whom breast cancer was diagnosed in British Columbia during each of three calendar years chosen to represent different province-wide treatment recommendations: 1974, when no adjuvant systemic therapy was recommended; 1980, when adjuvant chemotherapy was recommended only for premenopausal women with node-positive disease; and 1984, when adjuvant chemotherapy was also recommended for premenopausal women with node-negative disease and lymphatic, vascular, or neural invasion and tamoxifen was recommended for postmenopausal women with involved lymph nodes or lymphatic, vascular, or neural invasion unless their tumors were negative for estrogen receptors.
Results For women less than 50 years of age, disease-specific survival at seven years (i.e., with censoring of data on women who died from causes other than breast cancer) improved from 65.2 to 76.3 percent between 1974 and 1984 (P = 0.008), and overall survival improved from 64.8 to 74.6 percent (P = 0.02). For women from 50 through 89 years of age, disease-specific survival at seven years improved from 62.5 to 70.4 percent between 1980 and 1984 (P = 0.001), and overall survival improved from 53.9 to 58.3 percent (P = 0.05). The timing of the improvements in survival correlated with the introduction of adjuvant systemic therapy in each group.
Conclusions Survival among women with breast cancer improved significantly in a geographically defined population during the period when adjuvant systemic therapy became widely used.
In British Columbia, Canada (population in 1991, 3.3 million), there is a single centralized cancer agency that attempts to develop consensus, circulate province-wide recommendations for the treatment of cancer, and coordinate cancer-treatment services in the province. This system makes it possible to obtain information on such questions as when adjuvant systemic therapy for breast cancer was introduced and to determine what effect changes in treatment may have had on survival among women with breast cancer. In addition, a comprehensive provincial medical insurance plan guarantees that the inability to pay is not an impediment to treatment.
To determine whether the rate of survival among all women in the province in whom breast cancer was diagnosed had changed since the introduction of widespread adjuvant systemic therapy, we reviewed previous policy statements and then selected three cohorts, according to year of diagnosis, whom we studied to determine the effect of various treatment policies. We assessed the results of recommendations for the use of adjuvant systemic therapy within the province; we were not testing a specific treatment regimen.
Methods
Collection of Data
Data were obtained from the provincial cancer registry6 on all cases of breast cancer in women that were reported during the calendar years 1974, 1980, and 1984. Additional information was sought from the records of the cancer-treatment clinics, from the women's original physicians, and from other sources for women not known to have died for whom less than five years of follow-up data were available. The minimal information in all cases included the patient's name, date of birth, and date of diagnosis, the anatomical site, and the histologic findings. The registry does not record the stage of disease, so it was not possible to determine directly whether the distribution of stages was constant throughout the study period. The date of death and the principal and underlying causes of death were determined for each woman who had died. All deaths before January 1, 1992, were included in this analysis. A woman was considered to have died of breast cancer if the records of the cancer-treatment clinic or the death certificate listed breast cancer as the principal or underlying cause of death.
Selection of the Cohorts
The British Columbia Cancer Institute was established in 1939. In 1974 legislation was enacted creating a single provincial cancer authority. This organization has two treatment clinics in Vancouver and Victoria, provides all radiation-therapy services, and pays for the costs of all antineoplastic drugs used in therapy in the province, thus effectively controlling their use. Multidisciplinary panels regularly review and publish recommendations for treatment that are circulated to all physicians in the province.
Using policy manuals, archival memorandums, and minutes of meetings related to policies regarding systemic therapy from 1974 to 1985, we determined that adjuvant chemotherapy was first recommended for premenopausal women in late 1976. A woman was defined as premenopausal if she had menstruated in the 12 months before the diagnosis of breast cancer or, in the case of a woman who had undergone hysterectomy without oophorectomy, if she was younger than 50 years of age or had a serum follicle-stimulating hormone concentration in the premenopausal range.
Initially, adjuvant chemotherapy with the regimen of Bonadonna et al.,7 using cyclophosphamide, methotrexate, and fluorouracil (CMF), was offered to premenopausal women with newly diagnosed breast cancer who had pathologically involved lymph nodes. By the middle of 1977, cyclophosphamide (800 mg per square meter of body-surface area intravenously on day 1) was substituted for the oral doses. In 1981, the duration of chemotherapy was reduced to 6 months, with all three drugs given intravenously on day 1 of a 21-day cycle (600 mg of cyclophosphamide per square meter, 40 mg of methotrexate per square meter, and 600 mg of fluorouracil per square meter). Beginning in 1979, anthracycline-based chemotherapy was recommended for women with inflammatory breast cancer and for some patients with clinical stage III disease; by 1983 this recommendation had evolved into a specific neoadjuvant-therapy protocol8.
In July 1981 a policy of offering adjuvant tamoxifen therapy (20 mg per day for two years) was introduced for postmenopausal women with levels of estrogen receptors greater than 9 fmol per milligram of cytosol protein9 or with unknown estrogen-receptor levels and a high risk of relapse.
The definition of high risk was initially restricted to women with pathologically involved lymph nodes. Since 1982, women with negative nodes were also included if lymphatic, vascular, or neural invasion was seen in the primary tumor10. In all cases referred to the Vancouver clinic, the pathological findings were reviewed before treatment was planned.
A potential confounding influence affecting the interpretation of the data on survival might be an increase in the number of early cases detected by mammography or a greater awareness of breast cancer on the part of women or physicians2. This potential source of bias was considered when the cohorts were chosen for study. Figure 1 shows the annual incidence of breast cancer in British Columbia from 1971 through 1991. The rise after 1984 parallels but starts a few years later than a similar trend attributed to the increased use of screening mammography in the United States2,11,12. In 1976, 26 dedicated film-mammography units were installed in hospitals throughout the province, and indications for their use were circulated to all physicians13. Data from the provincial health insurance plan and the Screening Mammography Program14 demonstrate an increase of more than 15-fold in the use of mammography for diagnosis, screening, follow-up, and other reasons between the fiscal years 1980-81 and 1990-91. Figure 2 shows that 90 percent of this increase occurred after 1984. We chose 1984 as the year of the most recent cohort since it predated any measurable increase in the annual incidence of breast cancer that could be attributed to earlier diagnosis. In a further effort to reduce a possible effect of lead-time bias (improvement in survival due to earlier diagnosis), all cases of noninvasive breast cancer were eliminated from the analysis.
|
|
The Study Population
The female population of British Columbia was 1,184,800 in 1974, increasing to 1,338,700 in 1980 and to 1,445,300 in 198415. The median age of the female population increased from 31 to 33 years between 1974 and 1984. During the three calendar years selected for study, 3722 women younger than 90 years of age received a new diagnosis of breast cancer that was recorded in the provincial registry. Table 1 and Table 2 show the number of cases included in our study among women less than 50 years of age and from 50 through 89 years of age. Women 90 years of age or older at diagnosis (18, 36, and 28 women in 1974, 1980, and 1984, respectively) and women with noninvasive and nonepithelial cancers, those with cancer first diagnosed at the time of death, and those who died within 30 days after the diagnosis were excluded, as were the women (3.8 percent) who could not be traced beyond the date of diagnosis. Thus, a total of 3284 of the original 3722 women (88.2 percent) whose diagnoses were registered in the three calendar years were available for analysis. Only 49 women (1.5 percent), 11 of whom were younger than 50 years of age, were lost to follow-up before five years of follow-up were completed. These women were included in the analysis with follow-up data censored at the date of the last contact.
|
|
Statistical Analysis
Actuarial overall survival was calculated from the date of diagnosis. For the calculation of actuarial disease-specific survival, data were censored at the time of death if breast cancer was not the primary or underlying cause of death. Three women younger than 50 years of age (2 from the 1980 cohort and 1 from the 1984 cohort) and 18 women 50 through 89 years of age (6 from the 1974 cohort, 3 from the 1980 cohort, and 9 from the 1984 cohort) had a known date of death but an unknown cause of death. They were excluded from the analysis of disease-specific survival but were included in the analysis of overall survival.
Kaplan-Meier estimates of survival16 were calculated for each cohort. The difference between each pair of cohorts was tested in a two-tailed test using the log-rank statistic17.
Results
Outcome
Women Less Than 50 Years Old at Diagnosis
The disease-specific survival, regardless of the disease stage or the treatment delivered, among the women less than 50 years of age at the first diagnosis of breast cancer is shown in Figure 3A. Complete data on mortality are available through the end of 1991; thus, we have reported the actuarial overall and disease-specific survival at seven years (Table 3 and Table 4).
|
|
|
Women 50 through 89 Years Old at Diagnosis
The disease-specific survival rates among women 50 through 89 years of age are shown in Figure 3B. The curves for the 1974 and 1980 cohorts are almost identical. However, the seven-year disease-specific survival improved between the 1974 and 1984 cohorts, from 63.2 percent (95 percent confidence interval, 59.6 to 66.9 percent) to 70.4 percent (95 percent confidence interval, 67.5 to 73.4 percent; P = 0.01) (Table 3). Also, between the 1980 and 1984 cohorts it improved from 62.5 percent (95 percent confidence interval, 59.0 to 65.9 percent) to 70.4 percent (95 percent confidence interval, 67.5 to 73.4 percent; P = 0.001). There was thus a 20 percent reduction in the chance of dying from breast cancer between the 1974 and 1980 cohorts and the 1984 cohort.
The actuarial overall survival seven years after diagnosis for the 1980 cohort (53.9 percent [95 percent confidence interval, 50.5 to 57.3 percent]) was slightly worse than that for the 1974 cohort (56.0 percent [95 percent confidence interval, 52.3 to 59.6 percent; P not significant]) but improved to 58.3 percent (95 percent confidence interval, 55.2 to 61.4 percent) for the 1984 cohort (P = 0.05) (Table 4). A woman with breast cancer diagnosed in 1984 was 10 percent less likely to die of any cause within seven years than a woman with disease diagnosed in 1980.
Policy Changes and Survival
When one compares the two panels in Figure 3, it is clear that the improvements in survival for each age group occurred at different times and coincided with the introduction of policies that led to the widespread use of adjuvant systemic therapy. There was no improvement in disease-specific survival among women 50 to 89 years of age until the 1984 cohort. In contrast, a trend toward improved disease-specific survival among women less than 50 years of age occurred earlier, between 1974 and 1980, and became statistically significant with the 1984 cohort. The consistency of the survival rates among the older group in the 1974 and 1980 cohorts (who were treated with local-regional therapy alone) suggests that data collection by the registry and the types of cases diagnosed were essentially unchanged during this period.
Discussion
Significant improvements in survival occurred between 1974 and 1984 among women in British Columbia in whom breast cancer was newly diagnosed. Although this is a retrospective, observational study, we believe this improvement can be attributed to the use of adjuvant systemic therapy for selected women with a high risk of relapse and death from breast cancer.
The improved survival among these women during this period is particularly noteworthy because it was observed in an unselected population, rather than in selected subgroups or subgroups treated with particular regimens. All registered cases of breast cancer in women, regardless of the disease stage or the treatment actually received, in a large, well-defined geographic area have been included. Exclusions were restricted to women 90 years of age or older, those in whom the cancer was first reported at the time of death, those with noninvasive or nonepithelial histologic features, and those (3.8 percent) for whom no follow-up data were available after the date of diagnosis.
A 32 percent reduction in the chance of dying from breast cancer and a 28 percent reduction in the chance of dying from any cause within seven years of diagnosis were observed among women younger than 50 years of age. The magnitude of these reductions is consistent with the findings of the Early Breast Cancer Trialists' Collaborative Group (EBCTCG), which reported a reduction of 24 ±5 percent in the odds of dying from any cause within 10 years among women less than 50 years of age who had stage I or II disease that was treated with prolonged adjuvant combination chemotherapy in worldwide randomized trials initiated before 19851.
Among women 50 through 89 years of age, there was a 20 percent reduction in the chance of dying from breast cancer and a 10 percent reduction in the chance of death from any cause within seven years after diagnosis (in the 1984 cohort as compared with the 1980 cohort). The EBCTCG reported a 23 ±3 percent reduction in deaths from all causes among women 50 years of age or older who were randomly assigned to at least two years of tamoxifen therapy, as compared with the control groups1. The apparently lesser efficacy of tamoxifen in our study may partly be explained by our inclusion of women up to 89 years of age. Most randomized trials initiated before 1985 included only women who were younger than 70 years of age at the time of diagnosis. Women older than 70 are at higher risk than younger women for death from causes other than breast cancer. In addition, in 1984, tamoxifen was not offered to the 25 to 30 percent of postmenopausal women who were estrogen-receptor-negative,9,18 even if they were considered to have a high risk of relapse. Other women who were estrogen-receptor-positive, and who today would receive tamoxifen, were also not offered adjuvant systemic therapy because of the more restrictive definition of high risk used in 1984.
The most compelling evidence that the observed improvements in survival can be attributed to the difference in treatment is that the improvements in survival coincided with the introduction of recommendations that adjuvant systemic therapy be used in each age group. Disease-specific survival among women 50 through 89 years of age was virtually identical in the 1974 and 1980 cohorts but improved after tamoxifen began to be used as adjuvant therapy for high-risk postmenopausal patients. In contrast, among women younger than 50 years of age, disease-specific survival began to improve between 1974 and 1980, with a further increase observed by 1984. The decision to offer adjuvant chemotherapy to premenopausal women with pathologically involved lymph nodes was implemented in 1976, and the indications were broadened in 1982 to include women with negative nodes if lymphatic, vascular, or neural invasion was present in the primary tumor.
An alternative hypothesis that might explain the observed improvements in survival would be an increase in the use of mammography or a lower threshold for investigating breast abnormalities. We do not have information on disease stages for the women who were not referred to one of the treatment clinics, so we are not able to rule out the possibility that earlier diagnosis might have contributed to a stage drift over the period covered by this study. Such changes might have resulted in earlier diagnosis of cases, with a resulting increase in the annual incidence of breast cancer and an apparent improvement in survival due to lead-time bias2,11,12.
In an attempt to minimize lead-time bias in this population, we chose the most recent cohort after reviewing trends in the province-wide incidence of breast cancer and the use of mammography and by excluding cases of noninvasive breast cancer from our analysis. Increases in the incidence of breast cancer in British Columbia, starting in 1985 among women 50 years of age or older, and starting in 1988 among women 20 through 49 years of age, parallel similar trends attributed to screening mammography in the United States11,12. Although we cannot exclude a limited effect of earlier diagnosis, especially among women older than 50, lead-time bias is unlikely to have substantially altered survival patterns in the population we studied until after the 1984 cohort was enrolled.
Another possible explanation for the improvements in the more recent cohorts could be improvements in the treatment of women with established metastatic disease. Previous reports have shown either no survival advantage19 or a short-term improvement in survival from the date of the first metastasis, but no difference in survival from the date of diagnosis20 with the addition of multidrug chemotherapy to the treatment of women with metastatic breast cancer. Altered therapy for metastatic disease is thus unlikely to explain the observed improvements in survival.
In conclusion, we have demonstrated a modest but probably important improvement in survival between 1974 and 1984 among women in British Columbia with breast cancer. The magnitude of this benefit in women younger than 50 years of age at the time of diagnosis is consistent with that seen in the EBCTCG overview1. The occurrence of such an improvement in survival throughout a geographically defined population was correlated with the introduction of policies that broadened the use of adjuvant systemic therapy for breast cancer.
We are indebted to Eunice Dietrich, former head of the British Columbia Cancer Registry, and to Linda Lee for assistance in compiling this data; to Ferguson Neville, Jeanette Matheson, and Ramona Bianco for the preparation of the figures and tables; and to Drs. Mark Elwood and Martin Schechter for their helpful suggestions about the manuscript.
Source Information
From the Divisions of Radiation Oncology (I.A.O., S.M.J.), Epidemiology (C.D.B.), Medical Oncology (I.H.P., C.M.C., K.A.G., J.R.), and Laboratory Medicine (A.W.), British Columbia Cancer Agency and the University of British Columbia, Vancouver; and the Division of Medical Oncology, British Columbia Cancer Agency, Victoria, British Columbia (K.S.W.) -- both in Canada.
Address reprint requests to Dr. Olivotto at the British Columbia Cancer Agency, 600 W. 10th Ave., Vancouver, BC V5Z 4E6, Canada.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Adjuvant Systemic Therapy and Survival after Breast Cancer
Sankaranarayanan R., Black R. J., Evans R. A., Olivotto I.A., Bajdik C.D., Coppin C.M.
Extract |
Full Text
N Engl J Med 1994;
331:402-404, Aug 11, 1994.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
