Despite improvements in breast cancer survival over the past decade or so, the Surveillance, Epidemiology, and End Results (SEER) program, reports that 3.5% of the 200,000 women with newly diagnosed breast cancer in the United States present with stage IV disease, and have an intact primary tumor. This amounts to almost 7,000 women annually, for whom the traditional approach has been systemic therapy. Oncologic dogma has so far dictated that local treatment should be recommended only if the primary tumor is symptomatic, and that therapy directed to the primary tumor will not change the course of the distant disease.
In 2001, provocative data from the Southwest Oncology Group challenged this concept in a trial that examined the benefit of nephrectomy in patients with stage IV renal cell carcinoma, by randomizing patients to interleukin therapy with or without nephrectomy. An improvement in survival for the nephrectomy group was observed, with overall median survival increasing from 8 to 11 months, raising the possibility that aggressive local therapy may contribute to prolongation of survival in other tumors.
In the area of breast cancer, retrospective data have recently accumulated regarding the frequency with which women presenting with stage IV disease and an intact primary receive surgical resection of the primary tumor, and whether such ablation has any impact on survival. As reported in three separate retrospective studies in the past 4 years, substantial numbers of these women undergo surgical extirpation of the primary tumor (Table 1). These studies are based on data from the National Cancer Database (NCDB), representing a varied hospital setting in the United States,[2,3] a European tumor registry, and a large comprehensive cancer center. They show that 30% to 55% of women with de novo stage IV breast cancer undergo surgical therapy for the primary tumor. Data on the use of follow-up radiation therapy in this setting are very sparse. Two other studies have been reported in abstract form: one based on SEER data, and the other from the breast cancer database of the Baylor College of Medicine.[6,7]
These studies, summarized below, are remarkably consistent in the finding that women who undergo surgical therapy for the primary tumor in the setting of metastatic disease fare better than those whose primary tumor is left intact. However, the entire published literature on this topic consists of retrospective reviews, all subject to the same selection bias—ie, healthier women with lower disease burden were offered surgery. Despite the reassuring concordance in the studies reported so far, no retrospective study will settle this issue, and definitive proof of this concept awaits a randomized clinical trial testing systemic therapy alone against systemic therapy plus local therapy to the primary tumor.
National Cancer Database Study
The first formal analysis of local therapy in women with de novo metastatic disease was performed by Khan et al utilizing data from the National Cancer Database (NCDB) of the American College of Surgeons. They examined the current patterns of surgical care for women with stage IV breast cancer at initial presentation, analyzing the relationship between the use of surgical resection of the primary tumor and duration of survival.
The NCDB is a joint project of the Commission on Cancer of the American College of Surgeons and the American Cancer Society, providing information about cancer management and outcomes. In addition to demographics (age, race, etc), recorded tumor characteristics include primary site, histology, grade, clinical and pathologic AJCC (American Joint Commission on Cancer) stage, date and type of recurrence, and sites of distant metastasis. Treatment information includes the use of surgery, radiation, chemotherapy, hormonal agents, biologic modifiers, and other approaches; and follow-up data provide last contact date, vital status, and tumor status at last follow-up.
During the years examined in this study, the NCDB did not include detailed information regarding the type, dose, or duration of chemotherapeutic agents, although nonsurgical therapy was recorded as part of the primary treatment. Additionally, although the use of radiotherapy was recorded, it was not possible to determine whether this was directed at the primary tumor or at a metastatic site such as bone.
Over the 4-year period from 1990 to 1993, 16,024 women were identified (mean age: 62.5 years) who presented with stage IV breast cancer (4.1% of the total number of breast cancer cases accrued during this period). AJCC (5th ed) staging information was fairly complete for these women: The size of the primary tumor was unknown in 11.2% of women; the remainder was evenly divided between T1/2 (45.7%), and T3/4 (43.1%). Pathologic nodal status was unknown in a majority (63%) of patients. When reported, it was N0 in 7.1% of the total population, N1 in 20.8%, N2 in 8.0%, and N3 in 1.1%.
Of the 16,024 women identified, 6,861 (42.8%) received no surgical resection of the primary tumor other than a breast biopsy, and 9,162 (57.2%) underwent surgical resection of the primary tumor. Therapeutic breast procedures consisted of partial mastectomy in 3,513 (38.3%), and some form of total mastectomy in 5,649 (61.7%). Surgical margin data were available on 5,957 patients (69.5%) of the surgical therapy group. The free-margin rate was 45% in women undergoing partial mastectomy (913/2,025) and was significantly higher (63%) in the total mastectomy group: 2,483/3,932. Women undergoing total mastectomy were also much more likely to undergo nodal dissection, which was performed in 78.5% of the total mastectomy group and 18.9% of the partial mastectomy group.
Women who had single-organ involvement were significantly more likely to receive surgery than those with multiorgan involvement (P < .0001), as were those with soft-tissue/bone metastases rather than visceral metastases (see Figure 1). The majority of patients received systemic therapy: endocrine therapy alone (6,880 patients) and chemotherapy, either alone or in combination with other treatment modalities, in 7,779 patients.
Analysis of these data (with death from any cause as the endpoint) revealed a 3-year survival rate of 24.9% for the entire group. The mean survival duration was 19.3 months for the no-surgery group, 26.9 months for the partial mastectomy group, and 31.9 months for the total mastectomy group. Corresponding 3-year survival rates were 17.3%, 27.7%, and 31.8%, respectively. Further analysis of the benefits of total mastectomy over partial mastectomy demonstrated that the strength of the survival benefit experienced by surgically treated patients was entirely explained by the status of the pathologic resection margins. When margins of resection were involved, the 3-year survival rate was equivalent, regardless of the type of resection (26.4% and 26.1 %). When margins were free, the 3-year survival rates were also equivalent—34.7% for partial, and 35.7% for total mastectomy (see Figures 2A and 2B).
In the final model, therefore, the data for the partial and total mastectomy groups were collapsed by margin status, and the no-surgery group was compared to those who underwent surgery with involved margins and those who had surgery with free margins. A multivariate Cox proportional hazards model showed that in addition to surgical resection, three other factors were independently and significantly associated with survival: the administration of systemic therapy, the number of metastatic sites, and the type of metastatic disease. The adjusted hazard of death was significantly lower in the surgical than in the nonsurgical group: If margins were free, the hazard ratio (HR) was 0.61 (95% confidence interval [CI] = 0.58-0.65), and if margins were involved, the HR was 0.75 (95% CI = 0.71-0.79). In this model, the type of surgery (partial or total mastectomy), tumor size (AJCC T), the extent of surgical dissection of regional lymph nodes, or the number of involved axillary nodes were not independently predictive of survival.
Subsequently, the same group of investigators examined NCDB data for the 2-year period from 1994 to 1995, with identical findings. In this analysis, the HR was also adjusted for age, with no substantial differences in the results.