Benefits of High-Dose Chemotherapy in Women with Metastatic Breast Cancer: Is it all Selection Bias?


Synopsis: The eligibility requirements for women with metastatic breast cancer are appropriately selective. However, when those same criteria are applied to a series of women with metastatic disease treated with conventional dose chemotherapy, the apparent advantages for high-dose therapy decrease. Randomized studies might address the question but are difficult to administer.

Sources: Rahman ZU, et al. J Clin Oncol 1997;15: 3171-3177; Garcia-Carbonera R, et al. J Clin Oncol 1997;15:3178-3184.

Most of the data relating to high-dose chemotherapy (HDCT) for metastatic breast cancer has come from multiple phase II trials. The results of the HDCT studies have been compared to historical trials and in many cases appear to be superior. This has led to the acceptance of HDCT as a reasonable treatment option for women with metastatic breast cancer. Although we should already have learned our lessons regarding the unreliability of historical controls, two recent reports raise additional concerns about such an approach. Both studies raise the possibility that patient selection, and not superior treatment outcome, can be responsible for the improved response rates and survival observed when one phase II trial is compared with another, or with historical controls.

Rahman and colleagues at M.D. Anderson used their current criteria for selection of patients for HDCT (CR or PR to standard chemotherapy, age < 60 years, performance status < 2, bilirubin < 2 mg/dL, WBC > 2000/mcL, platelets 100,000/mcL, and absence of symptomatic cardiac dysfunction) to determine retrospectively how many of the 1581 women treated on 18 different protocols of standard therapy would have been eligible to receive HDCT and how they fared with standard dose chemotherapy. The 645 women who fit these criteria (candidates) had a higher response rate (both partial and complete) and longer progression-free and overall survival following conventional dose treatment than the patients receiving identical therapy who did not meet these criteria (noncandidates). The CR rate was 27% for candidates and 7% for noncandidates; median progression-free survival was 16 and eight months, and median overall survival was 30 and 17 months, respectively. The survival at five years and 10 years was 21% and 6% for HDCT candidates, and 7% and 2% for noncandidates, respectively. Thus, the major advantage for HDCT candidates was noted in five-year overall survival.

Garcia-Carbonera and colleagues addressed the issue of patient selection in 265 women with breast cancer and 10 positive lymph nodes. One hundred seventy-one women received conventional chemotherapy alone. Of these, 128 met their standard selection criteria for HDCT, (e.g., age < 60 years, no significant comorbid disease, and no progression during adjuvant therapy). Analyzing the entire group of women after a median follow-up of 4.4 years, the estimated disease-free survival was 32.3%, and overall survival was 49.4%. Both five-year disease-free and overall survival were significantly higher for patients who met HDCT criteria, 36.6% and 55.4%, than for patients who did not, 15.8% and 22.7%, respectively. When the 128 women who met HDCT criteria but received conventional dose treatment were compared to a cohort of 39 patients who actually received HDCT, there were no significant differences in disease-free or overall survival between the two groups. Thus it appears that meeting high-dose inclusion criteria is an independent indicator of favorable prognosis for both patients with high-risk or metastatic breast cancer.


A dose-response effect exists for most cytotoxic agents and dose intensification has been shown to improve the activity of chemotherapy in patients with breast cancer.1 Major improvements in supportive care (e.g., growth factors, stem cell technology, and antibiotics) and the large number of well-trained physicians have made it possible to perform HDCT in many community hospitals. We no longer need to ask the question "Can we?" However, we most certainly need to ask the question "Should we?" Breast cancer has become the number one diagnosis for which HDCT and hematopoietic support is used in the United States. Most troublesome is the fact that the overwhelming majority of these patients are treated outside a clinical trial. Young patients with metastatic breast cancer are routinely referred for HDCT even though there is only one randomized clinical trial (with significant flaws) that shows any significant survival advantage following HDCT.2 Patients with more than 10 involved axillary nodes also are routinely offered HDCT because of a single, non-randomized trial that shows a decreased frequency of relapse for transplanted patients compared to historical controls treated with standard chemotherapy.3 This is despite the ongoing national randomized trial design to compare HDCT to standard adjuvant therapy because we don’t know which treatment is better.

To complicate things further, there is now a randomized trial in women with 4-9 positive lymph nodes. Women in this situation are also being offered HDCT "off-protocol." For some reason, the tacit assumption is that HDCT must be better and both women and their physicians feel that the patient is being "deprived" of the best therapy if they must participate in a randomized trial where one of the arms is that awful "standard therapy." To what do we owe this phenomenon? This isn’t the first time that the technical ability to "do something" altered the way we practice medicine before we determined whether that procedure actually was beneficial.

But who can blame women or their physicians for their leap of faith to HDCT? There are a number of phase II studies showing that HDCT will result in an overall response rate of approximately 80% and a complete response rate of 5-15% in women with metastatic breast cancer. In some studies, there appears to be an improvement in the progression-free survival for patients who receive HDCT compared to conventional dose chemotherapy, and there is a suggestion of an improvement in the number of long-term disease-free survivors as well. There is no conventional dose therapy that produces a substantial number of long-term survivors among women with metastatic disease. Since we are all desperate for improved treatments for this disease, we have accepted these results as progress, and HDCT has been widely adopted as a new standard for the treatment of young women with metastatic breast cancer in many communities.

The two articles presented above suggest an alternative explanation for these benefits—cherry picking. The best patients are selected for HDCT and compared to a less favorable group of patients who received conventional therapy. It is not hard to figure out how this comparison will turn out. John Crowley, the director of the Southwest Oncology Group (SWOG) statistical center, illustrated the fallacy of this type of analysis in a presentation at a SWOG meeting a few years ago. He looked at the results of phase II studies of HDCT for multiple myeloma and showed the significant improvement in response rates and disease-free and overall survival seen with HDCT compared to the results obtained with conventional therapy over the previous two decades. However, when he performed a reassessment of the historical data after serially eliminating patients (one prognostic factor at a time) from the historical controls who would not be eligible for HDCT because of age, performance status, renal function, bone marrow compromise etc., he noted that there no longer was a significant benefit for the HDCT.

Some new treatments are so superior to the older standard treatment (combination chemotherapy vs single agent chemotherapy for Hodgkin’s disease comes to mind) that randomized trials are not necessary. Unfortunately, it is not entirely clear that HDCT for breast cancer is one of these treatments. At the crux of the matter is whether anyone with metastatic breast cancer is actually cured by HDCT. If so, then it becomes extremely difficult to deny patients even a small chance at cure when all other treatment approaches are clearly palliative. Despite the expense to society from using a costly treatment approach in 100 women to cure 10 or fewer, we feel compelled to use a potentially curative therapy. We certainly need additional information. We should all be a bit embarrassed by the lack of a definitive answer so many years after the pilot testing of this intervention and for our lack of support for the ongoing trials that might provide that answer.


1. Hryniuk W, Bush H. J Clin Oncol 1984;2:1281-1288.

2. Bezwoda WR, et al. J Clin Oncol 1995;13: 2483-2489.

3. Peters WP, et al. J Clin Oncol 1993;11:1132-1143.