The National Cancer Institute (NCI) has recognized the importance of establishing the role of high-dose chemotherapy for patients with breast cancer, both in the metastatic and adjuvant settings. Three large randomized trials have been selected by the NCI for its High-Priority Clinical Trials Program: Philadelphia Bone Marrow Transplant Group (PB)-01, "Maintenance Chemotherapy vs. High-Dose Chemotherapy with Transplant for Metastatic Breast Cancer"; Intergroup (INT)-0121 (Eastern Cooperative Oncology Group [ECOG] 2190), "Evaluation of High-Dose Consolidation Chemotherapy with Autologous Bone Marrow Transplantation for Stage II or Stage IIIA Breast Cancer"; and INT-0163 (Cancer and Leukemia Group B [CALGB]-9082), "High-Dose Chemotherapy with Stem Cell Support vs. Lower-Dose Chemotherapy for Stage II or Stage IIIA Breast Cancer." These are among the first clinical trials to directly compare high-dose chemotherapy with conventional treatment in similar patient groups. Accrual to these important studies has been slower than anticipated, partly because some patients and physicians believe that high-dose chemotherapy provides the best outcome [8-10].

Background

In patients with metastatic breast cancer who have had minimal previous therapy and respond to standard-dose induction chemotherapy, phase I and II trials have reported response rates with high-dose chemotherapy that are higher than those generally seen with standard-dose chemotherapy [5]. Studies with the longest follow-up also suggest that a few patients who receive high-dose chemotherapy, especially patients who have a complete clinical remission after receiving such therapy, achieve prolonged progression-free survival. Two-year progression-free survival has been reported in 14% to 27% of these patients [11-18]. Critics have suggested that similar results could be attained with standard-dose chemotherapy if the strict patient-selection criteria mandated in high-dose chemotherapy trials had been applied to historical controls derived from clinical trials of standard-dose chemotherapy [19, 20]. Clinical trials evaluating standard-dose chemotherapy usually did not require measurement of creatinine clearance or cardiac ejection fraction, pulmonary function tests (including tests for diffusing capacity), computed tomography of the brain, or examination of bone marrow before enrollment. In addition, older patients ( 75 years of age or no limit) and patients with suboptimal performance on electrocardiography were eligible for the standard-dose chemotherapy trials in the metastatic disease setting. However, patients with metastatic breast cancer who were being considered for the high-dose chemotherapy protocols were required to have objective evidence of good organ function (for example, creatinine clearance >1.00 mL/s, cardiac ejection fraction >50%, and pulmonary function test results >50% of predicted values), a bone marrow examination showing no evidence of tumor cells, and approval from their insurance companies.

The most important prognostic factor in primary breast cancer is the status of the axillary lymph nodes. Many studies have confirmed that the number of axillary lymph nodes affected by metastatic cancer independently predicts disease recurrence and overall survival [21-25]. Considered as a group, women whose axillary lymph nodes are affected have a 45% to 60% risk for relapse at 5 years and a 75% to 80% risk for relapse at 10 years [21-26]. Their overall survival is about 60% at 5 years and less than 40% at 10 years [21-26]. Although the relation between the number of affected axillary lymph nodes and prognosis is a continuum, patients in whom 10 or more axillary lymph nodes are affected have a particularly poor prognosis. At 5 years, disease-free survival is 13% to 45% and overall survival is 27% to 44%; at 10 years, disease-free survival is 10% to 30% and overall survival is 24% [22, 27].

Large, randomized clinical trials [28, 29] have convincingly shown that most women in the early stages of breast cancer can have breast-conserving surgery followed by adjuvant breast radiation therapy without compromising survival compared with the survival in patients who have mastectomy. Adjuvant therapy trials were developed for patients in the early stages of breast cancer in an effort to eradicate micrometastatic disease that was presumed to be present at the time of diagnosis. Tamoxifen therapy prolongs overall survival and disease-free survival in the early stages of breast cancer except in young women with estrogen receptor-poor tumors [30]. Several randomized trials of adjuvant chemotherapy have confirmed that systemic chemotherapy alters the natural history of the early stages of breast cancer by increasing disease-free survival and overall survival compared with survival in patients receiving primary therapy alone (for example, breast-conserving therapy or mastectomy) [30]. An estimate of the absolute magnitude of benefit (prolongation of disease-free survival or overall survival) conferred by adjuvant chemotherapy depends on the prognosis of the patient population. In a meta-analysis of randomized trials of adjuvant chemotherapy, the Early Breast Cancer Trialists' Collaborative Group [30] showed that combination chemotherapy reduces the annual odds of recurrence and death to a greater degree in women younger than 50 years of age than in women older than 50 years of age. As Gelber and colleagues [31] point out, the results of the overview by the Collaborative Group reflect outcomes that might be achieved for treated populations but provide no information on the distribution of benefit among individually treated patients. Nevertheless, most patients with high-risk, stage II breast cancer ( 10 affected axillary lymph nodes) will develop recurrent disease after receiving standard adjuvant chemotherapy.

Rationale for High-Dose Chemotherapy in the Early Stages of Disease

One of the fundamental observations obtained from experience with high-dose chemotherapy in metastatic disease is that a subset of patients with non-bulky disease, minimal previous therapy, and disease sensitive to chemotherapy appear to achieve the highest response rates and possibly the longest progression-free survival [11-1832, 33]. As a result, high-dose chemotherapy is being studied as adjuvant therapy for patients with high-risk stage II or stage IIIa breast cancer. Because these patients have subclinical disease and have not previously received therapy, administering high-dose chemotherapy as adjuvant therapy is theoretically sound. The critical question that remains unanswered is whether high-dose chemotherapy offers any advantage over several cycles of standard-dose chemotherapy or dose-intense chemotherapy that does not cause ablation of bone marrow.

Dose intensity is defined as the amount of drug administered per unit of time, and the dose intensity of a chemotherapy regimen that contains many drugs is determined by calculating the dose intensity of each drug relative to that of a standard reference regimen [34]. Retrospective reviews of adjuvant chemotherapy trials have reported conflicting results on the effect of dose intensity on patient outcome. Bonadonna and Valagussa [35] addressed this issue in a retrospective analysis of two clinical trials in which adjuvant cyclophosphamide, methotrexate, and 5-fluorouracil were administered. In that analysis, disease-free survival was longer in patients who received a higher percentage of the intended dose of chemotherapy than in patients who received a lower percentage. Henderson and colleagues [36] have reported no effect of dose intensity on patient outcome. These researchers have also cautioned that retrospective analyses may be biased because patients who can receive full doses of chemotherapy may have a better performance status and thus a better prognosis. Other investigators [37] have retrospectively analyzed dose rate rather than the cumulative dose of planned chemotherapy, but these analyses may also be biased [36].

Recent randomized studies of adjuvant chemotherapy in node-positive breast cancer shed some light on the issue of dose intensity. Wood and colleagues [38] reported the results of a randomized trial of different doses and dose intensities of adjuvant chemotherapy with cyclophosphamide, doxorubicin, and 5-fluorouracil in 1572 patients with axillary node-positive, stage II breast cancer. After a median follow-up of 3.4 years, the women who had received chemotherapy of high or moderate dose intensity had significantly longer disease-free survival and overall survival than did women who received chemotherapy of low dose intensity. The difference in survival between patients who received chemotherapy of medium dose intensity and those who received chemotherapy of high dose intensity was not statistically significant. Similarly, Tannock and colleagues [39] previously studied two dose levels of cyclophosphamide, methotrexate, and 5-fluorouracil in women with metastatic disease; this study showed a survival advantage for the group treated more dose-intensively. These findings are consistent with a minimal dose threshold effect, below which adjuvant chemotherapy provides no benefit. The treatment programs selected in these studies were all outpatient programs, and the designs did not assess whether further dose escalation would have produced more striking differences in patient outcome.

Because of their poor prognosis, patients in whom 10 or more axillary lymph nodes are affected have been selected to assess the efficacy of adjuvant high-dose chemotherapy. Two NCI-sponsored randomized clinical trials address this issue: The INT-0121 trial randomly assigns patients to receive high-dose chemotherapy or no additional therapy after six cycles of standard-dose, adjuvant chemotherapy; in INT-0163, standard-dose adjuvant chemotherapy is followed by random assignment to either high-dose chemotherapy or the same chemotherapy administered in the high-dose chemotherapy regimen but at lower doses that do not require stem-cell support. Patients in both studies receive irradiation of the chest wall and, if tumors test positive for hormone receptor, tamoxifen.

With ever-increasing pressure for improved adjuvant therapy, nonrandomized pilot trials have evaluated the potential benefit of adjuvant high-dose chemotherapy for patients in whom 4 to 9 axillary lymph nodes are affected [40-42]. The NCI-sponsored Cooperative Cancer Groups recently began a large-scale (1000 patients), randomized clinical trial (Southwest Oncology Group 9623) to evaluate the efficacy of high-dose chemotherapy in this subset of patients. In contrast to trials for women in whom 10 or more axillary lymph nodes are affected, this trial compares high-dose chemotherapy with dose-intense chemotherapy that does not cause ablation of bone marrow.

Data Supporting High-Dose Chemotherapy

The preliminary results of nonrandomized phase II studies have generated optimism about high-dose chemotherapy. Peters and colleagues [43, 44] recently provided an update on the largest experience with high-dose chemotherapy to date. After extensive eligibility evaluation that included assessment of cardiac and pulmonary function, bone marrow biopsy, computed tomography, and insurance screening, 85 patients initially received six cycles of chemotherapy with standard cyclophosphamide, doxorubicin, and 5-fluorouracil followed by high-dose chemotherapy. Patients were also required to receive irradiation of the chest wall and tamoxifen for 5 years. After a median follow-up of 6 years, the 5-year event-free survival rate was 64% and the overall survival rate was 75% [44]. These results were compared with those of historical controls from previous adjuvant chemotherapy trials in which standard-dose chemotherapy was administered. The 5-year event-free survival rate in the controls was 30% to 35%, and the overall survival rate was 37% to 48% [44]. This comparison suggests that adjuvant high-dose chemotherapy may be superior to standard-dose chemotherapy for patients with high-risk primary breast cancer (that is, patients in whom more than 10 axillary lymph nodes are involved).

Critics have countered that the superior results are primarily due to the strict selection criteria used in the adjuvant high-dose chemotherapy trials; these criteria were not used in older trials of adjuvant chemotherapy. Crump and colleagues [45] report that 23% of patients referred for a trial of adjuvant, high-dose chemotherapy were found to have metastatic disease, which made them ineligible. Metastatic disease in these patients would not have been detected with the eligibility criteria required by older trials of adjuvant chemotherapy. By implication, some of the historical controls in the trial by Peters and colleagues [43, 44] may have been found to have metastatic disease if they had been subjected to the same rigorous criteria. In addition, irradiation of the chest wall was required in the high-dose chemotherapy trial after disease on the chest wall recurred in several of the first patients treated. The routine administration of tamoxifen may have further reduced the risk for recurrence in patients receiving high-dose chemotherapy.

A recent report by Ung and colleagues [46] shows the potential danger of using historical controls. Sixty-four patients with extensive nodal involvement received three cycles of standard-dose induction chemotherapy and irradiation of the chest wall and regional nodes, followed by additional standard-dose adjuvant chemotherapy. The 5-year actuarial freedom from distant relapse rate and overall survival were 45% and 65%, respectively. These results are similar to those reported by Peters and colleagues [43, 44] for patients receiving adjuvant high-dose chemotherapy. The retrospective review by Ung and colleagues involved four successive treatment programs but did not require the extensive eligibility evaluation required in the trial of adjuvant high-dose chemotherapy by Peters and colleagues.

Early results were recently reported from the only randomized trial comparing conventional-dose chemotherapy with high-dose chemotherapy in patients with metastatic breast cancer [3]. Ninety patients were randomly assigned to receive 1) cyclophosphamide, mitoxantrone, and vincristine at doses that did not require stem-cell support or 2) high-dose cyclophosphamide, mitoxantrone, and vincristine followed by autologous bone marrow or peripheral stem-cell transplantation. The overall response rate in patients receiving the second regimen was 95%; 51% of patients achieved a complete remission. In contrast, the overall response rate in patients receiving the first regimen was 53%; only 4% of patients achieved a complete remission. The duration of survival was longer for patients receiving the second regimen [3]. Although these data are provocative, the trial was small, follow-up was short (median duration, 72 weeks), and the control group (which received the first regimen) did not fare as well as patients who received similar chemotherapy in a previous trial [47]. Oncologists have voiced concern that the results of this high-dose approach were not superior to those seen in their experience with conventional treatment [48, 49].

Peters and colleagues [6] recently presented results from a randomized trial. Although these results are available only in abstract form, this trial has been influential in convincing third-party payers to cover the cost of high-dose chemotherapy for women with metastatic disease. The study enrolled 422 women with metastatic breast cancer. All patients received two to four cycles of doxorubicin, 5-fluorouracil, and methotrexate, and 98 patients achieved a complete response. The complete responders were then randomly assigned to receive immediate high-dose chemotherapy or observation. Patients in the observation group received high-dose chemotherapy when relapse occurred. Disease-free survival was better in patients receiving immediate high-dose chemotherapy than in those in the observation group (0.32 compared with 1.85 years; P < 0.001). Surprisingly, however, overall survival favored the observed group (3.2 compared with 1.7 years; P = 0.04). The rate of death associated with high-dose chemotherapy was not high and thus does not explain the results, which appear to favor delaying high-dose chemotherapy until the occurrence of relapse in this subset of women. The study design does not allow a comparison with other standard therapies at the time of relapse. Thus, it remains uncertain whether high-dose chemotherapy would still have been considered as having prolonged survival had it been compared with other second-line treatments in these women.

Overcoming Patient and Physician Biases

Because of the rapid proliferation of high-dose chemotherapy programs in community hospitals, accrual to the NCI-sponsored high-priority trials has been slow. The NCI-sponsored trials involved several treatment centers in the United States and are designed to determine whether high-dose chemotherapy offers a long-term benefit compared with standard-dose adjuvant chemotherapy. Although few data support this conclusion, the perception of benefit based on results of phase II, single-institution trials has created an obstacle to the completion of these important trials. A recent survey of oncologists done at the 1995 American Society of Clinical Oncology meeting by the NCI's Office of Cancer Communication provides insight into the slow accrual [9, 10]. Several responders indicated that physicians and breast cancer advocates believe that the question has been answered and that a randomized trial is no longer needed. Others suggested that large data sets from previous trials can answer the question without the completion of a randomized trial. Many community physicians expressed frustration with the time and effort necessary to enroll a patient and with the follow-up data collection required in the randomized trials. These requirements have previously been cited as a cause for physicians' poor accrual of patients to other clinical trials [50]. Finally, many physicians indicated that patient preference strongly influenced treatment recommendations. Some were unwilling to accept random assignment to a treatment they believed was less effective, whereas others refused for fear of higher treatment-related mortality.

The NCI's Office of Cancer Communications has developed an educational campaign to counter false notions about high-dose chemotherapy. The Office has created materials that provide an appropriate medical background and describe the High-Priority Trials. The target audience for these materials includes patients with breast cancer, their families, advocacy groups, health professionals, third-party payers (including referring physicians), oncologists, nurses, social workers, data managers, and various professional organizations (such as The American Society of Clinical Oncology, The Oncology Nursing Society, The National Association of Oncology Social Workers, and The American Society of Blood and Bone Marrow Transplantation).

Conclusions

The long-term benefit of adjuvant, high-dose chemotherapy compared with that of standard adjuvant chemotherapy in patients with breast cancer has not been established. The ease with which community hospitals and entrepreneurial ventures can establish transplantation programs provides an economic incentive for the development of in-house, phase II, high-dose chemotherapy trials that directly compete with the randomized studies that have the statistical power to answer important questions.

How do we overcome the barriers to completion of the pivotal phase III trials? Although physicians have cited research cost as a barrier to participation, we believe that the essential obstacle is the notion of preconceived preference for therapy. Some have argued that physicians have an obligation to "play their best hunch" on the basis of available data from phase II trials [51]. As discussed earlier, however, phase II trials are small by definition and are usually single-institution studies that are highly susceptible to selection bias. The field of oncology is replete with examples of phase II trials with positive results that have not been confirmed in large, multi-institution, randomized phase III trials. A notable case proving this point occurred with non-Hodgkin lymphoma, a disease that is much more sensitive to chemotherapy than is breast cancer. On the basis of results of phase II lymphoma trials, many oncologists had replaced the standard chemotherapy regimen for this disease with more dose-intensive, so-called "third-generation" regimens. A large phase III trial did not confirm the putative advantage for dose intensity; such a trial was required to reverse the stampede toward adopting the routine use of the more toxic and costly third-generation regimens [52].

Such examples, as well as those from other fields of medicine [53], should remind clinicians and patients that intuition does not constitute scientific proof. As Freedman [54] has argued, clinical trials are necessary when "honest, professional disagreements among expert clinicians [exist] about the preferred treatment. At this point, a state of ‘clinical equipoise’ exists." There is clearly such disagreement about high-dose chemotherapy, and resolution requires dispassionate arbitration by clinical trials. Proponents and opponents of this approach to treating breast cancer must accept that the treatment they do not favor is, in Freedman's words [54], "preferred by colleagues whom they consider to be responsible and competent." Medical progress for all patients is best served by the willingness of physicians to recognize and test their biases. Successful completion of the NCI-sponsored, high-dose chemotherapy trials should allow us to replace our hunches with sound scientific data so that patients with breast cancer can benefit [55].

Dr. Abrams: National Cancer Institute, Clinical Investigations Branch, 6130 Executive Boulevard, #741, Bethesda, MD 20892.