Dose Density in Breast Cancer: A Simple Message?

EDITORIALS Dose Density in Breast Cancer: A Simple Message? Nancy U. Lin, Rebecca Gelman, Eric P. Winer “Everything should be made as simple as possible, but not one bit simpler.” —Albert Einstein In the 1970s, Norton and Simon (1), applying Gompertzian principles to cancer cell growth, hypothesized that maximal chemotherapy effectiveness could be achieved by scheduling the interval of chemotherapy to correspond to the period of most rapid tumor growth. As an outgrowth of this initial work, several pilot trials and two large randomized trials were conducted to test the feasibility and effectiveness of what has come to be called dose-dense chemotherapy. The Cancer and Leukemia Group B (CALGB) trial 9741, which was conducted in the North American Intergroup (2), provided support to Norton and Simon’s hypothesis by demonstrating that a change in the interval of anthracycline- and taxane-based chemotherapy, from every 3 weeks to every 2 weeks, improved disease-free and overall survival in women with lymph node–positive, early-stage breast cancer. In this issue of the Journal, Venturini et al. (3) report the results of a multicenter phase III study, led by the Gruppo Oncologico Nord Ovest-Mammella InterGruppo (GONO-MIG), comparing fluorouracil, epirubicin, and cyclophosphamide administered every 3 weeks (FEC21) versus the same regimen administered every 2 weeks with filgrastim support (FEC14) in patients with lymph node–positive or high-risk lymph node–negative breast cancer. They reported that FEC14 was both feasible and safe: Patients achieved an actual 48% increase in dose density, and no incident cases of acute myelogenous leukemia or myelodysplastic syndrome were observed. Only one case (0.2%) of grade 2 cardiac toxicity was reported on each arm. In contrast to the CALGB/Intergroup trial, however, the difference between the two arms did not reach traditional criteria for statistical significance for either recurrence (hazard ratio [HR] = 0.88, 95% confidence interval [CI] = 0.71 to 1.08) or death (HR = 0.87, 95% CI = 0.67 to 1.13). Overall, the study was thoughtfully designed, and the results are mature. The treatment arms were balanced in terms of patient and disease characteristics. The investigators recorded the incidence of treatment-related amenorrhea and found no difference between the two arms. As in the CALGB/Intergroup trial, the chemotherapy dose per cycle and total number of cycles were held constant, and only the interval between doses varied. Thus, the study is only one of two trials specifically designed to test the concept of dose density in the context of adjuvant breast cancer treatment. How should clinicians, researchers, and patients integrate the results of the present study into the overall body of evidence evaluating dose-dense treatment for breast cancer? Is dose density a unifying concept that should be applied whenever adjuvant chemotherapy is administered, or is it specific to a particular chemotherapeutic regimen or tumor subtype? How does 1712 EDITORIALS toxicity vary according to schedule, and in whom do the additional benefits of dose-dense chemotherapy outweigh the potential risks? On the surface, the conclusions from CALGB/Intergroup trial and the Venturini study appear to be discordant. There are several possible explanations, including subtle differences in the patient populations, differences in the chemotherapy regimens, and the play of chance. Of these possibilities, the absence of a taxane in the Italian study is the most compelling. A body of literature (4,5) in both the metastatic and preoperative settings suggests that more frequent paclitaxel scheduling can lead to meaningful differences in disease control. By extension, some researchers have hypothesized that most if not all of the benefit attributed to dose density in the CALGB 9741 trial could be related to the optimization of taxane scheduling. Several studies, however, argue against a specific effect of taxane scheduling. More than two decades ago, Bonadonna et al. (6) conducted a landmark study comparing four cycles of doxorubicin followed by eight cycles of cyclophosphamide, methotrexate, and fluorouracil (CMF) with one cycle of doxorubicin alternating with two cycles of CMF, for a total of 12 cycles of therapy. The trial was originally designed to ask whether alternating non–cross-resistant regimens would improve outcomes in women with lymph node–positive breast cancer, but, if viewed in a different light, it could be considered a trial evaluating dose density of both CMF and doxorubicin. At a medium follow-up of 9 years, there was a highly statistically significant difference in survival (58% versus 44%, difference = 14%; P = .002) in favor of the sequential arm. More recently, albeit in a very different setting, a German group (7) demonstrated that delivering cyclophosphamide, doxorubicin, vincristine, and prednisone every 2 weeks, compared with every 3 weeks, statistically significantly improved survival in elderly patients with aggressive lymphomas (relative risk reduction = 0.58, 95% CI = 0.43 to 0.79; P<.001). The results of each of these trials suggest that the benefits of a dose-dense treatment approach are not specific to a single class of drugs (i.e., taxanes). Despite the non–statistically significant results observed in the GONO-MIG trial, on closer examination, the findings are not entirely inconsistent with results of the CALGB/Intergroup study. There was a trend toward improved event-free and overall survival for women who received the every-14-day regimen Affiliation of authors: Department of Medical Oncology (NUL, EPW) and Department of Biostatistics and Computational Biology (RG), Dana-Farber Cancer Institute, Boston, MA. Correspondence to: Eric P. Winer, MD, Department of Medical Oncology, Dana-Farber Cancer Institute, 44 Binney Street, Boston, MA 02115 (e-mail: ). DOI: 10.1093/jnci/dji438 © The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: . Journal of the National Cancer Institute, Vol. 97, No. 23, December 7, 2005 (i.e., FEC14). The GONO-MIG trial observed a higher actuarial 10-year survival (78%, 95% CI = 74% to 82%) in the control group than was expected, but the number of total events and total deaths (and hence the power) in the GONO-MIG trial were similar to those in the CALGB/Intergroup trial. The GONO-MIG study actually had 80% power to detect a true 26% relative reduction in the risk of recurrence and a true 32% relative reduction in the risk of death, which are similar to the observed differences in the initial report of the CALGB study at a median follow-up of approximately 3 years. Using an exponential assumption, we estimated hazard ratios for event-free survival and overall survival from the GONO-MIG trial at a median follow-up of 3 years; the ratios for both event-free survival and overall survival appear to be within approximately 1% of those reported in the CALGB/ I (...truncated)