Taxanes in the Adjuvant Setting: Why Not Yet?

Martine J. Piccart, M.D., Ph.D., Caroline Lohrisch, and Luc Duchateau

Adjuvant chemotherapy with cyclophosphamide, methotrexate, and 5-fluorouracil (CMF) and anthracycline-based regimens is associated with significant reductions in breast cancer mortality (EBCTCG, 1998). Numerous other treatment strategies have been explored in attempts to further improve survival, including newer anticancer drugs with demonstrated activity in the metastatic setting, such as the taxanes (T). So far, two randomized trials of the adjuvant taxane paclitaxel have been reported. An Intergroup trial (Henderson, Berry, Demetri, et al., 1998) compared the efficacy of 4 AC with 4 AC followed by 4 paclitaxel (ACT) in node-positive cancer, while the M.D. Anderson Cancer Center trial compared 8 FAC to 4 T followed by 4 FAC (TFAC) in node-negative and node-positive breast cancer (Thomas, Buzdar, Theriault, et al., 2000). The positive results in the Intergroup trial (overall survival [OS] 95 percent versus 97 percent, p=0.04, and disease-free survival [DFS] 86 percent versus 90 percent, p=0.0008) led to approval by the Food and Drug Administration (FDA) of adjuvant paclitaxel for node-positive breast cancer. The absolute survival difference, however, is quite small, just below the 5 percent significance level. The Anderson trial, reported subsequently, showed nonsignificant results (DFS 81.5 percent versus 85.2 percent, p=0.2, and 15 deaths versus 13 deaths for TFAC and FAC, respectively).

The Intergroup and Anderson trials had important differences in design and patient eligibility. The proportion of patients in the Anderson trial with none, 1 to 3, or more than 3 positive nodes was approximately one-third each. This group therefore had a slightly better overall prognosis than the Intergroup population, in which no patients were node-negative and the proportion with 1 to 3 and more than 3 nodes was approximately even. More importantly, the two trials differed substantially with respect to total sample size and median followup. The Intergroup trial involved 3,170 patients, versus 524 in the Anderson trial, while the Anderson had more than double the median followup (18 months versus 43 months). In using a time-to-event analysis, the power of a trial to detect a significant effect rests primarily on the number of events. Thus, it is completely plausible that the Intergroup trial demonstrated a significant difference, given its approximately 374 events (DFS) and 125 deaths at 18 months, while the Anderson trial, with 75 events and 28 deaths, does not, despite longer followup. To what extent these differences influence the apparently disparate results of these two studies is not clear. Nevertheless, the conclusion must be that the results of the Anderson trial neither help nor hinder the case for adjuvant paclitaxel. There are, however, a number of reasons to support the view that the FDA may have made a hasty decision. Note that, after reviewing the same data, the European Regulatory Agency decided not to approve paclitaxel for adjuvant therapy.

Confidence in any treatment depends on whether its value is consistently reproducible. The general view within the scientific community is that a single randomized trial does not constitute sufficient level I evidence, no matter how compelling the results. This view arises from repeated observation that a particular regimen's superior results cannot always be duplicated in subsequent trials. Despite that well-known principle, the FDA seems to have been persuaded by the results of only one trial. Given that numerous ongoing randomized trials of taxanes will be reported in the next few years and that the absolute survival difference reported by the Intergroup trial was modest (2 percent), it would have been more prudent to await corroborative evidence before approving the routine use of adjuvant taxanes.

Design Limitations of Intergroup Trial

Furthermore, the Intergroup trial had design limitations, and it is unclear to what extent those limitations accounted for the treatment effect. A major potential confounder in this trial was the duration of therapy, which was 12 weeks longer in the taxol-containing arm. Several trials have demonstrated that duration of therapy may indeed influence outcome, but reports on the trials do not deal with the relative importance of duration of therapy and cumulative dose. Both of these factors may have biased the results in favor of the ACT arm, since the cumulative dose of doxorubicin in one-third of the study population was 240 mg/M2, which may be suboptimal.

The Oxford overview reported that anthracycline therapy may be associated with a small survival advantage over CMF (EBCTCG, 1998), but other studies have failed to demonstrate that anthracycline therapy beyond a threshold dose improves survival. There does seem to be a dose-response relationship below that threshold, with compromised efficacy of anthracyclines (Wood, Budman, Korzun, et al., 1994; Bonneterre, Roche, Bremond, et al., 1998). Thus, we are faced with the question of whether the cumulative dose of doxorubicin in 4 AC (240 mg/M2) was below this threshold, making it as effective as CMF but less effective than it could be. If the answer is yes, it may explain the apparent discrepancy in the results of NSABP B-15 (equivalence for 6 CMF and 4 AC 240 mg/M2 doxorubicin total dose), the National Cancer Institute of Canada (NCIC) comparison of CEF and CMF (CEF giving a 7 percent superior OS, 720 mg/M2 epirubicin total dose), and the Intergroup trial of CAF versus CMF (CAF giving a 2 percent superior OS) (Hutchins, Green, Ravdin, et al., 1998; Levine, Bramwell, Pritchard, et al., 1998). Although 4 AC (doxorubicin 60 mg/M2/cycle) is a standard adjuvant regimen in North America, many European clinicians give several cycles of CMF following 4 AC, or a higher total dose of anthracyclines, such as can be found in CAFFAC and CEFFEC regimens.

Another major concern with the Intergroup trial is the immaturity of the data. The initial results were reported 8 months after the accrual of the last patient (3-year accrual period), based on a preplanned interim analysis of 450 events. These significant results cannot be extrapolated to later points in time unless one assumes a constant proportional hazards model. If that assumption is not correct, however, more mature results could show a "reversal of fortune," as did the EORTC neoadjuvant breast cancer study (Sylvester, Bartelink, Rubens, 1994). An interim analysis demonstrated significant superiority for chemotherapy, but reanalysis 2.5 years after the last patient was accrued reversed the favorable outcome of hormonal therapy, thus demonstrating the limitations of analyses based on early data. The Intergroup analysis, carried out when, on average, 96 percent of the patients were still alive and 88 percent were disease-free, is no exception. One could even call the more recent analysis (at 30 months median followup) immature, given the recurrence rate in the control arm of 22 percent, since a higher final recurrence rate for a node-positive population treated with anthracyclines would be expected, based on previous trials (Fisher, Brown, Dimitrov, et al., 1990; Levine, Bramwell, Pritchard, et al., 1998).

Finally, there is the issue of whether all patients in the study derived equal benefit from the treatment. A subset analysis suggests that only patients with hormone receptor (HR)-negative tumors (one third of the study population) benefited from the addition of tamoxifen. For the 2,066 HR-positive patients, the hazard ratio for recurrence was 0.92 (95 percent CI 0.73-1.16) for ACT versus AC, while for HR-negative patients it was 0.68 (95 percent CI 0.55-0.85). A similar trend was observed in the Anderson trial: 58 percent of the population was HR-positive, and although not statistically significant, the absolute difference in DFS for FAC versus TFAC was 3 percent for HR-positive patients and 5 percent for HR-negative patients. There are several potential explanations for these findings. Either the baseline risk for HR-positive tumors is lower and therefore a benefit of tamoxifen is more difficult to demonstrate or does not exist, or the baseline risk is sufficiently lowered by AC and tamoxifen that the added benefit of tamoxifen, if it exists, cannot be demonstrated with this sample size and followup period. Another explanation may be that recurrences in the HR-positive population occur later, and a benefit may only become apparent with longer followup. Regardless of the reason, this subset analysis supports the contention that, based on the available evidence, sweeping generalizations about the value of adjuvant paclitaxel are premature.

Unsettled Issues

Despite the enormous strides made in adjuvant chemotherapy for breast cancer over the last 20 years, there are a number of unsettled issues. The NCIC trial faces the ambitious task of exploring the relative importance of cumulative dose, dose density, and noncross-resistant drugs added to anthracycline-based chemotherapy. The design calls for randomization between three arms: 6 FEC (Levine regimen), 4 AC followed by 4 T, and 6 EC every 2 weeks with G-CSF followed by 4 T. If the results were available now, we might be able to put to rest many of our reservations about the Intergroup trial, which leave us a little unsettled about the long-term reliability and generalizability of its results, regardless of how promising they may appear. Unfortunately, the results are not available, and the finding of nonsignificance in the Anderson study amplifies the uncertainty. It is necessary to wait for future results of ongoing trials before pronouncing judgment on the value of taxanes in the adjuvant setting.

It is also necessary to better define the population most likely to benefit from therapies of longer duration, intensification, and multiple regimens. It no longer is reasonable to judge all breast cancer patients as having equal probability of benefit from a given therapy. That was a paradigm that worked well when adjuvant chemotherapy for breast cancer was in its infancy and little was known about the molecular heterogeneity of breast cancer. It is now of critical importance to design trials with the aid of molecular tumor profiles with potential predictive value to prospectively identify the subgroup most likely to benefit from the addition to therapy of taxanes and other new drugs. This process has begun with the EORTC-Breast Cancer Cooperative Group trial in locally advanced breast cancerÑan attempt to examine the predictive value of p53 mutations in response to taxane chemotherapy.

It is to be hoped that the early promise of taxanes in adjuvant treatment of breast cancer will be confirmed, since there are few encouraging alternatives at this time. It is important to realize, however, that the data we have now only support their potential. Further followup, and trials that corroborate the results of the Intergroup trial of AC versus AC¨T, are essential to define the value of taxanes in early breast cancer.

References

Bonneterre J, Roche H, Bremond A, et al. Results of a randomized trial of adjuvant chemotherapy with FEC 50 vs FEC 100 in high risk node-positive breast cancer patients. [abstract]. Proc Am Soc Clin Oncol 1998;17:124a. Early Breast Cancer Trialists' Collaborative Group (EBCTCG). Polychemotherapy for early breast cancer: an overview of the randomised trials. Lancet 1998;352:930-42.

Fisher B, Brown AM, Dimitrov NV, Poisson R, Redmond C, Margolese RG, et al. Two months of doxorubicin-cyclophosphamide with and without interval reinduction therapy compared with 6 months of cyclophosphamide, methotrexate, and fluorouracil in positive node breast cancer patients with tamoxifen-nonresponsive tumors: results from the National Surgical Adjuvant Breast and Bowel Project B-15. J Clin Oncol 1990;8:1483-96.

Henderson IC, Berry D, Demetri G, et al. Improved disease-free and overall survival from the addition of sequential paclitaxel but not from the escalation of doxorubicin dose level in the adjuvant chemotherapy of patients with node-positive primary breast cancer. [abstract]. Proc Am Soc Clin Oncol 1998;17:101a.

Hutchins LF, Green S, Ravdin PM, et al. CMF versus CAF with and without tamoxifen in high-risk node-negative breast cancer patients and a natural history follow-up study in low-risk node-negative patients: first results of Intergroup trial INT 0102. [abstract]. Proc Am Soc Clin Oncol 1998;17:1a.

Levine MN, Bramwell VH, Pritchard KI, Norris BD, Shepherd LE, Abu-Zabra H, et al. Randomized trial of intensive cyclophosphamide, epirubicin, and fluorouracil chemotherapy compared with cyclophosphamide, methotrexate, and fluorouracil in premenopausal women with node-positive breast cancer. J Clin Oncol 1998;16:2651-8

. Sylvester R, Bartelink H, Rubens R. A reversal of fortune: practical problems in the monitoring and interpretation of an EORTC breast cancer trial. Stat Med 1994;13:1329-35.

Thomas E, Buzdar A, Theriault R, et al. Role of paclitaxel in adjuvant therapy of operable breast cancer: preliminary results of prospective randomized clinical trial. [abstract]. Proc Am Soc Clin Oncol 2000;19:74a.

Wood WC, Budman DR, Korzun AH, Cooper MR, Younger J, Hart RD, et al. Dose and dose intensity of adjuvant chemotherapy for stage II, node-positive breast carcinoma. N Engl J Med 1994;330:1253-9.*

Home · Search