Can We Abandon Anthracyclines for Early Breast Cancer Patients?
Can We Abandon Anthracyclines for Early Breast Cancer Patients?
Abstract: Anthracycline-containing regimens improve disease-free and overall survival of patients with early breast cancer, but the toxicity, especially the cardiotoxicity, of the anthracyclines make them unattractive in the adjuvant setting. Two large, randomized trials, one in unselected patients and one in those with HER2-positive tumors, suggest that a taxane combination without an anthracycline might be just as effective as more traditional regimens. These and other studies also suggest that the anthracyclines might reasonably be used only for those with more aggressive forms of breast cancer, as defined by molecular markers. The results of these studies are provocative but insufficient to justify the conclusion that anthracyclines can be either abandoned or used only for a very select group of patients.
The anthracyclines doxorubicin and epirubicin are among the most effective cytotoxic treatments developed for the treatment of breast cancer. In the adjuvant setting they significantly improve both disease-free and overall survival. Although all anthracycline regimens are not equally efficacious, when the results of all trials were pooled in an overview comparing anthracycline regimens to no chemotherapy at all, they reduced the odds of recurrence by 33% (standard error [SE] +/- 8%) and the odds of death by 26% (SE +/- 9%). Compared to chemotherapy regimens that did not contain an anthracycline, they reduced the odds of recurrence by an additional 11% (SE +/- 3%) and the odds of death by an additional 16% (SE +/- 3%) over what can be achieved with a CMF (cyclophosphamide, methotrexate, and 5-fluorouracil)-like combination.
The anthracyclines are also among the most toxic drugs ever developed. They induce nausea and vomiting that, in the days before the introduction of 5-HT3-receptor antagonist antiemetics, was severe enough to sometimes necessitate hospitalization and intravenous hydration. The majority of patients lose nearly all their scalp hair. Myelosuppression can be profound, although fortunately its time course is very predictable and short, so granulocyte colony-stimulating factors are rarely needed at doses that are of proven value. The incidence of leukemia associated with anthracycline use is clearly greater than that seen with older regimens such as CMF; depending on the acute and cumulative dose and type of anthracycline used, this incidence ranges between 0.4% and 1.7% in the 5 to 10 years following treatment. A cardiomyopathy related to the cumulative dose of the anthracycline limits the duration of drug treatment and is likely the greatest impetus to finding new, non-cardiotoxic alternatives. Adjuvant anthracycline regimens typically use 240 to 360 mg/m2 of doxorubicin or 240 to 720 mg/m2 of epirubicin over 4 to 6 cycles, and with these doses the percentage of patients who develop a cardiac event such as congestive heart failure within the 5 years following completion of chemotherapy ranges from 0.5% to 1.5%. Subclinical cardiac dysfunction has been reported to occur in 10% to 15% of patients. The frequency of cardiac events is about four times greater when trastuzumab (Herceptin) is administered with or following the anthracycline (Table 1). This cardiotoxicity is more often reversible and likely has a different underlying pathophysiology,[4, 5] but it is still a cause of considerable concern since it is as yet unknown whether the incidence of serious cardiac disease will expand as the normal risks of cardiovascular disease associated with aging are superimposed on the cardiac damage from anthracyclines and trastuzumab.
In the past decade it has been repeatedly shown that the addition of a taxane to an anthracycline combination reduces the hazard of recurrence by yet another 15%. However, the taxanes induce neurotoxicity, and in many cases the taxane regimen is more toxic because it is administered for a longer duration than the non-taxane combination. Most investigators have felt that this increased toxicity was justified by the increased benefit. Recently, a taxane has been substituted for the anthracycline in two trials in an effort to circumvent toxicity, especially cardiotoxicity.
Should these new, non-anthracycline, taxane combinations be used routinely in place of the older anthracycline or anthracycline/taxane regimens for all patients with early breast cancer? Or should they be used in selected patients for whom the anthracyclines might have only limited benefit? Or is it too early to use them at all outside of a clinical trial?
New Regimen #1: Docetaxel Plus Cyclophosphamide
In US Oncology trial 9735, 1016 relatively unselected patients were randomly assigned to receive either doxorubicin, 60 mg/m2, plus cyclophosphamide, 600 mg/m2, (AC) or docetaxel (Taxotere), 75 mg/m2, plus cyclophosphamide, 600 mg/m2 (DC) (Table 2).[7, 8] The duration of therapy in the two arms was identical. After a median follow-up of 7 years, the disease-free and overall survival of patients treated with the DC combination were significantly better than for those treated with AC. One patient in the AC arm died without relapse but with cardiomyopathy and congestive heart failure thought to be related to her exposure to doxorubicin. Significantly more patients in the AC arm had nausea and vomiting (Table 2). Significantly more patients in the DC arm had arthralgia, myalgia, and edema. Neutropenia and febrile neutropenia also occurred more frequently with DC, especially among women aged 65 or over. The incidence of neurotoxicity was not reported but may have been subsumed under the category “myalgia” and “arthralgia.” One older woman died of myelodysplasia and another of myelofibrosis. Neither of these women had relapsed, and the investigators thought their deaths were “probably related to treatment” with AC.
This was a well designed clinical study. Since the only difference in the two study arms was the substitution of docetaxel for doxorubicin, this was a true test of the relative efficacy of these two drugs in combination with intravenous cyclophosphamide.
Is DC superior? Equivalent? Possibly inferior to AC?
There was an increased number of deaths in the AC arm that cannot be attributed to either breast cancer or the complications of treatment. The investigators used a standard definition of event-free survival and included among the events either breast cancer relapse or death from any cause. There were 118 events in the AC arm and 88 events in the DC arm—a difference of 30. However, 25 patients in the AC arm and only 13 patients in the DC arm died without relapse—but one of the former 25 patients died of cardiac complications from doxorubicin. If we adjust events to include only breast cancer relapse or death from breast cancer, the difference in the number of events shrinks from 30 to 18. Thus, breast cancer–specific event-free and breast cancer–specific overall survival differences are not statistically significant. This does not negate the results of this study but should cause us to be less comfortable about accepting them without confirmation of the results in another trial.
If DC is not unequivocally superior to AC, can’t we conclude that it is at least equivalent to and less toxic than AC? It is not uncommon for investigators to design a superiority trial and then, if and when the trial fails to meet the primary endpoint, conclude that the experimental regimen is equivalent to the one it is being compared with. This happened in the analysis of National Surgical Adjuvant Breast and Bowel Project (NSABP) trial B-15 (see below). However, many statisticians, including those at the FDA, argue that this is an inappropriate practice. In a superiority trial it is assumed there is no difference between the two arms, and the trial is designed to reject that hypothesis and accept the alternative hypothesis that the treatments do differ. A non-inferiority or equivalence trial assumes there is a difference between the arms and is designed to reject this hypothesis and conclude that the treatments do not differ. Non-inferiority trials are usually much larger. This is because a non-inferiority trial usually compares a new treatment with an established treatment; physicians are unlikely to accept a new treatment that is otherwise “equivalent” if there is even a small decrease in an important endpoint, such as survival. A large trial is required to reliably detect small decreases of this sort. Conversely, a new treatment that is thought a priori to be better must usually be shown to be “a lot better” to replace an established therapy, and a smaller trial will reliably detect this larger benefit. Thus, a superiority trial that is interpreted as an inferiority trial is usually an underpowered inferiority trial. Practically, this means that the lower end of theconfidence interval is unreliable and that the two treatments may appear to be similar when in fact one treatment is less effective. Patients treated with the inferior regimen will be unknowingly shortchanged.
Is DC equivalent to the best anthracycline regimens?
Because of doxorubicin's unfavorable toxicity profile, its use in the adjuvant setting was initially met with considerable resistance. The first randomized trial evaluating an anthracycline in the adjuvant setting administered either 5 or 10 cycles of AC (cumulative doxorubicin doses of 225 or 450 mg/m2) to determine whether the shorter, less toxic regimen was as effective. This single-institution trial was seriously underpowered. Overall, the shorter arm was non-significantly better, but among premenopausal women there was a trend favoring the longer arm.
The efficacy results of this trial were not yet available when the first large randomized comparison of an anthracycline and CMF (NSABP B-15) was initiated with 4 cycles (3 months) of doxorubicin and cyclophosphamide administered intravenously once every 3 weeks at doses of 60 and 500 mg/m2, respectively. The trial was designed as a superiority trial. The null hypothesis could not be rejected, but the authors concluded that “. . . the outcome from AC and CMF was almost identical . . .” and that this “. . . indicate[s] the merit of 2 months of AC therapy for all positive-node breast cancer patients.” Following this report, 4 cycles of AC became one of the most popular adjuvant chemotherapy regimens in the United States.
A number of other anthracycline regimens have been developed subsequently. The most widely used of these are shown in Table 3. There are no direct randomized comparisons of these regimens, but the relative effectiveness of various regimens is suggested by the meta-analysis shown in Figure 1A. Six cycles of po FAC/FEC (CAF [cyclophosphamide, doxorubicin, and 5-fluorouracil] or CEF [cyclophosphamide, epirubicin, and 5-fluorouracil] with oral cyclophosphamide) appears to be superior to other regimens. In four studies, AC or EC (epirubicin plus cyclophosphamide) was compared to CMF; even when the results from these studies are pooled, there is no evidence that AC/EC imparts any benefit over classic CMF. Differences in the regimens that might account for differences in outcome include the route and/or frequency of cyclophosphamide administration, the duration of treatment, and the addition of 5-fluorouracil to the combination; the acute dose of doxorubicin is not an important variable (Table 4; also, see below).
Most anthracycline regimens in common use today include a taxane—either paclitaxel or docetaxel. In the first of these regimens, paclitaxel was given as a single agent for 4 cycles following 4 cycles of AC; this sequential treatment was compared to 4 cycles of AC. Five additional trials using a similar, sequential regimen have confirmed the results of the first study, and a meta-analysis of these 6 trials together demonstrates a reduction in the hazard of recurrence of 18% (Figure 1B). However, it has been argued that some of the benefit from the sequential regimen of AC followed by paclitaxel could be a result of the longer duration of treatment with the sequence.[6, 12] A meta-analysis of 5 studies using a taxane and anthracycline in sequence in which the duration of treatment was the same in both arms demonstrated a somewhat smaller reduction in risk (9%); however, there was still a statistically significant advantage for the taxane combination (Figure 1B).
DC as used in US Oncology trial 9735 is almost certainly inferior to the best anthracycline regimens shown in Figures 1A and 1B, whether these are combinations of doxorubicin, cyclophosphamide, and 5-fluorouracil—or doxorubicin, cyclophosphamide, and a taxane. The advantage of the best regimens may result from one or more of the factors listed in Table 4, but the duration of treatment is almost certainly one of the causes.
New Regimen #2: Docetaxel, Carboplatin, and Trastuzumab
In the second taxane/non-anthracycline study, Breast Cancer International Research Group (BCIRG) trial 006, 3222 patients with tumors amplified for HER2/neu were randomly assigned to treatment with one of three regimens: ACD (AC, 60 and 600 mg/m2 every 3 weeks for 4 cycles, followed by docetaxel, 100 mg/m2 every 3 weeks for 4 cycles); ACD-H (AC and docetaxel as in arm 1, with weekly trastuzumab starting after 4 cycles of AC and continuing every 3 weeks for a year); or DCarboH (docetaxel, 75 mg/m2, plus carboplatin AUC [area under the curve]=6, together every 3 weeks for 6 cycles, along with trastuzumab administered as in arm 2).[13, 14] Only arms 2 and 3, both of which included trastuzumab, are germane to this discussion. A full report of this study has not yet been published in the peer reviewed literature, but some of the data have been published, and the study has been presented twice at the San Antonio Breast Cancer Symposium.[13, 14] At the time of the third planned interim analysis in 2009, the median follow-up was 65 months and there were 348 deaths and 656 events (breast cancer relapses, second malignancies, and/or death). At 5 years, 84% of the patients treated with ACD-H and 81% of those treated with DCarboH were disease-free (P = .21); 92% of those treated with ACD-H and 91% of those treated with DCarboH were alive (P = .14). DCarboH was less toxic. Grade 3/4 arthralgia, myalgia, hand-foot syndrome, stomatitis, or vomiting occurred about half as often among patients treated with DCarboH as they did in those treated with ACD-H; however, the incidence of any one of these symptoms was less than 7% in both groups. Neuropathies, both sensory and motor, and leukopenia were also significantly less frequent with DCarboH, but anemia and thrombocytopenia occurred 2 to 3 times more frequently (Table 5). There were no cardiac deaths in the trial, but the incidence of grade 3/4 decreases in left ventricular cardiac function was significantly greater for patients treated with doxorubicin (n = 4 vs 20+; P < .001). The difference in cardiac function between patients treated with DCarboH and those treated with ACD (ie, the arm without trastuzumab) was not as striking; 4 and 7 patients, respectively, had grade 3/4 left ventricular cardiac function following treatment.
Is DCarboH equivalent to the best anthracycline regimens?
BCIRG 006 is a very large, well designed trial, and it involves a comparison with one of the most popular and more effective regimens currently in use to treat early breast cancer. The duration of chemotherapy was 6 weeks longer in the ACD-H arm than in the DCarboH arm, and it is plausible that if equal durations of chemotherapy had been used, results in the DCarboH arm would have been superior (although the differences in toxicity might have been less striking).
However, as with the US Oncology 9735 trial, this study was not designed as a non-inferiority trial, and the same limitations apply to the interpretation of “equivalency” (see above). The failure to demonstrate superiority does not mean that the DCarboH and ACD-H regimens are “equal” and “interchangeable.”
Further, only patients with HER2-positive cancers were enrolled in this study, and it is not clear that the results can be generalized to any other group of patients with early breast cancer.