点击显示 收起
American Society of Clinical Oncology 2005 Annual Meeting
2005年5月13－17日
美国佛罗里达州奥兰多
May 13 - 17, 2005, Orlando, Florida
Chemotherapy for Breast Cancer
Joseph A. Sparano, MD
Introduction
This year's American Society of Clinical Oncology (ASCO) meeting was dominated by the results of trials demonstrating striking benefits for integration of biological agents targeting HER-2/neu in operable breast cancer and vascular endothelial growth factor (VEGF) in metastatic breast cancer. Still, as an integral part of breast cancer treatment, chemotherapy remains an important focus of research, with many remaining questions regarding optimal sequencing and combination therapies in both operable and metastatic disease. Several important studies at ASCO addressed these questions; the results are presented and discussed here.
Two important reports focused on the integration of taxanes as components of adjuvant therapy. Goldstein and colleagues,[1] for the Eastern Cooperative Oncology Group (ECOG) and the North American Breast Cancer Intergroup, reported the results of E2197, which compared 4 cycles of a standard doxorubicin-cyclophosphamide (AC) combination given every 3 weeks with 4 cycles of doxorubicin plus docetaxel (AT) in patients with 0-3 positive lymph nodes. The trial was powered to detect a 25% reduction in the disease-free survival (DFS) hazard rate (from an anticipated 5-year DFS of 78% for the AC arm to 83% for the AT arm). All patients received tamoxifen 20 mg daily for 5 years following completion of chemotherapy. The treatment arms were well balanced with regard to median age (51 years), proportion of lymph node-negative disease (65%), and estrogen receptor (ER)-positive disease (64%).
The results outlined in Table 1 indicate no advantage for the AT combination, with a higher rate of febrile neutropenia for the AT arm despite the routine use of prophylactic ciprofloxacin in this arm (500 mg twice daily x 10 days beginning on day 8 of each cycle). Although subgroup analysis demonstrated a trend toward a lower recurrence rate for patients with ER-negative disease, use of the AT regimen as given in this study and in this population was not recommended. It is noteworthy that the DFS of 84% in the AC control arm was higher than the 78% anticipated for this arm based on the historical data, consistent with other reports indicating that the prognosis for patients with early-stage disease continues to improve.
AC = doxorubicin 60 mg/m2 and cyclophosphamide 600 mg/m2; AT = doxorubicin 60 mg/m2 and docetaxel 60 mg/m2; DFS = disease-free survival; OS = overall survival
Dr. Gianni reported the results of the European Comparative Trial in Operable Breast Cancer (ECTO),[2] a trial addressing a somewhat similar question, but which evaluated adding a taxane to doxorubicin therapy rather than substituting a taxane for an alkylator in combination with doxorubicin, as in the E2197 trial. The study included 1355 women with operable breast cancer and a primary tumor greater than 2 cm who were randomized to 1 of 3 treatment arms: surgery followed by sequential doxorubicin (given every 3 weeks) followed by cyclophosphamide/methotrexate/5-fluorouracil (CMF) (arm A) surgery followed by doxorubicin-paclitaxel followed by CMF (arm B) or the same regimen as arm B given prior to surgery (arm C). All patients received tamoxifen 20 mg daily for 5 years following completion of chemotherapy. The study was primarily designed to address whether the addition of paclitaxel to doxorubicin reduced the risk of recurrence when given with a sequential doxorubicin-CMF regimen.
After a median follow-up of 43 months, the results indicated that addition of paclitaxel reduced the risk of relapse by 34% (hazard ratio, .66; P = .01) (see Table 2). There was no difference in outcome when comparing whether the sequential AP → CMF regimen was given either pre- or postoperatively. Twenty-three percent of patients in arm C had a pathological complete response (pCR); those who had a pCR demonstrated improved relapse-free survival compared with those who did not (89% vs 74%; P = .005). There was no difference in the incidence of grade 3-4 cardiac toxicity in arm A compared with arms B and C (.7% vs .4%). These data are consistent with other reports indicating that integrating taxanes as a component of adjuvant therapy reduces the risk of recurrence in operable breast cancer.
HR = hazard ratio; S = surgery; A = doxorubicin 75 mg/m2s; P = paclitaxel 200 mg/m2 intravenously over 3 hours; CMF = cyclophosphamide/methotrexate/5-flurouracil; CI = confidence interval
N. Katsumata reported, on behalf of the Japanese Clinical Oncology Group (JCOG), the results of a phase 3 trial comparing doxorubicin (40 mg/m2) and cyclophosphamide (500 mg/m2) (AC) for 6 courses, single-agent docetaxel (D, 60 mg/m2) for 6 courses, or AC alternating with D (AC/D) for 6 courses.[3] Patients receiving D or AC were crossed over to the other arm at progression, whereas patients who progressed after AC/D were re-treated with the same alternating regimen. The trial was powered to detect a 3.5-month improvement in median time-to-treatment failure for the docetaxel or alternating arms.
There were no significant differences in response rate, median time-to-treatment failure, or median overall survival (see Table 3). An unusual and important feature of the trial was a mandated crossover to the other treatment arm for the D arm and the AC arm, and a requirement that patients on the AC/D arm be rechallenged with the same treatment. There was no difference in the crossover response rate among the 3 arms. In an updated survival analysis provided at the meeting, there was a marginal survival advantage for the initial docetaxel arm compared with the initial AC arm (26.7 vs 22.6 months) that was statistically significant (P = .04) -- but only if a 1-sided rather than a 2-sided significance test was used, a nonstandard approach to interpreting clinical trial results.
TTF = time-to-treatment failure; OS = overall survival
S. Chan reported the results of a phase 3 trial comparing docetaxel (75 mg/m2 every 3 weeks) in combination with either gemcitabine (1000 mg/m2 days 1 and 8) (DG) or capecitabine (1250 mg/m2 twice daily, days 1-14) (DC).[4] The trial was powered to detect a 33% improvement in median progression-free survival. There was no difference in median time-to-treatment failure or progression-free survival between the 2 arms (see Table 4). There was significantly more treatment discontinuation for adverse events in the DC arm compared with the DG arm (28% vs 13%), primarily for hand-foot syndrome, diarrhea, or mucositis.
TTF = time-to-treatment failure; PFS = progression-free survival; DG = docetaxel-gemcitabine; DC = docetaxel-capecitabine
A. Gennari reported the results of a phase 3 trial evaluating the role of paclitaxel as maintenance therapy.[5] All patients received a maximum of 8 courses of paclitaxel (200 mg/m2 every 3 weeks) and epirubicin (90 mg/m2). Patients with responding or stable disease were randomized to receive 8 additional cycles of paclitaxel (175 mg/m2 every 3 weeks) or no further chemotherapy. Patients with ER-positive disease were permitted to receive concurrent hormonal therapy (treating physician's choice). The trial was powered to detect a 30% improvement in median time to progression, with an initial accrual goal of 550 patients to be randomized. Because of slow accrual, a futility analysis was performed after the enrollment of 215 eligible patients to determine whether there was any reasonable likelihood of achieving the primary end point.
The analysis demonstrated no significant difference in median progression-free survival, and it was concluded that the benefit sought would not likely be reached even if the trial had completed accrual as initially planned (Table 5). The results of this trial are not consistent with the results of other trials evaluating the role of maintenance chemotherapy, but may have been confounded by the concurrent use of hormonal therapy, or the potent nature of the anthracycline-taxane combination used as induction therapy.
TTP = time to progression; OS = overall survival
Although the big headlines for breast cancer at the American Society of Clinical Oncology (ASCO) 2005 meeting were dominated by the "new" targeted therapies, such as trastuzumab and bevacizumab, the original "targeted" treatment -- hormonal therapy -- proved a venerable stalwart with many interesting presentations.
The historic standard for adjuvant endocrine therapy was 5 years of tamoxifen, and to date, studies of extended tamoxifen therapy beyond 5 years have not demonstrated substantial improvement in clinical outcomes. This is not the case for the use of aromatase inhibitors (AIs) in that setting. In MA17,[1] women finishing 5 years of tamoxifen were randomly assigned to placebo or to the AI letrozole. Letrozole therapy was associated with a substantial reduction in the risk of recurrence in this extended adjuvant setting.
Until ASCO, there had been no confirmatory results for use of AIs in this setting. That changed with the presentation of the ABCSG-6a trial from the Austrian Breast and Colorectal Cancer Study Group.[2] The original trial, ABCSG 6,[3] had 2 arms. Patients in arm 1 received 2 years of tamoxifen given in conjunction with the first-generation AI aminoglutethimide, followed in turn by 3 years of tamoxifen alone. Patients in arm 2 received 5 years of tamoxifen. There was no difference in outcome between these 2 arms. In Trial 6a, patients free of recurrence through 5 years of treatment were randomly assigned to receive extended adjuvant endocrine therapy with 3 years of anastrozole or no further treatment.
Trial 6a showed a 36% reduction in risk of recurrence in favor of extended therapy with anastrozole, as compared with no further planned endocrine treatment. Lower rates of local-regional and distant metastatic events were noted. There was no difference in overall survival.
Although modest in size (n = 856) compared with the MA17 trial (n ~ 5000), ABCSG-6a offers important confirmatory evidence supporting the use of extended adjuvant therapy with an AI following 5 years of tamoxifen treatment.
The annual ASCO meeting gave the Breast International Group (BIG) investigators an opportunity to reprise their presentation of the BIG 1-98 trial, first reported at the St. Gallen, Switzerland, meeting in January 2005. BIG 1-98[4] was a very large trial that compared tamoxifen vs the AI letrozole as primary treatment for postmenopausal, estrogen receptor-positive (ER+) breast cancer. Nearly 8000 women were randomly assigned to tamoxifen or letrozole; median age was 61 years; 41% of patients had node-positive tumors; and 25% received chemotherapy as well as endocrine therapy. The trial showed a 19% lower risk of disease recurrence among women taking letrozole compared with tamoxifen (hazard ratio, .81; P value < .05). Through a projected 5-year follow-up, this would translate into a 2.6% reduction in recurrence events. As yet, there is no significant survival advantage for use of the AI. In general, the side-effect profile reported for letrozole in this study is similar to that seen in previous, large AI trials. However, there was a slightly greater risk of ischemic cardiac events in the letrozole group. It is not clear what accounts for this difference.
Preoperative endocrine therapy, although not widely used in the United States, is a viable strategy for tumor shrinkage and has been shown to be safe and effective in many trials. Dr. Semiglazov from Moscow, Russia, reported on a small, randomized phase 2 study comparing tamoxifen vs the AI exemestane as preoperative endocrine treatment for postmenopausal women with primary breast tumors.[5] A total of 151 women were randomly assigned 1:1 to the AI or tamoxifen. Patients received 3 months of therapy before surgery. The principal findings are shown in Table 1.
Several interesting observations can be made on the basis of this provocative phase 2 trial. First, it is a reminder that neoadjuvant endocrine therapy can be quite useful for shrinking tumors and for facilitating better surgical options for patients. Second, there is the suggestion that exemestane may be superior to tamoxifen in producing tumor response and allowing breast-conserving surgery among these women.
AIs have been studied in 3 adjuvant settings: (1) primary treatment, as an alternative to tamoxifen, (2) as sequential treatment, following 2-3 years of tamoxifen, and (3) as extended treatment, following 5 years of adjuvant tamoxifen. As yet, there are no data comparing these strategies head-to-head, and thus the optimal treatment strategy remains unknown. It has been argued that the differences between tamoxifen and AI treatment upfront warrant the initial use of such therapies. By contrast, the differences in early use of AI therapy (15% to 20% reduction in risk compared with tamoxifen in the Arimidex or Tamoxifen Alone or in Combination [6] and BIG 1-98 trials) are smaller than the gains seen with later use of AIs vs tamoxifen (35% to 50% gains after several years of tamoxifen).
One way to think about the best strategies is to generate computer models that make assumptions about the relative risks of recurrence at different time points, and to let the models predict which strategy would be best.
Two such modeling efforts were presented at ASCO. Investigators[7] at Dana-Farber updated their models to incorporate recent information from the ATAC trial. These data suggested that women with ER+/progesterone receptor-negative (PgR-) tumors fared particularly well with anastrozole compared with tamoxifen; by contrast, the results were more similar between the AI and tamoxifen groups in women with ER+/PR+ tumors. When applied to the modeling efforts that looked at 10-year disease-free survival outcomes, the different subsets of tumors had different "best" treatment strategies. For patients with ER+/PR- tumors, the models predicted that upfront therapy with an AI would be the best strategy. By contrast, patients with ER+ PgR+ ER+/PR+ tumors fared best at 10 years with a sequential strategy consisting of 2.5 years of tamoxifen followed by an AI.
The ATAC investigators[8] reported a related but slightly different model, which employed somewhat different assumptions compared with the model described above. Overall, the ATAC team reported that the best strategy would be upfront use of an AI. However, when analyzing events for ER+/PR+ vs ER+/PR- tumors, these models also suggested that the ER+/PR+ patients might benefit from crossover treatment strategies.
These models highlight the challenges in rapidly integrating new data into clinical treatment. Although models have many limitations for clinical utility, they underscore the fact that data are not available for every treatment scenario in clinical medicine, and that some modeling efforts may help frame a clinical discourse for doctors and patients.
One of the most exciting diagnostic tools to emerge in recent years has been the OncotypeDX recurrence score assay from Genomic Health. This assay quantifies the expression of 21 genes in breast tumors based on RNA retrieved from paraffin-embedded, formalin-fixed tissue. The recurrence score has been translated into an absolute risk of tumor recurrence through 10 years of follow-up among women with ER+, node-negative breast cancer. Investigators[9] from the National Surgical Adjuvant Breast and Bowel Project (NSABP) have sought to determine whether this recurrence score, which appears to risk-stratify patients very successfully, can also predict which tumors benefit from tamoxifen. They characterized the recurrence score for tumors of patients in NSABP B-14, a randomized trial of placebo vs tamoxifen. Patients in B-14 had node-negative, ER+ tumors. Previous studies had shown that the lower the recurrence score, the lower the risk of recurrence. New data have used the recurrence score to determine which patients may benefit from tamoxifen (Table 2).
These data suggest the following: First, tamoxifen dramatically reduces the risk of recurrence in some, but not all breast cancers. Tumors with low or intermediate recurrence scores, which tend to have higher levels of ER expression, grade 1-2 features, and be HER-2-negative, derive tremendous benefit from tamoxifen treatment, with nearly 50% reduction in risk. By contrast, tumors with high recurrence scores, which tend to be lower ER expressors and to have poorly differentiated features and sometimes HER-2 expression, derive minimal benefit from tamoxifen. Previous research by the same group has shown that patients with these high-recurrence-score tumors derive substantial benefit from chemotherapy. It is important to note that these data refer only to patients treated with tamoxifen; whether the recurrence score is valid among patients treated with an AI is not known.
A key clinical point is that we can now refine our understanding of which patients have a more favorable prognosis due to tumor sensitivity to tamoxifen, and conversely, which patients may yet derive substantial benefit from chemotherapy because they are relatively insensitive to endocrine manipulation. Thus, this type of testing has the potential to dramatically refine patient selection for important adjuvant clinical treatments.
By far, the most anticipated and exciting reports on breast cancer at the American Society of Clinical Oncology (ASCO) 2005 meeting were the results from 3 trials looking at the use of the humanized monoclonal antibody trastuzumab in the adjuvant setting for HER-2-positive operable breast cancer. A special scientific symposium[1] was created in order to allow the results of the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-31, North Central Cancer Treatment Group (NCCTG) N9831, and HERceptin Adjuvant (HERA) trial to be reported simultaneously. Each of these trials was a study of chemotherapy alone vs chemotherapy with trastuzumab. Collectively, these data strongly suggest that patients with high-risk stage I or stage II or III breast tumors that are HER-2-positive should receive adjuvant trastuzumab as part of their multimodality treatment program.
One year ago, the National Cancer Institute encouraged the NSABP and the North American Breast Intergroup to pool their data to allow an early analysis of events in their related trastuzumab trials. Because these studies both used doxorubicin-cyclophosphamide (AC) x 4 cycles, followed by paclitaxel with or without trastuzumab, it was believed that pooling the data was a reasonable clinical effort, and it was hoped that by pooling the data, enough events would be noted earlier to make a determination of whether adjuvant trastuzumab was worthwhile. It must be acknowledged that as a consequence of this pooled analysis, neither study reached its primary end point of events by itself, and neither would have had enough events to report for at least another year.
The 2 trials differed slightly. NSABP B-31 was open to patients with HER-2-positive, lymph node-positive operable breast cancer. Patients had HER-2 testing done by a certified laboratory. Patients were randomly assigned to 1 of 2 treatment arms -- either AC at standard doses x 4 cycles followed by paclitaxel at 175 mg/m2 x 4 cycles (arm 1), or the same chemotherapy regimen with the addition of weekly trastuzumab beginning during the paclitaxel phase of treatment (arm 2). The NSABP trial was later amended to allow weekly paclitaxel at 80 mg/m2 x 12 weeks as therapy. The primary end point of this trial was survival. (It should be noted that disease-free survival was used as the end point for the pooled results.)
NCCTG N9831 was a 3-arm study for patients with high-risk node-negative tumors (1 cm or larger, if estrogen receptor [ER]-negative; 2 cm or larger, if ER-positive) that were HER-2-positive (either immunohistochemistry 3+ or fluorescence in situ hybridization (FISH)+ by central laboratory testing). Patients were randomly assigned to AC x 4 followed by weekly paclitaxel x 12 weeks (arm A), or the same chemotherapy followed by trastuzumab (arm B), or the same chemotherapy with initiation of trastuzumab during the paclitaxel treatment (arm C). The primary study end point was DFS.
In both trials, patients assigned to receive trastuzumab received 52 weeks of trastuzumab therapy.
Arm 1 and arm A from the NSABP and NCCTG studies were jointly analyzed as AC → paclitaxel, whereas arm 2 and arm C became AC → paclitaxel + trastuzumab. Based on this similarity, pooled analyses of the dichotomized question "trastuzumab: yes/no" were reported.
Ed Romond[2] presented the data on behalf of the NSABP and the Intergroup. Together, the studies involved 3351 patients (736 from B-31 and 1615 from arms A and C from the N9831), with data presented with a mean follow-up of only 2 years. In the pooled analysis, trastuzumab reduced the risk of tumor recurrence by 55% (hazard ratio, .45; P value < .0001). This translated into substantial differences in estimated 3- and 4-year DFS (see Table 1). Despite the short follow-up, the pooled data also suggested an improvement in overall survival (hazard ratio, .67; P = .015).
DFS = disease-free survival
Similar results were seen for the NSABP and NCCTG trials when analyzed separately. Subset analysis showed remarkably consistent benefit for all subgroups of patients. Indeed, benefit was seen regardless of ER status (ER-positive and ER-negative), nodal status, or tumor size.
Edith Perez[3] presented the data from the NCCTG experience. Because this was a 3-arm study, it was designed to address both whether or not trastuzumab improved outcome, and whether sequential or concurrent trastuzumab therapy was optimal. A consequence of the early analysis is that the sequential vs concurrent analyses are underpowered for extensive comparison. However, there were provocative findings, namely, that concurrent chemotherapy + trastuzumab seemed better than sequential treatment (see Table 2).
AC = doxorubicin-cyclophosphamide
These findings show that it was not clear that sequential trastuzumab improved on chemotherapy alone -- but conversely, it seemed that concurrent chemotherapy + trastuzumab was substantially better than sequential treatment. Based on this observation, patients being treated in arm B of this trial were notified to be treated in arm C after the data had become publicly available.
Martine Piccart-Gebhart[4] presented the results of the world-wide HERA Trial. This study was a 3-arm trial, in which patients received no trastuzumab, or 1 or 2 years of trastuzumab in the adjuvant setting. Results were presented for the "trastuzumab: yes/no" question, without any attempt to analyze any differences in outcome between the 1-year vs 2-year duration of therapy. These data on the duration of treatment are not expected for several years. The HERA study was open to patients with high-risk node-negative (stage T1c) or node-positive tumors that were HER-2-positive. HER-2 testing was performed by a central laboratory. In contrast to the North American trials, the HERA study had closed to accrual, having met its accrual goal with over 5000 women last year. Data were presented with a median follow-up of approximately 1-2 years.
An important difference between the HERA and NSABP/NCCTG trials is that patients were enrolled in the HERA Trial after they had concluded their adjuvant chemotherapy and radiation (if given) treatment. Patients were eligible provided that they had received at least 4 cycles of adjuvant chemotherapy; the regimen(s) were not specified, although 68% received anthracycline-based chemotherapy and 25% received anthracycline- and taxane-based chemotherapy. Patients could start treatment with trastuzumab on study within 6 months of finishing chemotherapy, although the vast majority proceeded directly from chemotherapy to study entry. Another difference between HERA and the North American trials is that patients in HERA received trastuzumab every 3 weeks (6 mg/kg) instead of the weekly (2-mg/kg) regimen.
Patients assigned to adjuvant trastuzumab had a significant improvement in their DFS compared with those not treated with trastuzumab. Again, the risk reduction was on the order of 50% (hazard ratio, .54; P value < .001). Again, subset analyses suggested benefit in ER-positive and ER-negative patients as well as lymph node-positive and node-negative patients. The estimated 2-year DFS for chemotherapy alone was 77% vs 86% with trastuzumab.
A major concern in the adjuvant trastuzumab studies is the risk of cardiac toxicity. Previous experience suggested a high rate of congestive heart failure among patients receiving trastuzumab in combination with anthracyclines. Therefore, all 3 trials conducted extensive cardiac surveillance throughout the studies, and reported these data at ASCO.
Patients were screened at baseline for eligibility, requiring a normal left ventricular ejection fraction (LVEF) at baseline, and followed frequently thereafter for changes in LVEF. Patients with cardiac symptoms or with significant asymptomatic declines in LVEF were taken off study.
Trastuzumab appears to increase the risk of clinically -apparent heart failure. In the pooled NSABP/NCCTG experience, the risk of cardiac events was 4% among those treated with chemotherapy and trastuzumab, as compared with .6% for those given chemotherapy alone. There are data suggesting that the risk is lower if sequential chemotherapy → trastuzumab is given (see Table 3) here are also suggestive data that older patients and those patients with borderline normal LVEF at baseline may be at greater risk for cardiac events (see Table 4).
NA = not available
LVEF = left ventricular ejection fraction
A fourth adjuvant trial of trastuzumab has yet to be reported. This study, BCIRG 006, compared AC → docetaxel vs AC → docetaxel + trastuzumab vs carboplatin-docetaxel + trastuzumab. In addition to asking whether adjuvant trastuzumab was beneficial, it will allow the determination of whether a non-anthracycline-based adjuvant chemotherapy regimen may offer clinical advantages in the adjuvant trastuzumab setting. At an Educational Session at ASCO, Dennis Slamon[5] related the absolute incidence of cardiac events reported to data in this trial (see Table 5). It would appear that the nonanthracycline docetaxel-carboplatin + trastuzumab regimen had the lowest risk of cardiotoxicity. It is not known whether the regimen has other side effects or comparable efficacy to AC → docetaxel-based treatments.
LVEF = left ventricular ejection fraction
Based on data from 3 large, prospective clinical trials, trastuzumab clearly improves outcomes for women with HER-2-positive, operable breast cancer. These findings have immediate significance for current clinical practice.
Patients with high-risk node-negative or node-positive tumors that are HER-2-positive clearly benefit. It is not known how the results translate to very low-risk patients (tumors less than 1 cm) or to patients not given adjuvant chemotherapy.
At the present time, 1 year is the established duration. It is not known whether such a long -duration of treatment is necessary, nor conversely, whether longer treatment might be beneficial.
The available data suggest that either IHC or FISH testing is appropriate, provided that the testing is done in experienced laboratories with well-documented false-positive and false-negative results. If such quality control is not available at local laboratories, then centralized laboratory testing is warranted.
The North American experience is structured around AC → paclitaxel therapy. For patients in North America, that remains the regimen of choice. It is noteworthy that the HERA Trial did not specify a chemotherapy regimen, and saw benefit regardless of the chosen regimen. It is probably the case that the modest differences between chemotherapy regimens are "trumped" by the dramatic effect of trastuzumab on reducing the risk of tumor recurrence. However, AC → paclitaxel is the regimen for which the most safety data are available.
This is a challenging question to address. Both strategies look to be useful: In the North American trials, concurrent therapy was optimal, but in the HERA Trial, sequential therapy was also quite effective. The comparison of sequential vs concurrent therapy in the NCCTG trial is provocative, and suggests at least that for women getting AC → paclitaxel, concurrent therapy with paclitaxel is better. But this comparison is limited by the small number of events and hence is underpowered. And the cardiotoxicity concerns for concurrent therapy remain.
At a minimum, patients need to be screened at baseline, and have LVEF checked again after AC chemotherapy, and again after several months of trastuzumab therapy. A high index of clinical suspicion is warranted for discontinuing treatment in the setting of cardiopulmonary symptoms. Patients should be carefully informed that although trastuzumab does seem to reduce the risk of breast cancer recurrence, it also carries a greater risk of cardiotoxicity. Older patients and those with borderline normal LVEF at baseline seem at particular jeopardy.
Given the wide availability of adjuvant trastuzumab trials, it seems reasonable to assume that most patients with HER-2-positive tumors have been offered participation in protocols over the past several years. Not all patients will have participated in these studies, however. The benefits of adding trastuzumab in the months or years after chemotherapy are not well established. The National Cancer Institute trials (NSABP and NCCTG) have recommended that patients who finished chemotherapy without trastuzumab within the past 6 months be offered trastuzumab therapy. This window of treatment is arbitrary although reasonable, and can be used to frame the issue with nonprotocol patients.
This year's American Society of Clinical Oncology (ASCO) annual meeting provided more than its share of buzz at the late-breaking session during which results of trial E2100 were presented. To a standing-room only crowd, investigators from the Eastern Cooperative Oncology Group (ECOG) reported positive results with first-line use of an antiangiogenic agent in metastatic breast cancer. Specifically, patients receiving the humanized monoclonal antibody bevacizumab plus a taxane enjoyed a longer period of progression-free survival compared with those who received the taxane alone. These groundbreaking results raise several questions, including how do they compare with prior findings and should the standard of care change? The study and its implications are discussed here.
Trial E2100[1] was a randomized study comparing paclitaxel alone vs paclitaxel plus bevacizumab as first-line chemotherapy for patients with locally recurrent or metastatic breast cancer. Bevacizumab is a humanized monoclonal antibody directed against the vascular endothelial growth factor (VEGF) receptor; it is believed to work by inhibiting vascular proliferation in tumors. In this study, patients received paclitaxel, 90 mg/m2, administered on days 1, 8, and 15 of a 28-day cycle and bevacizumab 10 mg/kg administered on days 1 and 15, or paclitaxel alone at the same dose.
Between 2001 and 2004, 715 eligible patients were randomized to one of these 2 treatments. Stratification factors included disease-free interval from diagnosis (greater or less than 24 months), number of metastatic sites (3 or fewer or greater than 3), having received adjuvant chemotherapy or not, and hormone receptor status. Patients had to have received no prior chemotherapy for metastatic disease, and if they received a taxane as adjuvant therapy, they must have had at least a 12-month disease-free interval. Patients were not eligible if they had central nervous system metastases, and could not be on therapeutic anticoagulation. Patients also could not have significant proteinuria.
Sixty-five percent of patients had received adjuvant chemotherapy at least 12 months prior to enrollment, and 63% had estrogen receptor-positive tumors. Although exact figures were not given, the initial review suggests that almost all of the patients had tumors that did not overexpress HER-2. Forty-one percent of patients had a disease-free interval from diagnosis of less than 24 months.
Objective overall response rates were 14.2% for paclitaxel alone and 28.2% for paclitaxel plus bevacizumab (P < .0001) (Figure). Progression-free survival (PFS) was also superior for patients receiving bevacizumab plus paclitaxel, with a median of 11 months for the combination vs 6.1 months for paclitaxel alone (log rank test P < .001). Overall survival (OS) was also improved with the addition of bevacizumab to paclitaxel, and although medians have not been reached, the hazard ratio for death is .674, significant at P = .01.
Figure. Bevacizumab and paclitaxel as first-line therapy for patients with metastatic breast cancer.
Toxicity was not trivial, but acceptable. As expected, hypertension (13% grade 3 and less than 1% grade 4) and proteinuria (.9% grade 3 and 1.5% grade 4) were seen in patients receiving bevacizumab. Surprisingly, more neuropathy was seen in patients receiving the combination therapy (9.9% grade 3) than in those receiving paclitaxel alone (13.6%; P = .01). Grade 4 neuropathy was seen in less than 1% of patients in either treatment arm. It is possible that patients in the combination arm had a greater incidence of neuropathy due to the fact that they had a greater than 4-month prolongation in PFS, and therefore received paclitaxel for a longer period of time. The association between the duration of paclitaxel treatment and neuropathy is well known. Bleeding and thromboembolic events were rare in both treatment arms. Therefore, significant benefit was achieved with the combination of paclitaxel and bevacizumab without a high price in toxicity.
How do we reconcile these data with the study performed by Miller and colleagues[2] comparing capecitabine with and without bevacizumab? In that study, 462 patients who had undergone 1 or 2 previous chemotherapy regimens for metastatic breast cancer were randomized to receive capecitabine 2500 mg/m2/day for 14 days of a 21-day cycle alone or with bevacizumab 15 mg/kg, administered on day 1 of each cycle. Although the response rate was higher in the arm receiving both bevacizumab and capecitabine, (19.8%) compared with those receiving capecitabine alone (9.1%), neither PFS (4.86 months for the combination vs 4.17 months for capecitabine alone) or OS (15.1 months for the combination vs 14.5 months for capecitabine alone) was improved with the combination therapy.
Could it be that bevacizumab is more active with paclitaxel than with capecitabine? This is unlikely based on data from the colorectal trials in which bevacizumab clearly has shown benefit in combination with 5-fluorouracil-based regimens. Although it is possible that the biology of breast cancer makes the combination of capecitabine and bevacizumab less active than in colorectal cancer, this is unlikely because capecitabine has significant antitumor activity against breast cancer. A more likely explanation is that in the capecitabine-bevacizumab study, approximately 45% of patients had received 1 previous chemotherapy regimen for metastatic disease and approximately 35% had received 2 such regimens. The ability of any agent to improve either PFS or OS in this setting is considerably less, and it is quite possible that if this study had been designed for first-line treatment of metastatic disease, a PFS and OS advantage might have been seen.
How does the benefit of the combination of bevacizumab and paclitaxel shown in this study compare to the benefit of trastuzumab combined with paclitaxel in patients with metastatic breast cancer that overexpresses HER-2? In Slamon's original publication of chemotherapy with and without trastuzumab as first-line chemotherapy for women with metastatic breast cancer, time-to-disease progression was extended from 2.9 months to 6.9 months by the addition of trastuzumab.[3] This 4-month benefit is very similar to that seen in the paclitaxel-bevacizumab trial in which the benefit was just over 4.5 months. Time-to-disease progression was shorter in the trastuzumab trial presumably because tumors that overexpress HER-2 are biologically more aggressive, with a shorter time-to-tumor progression. In addition, however, the response rates were very similar in the 2 studies. In the trastuzumab trial, the overall response rate to paclitaxel alone was 17%, and the overall response rate to paclitaxel combined with trastuzumab was 41%. In this trial, the response rate to paclitaxel alone was 15%, and to the combination of paclitaxel and bevacizumab, it was 28.2%.
It is also important to note that both in the case of trastuzumab combined with paclitaxel and bevacizumab combined with paclitaxel, we have seen significant prolongations in PFS and OS, whereas 30 years of clinical investigation with cytotoxic chemotherapy failed to significantly prolong either. In an analysis of Cancer and Leukemia Group B (CALGB) trials and those of the Autologous Blood and Marrow Transplant Registry, Berry and colleagues[4] found no evidence of significant extension of survival in both standard-dose and high-dose trials throughout the 1980s and 1990s. For 4 major combination chemotherapy trials performed by CALGB in the 1980s and 1990s, the overall survival was approximately 22 months in each. In a more recent trial studying doxorubicin and docetaxel, a combination that some believe is the most active chemotherapy regimen for patients with metastatic breast cancer, the time-to-treatment failure was just over 6 months, and the OS was 22.5 months.[5] Cytotoxic chemotherapy may be reaching its limits of benefit, and additional progress will be made by the addition of biological agents, such as bevacizumab.
The results of trial E2100 are quite dramatic and, in the context of a large, multicenter, randomized trial, are very convincing. Based on these results, should paclitaxel plus bevacizumab now be considered standard of care as first-line chemotherapy for patients with metastatic breast cancer whose tumors do not overexpress HER-2/neu? Although confirmatory studies add comfort to any decision to change practice, this trial is a large, cooperative group study that was well designed and executed with very convincing data, and although we will learn more from future studies, it is certainly appropriate to consider the use of bevacizumab together with paclitaxel as front-line chemotherapy for these patients.
Several other questions remain unanswered. Will bevacizumab have a role in the treatment of breast cancer with other chemotherapy agents, such as vinorelbine and capecitabine, and possibly doxorubicin? Will bevacizumab be safe and improve the efficacy of hormonal therapy, the true first-line therapy for the majority of patients with metastatic breast cancer? If the addition of bevacizumab improved time-to-treatment failure for hormonal agents, then the time to chemotherapy probably could be extended for these patients. In addition, the combination of bevacizumab with trastuzumab is currently under study, and those results may have important implications for patients with tumors that overexpress HER-2. Until data are available to answer these questions, it is probably wise to limit treatment with bevacizumab to the way it was used in this study. In addition, as illustrated in the previously mentioned study [2] that combined bevacizumab with capecitabine as second- or third-line therapy and failed to show benefit bevacizumab should only be used with front-line chemotherapy at this point, and probably should not be continued after there is progressive disease.
These data also demand that we study the use of this agent in the adjuvant setting, following the lead of trastuzumab, which has now been shown to have substantial benefit in the adjuvant as well as the metastatic setting. Improved outcome in the adjuvant setting translates into more women remaining disease-free and well, rather than merely prolonging time-to-tumor progression, with OS still measured in months.
Finally, cost cannot be ignored, and we need to at least consider the effects of this therapy on our healthcare system. A year of bevacizumab given as it was in this study is likely to cost approximately $100,000. Having said that, this study is the first to have demonstrated the effectiveness of antiangiogenesis in metastatic breast cancer and will clearly have a significant impact on patient care moving forward.