AIs should be considered as part of adjuvant endocrine therapy for postmenopausal women with HR-positive, early-stage breast cancer. Adjuvant studies of AIs can be separated into three categories: (1) initial therapy: head-to-head comparison of an AI vs tamoxifen(Drug information on tamoxifen) for 5 years; (2) sequential: comparison of tamoxifen for 2 to 3 years followed by an AI to complete 5 years vs tamoxifen for 5 years or vs an AI for 5 years; and (3) extended adjuvant: evaluation of the benefit of adding an AI after 5 years of tamoxifen.
The EBCTCG conducted a meta-analysis of randomized trials of AIs compared to tamoxifen (Table 1). In two randomized, prospective trials consisting of 9856 patients, AIs were compared to tamoxifen monotherapy. With a mean 5.8 years of follow-up, there was a significant reduction in recurrence but not in breast cancer mortality. In four trials (German Adjuvant Breast Cancer Group/Arimidex-Nolvadex [ARNO 95], Intergroup Exemestane(Drug information on exemestane) Study/Breast International Group [BIG] 02-97, Italian Tamoxifen Anastrozole(Drug information on anastrozole) [ITA], and Austrian Breast and Colorectal Cancer Study Group [ABCSG] 8), a total of 9015 patients were randomly assigned after 2 to 3 years of tamoxifen to either continue tamoxifen or switch to an AI, both for a total of 5 years of endocrine therapy. With a mean follow-up of 3.9 years, switching to an AI demonstrated a reduction in the risk of recurrence and mortality. There was no difference in benefit associated with age, nodal status, tumor grade, or PR status.
Another independent meta-analysis of studies comparing AIs to tamoxifen examined nine randomized controlled trials, which included 28,632 women and looked at three treatment strategies: monotherapy, sequential therapy (switching), and extended therapy (after tamoxifen) (Table 2). DFS was significantly better for AIs as monotherapy and sequential therapy. There was no difference in OS between monotherapy and extended therapy, but OS was prolonged for patients who switched from tamoxifen to AI therapy. The conclusion of the authors was that sequential therapy appears to be the preferred treatment.
AIs are not indicated for the adjuvant treatment of premenopausal women. The use of AIs with LHRH agonists is being studied, but data available to date do not suggest an advantage over tamoxifen and OvS/OA. ABCSG 12 randomly assigned 1803 premenopausal women with ER-positive, stage I/II breast cancer to receive either goserelin(Drug information on goserelin) (Zoladex) and tamoxifen or goserelin and anastrozole (Arimidex). At a median follow-up of 62 months, there was no difference in DFS (HR, 1.08; 95% CI, 0.81–1.44; P = .591), but OS was shorter with anastrozole than with tamoxifen (HR, 1.75; 95% CI, 1.08–2.83; P = .02). A subset analysis of this trial based on body-mass index (BMI) calculations showed that for overweight patients treated with anastrozole, there were significant increases in the risk of disease recurrence (HR, 1.49; 95% CI, 0.93–2.38; P = .08) and death (HR, 3.03; 95% CI, 1.35–6.82; P = .004) compared with patients treated with tamoxifen.
AIs as initial therapy
The Arimidex, Tamoxifen, Alone or in Combination (ATAC) trial was the first to show a benefit for an AI (anastrozole) over tamoxifen. In this double-blind, placebo-controlled trial, 9366 postmenopausal women were randomly assigned to receive anastrozole alone, tamoxifen alone, or the combination of both drugs for 5 years (Table 3). The combination arm was closed due to inferiority. With a median follow-up of 120 months, the primary endpoint of DFS was significantly longer with anastrozole. Anastrozole was also associated with a longer time to recurrence, longer time to distant recurrence, and decreased contralateral breast cancers. The greatest relative reductions in DFS, time to recurrence, and contralateral breast cancer were in the first 2 years of active therapy, but these differences were sustained throughout the entire follow-up period, and persisted after treatment completion. There was no significant difference in OS.
The BIG 1-98 trial showed that letrozole (Femara) was more beneficial than tamoxifen (Table 4). In this phase III, double-blind trial, 8010 postmenopausal women with ER-positive and/or PR-positive breast cancer were randomly assigned to 5 years of hormonal therapy in one of four arms: tamoxifen, letrozole, tamoxifen for 2 years followed by letrozole for 3 years, or letrozole for 2 years followed by tamoxifen for 3 years. The primary endpoint was DFS. The monotherapy arms included 4922 women. With a median follow-up of 51 months, there was an advantage in DFS for letrozole compared to tamoxifen. An analysis at 71 months of follow-up showed a nonsignificant difference in OS favoring letrozole. When the initial results were announced in January 2005, 25.2% of the women randomized to tamoxifen monotherapy crossed over to letrozole. With a median of 76 months of follow-up, inverse probability of censored weighted (IPCW) modeling was used to gain better estimates of the relative treatment effects in the presence of the selective crossover. The DFS benefit for letrozole monotherapy was confirmed, and there was a benefit in OS as well. Letrozole was favored in nearly all subgroups.
Sequential use of tamoxifen and AIs
Some trials have compared switching to an AI after 2 to 3 years of tamoxifen to tamoxifen alone, for a total of 5 years of therapy. In the International Exemestane Study (IES), 4724 postmenopausal women with ER-positive or ER- unknown tumors, who were disease-free after 2 to 3 years of tamoxifen therapy, were randomly assigned to continue tamoxifen or switch to exemestane (Aromasin) (Table 5). At a median follow-up of 55.7 months, DFS was superior for those who switched. The OS was not significantly improved for the entire group of patients who switched, but there was a significant improvement in OS when the 122 patients who had ER-negative tumors were excluded (HR, 0.83; 95% CI, 0.69–1.00; P = .05). In the most recent publication of data from the trial, at a median follow-up of 91 months, the benefit in those patients who switched was sustained. There continued to be a benefit in breast cancer–free survival (BCFS) in the ER-positive and ER-unknown group, and an OS benefit was seen for the patients who switched to exemestane.
A meta-analysis was performed on three trials that compared switching to anastrozole after 2 to 3 years of tamoxifen with 5 years of tamoxifen (Table 6). The studies included ABCSG 8, ARNO 95, and ITA. These trials enrolled a total of 4,006 postmenopausal women with hormone-sensitive, early-stage breast cancer. In ARNO 95 and ITA, only patients who were relapse-free after 2 to 3 years of tamoxifen were randomized. In contrast, the ABCSG 8 study randomized patients at diagnosis. In this meta-analysis, the switch to anastrozole demonstrated an advantage in DFS, as well as significant improvements in event-free survival, distant disease-free survival (DDFS), and OS. Switching to anastrozole was beneficial, irrespective of nodal status, PR status, previous chemotherapy, or tumor size. An update of the ABCSG 8 trial with a median follow-up of 60 months showed a modest, statistically nonsignificant improvement in the primary endpoint of RFS, and a significant improvement in the defined exploratory endpoint of distant relapse-free survival (DRFS) (see Table 6).
The National Surgical Adjuvant Study Breast Cancer 03 (N-SAS BC03) trial involved 706 Japanese women randomly assigned after 1 to 4 years of tamoxifen to either continue tamoxifen or switch to anastrozole, for 5 years of total adjuvant endocrine therapy. At a median follow-up of 42 months, the primary endpoint of DFS as well as a secondary endpoint of RFS favored a switch to anastrozole (see Table 6).
Some trials have compared a switching strategy to AI monotherapy (Table 7). In the BIG 1-98 trial, the two sequential arms (tamoxifen for 2 years followed by letrozole for 3 years; letrozole for 2 years followed by tamoxifen for 3 years) were compared to letrozole monotherapy. With a median follow-up of 71 months, there was no significant difference in DFS for either switching arm compared to monotherapy.
The Tamoxifen Exemestane Adjuvant Multinational (TEAM) phase III trial was initially designed to compare 5 years of exemestane monotherapy to 5 years of tamoxifen monotherapy. Women were randomly assigned to receive one agent or the other at treatment initiation. However, based on favorable results from the IES trial, the study design was amended to compare tamoxifen for 2 to 3 years followed by exemestane until the completion of 5 years vs exemestane for 5 years. This trial randomized 9229 postmenopausal women, and at a median follow-up of 5.1 years, it showed no difference in the primary endpoint of DFS.
The designs of these switching trials must be taken into consideration. In the IES, ARNO 95, and ITA trials, patients were randomized only if they were disease-free at 2 to 3 years, thereby excluding patients with a poor prognosis whose disease would have recurred before the randomized switch. A true test of the switching strategy requires randomization at baseline, as in the ABCSG 8, BIG 1-98, and TEAM trials. Furthermore, this is the only study design that helps the practicing oncologist, who cannot anticipate whether a patient's disease is going to recur in the first 2 to 3 years of tamoxifen treatment.
Extended adjuvant use of AIs
One large trial evaluated the use of AIs in the extended adjuvant setting (Table 8). MA.17 was a phase III, randomized, double-blind, placebo-controlled trial in which 5187 postmenopausal women who had completed about 5 years of adjuvant tamoxifen were randomly assigned to receive another 5 years of therapy with letrozole or placebo. At a median of 2.5 years, there were significant improvements in the primary endpoint of DFS as well as DDFS for the women who received letrozole. There was no difference in OS between the two groups overall, but for the subset of lymph node–positive patients OS was statistically significantly improved for those who received letrozole. Because the initial interim analysis showed an improvement in DFS for the letrozole arm, the independent data and safety monitoring committee recommended unblinding the study. Patients in the placebo arm were offered letrozole, at a median of 2.8 years after the completion of tamoxifen. Of the 2594 patients in the placebo arm, 66% (1579 patients) chose to start letrozole. When compared to the women in the placebo arm who chose no further therapy, these women had an improvement in DFS and DDFS.
In an intent-to-treat (ITT) analysis with a median follow-up of 64 months, there was still a significant difference in DFS, but no significant differences in DDFS or OS, between the groups created in the original randomization to letrozole or placebo. These results were difficult to interpret and likely underestimated the benefit of letrozole, given that nearly two-thirds of patients randomly assigned to placebo crossed over to active treatment. In the most recent analysis, two approaches were used to adjust for the crossover from placebo to active treatment: IPCW modeling and a Cox model with a time-dependent treatment covariate (see Table 8). Both approaches suggested that letrozole showed statistically significant improvements in DFS, DDFS, and OS that had not been significant in either the first interim analysis or the post-blinding ITT analysis.
Two smaller studies also evaluated AIs in the extended adjuvant setting. NSABP B-33 was a randomized, double-blind, phase III trial in which patients with T1-3, N1, M0 breast cancers who were disease-free after 5 years of tamoxifen were randomly assigned to 5 years of exemestane or placebo. The primary endpoint was DFS. Accrual was terminated early, patients were unblinded, and the patients receiving placebo were offered the opportunity to cross over to exemestane after October 2003, when the MA.17 trial data were released. At that time, of the 1598 patients on trial, 72% of those randomized to exemestane continued therapy, whereas 44% of those on placebo chose to start exemestane. In an ITT analysis, there was a borderline statistically significant improvement in 4-year DFS favoring the AI, 91% for exemestane vs 89% for placebo. In a second study, ABSCG 6a, 856 patients who were disease-free at the end of the ABCSG 6 trial (which compared 5 years of tamoxifen with or without aminoglutethimide [Cytadren] for the first 2 years) were randomly assigned to receive 3 years of anastrozole or no further therapy. With a median follow-up of 62.3 months, there was a statistically significant reduction in the risk of recurrence for patients who received anastrozole.
Thus, on the basis of these three clinical trials, there is strong evidence for the value of extended adjuvant therapy with AIs following 5 years of tamoxifen.
Comparison of different AIs
The benefits of AI therapy appear to be a “class effect,” and the different AIs appear to have equivalent efficacy. This was confirmed in the MA.27 trial, which compared anastrozole to exemestane as initial adjuvant therapy for 5 years in 7576 postmenopausal women with HR-positive primary breast cancer. There was no difference between the two arms for the primary endpoint, event-free survival (HR, 1.02; P = .85), or for the secondary endpoints of OS, DDFS, or contralateral breast cancer.
Biomarker analysis and benefit from AIs
An initial exploratory analysis from the ATAC trial suggested that the benefit of anastrozole compared to tamoxifen was substantially greater in the PR-negative subgroup. A subsequent central analysis of ER, PR, and human epidermal growth factor receptor 2 (HER2) expression in 2006 of the 5880 specimens showed that quantitative expression of ER, PR, and HER2 did not identify patients with differential relative benefits from anastrozole. In the BIG 1-98 study, central review of ER, PR, HER2, and Ki-67 expression was performed on 84% of tumors. Much as in the ATAC trial, in patients with at least some ER expression, PR levels were not predictive of a relative benefit of letrozole over tamoxifen monotherapy. When analyzed for a differential treatment effect, each of the four markers showed a trend favoring letrozole monotherapy at the “higher end” of its spectrum, but none of the markers significantly predicted treatment selection. A composite measure of prognostic risk that incorporated clinical, pathological, and biological markers was better able to predict the relative benefits of the different regimens. For patients at the highest risk, the best outcome was achieved with letrozole monotherapy. For patients at intermediate risk, any of the three letrozole-containing regimens performed equally well. Patients at the lowest risk did equally well with letrozole monotherapy, either sequence of letrozole and tamoxifen, or tamoxifen monotherapy. A biomarker analysis of the TEAM trial showed that both ER and PR quantitative expression were associated with DFS and that the relative risk of relapse increased proportionally with decreased expression of either. PR status did not predict an additional benefit of exemestane over tamoxifen.