Results
Efficacy Analysis
Between June 1998 and June 2001, 2887 patients from 173 centers in 21 countries were enrolled. Patient characteristics are summarized in Table 1 and were well balanced between the treatment arms. Intervention and follow-up are summarized in Supplementary Table S1,availableatAnnals of Oncology online. In May 2009, the median follow-up was 93.4 months with 916 DFS events and 566 deaths. The actual event rate was lower than anticipated.
For the primary analysis (A-T + AT versus A + AC), addition of docetaxel did not improve DFS (HR = 0.91, 95% CI = 0.80–1.05, P = 0.187) or OS (HR = 0.91, 95% CI = 0.77–1.08, P = 0.28). For secondary end points, sequential docetaxel (A-T) improved DFS compared with sequential control (A) (HR = 0.81, 95% CI = 0.67–0.99, P = 0.036). Sequential docetaxel (A-T) was superior to concurrent docetaxel (AT), for both DFS (HR = 0.84, 95% CI = 0.72–0.99, P = 0.035) and OS (HR = 0.79, 95% CI = 0.65–0.98, P = 0.028), see Table 2 and Figure 1.
(Enlarge Image)
Figure 1.
Disease-free survival (DFS) and overall survival (OS) at 8-year median follow-up divided by the treatment arm (Kaplan–Meier plots). (A) DFS; (B) OS. A: doxorubicin (A) → cyclophosphamide, methotrexate, 5-fluorouracil (CMF). AC: AC → CMF. A-T: A → docetaxel (T) → CMF. AT: AT → CMF.
Exploratory Biological Subtype Analysis
FFPE primary tumors were provided by 2172 patients (75%). Central laboratory reassessment and categorization by IHC-defined biological subtypes were possible in 1777 patients (62%). The remaining 395 patients had inadequate tissue quantity and/or quality.
The substudy patients were representative of the entire population with no substantial differences in patient and tumor characteristics, or DFS and OS, compared with patients not included (data not shown). Patients per tumor subtype: luminal-A N = 294 (17%); luminal-B N = 1034(58%); HER2 positive N = 149 (8%) and triple negative N = 300 (17%). Of luminal-B tumors, 181 of 1034 (18%) were positive for HER2. Within triple-negative disease, 146 of 300 (49%) were basal-like and 42 (14%) were non-basal-like. The remaining 112 (37%) triple-negative tumors had inadequate tissue for EGFR, CK5/6 and/or CK14 determination. Patient characteristics according to subtype are summarized in Table 3.
Subtypes were prognostic for DFS (see Figure 2). DFS was most favorable for luminal-A, and least favorable for triple negative and HER2 positive. HR derived from pairwise comparison with luminal-B revealed HR = 0.66 (95% CI = 0.50–0.86, P = 0.0018) for luminal-A, HR = 1.75 (95% CI = 1.35–2.28, P < 0.0001) for HER2 positive and HR = 1.37 (95% CI = 1.11–1.69, P = 0.0039) for triple negative.
(Enlarge Image)
Figure 2.
Prognostic evaluation of molecular subtypes: disease-free survival (DFS) analysis (Kaplan–Meier plots).
In the luminal-B subdivision, the 8-year DFS rates were 68.0% for luminal-B HER2 negative and were 58.5% for luminal-B HER2 positive. In the triple-negative subdivision, the 8-year DFS rates were 62.5% for basal-like and 47.1% for non-basal-like.
A multivariate DFS analysis included the following variables: age (<50 versus ≥50 years), body mass index (<30 versus ≥30), menopausal status, histopathological type, tumor size, grade, number of positive nodes (1–3 versus ≥4), mastectomy, chemotherapy (sequential versus concurrent), treatment (A + AC versus A-T + AT), radiotherapy, hormonotherapy usage and subtype (with comparison to luminal-B). After adjustment of statistically significant prognostic factors, luminal-A maintained a significantly better prognosis (HR = 0.74, 95% CI = 0.57–0.97, P = 0.03).
The substudy patients showed a trend favoring sequential docetaxel over control therapy for improved DFS (HR = 0.80, 95% CI = 0.65–1.00). The HR favored docetaxel in all subtypes, except luminal-A, as expected. The test for heterogeneity, however, failed to reach statistical significance (P = 0.38) (see Figure 3A).
(Enlarge Image)
Figure 3.
DFS analysis: predictive evaluation of molecular subtypes: (A). Predictive value of breast cancer subtypes comparing combined control arms (A + AC) with the sequential docetaxel arm (A-T). (B). Predictive value of basal-like and non-basal-like triple-negative subsets comparing concurrent control (AC) with the sequential control arm (A). A: doxorubicin (A) → cyclophosphamide, methotrexate, 5-fluorouracil (CMF); AC: AC → CMF; A ->T: A → docetaxel → CMF; CI: confidence interval; HR: Hazard ratio; N: number of patients exposed to the risk; O: number of observed events.
In triple-negative basal-like and non-basal-like subsets, concurrent control (higher cyclophosphamide dosing) was compared with sequential control. In both the groups, the HR favored concurrent AC, as expected. However, patient numbers were quite small, particularly for non-basal-like (see Figure 3B).