Baseline patient and tumor characteristics differentially affected type of death in the MA.17 placebo-controlled letrozole trial where cardiovascular death was not separately identified. The MA.27 trial allowed competing risks analysis of breast cancer (BC), cardiovascular, and other type (OT) of death. MA.27 was a phase III adjuvant breast cancer trial of exemestane versus anastrozole. Effects of baseline patient and tumor characteristics were tested for whether factors were associated with (1) all cause mortality and (2) cause-specific mortality. We also fit step-wise forward cause-specific-adjusted models. 7576 women (median age 64 years; 5417 (72 %) < 70 years and 2159 (28 %) ≥ 70 years) were enrolled and followed for median 4.1 years. The 432 deaths comprised 187 (43 %) BC, 66 (15 %) cardiovascular, and 179 (41 %) OT. Five baseline factors were differentially associated with type of death. Older patients had greater BC (p = 0.03), cardiovascular (p < 0.001), and other types (p < 0.001) of mortality. Patients with pre-existing cardiovascular history had worse cardiovascular mortality (p < 0.001); those with worse ECOG performance status had worse OT mortality (p < 0.001). Patients with T1 tumors (p < 0.001) and progesterone receptor positive had less BC mortality (p < 0.001). Fewer BC deaths occurred with node-negative disease (p < 0.001), estrogen receptor-positive tumors (p = 0.001), and without adjuvant chemotherapy (p = 0.005); worse cardiovascular mortality (p = 0.01), with trastuzumab; worse OT mortality, for non-whites (p = 0.03) and without adjuvant radiotherapy (p = 0.003). Overall, 57 % of deaths in MA.27 AI-treated patients were non-breast cancer related. Baseline patient and tumor characteristics differentially affected type of death with women 70 or older experiencing more non-breast cancer death.
Keywords: Breast cancer death; Cardiovascular death; Competing risks; Elderly.