Background
EBCTCG individual-patient-data meta-analysis of 31,920 postmenopausal women with HR+ early breast cancer, pooling randomized trials comparing aromatase inhibitor (AI) versus tamoxifen-based adjuvant endocrine therapy.
Interventions and follow up
Design: EBCTCG meta-analysis across three comparison cohorts (5yr AI vs 5yr TAM; 5yr AI vs 2-3yr AI then TAM; 2-3yr TAM then AI vs 5yr TAM)
Primary endpoint: Breast cancer recurrence and breast-cancer mortality
Other outcomes: death without recurrence, all-cause mortality
Primary endpoint: Breast cancer recurrence and breast-cancer mortality
Other outcomes: death without recurrence, all-cause mortality
Results
Recurrence (during AI exposure years): AI reduced recurrence by ~30% vs tamoxifen (RR ~0.70).
10-yr breast-cancer mortality, AI vs TAM (5yr vs 5yr cohort): 12.1% vs 14.2%; RR 0.85, 95%CI 0.78-0.93, P=.001.
Cohort 1 (5yr AI vs 5yr TAM): 10-yr breast-cancer mortality 12.1% vs 14.2% (RR 0.85, 95%CI 0.75-0.96, P=.009).
Cohort 2 (5yr AI vs 2-3yr AI then TAM): recurrence favored AI (RR 0.90, 95%CI 0.81-0.99, P=.045); breast-cancer mortality not significant (RR 0.89, 95%CI 0.78-1.03, P=.11).
Cohort 3 (2-3yr TAM then AI vs 5yr TAM): 10-yr breast-cancer mortality 8.7% vs 10.1%, P=.015.
All-cause mortality: reduced with AI-containing strategies.
10-yr breast-cancer mortality, AI vs TAM (5yr vs 5yr cohort): 12.1% vs 14.2%; RR 0.85, 95%CI 0.78-0.93, P=.001.
Cohort 1 (5yr AI vs 5yr TAM): 10-yr breast-cancer mortality 12.1% vs 14.2% (RR 0.85, 95%CI 0.75-0.96, P=.009).
Cohort 2 (5yr AI vs 2-3yr AI then TAM): recurrence favored AI (RR 0.90, 95%CI 0.81-0.99, P=.045); breast-cancer mortality not significant (RR 0.89, 95%CI 0.78-1.03, P=.11).
Cohort 3 (2-3yr TAM then AI vs 5yr TAM): 10-yr breast-cancer mortality 8.7% vs 10.1%, P=.015.
All-cause mortality: reduced with AI-containing strategies.
Adverse events
AI vs tamoxifen:
Gynecologic/oncologic: endometrial cancer lower with AI (10-yr incidence 0.4% vs 1.2%; RR 0.33, 95%CI 0.21-0.51).
Bone: fractures higher with AI (5-yr risk 8.2% vs 5.5%; RR 1.42, 95%CI 1.28-1.57).
Non-breast-cancer mortality: similar between strategies.
Gynecologic/oncologic: endometrial cancer lower with AI (10-yr incidence 0.4% vs 1.2%; RR 0.33, 95%CI 0.21-0.51).
Bone: fractures higher with AI (5-yr risk 8.2% vs 5.5%; RR 1.42, 95%CI 1.28-1.57).
Non-breast-cancer mortality: similar between strategies.
Conclusions
In postmenopausal women, 5yr of AI reduces 10-yr recurrence by ~30% and breast-cancer mortality by ~15% vs 5yr tamoxifen (and by ~40% vs no endocrine therapy). AI carries higher fracture risk but lower endometrial cancer risk; patients with poor AI tolerance/adherence may switch to tamoxifen.
Key Limitations
Pooled trials span different eras, AI agents, and durations; benefit largely confined to years of AI exposure. Adherence varied across constituent trials. Constituent trials included ATAC, BIG 1-98/IBCSG 18-98, FEMTA, HOBOE, PROACT, TEAM, ARNO 95, IES, ITA, N-SAS-BC 03, University of Siena Exemestane Trial, ABCSG VIII.
Clinical Context
This EBCTCG analysis is the principal evidence base for preferring an AI-containing regimen over tamoxifen alone in postmenopausal HR+ disease. ASCO and ESMO guidelines recommend incorporating an AI for ~5yr (upfront or sequential) with bone-health monitoring; tamoxifen remains an option for AI-intolerant patients.
References