Abstract S1-2: ATLAS – 10 v 5 years of adjuvant tamoxifen (TAM) in ER+ disease: Effects on outcome in the first and in the second decade after diagnosis

Abstract

Abstract Background: In ER+ early breast cancer, 5 years of tamoxifen greatly reduces recurrence throughout the first decade (years 0–9) with little further gain later, and reduces breast cancer mortality (BCM) substantially throughout years 0–14 (EBCTCG, Lancet 2011; 378: 771–84 – see Table). It is not known how 10 years TAM compares with the current standard of just 5 years TAM. Methods: In 1996–2005 the international ATLAS trial randomized 6846 women with ER+ disease who had had ∼5 years of adjuvant TAM to continue another 5 years (to year 10) or stop at year 5 (control). Annual follow-ups recorded compliance, hospital admissions, breast cancer recurrence (including new contralateral), any other new primary cancer and cause of death. Results: Compliance was ∼80%, as after 2 years 84% of those allocated continue and 4% of those allocated stop were still taking endocrine treatment (>99% TAM). With mean 7.1 woman-years follow-up (30,000 w-y in years 5–9, 16,000 in years 10–14, 2000 later), 1328 recurrences were reported (900 in years 5–9, 379 in years 10–14). Recurrence was significantly lower with10 than 5 years TAM (Table: logrank 2p = 0.002, rate ratio (RR)=0.90 se 0.06 in years 5–9 and 0.75 se 0.08 in years 10+). So were both BCM (2p = 0.01, RR = 0.97 se 0.10 in years 5–9 and 0.71 se 0.09 in years 10+) and all-cause mortality (2p = 0.01, with no increase in non-BCM). Proportional risk reductions were homogeneous by country, age and stage. Kaplan-Meier risks in years 5–14 (K-M) were: recurrence 21.4 vs 25.1%, BCM 12.2 vs 15.0%. Uterine cancer K-Ms in those randomized at age 50+ were: incidence 2.6 vs 1.6% (2p = 0.08), mortality 0.2 vs 0.2%. In pre-menopausal women (where AIs are not an alternative to TAM) there was no apparent excess of uterine cancer. Discussion: Compared with just 5 years TAM, continuing TAM to year 10 safely protects further against recurrence and, particularly during the second decade, BCM. Combining results from ATLAS and the EBCTCG meta-analyses of 5 years TAM vs none (both had ∼80% compliance), in a hypothetical trial of 10 vs 0 years TAM with ∼80% compliance, 15-year BCM would be reduced by at least one-third. Hence, full compliance with 10 years TAM would yield even greater benefit (Table). Further follow-up of ATLAS will assess more reliably the apparently substantial mortality reduction in the second decade after diagnosis. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr S1-2.