[705] A meta-analysis of phase III trials evaluating the predictive value of HER2 and topoisomerase II alpha in early breast cancer patients treated with CMF or anthracycline-based adjuvant therapy.

Di Leo A, Isola J, Piette F, Ejlertsen B, Pritchard KI, Bartlett JMS, Desmedt C, Larsimont D, Tanner M, Mouridsen H, O'Malley FP, Twelves C, Cardoso F, Poole CJ, Piccart MJ, Buyse ME HER2/TOP2A Meta-Analysis Study Group, Prato, Italy; HER2/TOP2A Meta-Analysis Study Group, Finland; HER2/TOP2A Meta-Analysis Study Group, Belgium; HER2/TOP2A Meta-Analysis Study Group, Denmark; HER2/TOP2A Meta-Analysis Study Group, Canada; HER2/TOP2A Meta-Analysis Study Group, United Kingdom

Background: Retrospective studies have suggested that HER2 and topoisomerase II (TOP2A) might predict sensitivity to anthracyclines (A)
Methods: Four phase III trials comparing A with CMF in early breast cancer (EBC) patients (pts), and with available primary tumor samples, were identified. HER2 and TOP2A genes were evaluated locally by FISH (amplification if ratio 2). Data were centralized at the statistical office (IDDI, Belgium). On-site visits were performed to check data quality and laboratory procedures. HER2 and TOP2A local scores were validated by submitting randomly selected samples to a central lab (CL) (University of Tampere Finland), for a 3 color FISH test (HER2, TOP2A, centromere 17, Abbott Labs, IL, USA). Estrogen (ER), progesterone (PgR) receptors, and grade (G) were evaluated locally (no score validation at the CL).
Results: We present the results of the planned interim analysis on 1944 pts. Final results ( 3500 pts) will be presented when tumor sample collection for the UK trial is complete. HER2 local scores were validated at the CL in 137 cases (discordance: 8/137, 5.8%). TOP2A local scores were validated in 123 cases (discordance 38/123, 30.8%). Half of the TOP2A discordant cases were locally deleted-centrally normal or vice versa.

A planned exploratory analysis evaluated the TOP2A predictivity in 4 biologically homogeneous groups: highly or moderately hormone sensitive, HER2+ and ER/PgR negative, triple negative (TN). This analysis did not enhance the TOP2A predictivity in any of the 4 groups. In the TN group (294 pts), the DFS HR was 0.77 (0.54-1.09), suggesting that benefit from A might not be confined to HER2+ pts.
Conclusions: The interim analysis shows that HER2 and TOP2A have a clinically modest and a statistically borderline predictive value. In triple negative disease A may be superior to CMF.
Acknowledgments: Abbott Laboratories, Belgian Federation Against Cancer, Cancer Research UK, Les Amis de l'Institut Bordet, Pfizer, Scottish BC trials group.

Saturday, December 13, 2008 7:00 AM

Poster Discussion Session: HER2 as a Biomarker (7:00 AM-9:00 AM)