Choice of primary endpoints in first-line phase III trials of HER2-negative or HER2-unknown metastatic breast cancer (MBC).

Abstract

607 Background: OS is considered the most clinically relevant endpoint in cancer therapy trials, but OS can be confounded by post-trial therapy, particularly in settings with long post-progression survival (PPS). Based on statistical modelling with 50,000 simulated trials, Broglio & Berry [JCNI 2009] suggested the association between improved progression-based endpoints (eg, PFS) and OS is weak in settings with PPS >12 months and can only be shown in very large trials. To test this hypothesis, we analysed efficacy outcomes and choice of primary endpoints in first-line MBC trials. Methods: Data wereanalysed from:randomised, controlled phase III trials comparing systemic chemotherapies and/or targeted agents; enrolling ≥150 pts; published in peer-reviewed journals from 2000–2011. Trials that enrolled only HER2-positive MBC pts or evaluated non-EU approved agents were excluded. Results: Of27 trials, 1 (3.7%) had OS as the primary endpoint, while 18 (66.7%) had progression-based parameters, commonly PFS (9 trials, 33.3%). A significant increase in progression-based parameters was seen in 14 trials (51.9%). All 5 trials (18.5%) showing a significant OS benefit had a PPS <12 months. Mean PPS was considerably longer in trials published in 2006–2011 vs 2000–2005 (14.1 ± 4.0 vs 9.7 ± 2.3 months, respectively), as was mean PFS (8.3 ± 2.2 vs 6.6 ± 1.6 months) and mean OS (25.4 ± 5.6 vs 18.6 ± 3.0 months). A weak correlation was seen between OS and PFS (Pearson’s coefficient r=0.32), but a higher correlation was seen in treatment arms with an OS benefit (r=0.59), as well as in treatment arms with a PPS <12 months (r=0.43) or older studies (r=0.51). Data were insufficient to allow a valid analysis of the effect of further-line therapies or cross-over on OS (16 trials reported further-line therapies; 9 cross-over therapy). Conclusions: In support of the hypothesis by Broglio & Berry, a significant OS advantage was shown exclusively in trials with a PPS <12 months. Due to the weak correlation between PFS and OS, it is difficult to determine the possible surrogacy of PFS for OS. Thus, as an OS benefit might be increasingly difficult to attain, progression-based parameters appear to be valid endpoints in MBC clinical trials.