Comparison of progression-free survival (PFS) with best or confirmed response (BR, CR) as an endpoint for overall survival (OS) in advanced non small cell lung cancer (A-NSCLC): A North Central Cancer Treatment Group (NCCTG) investigation

Abstract

8021 Background: Historically, tumor response has been the primary endpoint in Phase II (P2) trials in A-NSCLC. With targeted therapies that prolong disease stabilization, PFS rate has become an accepted alternate endpoint. We investigated the relationship between PFS, BR, and CR with OS to assess their suitability as endpoints in P2 trials. Methods: Individual patient (pt) data from 343 pts from 4 first line A-NSCLC P2 NCCTG treatment trials were pooled. The trial regimens included Docetaxel, Gemcitabine, Pemetrexed, Temsorilimus and Sorafenib. Any pt who progressed prior to 6 months (m) but lived beyond 12 m or progressed after 6 m but died before 12 m was considered discordant (D). Percent agreement (PA) between PFS rate, BR, and CR with OS at 12 m was calculated on a per-pt basis, and the predictive utility assessed using the area under the receiver-operating characteristic curve (A-ROC) in logistic models. Patterns in baseline pt and tumor characteristics and subsequent therapy (ST) between the concordant (C) and D subgroups were evaluated. Results: The 12 m OS and the 6 m PFS rates were 40% and 33%. The median time from progression to death was 5.4 m. The PA (A-ROC) for PFS at 6 m, BR and CR was 78% (0.76), 63% (0.58) and 65% (0.59) respectively. 72% of pts in the D group versus only 61% in the C group had ST (p=0.09). Pts with discordant outcomes had a lesser median tumor burden at baseline (6.0 vs. 6.6, p=0.2) and longer median time to progression (TTP) (136 vs. 116 days; log rank p=0.03). No differences by age, gender, ECOG performance status or tumor stage were found. Conclusions: PFS is more predictive of OS than BR or CR for P2 trials in A-NSCLC. Discordant pts had lesser tumor burden, longer TTP on first line treatment and more ST. Such heterogeneity in pt population makes randomized P2 trials more suitable in A-NSCLC. No significant financial relationships to disclose.