Systematic evaluation of the impact of disease progression (DP) date determination on progression-free survival (PFS) in advanced lung cancer: A joint North Central Cancer Treatment Group (NCCTG) and Southwest Oncology Group (SWOG) investigation.

Abstract

7005 Background: In advanced lung cancer, overall survival is largely influenced by the DP status. At ASCO 2010, we reported that the PFS estimates are significantly impacted by the approach used for declaring DP date; the magnitude of difference being large enough to alter trial conclusions. Here, we present updated results from NCCTG and further validate our findings using data from SWOG. Methods: Individual patient (pt) data from 14 NCCTG and 7 SWOG trials were used. Reported progression date (RPD) was either the scan date or the clinical deterioration date. DP date was determined using one of the 4 common methods (M): RPD (M1), one day after last progression-free (PF) scan (M2), midpoint between last PF scan and RPD (M3), or interval censoring (M4). PFS was defined as the earlier of the time to DP or death from any cause, and estimated using the method of Kaplan-Meier (M1-3) or maximum likelihood (M4). Results: Within NCCTG (SWOG), 67% (78%) of pts reported DP during treatment, 11% (24%) of pts died without DP, and the median time from the last PF scan to DP during treatment and follow-up was 1.4 (1.3) mos, and 3.0 (1.7) mos respectively, for NSCLC. Results were similar for SCLC trials. PFS estimates using RPD was the highest, with method 2 being the most conservative. M3 and M4 were similar since majority of DP occurred during treatment (frequent disease assessments). For the 3 randomized trials, the 4 methods resulted in the same overall conclusion. Conclusions: Although NCCTG and SWOG had different data collection requirements for monitoring DP, the results are consistent with % difference in PFS estimates being large enough to conclude efficacy in advanced lung cancer. Thus, standards for declaring DP date and sensitivity analysis when analyzing PFS data are critical when reporting efficacy or comparing trial results. DP date method Median PFS (mos) NSCLC SCLC NCCTG (n=660) SWOG (n=297) NCCTG (n=116) SWOG (n=131) 1 4.3 3.1 2.7 1.3 2 1.8 1.5 0.03 0.03 3 3.3 2.2 1.8 0.7 4 (lower, upper) 3.52, 3.55 2.86, 2.89 1.45, 1.48 0.03, 0.16 % Difference Method 1 vs. 2 Method 1 vs. 3 58% 23% 52% 29% 99% 33% 98% 46%