Relationship between progression-free survival and overall survival in patients with advanced non-small cell lung cancer treated with anticancer agents after first-line treatment failure.

Abstract

The hazard ratio of progression-free survival (PFS-HR) generally does not reflect that of overall survival (OS-HR) in advanced non-small cell lung cancer (NSCLC) patients treated with first-line therapy. Short survival post-progression (SPP) better reflects the PFS-HR and OS-HR in simulations. We aimed to evaluate whether the PFS-HR reflects the OS-HR in NSCLC clinical trials for post-first-line treatments. We reviewed clinical studies of post-first-line anticancer agents for NSCLC. We examined the sample size of the experimental arm (EA), median PFS (mPFS) or median time to progression in the EA, median overall survival (mOS) in the EA, the PFS-HR and the OS-HR. SPP was defined as the difference between mOS and mPFS. The association between mPFS and SPP, mPFS and mOS, and the PFS-HR and OS-HR was tested. We sought for the optimal point of correlation of PFS-HR and OS-HR by every 1 month of SPP. We identified 32 trials (34 arms). mPFS and mOS were weakly correlated (correlation coefficient [r] = 0.376; P = 0.0286). The PFS-HR and OS-HR were also moderately correlated (r = 0.415; P = 0.015). The maximum r value was 0.770 (SPP < 6 months; P < 0.0001) when we tested the associations between the PFS-HR and OS-HR for SPP using 1-month increments. The estimated regression equation at this point was OS-HR = 0.679 × (PFS-HR) + 0.349. The PFS-HR and OS-HR were strongly correlated in advanced NSCLC patients treated with post-first-line anticancer agents, with a SPP of less than 6 months.