In advanced non-small cell lung cancer (NSCLC), overall survival (OS) has been the most commonly used end point in phase III trials. However, effect of subsequent chemotherapy on OS has the potential to underestimate the efficacy of an experimental arm when compared with control arm. Although the effect of therapies instituted after disease progression is thus of interest, little is known about postprogression survival (PPS).Methods
A literature search identified 87 trials that were published during the past decade. Trials in first-line and salvage setting were 69 and 18, respectively. We partitioned OS into PFS and PPS and evaluated the relation between OS and either PFS or PPS. We also examined strength of the association between the treatment effect of PFS and OS in trial-level.Results
The average PPS for trials in first-line setting was same as trials in second -line setting (5.4 months). In both of groups, PPS was strongly associated with OS (in first-line and second-line setting; r = 0.82 and 0.94, respectively). The correlation between OS and PPS in recent trials (2003 and later) was stronger than that in older trials (2002 and earlier) (r = 0.96 and 0.68), whereas the association between treatment benefit on PFS and on OS was less pronounced in recent trials than in older trials. Additionally, the induction rate of subsequent chemotherapy after study treatment is moderately associated with PPS. Conclusions: PPS is highly associated with OS in advanced NSCLC. Late-line therapies are likely attributable to the prolongation of PPS. Our findings indicate that treatment benefit on OS in advanced NSCLC patients can be skewed by the effects of subsequent therapies. Considering the increase of the opportunity for receiving the subsequent chemotherapy, detection of a statistically significant benefit in OS may be too high a hurdle in future clinical trial for advanced NSCLC.