OBJECTIVES: We aimed to identify baseline population characteristics that may affect the association between progression-free survival (PFS) and overall survival (OS) in advanced non-small cell lung cancer (NSCLC) patients.

METHODS: Based on a previously conducted systematic literature review, we used data from 928 individual treatment arms (119,603 NSCLC patients) of RCTs in first- and further-line therapy setting. All analyses were weighted by arm sample size and carried out using absolute values: median PFS and median OS in months. First, we fitted multivariate weighted linear regressions with backward elimination. Second, the relationship was tested using the instrumental variable 2-stage least-squares regression.

RESULTS: Average median PFS and median OS were 4.3 months (standard deviation (SD): 1.9 months) and 9.4 months (SD: 2 months), respectively. Median OS had a moderate (0.4 > r > 0.6) association with median PFS (weighted Pearson r = 0.60). The results of the multivariate regression show the increment of 1.43 months (95% confidence interval (CI); 1.197 to 1.658) of OS with each additional month of PFS. The instrumental variable approach shows 1.58 months (95% CI; 1.216 to 1.954) of OS estimated for each additional month of PFS. In both analyses, the presence of mutations, histology, disease stage, and treatment type remained significant predictors of median OS.