OBJECTIVES:

To investigate disease-free survival (DFS) as a potential surrogate endpoint for overall survival (OS) among patients with early-stage resectable non-small cell lung cancer (ES-NSCLC).

METHODS:

A systematic literature review (SLR) was conducted to identify randomized controlled trials (RCTs) reporting data for OS and DFS for patients with resectable ES-NSCLC receiving treatment in the adjuvant setting. Embase, MEDLINE, and Cochrane CENTRAL were searched via Ovid to identify studies of interest. Chemotherapy and immunotherapy interventions currently used in the adjuvant setting were included. Trials included in the SLR were evaluated in a feasibility assessment to determine the base-case dataset for surrogacy analyses. Bivariate meta-analyses (BMAs) models, including Daniels and Hughes (DH), and the bivariate random-effects model (BRMA) in both the standard form and product normal formulation (PNF) were used to assess DFS as a surrogate endpoint for OS.

RESULTS:

The database searches identified 636 unique citations, of which 146 full-text publications were screened. Ultimately, 44 publications reporting data for 25 RCTs were included in the SLRs; the base-case BMA included 20 RCTs for DFS-OS surrogacy. The correlation between DFS hazard ratios (HR) and OS HRs was high (~0.8). BMA results were consistent across different models, with BRMA-PNF and BRMA standard accounting for more uncertainty. The models showed DFS to be a significant predictor of OS across a wide range of DFS HR values. Our results suggest that the surrogate threshold effect (STE) predicted by the BMA models (0.68: BMA-DH, 0.59: BRMA-PNF, 0.6: BRMA standard) is somewhat attainable in this disease area with currently available treatments. Take-one-out cross-validation showed that the observed OS HR fell within the study-specific prediction intervals for all 20 studies included for all 3 BMA models.

CONCLUSIONS:

These findings support the validity of DFS as a surrogate endpoint for OS within the ES-NSCLC patient population.