Efficacy, safety and predictive biomarker results from OAK, a randomized phase III study comparing atezolizumab with docetaxel in patients with advanced NSCLC

Background: The development of cancer immunotherapy represents a key breakthrough in the treatment of NSCLC. Atezolizumab (atezo) inhibits programmed death-ligand 1 (PD-L1) signaling and can restore anti-tumor T-cell activity. The phase II POPLAR study showed that atezo significantly improves overall survival (OS) vs. docetaxel (doc) in previously treated patients (pts) with NSCLC. OS benefit increased with increasing PD-L1 expression on tumor cells (TC) and tumor-infiltrating immune cells (IC). We will present efficacy, safety and pre-planned predictive biomarker data of atezo vs. doc from OAK, the first phase III study of an anti-PD-L1 antibody in pts with NSCLC. Methods: OAK is a global, multicenter, randomized, open-label study evaluating atezo vs. doc after failure of platinum-containing chemotherapy in a biomarker-unselected NSCLC population. Pts werestratified by PD-L1 expression, prior chemotherapy regimens (1 vs. 2) and histology (nonsquamous vs. squamous) and randomized 1:1 to either atezo (1200 mg IV q3w) or docetaxel (75 mg/m2 IV q3w). Key eligibility criteria were measurable disease, ECOG status 0/1 and ≥1 prior platinum-based treatment. PD-L1 expression was centrally evaluated using the VENTANA SP142 IHC assay; pts were scored as TC0, 1, 2 or 3 and IC0, 1, 2 or 3. The primary population of the study will be the first 850 of 1225 randomized patients. The primary endpoint, OS, will be compared between the treatment arms within the ITT and TC1/2/3 or IC1/2/3 populations using stratified log-rank test; Kaplan-Meier methodology will be used to construct survival curves and estimate median OS. Stratified Cox regression proportional hazard models will be used to estimate HR. Secondary endpoints include PFS, ORR, DOR and safety. An estimate of ORR per RECIST v1.1 and its 95% CI will be calculated using the Clopper-Pearson method for each arm. Anticipated data: Enrollment was completed in Apr 2015. The primary endpoint analysis will be performed when ≈ 595 events occur in the first 850 patients, which is expected in Aug 2016. Results of the primary and secondary endpoint analyses for the ITT population and PD-L1 subgroups will be presented.