Background: Outcomes are poor for patients with previously treated, advanced or metastatic non-small-cell lung cancer (NSCLC). The anti-programmed death ligand 1 (PD-L1) antibody atezolizumab is clinically active against cancer, including NSCLC, especially cancers expressing PD-L1 on tumour cells, tumour-infiltrating immune cells, or both. We assessed efficacy and safety of atezolizumab versus docetaxel in previously treated NSCLC, analysed by PD-L1 expression levels on tumour cells and tumour-infiltrating immune cells and in the intention-to-treat population.

Methods: In this open-label, phase 2 randomised controlled trial, patients with NSCLC who progressed on post-platinum chemotherapy were recruited in 61 academic medical centres and community oncology practices across 13 countries in Europe and North America. Key inclusion criteria were Eastern Cooperative Oncology Group performance status 0 or 1, measurable disease by Response Evaluation Criteria In Solid Tumors version 1.1 (RECIST v1.1), and adequate haematological and end-organ function. Patients were stratified by PD-L1 tumour-infiltrating immune cell status, histology, and previous lines of therapy, and randomly assigned (1:1) by permuted block randomisation (with a block size of four) using an interactive voice or web system to receive intravenous atezolizumab 1200 mg or docetaxel 75 mg/m(2) once every 3 weeks. Baseline PD-L1 expression was scored by immunohistochemistry in tumour cells (as percentage of PD-L1-expressing tumour cells TC3≥50%, TC2≥5% and <50%, TC1≥1% and <5%, and TC0<1%) and tumour-infiltrating immune cells (as percentage of tumour area: IC3≥10%, IC2≥5% and <10%, IC1≥1% and <5%, and IC0<1%). The primary endpoint was overall survival in the intention-to-treat population and PD-L1 subgroups at 173 deaths. Biomarkers were assessed in an exploratory analysis. We assessed safety in all patients who received at least one dose of study drug. This study is registered with ClinicalTrials.gov, number NCT01903993.

Findings: Patients were enrolled between Aug 5, 2013, and March 31, 2014. 144 patients were randomly allocated to the atezolizumab group, and 143 to the docetaxel group. 142 patients received at least one dose of atezolizumab and 135 received docetaxel. Overall survival in the intention-to-treat population was 12·6 months (95% CI 9·7-16·4) for atezolizumab versus 9·7 months (8·6-12·0) for docetaxel (hazard ratio [HR] 0·73 [95% CI 0·53-0·99]; p=0·04). Increasing improvement in overall survival was associated with increasing PD-L1 expression (TC3 or IC3 HR 0·49 [0·22-1·07; p=0·068], TC2/3 or IC2/3 HR 0·54 [0·33-0·89; p=0·014], TC1/2/3 or IC1/2/3 HR 0·59 [0·40-0·85; p=0·005], TC0 and IC0 HR 1·04 [0·62-1·75; p=0·871]). In our exploratory analysis, patients with pre-existing immunity, defined by high T-effector-interferon-γ-associated gene expression, had improved overall survival with atezolizumab. 11 (8%) patients in the atezolizumab group discontinued because of adverse events versus 30 (22%) patients in the docetaxel group. 16 (11%) patients in the atezolizumab group versus 52 (39%) patients in the docetaxel group had treatment-related grade 3-4 adverse events, and one (<1%) patient in the atezolizumab group versus three (2%) patients in the docetaxel group died from a treatment-related adverse event.

Interpretation: Atezolizumab significantly improved survival compared with docetaxel in patients with previously treated NSCLC. Improvement correlated with PD-L1 immunohistochemistry expression on tumour cells and tumour-infiltrating immune cells, suggesting that PD-L1 expression is predictive for atezolizumab benefit. Atezolizumab was well tolerated, with a safety profile distinct from chemotherapy.

Funding: F Hoffmann-La Roche/Genentech Inc.