ImportanceImmune-related adverse events (irAEs) arising from immune checkpoint inhibitor (ICI) cancer therapy may potentially predict improved outcomes.ObjectiveTo evaluate the association between irAEs and atezolizumab efficacy in patients with advanced non–small cell lung cancer (NSCLC) using pooled data from 3 phase 3 ICI studies.Design, Setting, and ParticipantsIMpower130, IMpower132, and IMpower150 were phase 3, multicenter, open-label, randomized clinical trials to evaluate the efficacy and safety of chemoimmunotherapy combinations involving atezolizumab. Participants were chemotherapy-naive adults with stage IV nonsquamous NSCLC. These post hoc analyses were conducted during February 2022.InterventionsEligible patients were randomly assigned 2:1 to receive atezolizumab with carboplatin plus nab-paclitaxel, or chemotherapy alone (IMpower130); 1:1 to receive atezolizumab with carboplatin or cisplatin plus pemetrexed, or chemotherapy alone (IMpower132); and 1:1:1 to receive atezolizumab plus bevacizumab plus carboplatin and paclitaxel, atezolizumab plus carboplatin and paclitaxel, or bevacizumab plus carboplatin and paclitaxel (IMpower150).Main Outcomes and MeasuresPooled data from IMpower130 (cutoff: March 15, 2018), IMpower132 (cutoff: May 22, 2018), and IMpower150 (cutoff: September 13, 2019) were analyzed by treatment (atezolizumab-containing vs control), irAE status (with vs without), and highest irAE grade (1-2 vs 3-5). To account for immortal bias, a time-dependent Cox model and landmark analyses of irAE occurrence at 1, 3, 6, and 12 months from baseline were used to estimate the hazard ratio (HR) of overall survival (OS).ResultsOf 2503 randomized patients, 1577 were in the atezolizumab-containing arm and 926 were in the control arm. The mean (SD) age of patients was 63.1 (9.4) years and 63.0 (9.3) years, and 950 (60.2%) and 569 (61.4%) were male, respectively, in the atezolizumab arm and the control arm. Baseline characteristics were generally balanced between patients with irAEs (atezolizumab, n = 753; control, n = 289) and without (atezolizumab, n = 824; control, n = 637). In the atezolizumab arm, OS HRs (95% CI) in patients with grade 1 to 2 irAEs and grade 3 to 5 irAEs (each vs those without irAEs) in the 1-, 3-, 6-, and 12-month subgroups were 0.78 (0.65-0.94) and 1.25 (0.90-1.72), 0.74 (0.63-0.87) and 1.23 (0.93-1.64), 0.77 (0.65-0.90) and 1.1 (0.81-1.42), and 0.72 (0.59-0.89) and 0.87 (0.61-1.25), respectively.Conclusions and RelevanceIn this pooled analysis of 3 randomized clinical trials, longer OS was observed in patients with vs without mild to moderate irAEs in both arms and across landmarks. These findings further support the use of first-line atezolizumab-containing regimens for advanced nonsquamous NSCLC.Trial RegistrationClinicalTrials.gov Identifiers: NCT02367781, NCT02657434, and NCT02366143
