Clinical efficacy of atezolizumab (Atezo) in PD-L1 subgroups defined by SP142 and 22C3 IHC assays in 2L+ NSCLC: Results from the randomized OAK study

Abstract

Background: In the Phase III OAK trial, patients (pts) with previously treated advanced NSCLC had improved median overall survival (OS) with atezo vs docetaxel (doc), regardless of PD-L1 expression (per VENTANA PD-L1 SP142 IHC assay). Although efficacy correlated with PD-L1 expression on tumor cells (TC) and tumorinfiltrating immune cells (IC), an OS benefit was also observed in pts with PD-L1- negative tumors (i.e., TC0 and IC0; HR, 0.75 [95%CI: 0.59, 0.96]). To determine whether these results were consistent across PD-L1 IHC assays, we assessed atezo efficacy in PD-L1 subgroups as defined by SP142 and Dako 22C3 pharmDx PD-L1 IHC assays. Methods: PD-L1 expression was assessed prospectively with SP142 and retrospectively with 22C3. The SP142 assay measured PD-L1 expression on TC and IC, while the 22C3 assay gave a tumor proportion score (TPS) based on TC membrane staining. Results: Among the primary population of 850 pts (ITT850), 400 had results from the 22C3 assay (biomarker-evaluable population [BEP]). Clinical outcomes in the BEP vs ITT850, and prevalence in PD-L1 subgroups are summarized (Table). Among pts with tumors negative by SP142 (TC0 and IC0), most (77%) were also negative by 22C3 (TPS <1%). Comparable OS benefit with atezo was seen in PD-L1-negative subgroups defined by both assays. Improved clinical benefit was observed in pts with the highest PD-L1 expression by either assay (TC3 or IC3 by SP142, or TPS ≥ 50% by 22C3; Table). Conclusions: Prevalence of PD-L1 subgroups in the BEP was consistent with previous reports for both assays. Most tumors considered negative by SP142 were also negative by 22C3. An OS benefit (atezo vs doc) was observed in PD-L1-negative subgroups defined by either assay and was consistent with the overall population results from OAK. These data provide evidence of atezo OS benefit in pts with PD-L1-negative tumors irrespective of the PD-L1 IHC assay used. (Table Presented).