Background: Next generation sequencing (NGS) is used widely to identify somatic oncogenic driver mutations in aNSCLC to guide treatment. Comprehensive profiling has led to identification of multiple gene mutations of unclear significance, including germline genes associated with cancer risk, termed CSG. Using data from two randomized trials comparing atezolizumab to docetaxel, we investigated the prognostic and predictive values of CSG in aNSCLC. Methods: We used publicly available data from the OAK (NCT02008227) and POPLAR (NCT01903993) trials. At baseline, plasma was analyzed for cfDNA using FoundationOne CDx NGS assay. We defined CSGs as pathogenic variants of APC, BAP1, BRCA1/2, BRIP1, CDH1, CDKN2A, CHEK2, FH, FLCN, MEN1, MET, MSH2/6, MLH1, PMS2, PALB2, PTCH1, PTEN, RAD51, RB1, SDHB, SMARCA4, STK11, TSC1/2 and VHL. Cox models with treatment covariate, CSG status (mutant [mt] vs wild-type [wt]) and their interaction was used to assess the predictive value by treatment of CSG mt for overall survival (OS) and progression free survival (PFS) using univariate and multivariate models. Similar analysis was performed for objective response rate (ORR). Results: Of 1137 patients, 853 with sufficient tumor content formed the analysis population. In total, 295 (35%) had a known/likely pathogenic CSG mt. The variant allele frequencies (vAF) of CSG mt ≥30%, 10-30%, and <10% were 8% (N=23), 25% (N=73), and 67% (N=199) respectively. Patients with CSG mt were more likely to be smokers (89% vs 81%, P=0.005), had squamous tumors (37% vs 26%, P=0.001) with higher blood-based tumor mutation burden (mean 13.8 vs 10.2 per megabase, P<0.001). CSG mt was not predictive of greater OS benefit with atezolizumab over docetaxel (Table). CSG mt was associated with 35% increase risk of death in univariable analysis (HR 1.35, 95% CI 1.15-1.59). CSG mt was associated with inferior OS in multivariable analysis adjusting for performance status, smoking status, tumor histology, age, sex and number of organ sites of metastasis (HR 1.26, 95% CI 1.07-1.48). Conclusions: Plasma CSG mt is an independent poor prognostic factor in two large aNSCLC clinical trial datasets. Majority of vAF of CSG were low, suggesting that very few were potentially germline in origin, but dedicated sequencing for confirmation will be required. If confirmed, CSG status could be used as a stratification factor in future aNSCLC trials.

*89 patients were excluded (atezolizumab, N=34; docetaxel, N=55) from analysis for ORR because of non-measurable disease.