Purpose: Next-generation sequencing in non-small cell lung cancer (NSCLC) identifies somatic genomic variants (SGVs) in cancer susceptibility genes (CSGs). We hypothesized that SGVs would be associated with poorer overall survival (OS) but greater benefit with immune checkpoint inhibitors over chemotherapy. We investigated the prevalence and predictive value of SGVs, using data from OAK and POPLAR trials comparing atezolizumab with docetaxel.
Methods: We curated a list of SGVs (excluding TP53, EGFR, ALK, and ROS1) on the basis of CSGs associated with tumorigenesis. We classified participants as SGV mutant or wild-type using baseline plasma analyzed by the FoundationOne Liquid CDx assay. Cox regression analyses and interaction tests between SGV status and treatment were performed.
Results: Of 762 participants, 29% harbored an SGV. The SGV mutant group had worse OS (hazard ratio [HR], 1.28, 95% CI, 1.06 to 1.54), and within each treatment arm (docetaxel: HR, 1.31; atezolizumab: HR, 1.27). In the atezolizumab arm, the SGV mutant group compared with wild-type had worse OS in the PD-L1 high (HR, 1.31 [95% CI, 0.59 to 2.91]) and low (HR, 1.38 [95% CI, 0.98 to 1.93]) subgroups. SGV with missense, splice, and nonsense mutations had significantly worse OS than wild-type in the docetaxel arm (log-rank P = .01) but not in the atezolizumab arm (log-rank P = .33). SGV status did not predict greater OS benefit with atezolizumab over docetaxel (interaction P = .67).
Conclusion: In advanced NSCLC after chemotherapy progression, plasma-detected SGVs are common, and associated with inferior OS. Plasma SGV status should be considered as a stratification factor in future trials.