MA09.07 Genomic Landscape and Clinical Outcomes With Immune Checkpoint Inhibitors in NF1-Mutant NSCLC

Abstract

NF1 encodes for neurofibromin 1, a RAS GTPase-activating protein (GAP) that regulates RAS nucleotide cycling and when inactivated can promote aberrant RAS signaling and carcinogenesis. Deleterious NF1 mutations have been previously reported in non-small cell lung cancer (NSCLC), particularly in current/former smokers, but the genomic landscape of this molecular subgroup remains incompletely characterized. Furthermore, the clinical outcomes of NF1-mutant (NF1-mut) NSCLC upon treatment with immune checkpoint inhibitors (ICI) remain unknown. The Foundation Medicine Inc. (FMI) NSCLC dataset (FoundationCore – N=62845) was used to analyze NF1 mutations, cancer gene co-mutations, tumor mutational burden (TMB) and PD-L1 expression in NF1-mut NSCLC. A proprietary, artificial intelligence-supported algorithm was employed to infer pathogenicity of individual NF1 mutations. Clinical outcomes upon treatment with single-agent ICI were assessed using two publicly available clinical cohorts: OAK/POPLAR (N=765) and Rizvi (N=185). Patients with known EGFR/ALK alterations were excluded. PFS was used for outcome analysis. In the FMI dataset, inactivating somatic NF1 mutations were present in 6.9% (2805/40388) of lung adenocarcinomas (LUAD) and 7.7% (855/11041) of lung squamous cell carcinomas (LUSC). NF1-mut LUAD were significantly enriched for high TMB (TMB≥16 mutations/Mb: NF1-mut 31.3% and NF1-wild type [NF1-wt] 14.9%; P= 8E-141) and PD-L1 TPS≥1% (P=6E-10), and both LUAD and LUSC were enriched for TP53 co-mutations (LUAD P=3E-79; LUSC P=1E-5). Co-alterations in TP53 and/or CDKN2A/B were highly prevalent in NF1-mut LUAD (82.0%) and near universal in NF1-mut LUSC (95.4%). The association between NF1 mutations and high TMB was further validated in the OAK/POPLAR (P=0.002) and Rizvi cohorts (P=0.008). Importantly, in the OAK-POPLAR cohort TMB-high NSCLC patients bearing NF1-mut tumors treated with ICI exhibited markedly longer PFS compared with those harboring TMB-high NF1-wt tumors (cutoff of TMB>=16mut/Mb: PFS 9.63 vs 2.84 months, HR 0.55, P=0.054; cutoff of TMB>=10mut/Mb: PFS 4.14 vs 2.53 months, HR 0.69, P=0.119). This finding was further validated in the Rizvi cohort (cutoff of TMB>=10mut/Mb: PFS 8.33 vs 3.07 months, HR 0.52, P=0.070). Despite the small sample size, a trend for longer PFS in NF1-mut-TMB high NSCLC was also observed in the subpopulation harboring STK11/KEAP1 co-mutations (TMB>=16mut/Mb 14.36 vs <16mut/Mb 1.40 months, HR 0.13, p=0.067). Among NSCLC with high TMB, inactivating NF1 somatic genomic alterations identify a subgroup with markedly longer PFS to ICI therapy. These findings further highlight the impact of cancer gene mutations and underscore a context dependent effect of TMB on clinical outcomes from immunotherapy.