2398 - KEAP1/NRF2 Mutation Phenotype Predicts Local Control in Early-Stage Lung Cancer Patients Treated with Stereotactic Ablative Radiotherapy

Purpose/Objective(s): Mutations in KEAP1 and NFR2 occur in roughly 20% of non-small cell lung cancer (NSCLC) patients and have been shown to impart resistance to systemic and radiation therapy. We hypothesized some mutations are benign and do not affect outcomes. We determined the phenotypes of all possible KEAP1 and NRF2 missense and nonsense mutations and examined the impact of these mutational calls on the outcomes of a cohort of early-stage NSCLC patients treated with stereotactic ablative radiotherapy (SABR).

Materials/Methods: KEAP1 mutation phenotypes were determined through deep mutational scanning (DMS) experiments where each codon position of the KEAP1 ORF was mutated to all 20 possible amino acids plus a stop codon, for a total of 13,076 variants. The DMS library was then transduced into H1299-KEAP1^NULL cells using a lentivirus vector, subjected to selective pressure with H₂O₂ or ionizing radiation and surviving variants were quantified using next-generation sequencing (NGS). NRF2 mutation phenotypes were assigned by examining the impact of variant sequence position on an RNA-seq NRF2 signaling score in TCGA patient samples. A retrospective cohort of NSCLC patients was created by querying a single institutional database for individuals that underwent targeted NGS of tumor tissue and were treated with SABR. Clinical and treatment characteristics were extracted from the EMR under an IRB approved protocol and local recurrence (LR) was evaluated using a Fine-Gray competing risks model.

Results: Enrichment experiments using the KEAP1 DMS library identified two groups of mutations, one enriched for silent mutations and the other for nonsense mutations. A threshold of 95% specificity for silent mutations was selected to differentiate benign and pathogenic single amino acid substitutions, which identified 88% of previously characterized pathogenic mutations. For NRF2, mutations in the KEAP1 interacting DLG and ETGE motifs resulted in high RNA-seq signaling scores and were assigned pathogenic, while variants in the remaining sequence positions were considered benign. For the SABR cohort, 40 KEAP1/NRF2 wild type patients and 32 patients with KEAP1/NRF2 mutated tumors were identified of which 11 were classified as benign and 21 as pathogenic. There was no significant difference in age, gender, smoking history, histology, radiation dose, and gross tumor volume between the groups. One- and three-year LR rates were 3.3% and 10.7% for the wild type, 0% and 0% for the benign, and 20.8% and 52.9% for the pathogenic groups, which was significantly different for pathogenic vs. wild type (P = 0.001) or benign (P = 0.011).

Conclusion: We generated a comprehensive list of pathogenic and benign KEAP1/NRF2 variants and showed that pathogenic mutations significantly increase the risk of LR following SABR, while benign mutations do not. These results highlight the importance of molecular phenotyping of tumor mutations and may enable future personalized radiotherapeutic approaches.