Prevalence and association of ARID1A with driver alterations and immune checkpoint inhibitor (ICPi) biomarkers in cell-free circulating tumor DNA (ctDNA) from 27,000 non-small cell lung cancer (NSCLC) patients.

Abstract

9523 Background: Recent data suggest that the tumor suppressor gene ARID1A is associated with high anti-tumor immunity and may have value as a predictive biomarker for response to ICPi therapy in NSCLC. We examined ARID1A alterations detected in ctDNA from a large cohort of advanced NSCLC patients using a commercially available liquid biopsy assay and explored associations of ARID1A with lung cancer driver alterations and other putative ICPi mutational biomarkers. Methods: Consecutive samples from stage IIIB/IV NSCLC patients tested from March 2016 - August 2019 using a 73- to 74-gene targeted next-generation sequencing ctDNA assay (Guardant360) were queried. Testing included analysis of single nucleotide variants, insertions/deletions, fusions, and amplifications ( KEAP1 not tested). Mutation frequencies were compared using Fisher’s exact test, with variants of uncertain significance and synonymous variants excluded. Results: Of 27,776 NSCLC patients with >1 ctDNA alteration detected, 1,094 (3.9%) had >1 functional ARID1A (f ARID1A) mutation. f ARID1A mutations were significantly more common in patients with squamous histology compared to adenocarcinoma (5.1% vs 3.8%, p = 0.0007). There were significantly fewer EGFR exon 19 deletion mutations (4.9% vs 11.1%; p < 0.0001) and EGFR L858R mutations (4.0% vs 7.0%; p < 0.0001), and significantly more BRAF V600E alterations (2.2% vs 1.4%; p = 0.0338) in patients with f ARID1A. There was no significant difference in the frequency of ALK and ROS1 fusions, nor STK11 mutations between patients with and without f ARID1A (8.0% vs 6.8%; p = 0.126). Activating KRAS mutations were significantly more frequent in patients with f ARID1A (31.1% vs 19.4%; p < 0.0001), including KRAS G12C (10.9% vs 7.0%; p < 0.0001). Conclusions: These data provide a mutational landscape for f ARID1A mutations in NSCLC. f ARID1A was associated with significant differences in the frequency of multiple lung cancer driver alterations, of particular interest in the EGFR-mutated cohort, where ICPi efficacy is low. The frequency of STK11 mutations, a possible negative predictor of ICPi efficacy, was not significantly different. KRAS mutations were significantly more frequent in patients with f ARID1A, notable given recent data reporting that KRAS mutations, particularly KRAS G12C, may be a positive predictor of ICPi response in NSCLC. Determination of ICPi efficacy in patients with f ARID1A is in-process.