Somatic ARID1A mutation stratifies patients with gastric cancer to PD-1 blockade and adjuvant chemotherapy.

Background: AT-rich interaction domain 1A (ARID1A) encodes a vital component of switch/sucrose non-fermentable chromatin-remodeling complex. Given its association with genomic instability, we conducted this study to determine whether ARID1A mutation status had an impact on therapeutic responsiveness in gastric cancer (GC), especially combinatory chemo-immunotherapy.

Methods: We retrospectively enrolled a total of 1162 patients from five independent cohorts. ZSHS Cohort and TCGA Cohort were designed to inform chemotherapeutic relevance and immunobiology of ARID1A-mutant GC based on tissue samples and sequencing data, respectively. MSKCC Cohort, mGC Cohort, and Melanoma Cohort were utilized to interrogate the predictive efficacy of ARID1A mutation to programmed cell death protein 1 (PD-1) blockade.

Results: ARID1A mutation was enriched in EBV-positive, hypermutated-single nucleotide variant and microsatellite-unstable subtype GC, and was predictive of responsiveness to both fluorouracil-based chemotherapy and PD-1 blockade. Specifically, ARID1A mutation score was a highly sensitive indicator (91%) of response to pembrolizumab. Mechanistically, ARID1A mutation correlated with extensive DNA damage repair deficiency and immunogenic tumor microenvironment (TME) featured by elevated activated subsets of CD8 T cells, CD4 T cells, and NK cells. Type 17T helper cells were typically abundant in ARID1A-mutant GC and might be a precondition for chemosensitivity conferred by ARID1A mutation. Furthermore, ARID1A mutation indicated elevated expression of VEGFA and CLDN18, as well as over-representation of ERBB2 and FGFR2 signaling pathway.

Conclusions: ARID1A-mutant GC displayed immunogenic TME and might be a candidate for both monotherapy and the combination of frontline chemotherapy and PD-1 blockade.