Genomic aberration of chromatin regulatory BAF complex as predictive biomarker for immunotherapy in gastrointestinal adenocarcinoma.

Abstract

4035 Background: SNF/SWI, a large ATP-dependent chromatin remodeling complex, is required for transcriptional activation of genes normally repressed by chromatin, and critical to tumor initiation and progression. Here, we analyzed the predictive utility of the mutations of the SNF/SWI members involved in BAF and PBAF complexes, and sought to explore the potential mechanisms. Methods: Clinical, genomic, transcriptional, and immunohistochemical data from immunotherapeutic cohort (MSKCC, n=185), Cancer Cell Line Encyclopedia (CCLE, n=92), The Cancer Genome Atlas (TCGA, n=925), and 3D Medicines database (3DMed, n=1812) were analyzed to explore the predictive effect of genomic aberration of BAF complex on the benefit from immunotherapy in patients with gastrointestinal adenocarcinoma. Results: In the MSKCC cohort involving 185 patients with gastrointestinal adenocarcinoma, the mutation of any member involved in BAF complex ( ARID1A, ARID1B, SMARCA4, SMARCB1, and SMARCD1) was significantly associated with prolonged OS of ICI treatment (HR 0.53, 95%CI 0.31-0.90, P=0.019), instead of the mutations of PBAF members including PBRM1 and ARID2. In addition, BAF mutation was not linked with better prognosis in TCGA database, indicating its predictive, not prognostic efficacy of immunotherapy. BAF-mutated samples exhibited higher tumour mutational burden (TMB, P<0.05, Table), and increased mRNA expression of immune-related genes including chemokines and granzyme A. In the 3DMed cohort where tumour samples received both genomic sequencing and PD-L1 immunohistochemical staining, BAF mutation was associated with higher PD-L1 positive rate in tumour cells (P<0.05, Table). Conclusions: Genomic aberration of members in chromatin regulatory BAF complex may serve as a predictive, not prognostic biomarker of ICI benefit in patients with gastrointestinal adenocarcinoma, partially underlying the mechanisms including higher mutational burden, transcription of immune-related genes, and protein-level PD-L1 expression. [Table: see text]