Background: Alterations in ARID1A, PBRM1 and BAP1 subunits of the SWI/SNF complex are common in biliary tract cancers (BTCs) and have been implicated in tumor microenvironment (TME) immunomodulation. However, there is heterogeneity in outcomes, with PBRM1 and BAP1 conferring improved survival vs worse survival with ARID1A. We explored relationships between ARID1A, PBRM1 and BAP1 alterations with putative immunotherapy biomarkers and genomic co-alterations. Methods: We examined BTC patients in the Tempus database with ARID1A, PBRM1 and BAP1 alterations. Co-alterations were identified through next-generation sequencing (NGS) using the Tempus xT assay, a targeted, tumor-normal-matched DNA panel that detects SNVs, indels, and CNVs in 648 genes, as well as chromosomal rearrangements in 22 genes with high sensitivity and specificity. Samples additionally underwent whole transcriptome RNA-seq via the Tempus xR assay (Tempus Labs, Chicago, IL). Immune biomarkers explored include MSI-H, PD-L1, Tumor Mutation Burden (TMB), and immune cell infiltration by RNA sequencing. Parameters were analyzed between two groups: BTC with ARID1A, vs BTC with PBRM1 and/or BAP1 alterations. Results: 696 BTC patients with ARID1A, PBRM1 and BAP1 alterations were included. Overall, there was low frequency of MSI-H (3.2%), PD-L1-positivity (5.7%) and TMB >10 mt/Mb (4.7%). Among the two groups, there was a higher proportion of MSI-H patients in ARID1A BTC vs PBRM1 and/or BAP1 BTC (4.9% vs 1.4%, p=0.01). Macrophages were the predominant immune cell (46%), followed by CD4+ T-cells (23%) and NK cells (12%). CD8+ T cells were the least encountered (5%). B-Cells were seen more in ARID1A vs PBRM1 and/or BAP1 (9% vs 6%, p <0.001) while macrophages were more frequent in PBRM1 and/or BAP1 vs ARID1A (49% vs 44%, P <0.001). Co-alterations in SMAD4, TP53, ARID2 and KRAS were more frequent in ARID1A BTC while IDH2 and NRAS were more common in PBRM1 and/or BAP1. Within the PBRM1 subgroup, TP53 was most frequently encountered (33%), while CDKN2A alterations were most frequently encountered in BAP1 BTC (35%) (Table 1). Conclusions: This is the largest known data set exploring the genomic and immune landscape of BTC with ARID1A, PBRM1 and BAP1 alterations. Macrophages were the dominant immune cell TME and may be a target of interest. Co-alteration profile is distinct between ARID1A- vs PBRM1- and/or BAP1-altered BTC.

¹Pearson’s Chi-squared test; Fisher’s exact test. ²FDR correction for multiple testing.