Genomic predictors of sensitivity to chemotherapy and immunotherapy in cholangiocarcinoma.

Abstract

540 Background: Cholangiocarcinoma (CCA) has seen a surge immunotherapy (IO) use. Data on the sensitivity and resistance to chemotherapy and IO in CCA is limited, making it difficult to identify predictive biomarkers for guiding treatment. We hypothesize that transcriptional signatures differentiating PD-L1 high from PD-L1 low patients (pts), can be leveraged to better understand IO sensitivity. These signatures could potentially facilitate personalized treatment strategies and mitigate therapeutic resistance. Methods: We explored transcriptomic signatures related to chemotherapy and IO sensitivity in the CCA cohort of The Cancer Genome Atlas (TCGA-CHOL). We divided the pts (n=36) into 2 groups based on PD-L1 (CD274) expression levels. Pts were designated as PD-L1 high (n=11) if PD-L1 ≥ average PD-L1 expression, and PD-L1 low (n=25) if PD-L1 ≤ average expression PD-L1in the cohort. Transcriptional dysregulation was assessed by comparing transcriptional profiles of the 2 groups. Differential expression analysis, using Limma, was utilized for Gene Set Enrichment Analysis (GSEA), to assess the predicted function of transcriptional dysregulation that occurs between the 2 groups. Immune deconvolution by using the Tumor IMmune Estimation Resource (TIMER) deconvolution software to assess differences in immune infiltration and resulting scores from this analysis were used to assess how immune cell population estimates correlated with immune factors, like PD-L1 expression. Results: GSEA revealed PD-L1 low pts showed significant alterations in metabolic pathway genes that regulate cholesterol, amino acid transport and cellular respiration. As expected, GSEA revealed alterations to immune cell pathways associated with PD-L1 expression, assessed using TIMER revealed several significant correlations between immune cell deconvolution scores: CD4+ and CD8+ T-Cell (R=-0.35, p=0.02), CD8+ and neutrophils (R=0.42, p=0.009), and CD8+ and myeloid dendritic cells (R=0.3348, p=0.04). We noted significant correlations between immune cells deconvolution scores with PD-L1 expression, including PD-L1 and B Cells (R=0.54, p=0.0005), macrophages (R=0.41, p=0.01), CD8+ T Cells (R=0.43, p=0.008), neutrophils (R=0.56, p=0.0003), and myeloid dendritic cells (R=0.48, p=0.002). Significant differences in immune deconvolution scores of CD8+ T Cells, macrophages, neutrophils were noted between PD-L1 high and low pts. Conclusions: PD-L1 high pts with CCA exhibit expected differential enrichment of immune signatures. Notably, we report for the first time that CCA PD-L1 low patients uniquely enrich for differential metabolic signatures. This could lay the groundwork for understanding mechanisms that underlie IO response differences, provide valuable insights if confirmed in IO treated cohorts and serve as evidence for the utility of metabolism-targeted therapies to overcome therapeutic resistance in CCA.