Abstract

Abstract With recent advances in distinguishing Pancreatic Ductal Adenocarcinoma (PDAC) subtypes based on transcriptional profiles, it is becoming clear that different subtypes have distinct therapeutic responses– e.g. Basal-like tumors tend to show less response to chemotherapy than Classical tumors. Here, we asked if the Basal-like, Classical A, and Classical B PDAC subtypes show distinct immuno-suppressive molecular profiles of relevance for targeted immunotherapy of PDAC. Toward this goal, we leveraged two large, public sequencing cohorts – TCGA and ICGC. To accommodate the large variation in cell type composition among the tumors across these cohorts, we applied the novel expression deconvolution (XDec) method to computationally deconvolute the bulk RNA-seq profiles. The method identified cell-type composition of each tumor and identified distinct Basal-like, Classical A, and Classical B subtypes, thus allowing classification of each tumor based on the state of its constituent cancer cell fraction. We further extended the method to also infer cell-type specific gene expression for each tumor sample and to identify correlation of cell-type specific gene expression levels across cohorts. Using the extended method, we determined that Basal-like cancer cells are CD274 high, Classical B cancer cell CD274 expression correlated strongly with MSLN and RELA (a recently proposed mechanism controlling PD-L1), and Classical A cancer cells had low CD274 expression. Because heterotypic interactions may also contribute toward an immuno-suppressive microenvironment, we next correlated expression levels between cancer and stromal cell types. Cancer cell MSLN expression correlated strongly with stromal CD274 expression which also correlated with stromal IFNGR1 and IFNγ. Using co-culture experiments we validated the causative relationship between cancer cell MSLN and stromal INFGR1 and CD274. Taken together, our results suggest that PDAC subtypes have distinct therapeutic vulnerabilities, with Basal-like and MSLN-high Classical B patients more likely to respond to PD-L1 targeting therapies. Moreover, we provided a methodological advance by extending the XDec method and by making it web-accessible, thus allowing the community to derive hypotheses testable in cell line models by performing correlations of cell-type specific gene expression levels deconvoluted from bulk RNA-seq profiles of PDAC patient tumors. Citation Format: Emily L. LaPlante, Dongliang Liu, Aleksandar Milosavljevic, Qizhi Yao. Basal-like, Classical A, and Classical B subtypes of pancreatic cancer show distinct immuno-suppressive molecular profiles [abstract]. In: Proceedings of the AACR Virtual Special Conference on Pancreatic Cancer; 2021 Sep 29-30. Philadelphia (PA): AACR; Cancer Res 2021;81(22 Suppl):Abstract nr PO-004.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call