Transcriptional and immunohistological assessment of immune infiltration in pancreatic cancer

• Published in: PloS one 2020-08, Vol.15 (8), p.e0238380-e0238380
• Main Authors: Bernard, Brady, Rajamanickam, Venkatesh, Dubay, Christopher, Piening, Brian, Alonso, Emilio, Jutric, Zeljka, Tang, Ephraim, Newell, Pippa, Hansen, Paul, Medler, Terry, Gunderson, Andrew, Young, Kristina, Bifulco, Carlo, Pucliowska, Joanna, Crittenden, Marka R, Gough, Michael J
• Format: Article
• Language: English
• Subjects: Adenocarcinoma; Automation; Biology and Life Sciences; Cancer; Cancer research; Complexity; Development and progression; Diagnosis; Diagnostic immunohistochemistry; Gene expression; Genetic aspects; Immune system; Infiltration; Liver; Lymphocytes; Lymphocytes T; Medicine and Health Sciences; Metastases; Methods; Mutation; Pancreatic cancer; Patients; Ribonucleic acid; RNA; RNA sequencing; Surgery; Transcription; Transcription (Genetics); Tumors
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0238380

Pancreatic adenocarcinoma is characterized by a complex tumor environment with a wide diversity of infiltrating stromal and immune cell types that impact the tumor response to conventional treatments. However, even in this poorly responsive tumor the extent of T cell infiltration as determined by quantitative immunohistology is a candidate prognostic factor for patient outcome. As such, even more comprehensive immunophenotyping of the tumor environment, such as immune cell type deconvolution via inference models based on gene expression profiling, holds significant promise. We hypothesized that RNA-Seq can provide a comprehensive alternative to quantitative immunohistology for immunophenotyping pancreatic cancer. We performed RNA-Seq on a prospective cohort of pancreatic tumor specimens and compared multiple approaches for gene expression-based immunophenotyping analysis compared to quantitative immunohistology. Our analyses demonstrated that while gene expression analyses provide additional information on the complexity of the tumor immune environment, they are limited in sensitivity by the low overall immune infiltrate in pancreatic cancer. As an alternative approach, we identified a set of genes that were enriched in highly T cell infiltrated pancreatic tumors, and demonstrate that these can identify patients with improved outcome in a reference population. These data demonstrate that the poor immune infiltrate in pancreatic cancer can present problems for analyses that use gene expression-based tools; however, there remains enormous potential in using these approaches to understand the relationships between diverse patterns of infiltrating cells and their impact on patient treatment outcomes.