Weighted gene co-expression network analysis reveals key stromal prognostic markers in pancreatic cancer

Abstract

In recent years, it has been shown that stroma compartment can favor tumor proliferation and aggressiveness. Although extensive research with network analyses such as Weighted Gene Co-expression Network Analysis (WGCNA) has been conducted on pancreatic cancer and its stromal components, WGCNA has not previously been applied to isolate and identify genes associated with the abundance of stroma and survival outcome from bulk RNA data. We investigated the gene expression profile and clinical information of 140 pancreatic ductal adenocarcinoma patients from TCGA. Network analysis was performed using WGCNA and four modules were found to be associated to patients’ clinical traits. Specifically, one module of 2459 genes, was associated to stromal sample content. Subsequently, those genes were further analyzed for survival association through log-rank test and Cox regression. HPGDS and ITGA9-AS1 emerged as significant indicators of favorable prognosis while KCMF1 and YARS1 were implicated in poorer prognostic outcomes. Importantly, HPGDS was found to be stromal-specific in the TMA cohort of Human Protein Atlas. Single sample GSEA showed that the stromal module is enriched for stromal signature of Moffitt and Puleo. These findings suggest that we uncovered a stromal specific signature through WGCNA and found putative prognostic markers.