A tumor microenvironment integrated (TMI) staging system for pancreatic ductal adenocarcinoma based on cancer-associated fibroblast and molecular consensus signatures

Background Pancreatic ductal adenocarcinoma (PDAC) is an aggressive solid tumor with poor prognosis. Existing prognostication systems rely on clinical staging, but systems of tumor classification through molecular signatures and the association between cancer-associated fibroblast (CAF) subtypes and clinical outcomes suggest that prognostic systems could be improved by integrating these data. Methods We clustered 320 PDAC tumors from The Cancer Genome Atlas (TCGA) repository into basal and nonbasal subsets by applying a previously defined molecular signature. We performed deconvolution using data from 5 single-cell RNA-seq studies to estimate the proportion of different CAFs in each TCGA tumor sample. We defined a poor prognosis phenotype based off calculated proportions of myofibroblastic CAFs (myCAFs) and basal-type molecular signatures and integrated these data into the American Joint Committee on Cancer (AJCC) clinical staging system to generate a tumor microenvironment integrated (TMI) staging system. We compared TMI and AJCC staging in predicting median overall survival (OS) and 5-year OS in the TCGA cohort. Results Overall, TMI staging overall outperforms AJCC staging within the TCGA cohort, particularly when accounting for survival rate variations within each stage and achieves a greater overall balance in the distribution of TCGA cohort tumors among stages. AJCC staging remained stronger in hazard consistency, a surrogate of the similarity of survival rates within subgroups of each defined stage. We further identified three families of extracellular matrix proteins – collagen, fibronectin 1 (FN1) and laminins – involved as key signaling pathways within PDAC tumors, particularly between myCAFs and ductal epithelial cells (including tumor cells), that were additionally enriched in basal tumors. Conclusion The integration of molecular signatures and CAF make-up with clinical parameters improves prognostication in PDAC. TMI staging also provides additional biological information on tumor subtypes and stromal components, which may impact on therapeutic selection in future.