Pervasive subtypes of pancreatic ductal adenocarcinoma (PDA) and their differing response to therapy.

Pancreatic ductal adenocarcinoma (PDA) carries a dismal prognosis and current treatments are only modestly effective. We present evidence that this variation is caused in part by recurrent, pervasive molecular differences between tumors. mRNA expression profiles measured using microdissected PDA clinical samples reveal three dominant subtypes of disease; epithelial, mesenchymal and acinar-like. The classical and quasi-mesenchymal subtypes are observed in human and mouse PDA cell lines. Importantly, responses to cytotoxics and KRAS depletion in human PDA cell lines differ substantially between subtypes, and in opposing directions. Integrated genomics implicate and functional studies support overexpression of the trancription factor GATA6 as a driver of the epithelial subtype. These results provide a molecular framework for evaluating the prospects of personalized treatment in PDA. RNA was extracted from archival patient FFPE PDA samples and hybridized on Affymetrix U133 plus 2.0 microarrays. The CEL files were processed using R based Bioconductor and normalized values were obtained using RMA. Note: These patient sample microarrays were done in different batches and were batch corrected prior to further analysis as described in our original paper and Nature Protocols Community Contributed Protocol Exchange 04/04/2011 - doi:10.1038/protex.2011.231. RNA was extracted from human PDA cell line samples and hybridized on Affymetrix U133 plus 2.0 microarrays. The CEL files were processed using R based Bioconductor and normalized values were obtained using RMA. Note: Cell line microarrays were done separately. Our preprocessed and normalized data provided here represent analyses performed separately for patient samples and cell lines. RNA was extracted from mouse PDA cell lines and hybridized on Affymetrix Mouse 430a 2.0 microarrays. The CEL files were processed using R based Bioconductor and normalized values were obtained using RMA.