Pancreatic cancer is the third deadliest cancer annually in the United States. Although the vast majority of pancreatic cancer belongs to a single type called pancreatic ductal adenocarcinoma (PDAC), tremendous heterogeneity exists within and between PDACs in their biology and clinical behavior, making it difficult to optimize treatment strategies and therapeutics research. The possibility exists that the heterogeneity results from the fact that PDACs actually encompass several distinct subtypes. Recent research has uncovered much evidence for such subtypes, but so far, the research has not produced clear definitions of the subtypes or associated biomarkers that define them. PDACs express a unique set of glycans derived largely from their origins as duct cells with a protective glycocalyx, including the CA19-9 antigen sialyl-Lewis A (sLeA), which serves as the only approved biomarker of pancreatic cancer, and its near relative sTRA. I hypothesized that the neoplastic cells of pancreatic ductal adenocarcinoma can be separated into subpopulations by their specific glycan expression of sTRA and CA19-9 and that these subpopulations have different functional characteristics and risk for disease dissemination. To test this hypothesis, I used several methods involving both primary specimens and model systems. First, I used multimarker immunofluorescence to detect sTRA and CA19-9 and compare their cellular locations, morphologies, and protein co-expression in tumor and matched adjacent uninvolved tissue, lymph nodes, and metastases. Immunofluorescence was detected by automated microscopy and quantified by novel automated software developed specifically for this project. Clear differences were observed between cancer cells that expressed only CA19-9 and those that expressed only sTRA, as well as a third cell subpopulation represented by dual expression. Dual expression represented a well differentiated epithelial population of cells in well-formed glandular tissue; CA19-9-only expression represented poor to moderately differentiated cell subpopulations of epithelial and flat (mesenchymal) characteristics; and sTRA-only expression represented poor to moderately differentiated cell subpopulations present in "foamy cytoplasm" and flat (mesenchymal) cell features. The co-expression of MUC5AC and beta-catenin was different between the subsets, indicating differences in differentiation. The differences were preserved in cell-line and patient-derived mouse xenografts. I next tested for differences in metastatic propensity. Xenograft tumors expressing sTRA were more strongly correlated with metastasis than those expressing CA19-9, and primary tumors showed differential correlations with lymph-node or liver metastasis depending on glycan expression. Finally, we tested whether blood plasma levels of these glycans correlate with tissue expression and whether elevations occur in distinct subpopulations of patients. The secretion of glycans into cell-culture media, mouse sera, or plasma from human patients generally correlated with glycan expression in the cancer cells, indicating the value of the glycans as serological biomarkers to indicate the tumor type. Certain tissues expressing only CA19-9 did not secrete to blood plasma, particularly in hyperglandular and very high stromal tissue, suggesting a new cause of false negative CA19-9 patients in PDAC detection. CA19-9 and sTRA were elevated in separate subgroups of patients, each with low false-positive rates. As a result, CA19-9 and sTRA together gave better accuracy of PDAC diagnosis than CA19-9 alone (97% specificity, 65% sensitivity vs. 96% and 46%). In summary, these studies support the concept that distinct subtypes of PDAC can be identified by the expression of sTRA or CA19-9. Additionally, sTRA co-expression with CA19-9 also identified a third subpopulation of PDAC with different morphology, likely aggressiveness, and secretion characteristics. Clinical translation is potentially enabled by the detection of these biomarkers in blood plasma, which provides a new approach to improve diagnosis, prognosis and treatment development.
Read
