Abstract 473: Glycomic characterization of carbohydrate antigen-defined pancreatic cancer tissues using lectins and imaging mass spectrometry

Abstract

Abstract N-glycosylation alterations in tumors are now recognized as cancer hallmarks for its role in oncogenesis, cancer signaling and metastasis. For pancreatic adenocarcinoma (PDAC), the sole FDA-approved biomarker is the carbohydrate antigen 19-9 (CA19-9). Because of CA19-9's association with pancreatitis and other gastrointestinal cancers, it is used for surveillance of disease progression rather than for diagnosis. A recently identified carbohydrate antigen, sialylated tumor-related antigen (sTRA) recognized by the TRA-1-60 mAb, is expressed in a non-overlapping subset of pancreatic cancer and is under consideration for use alongside CA19-9. To better understand these carbohydrate antigens in relation to PDAC pathogenesis, complementary N-glycan analysis approaches were applied to a cohort of 53 patient-matched tumor and normal tissue samples represented by tissue microarrays (TMA) and whole-tissue FFPE samples. The distribution of N-glycans in these tissues were defined using a matrix-assisted laser desorption ionization imaging mass spectrometry (MALDI-IMS) workflow. Over 100 individual N-glycan species were mapped to their histopathology locations. The same cohort was characterized by multi-round immunofluorescence for CA19-9 and sTRA expression as well as two lectins, PHA-E and GSL-II,. N-glycan IMS data from the TMAs were evaluated by LASSO-regularized logistic regression for the identification of N-glycan analytes with predictive value in differentiating between tumor and normal tissue. LASSO selected N-glycan masses were incorporated alongside biomarker expression data to construct a model for PDAC tissue classification. N-glycans associated with tumor tissue tended to be bi-, tri- and tetra-antennary complex structures with multople fucose, terminal sialic acids and bisecting GlcNAc. Quantitative analysis of the IMS data by LASSO regression identified 9 differentiating N-glycan masses, 7 from tumor cores and 2 from normal cores. When both IMS glycan masses and biomarker expression data for CA19-9 and sTRA were integrated into a single linear regression model, receiver operator curve (ROC) accuracy and AUC improvements, +0.172 and +0.221 respectively, were observed in comparison to models utilizing either dataset individually. Critically, this model was able to rescue cases where biomarker data was missing or insufficient so that they could be correctly classified as tumor or normal tissue. Analysis of this PDAC tissue cohort using MALDI-IMS illustrated the utility of investigating specific N-glycans and N-glycan structural classes correlated with carbohydrate PDAC biomarkers. Integration of N-glycan IMS and biomarker lectin immunostaining bolsters pancreatic cancer identification and posits expanded utility for each approach for PDAC surveillance and early identification. Citation Format: Colin McDowell, Zachary Klamer, Johnathan Hall, Luke Wisniewski, Peggi Angel, Anand Mehta, Brian Haab, Richard R. Drake. Glycomic characterization of carbohydrate antigen-defined pancreatic cancer tissues using lectins and imaging mass spectrometry [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 473.