Beschreibung:
<jats:title>Abstract</jats:title>
<jats:p>Introduction: In up to 10% of individuals newly diagnosed with type 2 diabetes (T2DM), diabetes is secondary to pancreatic disease, of which 10% is pancreatic cancer-related. Individuals over the age of 50 yr with new-onset diabetes are considered to be the highest risk group for pancreatic cancer (PDAC), with diabetes presenting as a paraneoplastic symptom and occurring, on average, 13 months prior to PDAC being identified. Screening in this population could facilitate earlier cancer detection, however, there is currently no test to distinguish PDAC-related diabetes (PDAC-DM; a form of type 3c diabetes) from T2DM. The aim of our study was to develop protein biomarkers capable of identifying T3cDM from T2DM to facilitate the identification of a PDAC enriched population suitable for screening. Methods: Mass spectrometry (SWATH)- and aptamer-based proteomics workflows were employed for the analysis of 210 plasma samples, comprising seven case and control groups, stratified according to diabetes status. Groups included PDAC with/without diabetes, chronic pancreatitis with/without diabetes, new-onset diabetes (&lt;3 yr post-diagnosis) and long-standing diabetes (&gt;3 yr post-diagnosis). Raw SWATH data was processed using a publicly available in-silico peptide library (generated by DIA-NN), with filtering for plasma proteins. Data was split into training and test sets prior to analysis and a T3cDM group was generated by combining PDAC- and chronic pancreatitis-related diabetes sample data. Linear models were created to identify proteins that were differentially expressed in multiple comparisons. Three main comparisons of interest were proteins differentially expressed between 1) T3cDM and new-onset diabetes 2) PDAC and new-onset diabetes and 3) long-standing diabetes and new-onset diabetes. From the differentially expressed proteins a classifier was trained for the distinction of T3cDM from T2DM (allowing identification of individuals whose diabetes was secondary to underlying pancreatic disease). An ensemble of models was created using a bootstrapping approach, taking forward candidate biomarkers appearing in 90% of runs for the creation of a composite biomarker. Results: Over 7500 proteins, from &gt;6500 genes, were identified per sample. Our linear models identified distinct T3cDM (28 and 105 differentially expressed proteins from SWATH- and aptamer-platforms, respectively) and PDAC signals (72 and 102 differentially expressed proteins from SWATH- and aptamer-based platforms, respectively). These had minimal overlap with the long-standing diabetes comparison. Conclusion: We have undertaken the deepest reported interrogation of the PDAC plasma proteome to-date. The robustness of our approach is supported by the identification of both novel and existing biomarkers of PDAC. Further work is underway to interrogate the molecular pathways contributing to the pathophysiology of diabetes in PDAC. This knowledge will support the development of new biomarkers for earlier detection and will inform future treatment strategies for early disease.</jats:p>
<jats:p>Citation Format: Lucy Oldfield, Emily Johnson, Eva Caamaño Gutiérrez, Martyn Stott, William Greenhalf, Christopher Halloran, Eithne Costello. Deep proteomics identifies novel biomarker candidates and molecular pathways of pancreatic cancer-related diabetes [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Pancreatic Cancer; 2023 Sep 27-30; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(2 Suppl):Abstract nr PR02.</jats:p>