Identification of a serum proteomic biomarker panel using diagnosis specific ensemble learning and symptoms for early pancreatic cancer detection

PLoS Computational Biology. 2024. Vol. 20. No. 8. Article e1012408.

Ney A., Nené N., Sedlak E., Blyuss O., Whitwell H., Costello E., Gentry-Maharaj A., Williams N., Menon U., Fusai G., Zaikin A., Pereira S.

Pancreatic ductal adenocarcinoma (PDAC) has one of the lowest 5-year survival rates among cancers, primarily due to its complex biology and late-stage diagnosis. Early symptoms often mimic benign gastrointestinal conditions, complicating timely detection. The standard biomarker, carbohydrate antigen (CA) 19–9, is not reliable due to its suboptimal performance and elevation in benign conditions, highlighting the need for better diagnostic tools. In our study, we aimed to develop a biomarker signature to distinguish between benign pancreatic/biliary conditions and PDAC using serum samples from the Accelerated Diagnosis of neuro Endocrine and Pancreatic Tumours and UK Collaborative Trial of Ovarian Cancer Screening studies.

We screened 539 patient serum samples with state-of-the-art biomarker panels and developed a robust predictive signature by applying specialized machine learning methods. This new signature significantly outperformed CA19-9 and other panels reported in the literature.

Our findings suggest that this new biomarker signature can improve early detection of PDAC in patients with ambiguous clinical symptoms, potentially leading to better outcomes for those at risk.

Keywords: biomarkers, CA19-9, Pancreatic ductal adenocarcinoma

Publication based on the results of: Study of neuronal processes in cognitive and fundamental and applied neuroeconomic tasks, taking into account individual differences (2024)