Novel mRNA-based Immunological Biomarker for Early Detection of Resectable Pancreatic Cancer

Abstract Background: In this study, we aimed to develop and validate a novel mRNA-based immunological biomarker panel for the early detection of resectable pancreatic cancer (RPC). Methods: Blood samples and clinical information of a cohort of 297 individuals (102 patients with PC and 195 non-disease controls) were obtained between 2015 and 2020. mRNA levels of candidate immunological biomarkers identified through an extensive literature review in buffy coat samples were measured using quantitative polymerase chain reaction. The final biomarker panel was confirmed using biological feature selection. Subsequently, the marker panel was trained and validated through artificial intelligence (AI)-based cross-validation. Results: Among 55 candidate biomarkers, eight mRNAs (CCL2, CCL5, CXCR2, IFN-γ, IFN-λ1, PTGES2, SLC27A2, and TNF) were selected based on biological and statistical performance to develop a biomarker panel. The octet-biomarker panel successfully identified all stages of PC with an area under the curve (AUC) of 0.925 in the training set (AUC = 0.922 in the validation set). In patients with RPC, the AUC for the panel was 0.973, whereas that for CA19-9 level was 0.809. For the combination of the panel and CA19-9 level, the AUC was 0.990 in patients with RPC. Furthermore, even in patients with RPC and normal CA19-9 level, the AUC for the octet-biomarker panel was 0.969, whereas that for CA19-9 level alone was 0.787 and for the combination of the panel and CA19-9 level was 0.977. Conclusion: Compared with CA19-9, the mRNA-based immunologic octet-biomarker panel improved diagnostic performance, especially in RPC with normal CA19-9 levels.