METHODS: Here, we used high-throughput, custom cancer antigen microarrays to identify a clinically relevant autoantibody biomarker combination able to differentially detect PDAC. Specifically, we quantified the serological AAb profiles of 94 PDAC, chronic pancreatitis (CP), other pancreatic- (PC) and prostate cancers (PRC), non-ulcer dyspepsia patients (DYS), and healthy controls (HC).

RESULTS: Combinatorial ROC curve analysis on the training cohort data from the cancer antigen microarrays identified the most effective biomarker combination as CEACAM1-DPPA2-DPPA3-MAGEA4-SRC-TPBG-XAGE3 with an AUC = 85·0% (SE = 0·828, SP = 0·684). Additionally, differential expression analysis on the samples run on the iOme™ array identified 4 biomarkers (ALX1-GPA33-LIP1-SUB1) upregulated in PDAC against diseased and healthy controls. Identified AAbs were validated in silico using public immunohistochemistry datasets and experimentally using a custom PDAC protein microarray comprising the 11 optimal AAb biomarker panel. The clinical utility of the biomarker panel was tested in an independent cohort comprising 223 PDAC, PC, PRC, colorectal cancer (CRC), and HC samples. Combinatorial ROC curve analysis on the validation data identified the most effective biomarker combination to be CEACAM1-DPPA2-DPPA3-MAGEA4-SRC-TPBG-XAGE3 with an AUC = 85·0% (SE = 0·828, SP = 0·684). Subsequently, the specificity of the 11-biomarker panel was validated against other cancers (PDAC vs PC: AUC = 70·3%; PDAC vs CRC: AUC = 84·3%; PDAC vs PRC: AUC = 80·2%) and healthy controls (PDAC vs HC: AUC = 80·9%), confirming that this novel AAb biomarker panel is able to selectively detect PDAC amongst other confounding diseases.