e16360 Background: Early detection of PDAC can improve survival but is hampered by the lack of recognizable symptoms in early disease. Imaging has been the mainstay for surveillance of high-risk individuals but has proven to be cumbersome and may lack sensitivity to detect small tumors. CA19-9 is the only available blood biomarker for PDAC, but it is neither sensitive nor specific enough for it to be recommended for high-risk surveillance. A sensitive and specific blood biomarker for PDAC could positively impact this situation. Methods: Serum samples from patients across multiple centers with stage I and II PDAC (n = 75) and matched controls at high risk for PDAC (genetic and familial) (n = 83) were assessed for more than 2,900 protein analytes using O-link multiplex technology and conventional immunoassays for selected protein candidates. Machine learning was used to combine the biomarker candidates to create 4-plex signatures discriminating PDAC and controls. Results: ROC (receiver operating characteristic) curve analysis of these revealed signatures with performance superior to CA19-9 and a previously commercially available assay, IMMRay PanCan-d (AUC of 0.97 and 0.94 vs. 0.87 for CA19-9 and 0.88 for IMMRay). A subset of these candidate biomarkers has now been transitioned to standardized ELISA assays which are being used to construct a final signature for clinical validation in an independent set of PDAC and high-risk control samples. Conclusions: A novel multi-analyte blood test utilizing newly-discovered PDAC biomarkers has the potential to improve early detection of PDAC in high-risk individuals.