Abstract BACKGROUND: A sensitive and specific blood biomarker test is needed to improve survival in PDAC through early detection. An initial discovery study analyzing more than 3000 proteins using both pre-selected immunoassays and the Olink proteomics platform identified 15 promising protein candidates, of which ten were successfully pre-verified for ELISA immunoassay analysis. This model-development study was conducted to develop and pre- validate a specific and sensitive biomarker signature for the early detection of PDAC. METHODS: This multicenter case-control study was performed using existing bio-banked pretreatment samples from patients with newly diagnosed PDAC (Stages I-II) and participants in PDAC high-risk surveillance programs as controls. Blinded quantitative measurement of protein biomarkers was performed using ELISA-based immunoassays. CA19-9 was measured on a Roche COBAS instrument. An AI-derived predictive mathematical biomarker signature incorporating multiple protein biomarkers was developed and assessed using standard cross- validation techniques. RESULTS: 128 Stage 1 and 2 PDACs and 495 high-risk control serum samples were analyzed with 10 predetermined biomarkers to develop a four-protein biomarker signature. The model predicted PDAC versus non-PDAC with 84% sensitivity at 98% specificity. Model performance was significantly better when compared to CA19-9 alone (sensitivity 65% at 98% specificity, p<0.001). In those patients with low CA19-9 (<37 U/ml), the model predicted PDAC with 62% sensitivity at 98% specificity, performing significantly better than CA19-9 alone (9% sensitivity at similar 98% specificity (p<0.001). Additionally, in patients >65 years old, the signature predicted PDAC with 90% sensitivity at 98% specificity, which was significantly more sensitive than CA19-9 alone (75% sensitivity at 98% specificity, p<0.001). Test accuracy was not affected by diabetes status. CONCLUSIONS: This blood-based, multiplexed biomarker signature showed high sensitivity and specificity and significantly outperformed CA19-9 for early-stage PDAC. This biomarker model has the potential to improve outcomes in PDAC patients through early detection. Analytical and clinical validation studies are planned. Citation Format: Norma Palma, Ralph Schiess, Thomas King, Alcibiade Athanasiou, Natasha Kureshi, Lisa Ford, Aimee Lucas, Bryson Katona, Randall Brand, Diane Simeone. A blood-based, multiplex protein biomarker signature shows superior performance in detection of early-stage pancreatic ductal adenocarcinoma (PDAC) [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Pancreatic Cancer Research; 2024 Sep 15-18; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2024;84(17 Suppl_2):Abstract nr B027.