Abstract Introduction: Pancreatic cancer has a very poor prognosis, with no established prognostic biomarkers in clinical use. This project aims to identify a prognostic proteomic-based signature for pancreatic adenocarcinomas. Methods: Fresh frozen tumors and matched normal samples from 125 patients were prepared for proteomic analyses using data-independent acquisition mass spectrometry (DIA-MS). Differential expression analyses were conducted on the normalized protein matrix to identify the top differentially expressed proteins (DEP) within the tumor samples. DEP were subjected to crosstalk and pathway enrichment analysis (PEA). Survival analysis based on initial univariate and subsequent 100 runs of multivariate Cox regression with Least Absolute Shrinkage and Selection Operator (LASSO) was performed to obtain a reduced list of candidate proteins associated with Overall Survival (OS). The proteins that appeared in greater than 95% of the LASSO runs were then used in a multivariate Cox model with recursive feature selection, which yielded the final 29 proteins. A risk score was built from the final 29 proteins. Consensus clustering was performed on the median absolute deviation-based top 20% highly variable proteins in tumor samples to identify proteomic-based subtypes. Results: Proteomic analyses revealed 5614 proteins identified from 599 sample runs. Differential expression analyses revealed 398 DEP in tumor samples (FDR-adjusted p-value <0.05, and |logFC|>1). PEA showed that these proteins were related to focal adhesion, extracellular matrix interaction (ECM), angiogenesis, and PI3K signaling pathways. A total of 803 proteins were significantly associated with OS in a univariate Cox regression analysis (p<0.05). PEA on the top 200 proteins associated with poorer OS revealed pathways related to focal adhesion, PI3K signaling, ECM and hypoxia-induced factor-1. Using LASSO multivariate Cox regression modeling, a 29-protein signature was identified, from which a risk score was calculated that dichotomized patients into high- and low-risk groups in terms of OS (Hazard ratio (HR) 2.8, 95% Confidence Interval (CI) [2.3, 3.3], concordance index of 0.91). This risk score was also prognostic for recurrence and three-year survival (both p<0.0001). A multivariate Cox regression model adjusted for other clinical variables revealed a significant association of the risk score with OS (HR 2.91, 95% CI [2.4, 3.5], p<0.001) while maintaining the concordance index (0.907). Consensus clustering analyses revealed four proteomic-based clusters, with cluster 3 showing the worst OS (p<0.001), independent of other clinical variables. PEA on the DEP within cluster 3 showed upregulation of proteins related to cell adhesion, angiogenesis, and immune-related pathways. Conclusion: A 29-protein signature identified a sub-group of patients with pancreatic adenocarcinoma with a poorer prognosis independent of clinical variables. Citation Format: Adel T. Aref, AKM Azad, Asim Anees, Mohashin Pathan, Jason Grealey, Daniela-Lee Smith, Erin M. Humphries, Daniel Bucio-Noble, Jennifer M. Koh, Erin Sykes, Steven G. Williams, Ruth Lyons, Natasha Lucas, Dylan Xavier, Sumit Sahni, Anubhav Mittal, Jaswinder S. Samra, John V. Pearson, Nicola Waddell, Peter G. Hains, Phil J. Robinson, Qing Zhong, Roger R. Reddel, Anthony J. Gill. A proteomic-based prognostic signature of pancreatic adenocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2209.