Poster Presentation 31st Annual Lorne Proteomics Symposium 2026

A Window of Opportunity: Liquid Biopsy Proteomics for Early Detection of Pancreatic Cancer (#27)

Jumana Yousef 1 2 , Susanne Wudy 1 2 , Sukhdeep Spall 1 2 , Belinda Lee 1 2 3 4 , Samantha Emery-Corbin 1 2 , Ka Yee Fung 1 2 , Peter Gibbs 1 2 5 , Tracy Putoczki 1 2 , Laura F Dagley 1 2
  1. The Walter and Eliza Hall Institute of Medical Research, Parkville, VIC, Australia
  2. Department of Medical Biology, University of Melbourne, Parkville, VIC, Australia
  3. Peter MacCallum Cancer Centre, Parkville, Victoria, Australia
  4. Northern Health, Melbourne, Victoria, Australia
  5. Western Health, Melbourne, Victoria, Australia

Background
Pancreatic ductal adenocarcinoma (PDAC) has a 5-year survival rate of 11.5%, largely due to late-stage diagnosis. Non-invasive biomarkers are urgently needed to enable earlier detection and improved outcomes. A liquid-biopsy based screening panel involving plasma and/or urine could provide a non-invasive test. This study leverages the PURPLE registry—a real-world cohort of consecutive PDAC cases with matched plasma, urine, and tissue bio-samples—to integrate multi-fluid proteomics with clinical data for translational biomarker discovery.

Objectives
To identify plasma and urine protein biomarkers capable of distinguishing early from late PDAC and non-PDAC individuals, supporting early screening, prognostication, and therapeutic target discovery through data-independent acquisition mass spectrometry (DIA-MS) and machine learning.

Methods
Optimised on-bead enzymatic digest protocols1 and diaPASEF MS workflows were applied to plasma and urine samples using Bruker timsTOF Pro and HT instruments with 15–25 cm IonOpticks Aurora columns and 30–45 min gradients. Data were processed in Spectronaut v20 and analysed using limpa2. Clinicopathological and survival data were integrated from the PURPLE registry. Machine learning (xGBoost) was used to classify disease stage. A subset of 20 plasma samples (7 late PDAC, 6 early PDAC, 7 healthy) were cross-validated using the NULISA3 Inflammation 250 panel (Alamar).

Results
Discovery plasma proteomics (n = 203; 120 PDAC, 14 non-PDAC, 69 healthy) identified candidate proteins enriched in early PDAC. Two independent validation cohorts (n = 185; 86 PDAC, 50 healthy, 45 non-PDAC and n = 90; 34 PDAC, 56 healthy) confirmed a reproducible plasma signature distinguishing early versus late PDAC with high specificity and sensitivity. Matched urine proteomics from the final validation cohort supported complementary biomarker trends. Machine learning on the plasma data demonstrated high specificity and sensitivity in PDAC stage classification.  

Conclusions
A robust plasma protein signature for PDAC was identified and validated across independent cohorts. This biomarker panel offers potential for clinical translation as a minimally invasive screening test and provides insights into PDAC biology relevant to drug target selection and clinical trial design.

  1. 1. Dagley LF, Infusini G, Larsen RH, Sandow JJ, Webb AI. Universal Solid-Phase Protein Preparation (USP(3)) for Bottom-up and Top-down Proteomics. J Proteome Res. Jul 5 2019;18(7):2915-2924. doi:10.1021/acs.jproteome.9b00217
  2. 2. Li M, Cobbold SA, Smyth GK. Quantification and differential analysis of mass spectrometry proteomics data with probabilistic recovery of information from missing values. bioRxiv. 2025:2025.04.28.651125. doi:10.1101/2025.04.28.651125
  3. 3. Feng W, Beer JC, Hao Q, et al. NULISA: a proteomic liquid biopsy platform with attomolar sensitivity and high multiplexing. Nat Commun. Nov 9 2023;14(1):7238. doi:10.1038/s41467-023-42834-x