Introduction: Pancreatic cancer was responsible for almost 500,000 deaths globally in 2020 according to GLOBOCAN. Pancreatic cystic lesions (PCLs) are fluid-filled protrusions either on or inside the pancreas and can either be benign or pre-malignant. Current clinical guidelines to risk stratify PCL patients are imperfect. Multi-omic profiling of pancreatic cyst fluid (PCF) could aid in the identification of a novel biomarker panel to improve PCL risk stratification.

Methods: PCF was collected from 40 patients by EUS-FNA, with matched serum collected prior to EUS. Patients were stratified using the 2018 European evidence-based guidelines into low-risk (n=15), high risk (n=15) and no-risk or pseudocyst (n=10). PCF was sonicated and subsequently processed using an SP3 paramagnetic bead protocol prior to LC-MS. MS-generated LFQ intensity data were analyzed in Perseus (v1.6.13.0) and STRING (v11.5). HTG microRNA whole transcriptome sequencing was run on whole PCF. MiRNA sequencing data were analyzed using HTG EdgeSeq Reveal (v3.1.0). Spearman correlations were generated using R packages ‘Hmisc’ (v4.5-0) and ‘corrplot’ (v0.90).

Results: MS-analysis of PCF revealed 8 proteins to be significantly upregulated in high-risk compared to low-risk (p<0.05, FDR=0.05, s0=0.1). All 8 proteins had significantly positive correlations with patient risk and expression of the other seven. LCN-2, REG1A, LGALS3, PIGR and S100A8 have been shown to be elevated in the blood of pancreatic cancer patients. PRSS8 is known to be elevated in the serum of ovarian cancer patients, while MUC6 and TCN1 have not been shown to be differentially expressed in the circulation of cancer patients. STRING analysis revealed 11.8% and 6.8% of the proteins identified to be involved in the innate and adaptive immune responses, respectively. Significant positive correlations were found between 11 immune-associated proteins and patient risk classification (p<0.05). Whole transcriptome sequencing revealed 3 miRNA (miR-216a-5p, miR-216b-5p and HK_SKORA66) to be significantly upregulated in high-risk PCF compared to low-risk, and 5 miRNA (miR-197-5p, miR-6741-5p, miR-3180-3p, miR-3180 and miR-6782-5p) to be significantly upregulated in matched high-risk serum compared to low-risk (adj-p<0.05, FDR=0.05, s0=0.1). Unsupervised hierarchical clustering of patients using the 8 differentially expressed proteins and 3 miRNA from the PCF gave a clustering accuracy of 95.8%, with just 1/24 patients being misclassified.

Conclusion: We have identified a putative multi-omic biomarker panel for PCL patient risk stratification. Practically, further refinement of this panel through the inclusion of additional biological compartments is required. These data will be validated in a larger patient cohort, with the aim of generating a less invasive blood-based panel that will aid in the improvement of risk stratification.

Citation Format: Laura E. Kane, Gregory S. Mellotte, Rebecca G. Lyons, Eimear Mylod, Simone Marcone, Paul F. Ridway, Finbar MacCarthy, Kevin C. Conlon, Joanne Lysaght, Barbara M. Ryan, Stephen G. Maher. Establishment of a novel multi-omic biomarker panel in cyst fluid and blood for stratifying patient risk of pancreatic cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3381.