Background: The accurate identification of high-grade dysplasia (HGD) or invasive cancer (advanced neoplasia, AN) in pancreatic cystic lesions (PCLs) is needed to identify PCLs that warrant surgical intervention. Cytologic evaluation of cyst fluid is widely-used but has poor sensitivity. We previously found that activity of the lysosomal serine protease tripeptidyl peptidase 1 (TPP1) is associated with mucinous PCLs that harbor AN (AUC 0.72). We aimed to identify additional functional biomarkers that would improve performance of our classifier for dysplastic grade in mucinous PCLs.

Methods: We used a combination of global protease activity profiling and shotgun proteomics to identify differentiating markers between PCLs with low-grade dysplasia (LGD, n=3) and HGD (n=3). Candidate biomarkers were validated in an independent cohort of 28 clinically-annotated mucinous PCLs (HGD n=12, LGD n=16), by measuring both protein concentration (using ELISA) and enzymatic activity (using internally-quenched fluorescent substrates). We used a nested cross-validation approach to iteratively split our cohort into training and validation sets to predict HGD. Model accuracy was evaluated using area under the curve (AUC), sensitivity, specificity, and accuracy.

Results: We identified 8 proteins that were highly abundant in mucinous PCLs with HGD. Among these were the inflammatory enzymes myeloperoxidase (MPO) and neutrophil elastase (ELANE), suggesting that differential activity levels could be exploited that would minimize fluid requirements and cost. Among 28 mucinous PCLs, a composite score that included both activity and mass of MPO performed the best for HGD in mucinous cysts (AUC 0.955, sensitivity 98%, specificity 93%). ELANE activity (AUC 0.76, sensitivity 69%, specificity 59%) and TPP1 activity (AUC 0.765, sensitivity 71%, specificity 74%) had modest performance.

Conclusions: Our functional biomarkers accurately classified HGD among mucinous cysts. Functional biomarkers that require low sample volumes (<10μl) of cyst fluid have the potential to substantially improve clinical decision-making for patients with PCLs.

Citation Format: Francesco Caiazza, Andre Lourenco, Patricia Conroy, Thomas Hoffmann, Sam L. Ivry, Tyler York, Gina Zhu, Audrey Mustoe, Anthony J. O'Donoghue, Charles S. Craik, Kimberly Kirkwood. Using functional biomarkers to accurately predict advanced neoplasia in pancreatic cystic lesions [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 2226.