Abstract 3674: Blood-based differentiation of malignant and benign pancreatic lesions using analysis of fragmentation patterns in cell-free DNA

Abstract

Abstract Pancreatic ductal adenocarcinoma (PDAC) remains lethal, with a five-year survival rate of less than 10%. For patients with pancreatic cystic lesions (PCL), analysis of DNA from cyst fluid has shown promise in determining the risk of malignant transformation. However, obtaining cyst fluid requires an invasive procedure. We recently developed an approach for blood-based cancer detection by analyzing fragmentation characteristics in plasma cell-free DNA (cfDNA). In this study, we evaluate the performance of this method in differentiating between patients with malignant and benign pancreatic lesions. We collected plasma samples from 81 patients at the time of endoscopic evaluation or surgical resection of a PCL. We simultaneously obtained plasma samples from 209 PDAC patients and 56 healthy individuals. Using plasma cfDNA whole genome sequencing data from patients with PDAC and healthy individuals, we trained and cross-validated an ensemble machine learning model based on 10 genomic features capturing plasma cfDNA fragmentation patterns. This model was applied to sequencing data from patients with PCL for independent evaluation of diagnostic performance. In training and cross-validation, the model demonstrated an area under the receiver operating characteristic curve (AUROC) of 0.90 for differentiating PDAC samples from healthy individuals. Of the 81 PCL patients, ten patients (12.3%) were found to have a malignant PCL, with either high-grade dysplasia or invasive carcinoma. The trained model showed an AUROC of 0.78 for differentiating malignant from benign lesions, achieving 50% sensitivity and 62.5% positive predictive value at 95% specificity. Our results are a novel demonstration that a peripheral blood test based on plasma cfDNA analysis can enable differentiation between malignant and benign pancreatic lesions. This approach may improve risk-stratification and clinical decision-making regarding the necessity for surgical resection. Larger studies and clinical trials to validate these results and evaluate their impact on outcomes are warranted. Citation Format: Clayton T. Marcinak, Michelle D. Stephens, Bradon R. McDonald, Nabeel Merali, Stephanie M. McGregor, Sharon M. Weber, Adam E. Frampton, Shivan Sivakumar, Rebecca M. Minter, Muhammed Murtaza. Blood-based differentiation of malignant and benign pancreatic lesions using analysis of fragmentation patterns in cell-free DNA [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3674.