Abstract
Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy, often associated with a poor prognosis for patients. One of the major challenges in managing PDAC is the difficulty in early diagnosis, owing to the limited and invasive nature of current diagnostic methods. Recent studies have identified the oral microbiome as a potential source of non-invasive biomarkers for diseases, including PDAC. In this study, we focused on leveraging the differential expression of virulence factors (VFs) encoded by the oral microbiome to create a diagnostic tool for PDAC. We observed a higher alpha diversity in VF categories among PDAC patients compared to healthy controls. We then identified a panel of VF categories that were significantly upregulated in PDAC patients, these being associated with bacterial adherence, exoenzyme production, and nutritional/metabolic processes. Moreover, Streptococcus-derived VFs were notably enriched in PDAC patients. We developed a diagnostic model using random forest analysis based on the levels of these VFs. The model's diagnostic accuracy was evaluated using receiver operating characteristic (ROC) curve analysis, with an area under the curve (AUC) of 0.88, indicating high accuracy in differentiating PDAC patients from healthy controls. Our findings suggest that VFs encoded by the oral microbiome hold potential as diagnostic tools for PDAC, offering a non-invasive approach that could significantly enhance early detection and prognosis, ultimately leading to improved patient outcomes.