Combining Machine Learning and Nanofluidic Technology for The Multiplexed Diagnosis of Pancreatic Adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer-related death in the United States with an overall 5-year survival of 9%. Diagnosis and staging continue to rely on endoscopic biopsy and imaging, and as such most patients are diagnosed at an advanced stage. Sufficiently sensitive and specific screening tests for early disease remain elusive.

Moreover, while curative-intent surgery is an option for patients whose disease is confined to the pancreas, distinguishing patients with metastases who are unlikely to benefit from surgery, remains challenging due to occult metastases not detectable by imaging. To address these challenges, several blood-based liquid biopsy biomarkers have been developed but show low

sensitivity for detection of early-stage disease. We have recently shown that circulating tumor derived extracellular vesicles(EVs) can be isolated from blood and their RNA cargo used to diagnose early pancreatic cancer and stage disease. These findings suggest an opportunity to improve patient outcomes through development of a non-invasive diagnostic for pancreatic

cancer. However, as has been well documented, EVs are highly heterogeneous in their expression of protein surface markers and their nucleic acid and protein cargo, and originate from multiple cell types in the tumor micro environment (TME) (e.g. tumor cells, tumor associated macrophages). The ultimate goal of this proposal is to address a fundamental

technological unmet need in EV diagnostics, by further developing our new approach to EV subpopulation isolation using magnetic nanopores, which combines the benefits of nano-scale sorting with sufficiently fast flow rates (106x faster than typical nanofluidic approaches) to be practical for clinical diagnostics. In this R33, we develop this approach into a multiplexed EV

assay that will allow multiple unique EV sub-populations - based on surface marker expression- to be isolated and their RNA cargo profiled. Building on our prior work that demonstrated the value of analyzing single EV-subpopulations, and improved sensitivity of a multi-analyte vs single analyte test, we will develop a multi-analyte EV-based assay that algorithmically

combines tumor associated EV RNA from multiple circulating EV isolates from the TME, as well as Circulating cell-free DNA (ccfDNA) concentration, circulating tumor DNA-based KRAS mutation detection, and CA19-9 using machine learning.