We furthermore hypothesized that mutated oncoproteins (e.g., KRAS mut) or tumor suppressor genes (e.g., P53 mut) can potentially be detected in single EV, while the same mutations occurring in cancer cells are causal in increased tumoral EV shedding ( 21– 24). We argue that one way to address the current question is to perform sEVA because the identification of a small number of tumor-originating EV (such as those found in microscopic cancers) in a background of host EV may be impossible by bulk methods. One of the key questions that have thus emerged is whether this failure is due to the choice of a given molecular biomarker, the extreme heterogeneity of PDAC, or the inability to accurately detect very small numbers of biomarker-positive EV in a population of normal EV. While research is still ongoing, it has also become apparent that many of the described biomarkers have failed in larger clinical validation cohorts ( 18– 20). Irrespective of the specific physical measurement, the next question is which molecular biomarker to assay. Many of the earlier techniques relied on bulk measurements requiring 10 2 to 10 6 EVs for a single measurement while single EV methods are being actively developed, they measure different parameters (often size rather than molecular biomarkers) ( 16, 17). We and others have developed a number of methods to analyze EV ( 12– 15). Especially in early disease, the lower diagnostic accuracy is likely due to the large stromal component of PDAC, coexistence with focal areas of pancreatitis indistinguishable from tumor-free pancreatitis, and the low quantity of circulating biomarkers at early stages ( 7). Although EV, cfDNA, and metabolism markers have already shown promise for PDAC diagnostics, their current detection sensitivities require further improvements to be clinically useful. These “liquid biopsy” markers include mutated or methylated cell-free DNA (cfDNA) ( 5– 8), tumor-associated extracellular vesicles (EVs) ( 9), circulating tumor cells ( 10), and metabolic parameters (e.g., onset of diabetes and muscle wasting) ( 11). Technological advances continue to rapidly improve the capacity to detect circulating biomarkers in blood samples ( 3, 4). High-resolution, contrast-enhanced CT and MRI imaging are routinely obtained, but their sensitivity in detecting minimal disease is often limited especially in the neoadjuvant ( 2) and postoperative settings. If we are to cure more patients with pancreatic cancer, new diagnostic tools are necessary to identify patients with preclinical and truly localized disease and to explore novel therapeutic approaches to allow for more R0/N0 resections.

Despite state-of-the-art computed tomography (CT), magnetic resonance imaging (MRI), and endoscopic ultrasound for patients with “localized disease by imaging” and who undergo surgery, the rate of pathologically positive margins is 40 to 70% and the rate of pathologically positive peri-pancreatic nodal disease is 60 to 80%. Under the current therapeutic paradigm, only 20% of patients will present with surgically resectable disease, while 40% will present with localized but surgically unresectable disease and 40% will present with metastatic disease. The only way to cure pancreatic cancer is to surgically resect it when it is still localized. In 2021, 60,430 Americans are expected to develop pancreatic cancer and an estimated 48,220 Americans will die from the disease ( ). Pancreatic ductal adenocarcinoma (PDAC) currently represents the third leading cause of cancer death in the United States, and it is expected to become the second leading cause by 2030 ( 1).

Dead cells mutations