Efficacy of Liquid Biopsies Using Extracellular Vesicles for Patients with PDAC

Our technology surpasses traditional methods like ultracentrifugation, considered the gold standard, in terms of yield. Through our technology, the isolation of exosomes results in enhanced purity and reduced background noise. This heightened purity is crucial for subsequent steps in the process. The isolated exosomes are then introduced into a multiplex ELISA to quantify their amount. Following this, a machine learning algorithm is applied to analyze performance curves, including ROC curves, predicting the likelihood of developing pancreatic cancer—whether it falls into a high or low likelihood category.

How did this trial incorporate machine learning, and what information is included in the algorithm used to identify PDAC?

Dr. Dhani: Currently, our proprietary algorithm relies on seven protein biomarkers, specifically measuring the proteins expressed on the surface of exosomes or extracellular vesicles. These biomarkers, which align with existing literature, comprise CA 99, CA 153, prolactin, leptin, ferritin, sHER3, and FGF-2. These seven biomarkers were identified through a meticulous process.

In the realm of machine learning, our algorithm initially considers the 52 biomarkers obtained from the multiplex ELISA. Subsequently, we employ feature selection to identify the most optimal biomarkers, striking a balance between sensitivity and specificity. This meticulous approach results in the final selection of seven protein bio-signatures constituting our proprietary algorithm. The selection is based on both the training set and the validation set, ensuring robustness and reliability.

What were your overall findings of the sensitivity and specificity on liquid biopsies using EVs in detecting PDAC?

Dr. Dhani: A critical aspect to grasp here is the insufficient sensitivity prevalent in current clinical practices, hindering the widespread adoption of modalities. Our emphasis on sensitivity, as validated through training and testing, sets our approach apart. The training set, encompassing approximately 650 patients, featured 105 early-stage pancreatic cancer patients (stages one and two) alongside 545 patient controls. The area under the curve (AUC) impressively yielded 97.1%, showcasing the robustness of our model. Overall sensitivity reached 93.3%, with a specificity of 91%.

However, a comprehensive assessment demands an independent cohort for testing—enter the validation study. This subsequent evaluation involved 113 patients, including 30 documented cases of early-stage pancreatic cancer and 83 controls. Notably, this cohort also incorporated patients with benign diseases, specifically 11 individuals with acute or chronic pancreatitis. The results from the validation cohort demonstrated a sensitivity of approximately 90% and a specificity of 92.8%, reinforcing the reliability and applicability of our model beyond the initial training set.

What further information or studies are needed to confirm the efficacy of a blood test analyzing EVs for early detection in patients with PDAC?

Dr. Dhani: I believe this marks the beginning of an exciting journey in the exosome science space, highlighting the potential of our technology platform and its application in the realm of pancreatic disease. Currently, we are actively conducting an ex-illuminate registry trial, with me serving as the principal investigator for Biological Dynamics. The trial is well underway, aiming to enroll between 1,000 to 5,000 subjects. Our focus is on prospectively collecting blood from high-risk patients predisposed to developing pancreatic cancer.

These high-risk cohorts include patients with pancreatitis (acute or chronic), hereditary pancreatitis, those with genetic mutations, and individuals with a strong family history of pancreatic cancer. Of particular interest are patients with intraductal papillary mucinous neoplasms (IPMNs), characterized by cysts on the pancreas. Addressing a significant unmet need, our goal is not only to identify high-risk individuals but also to distinguish between patients with cysts who truly require surgery and those who may benefit from a watch-and-wait protocol.

The challenge lies in providing clinical providers, including surgeons and gastroenterologists, with a tool to differentiate between high-grade dysplasia and low-grade dysplasia in patients with pancreatic cysts. This differentiation is crucial for guiding appropriate clinical interventions—ensuring that patients in need of surgery receive timely treatment, while those suitable for a watch-and-wait approach are carefully monitored in accordance with an accurate clinical care plan.

Efficacy of Liquid Biopsies Using Extracellular Vesicles for Patients with PDAC

Related Posts