University of Pennsylvania Researchers Develop Machine-Learning Influenced Liquid Biopsy to Detect Pancreatic Cancer

University of Pennsylvania Researchers Develop Machine-Learning Influenced Liquid Biopsy to Detect Pancreatic Cancer

A multi-department-based group of University of Pennsylvania researchers developed a machine-learning-based, liquid biopsy to detect early signs of the most prevalent form of pancreatic cancer by identifying its biomarkers in an individual sample. After conducting a blinded study, this multi-analyte test exhibits superior accuracy than the sole biomarker test when identifying pancreatic ductal adenocarcinoma (PDAC) and exhibited improvement for staging disease than typical imaging.

Investigators were quoted in the report, “Although a larger validation study is needed, this test may improve early disease detection and, when performed in addition to diagnostic imaging, patient selection for curative intent surgery.” TitledA Multi-analyte panel consisting of extracellular vesicle miRNAs and mRNA, cfDNA and CA19-9 shows utility for diagnosing and staging of pancreatic cancer, the team was led by the Perelman School of Medicine, the Abramson Cancer Center and the School of Engineering and Applied Science.

The Challenge: pancreatic ductal adenocarcinoma (PDAC)

The third leading cause of death by cancer in America, the diseases’ survival rate stands at 9%. Typically, the disease is not caught until it has substantially spread through the body, hence the poor prognosis for patient life. PDAC accounts for about 90% of all pancreatic cancer cases.

However, with early diagnosis there are actions oncologists can take to improve the odds. For example, in scenarios where the disease that hasn’t spread beyond the pancreas—and not surgery candidates—other treatment such as chemo-or-radiation therapy and thereafter surgery can be considered. There are no treatment options for those whose disease has spread pervasively throughout the body.

The Goal

In most cases, once a diagnosis occurs, the patient has already moved towards a metastatic condition. Hence the importance of finding ways to not only detect the disease earlier on, but also assess the optimal treatment option at that point in time. Erica Carpenter, PhD, director of the Liquid Biopsy Laboratory and a research assistant professor of medicine, led the study with David Issadore, PhD, associate professor of bio-engineering and electrical and systems engineering. Carpenter reported, “Right now, the majority of patients who are diagnosed already have metastatic disease, so there is a critical need for a test that can not only detect the disease earlier but also accurately tell us who might be at a point where we can direct them to potentially curative treatment.”

The Penn Breakthrough

The current batch of developed blood-based liquid biopsy markers haven’t demonstrated superiority at diagnosing early-stage PDAC. But the Penn team reported a breakthrough with a panel of PDAC diagnostic biomarkers—such as tumor-associated extracellular vesicle miRNA, carbohydrate antigen 19-9 (CA19-9), and circulating cell-free DNA—combined, the Penn group articulated can more precisely recognize patients with PDAC.

How did they do it?

The team developed a machine learning algorithm and associated models that when employed on data including the biomarker, data can “generate a predictive panel of biomarkers.” Hence the team employed machine learning on a panel of 14 biomarkers—setting up and training the machine with 15 healthy controls, 12 disease controls (3 intraductal papillary mucinous neoplasm, and 9 pancreatitis), and 20 patients with PDAC of various stages.”

In running a blinded study, when executing the machine to identify and detect pancreatic cancer, the team reported a 92% accuracy result. Encouragingly, when assessing disease staging accuracy, the algorithms performed well at 84% accuracy—this is compared to the 65% accuracy associated with imaging alone.

Lead Research/Investigator

Erica Carpenter, PhD, director of the Liquid Biopsy Laboratory and a research assistant professor of medicine

David Issadore, PhD, associate professor of bio-engineering and electrical and systems engineering. 

Call to Action: The study investigators understand the need for studies on larger groups (e.g. larger cohorts) to fully validated the breakthrough tests.