Precision Medicine Tool May Enhance Pancreatic Cancer Outcome Prediction

The use of a precision medicine tool designed by Cedars-Sinai researchers has led to the development of a blood test that outperforms the only United States Food and Drug Administration (FDA)-approved test to forecast pancreatic cancer outcomes, according to a study published this week in Nature Cancer.

The tool, known as the Molecular Twin Precision Oncology Platform, used artificial intelligence (AI) to analyze molecular and genetic information from 74 patients with resected pancreatic ductal adenocarcinoma in order to predict disease survival.

To build the tool, the research team combined 6,363 blood- and tissue- related data points into a machine learning algorithm, which was able to accurately predict cancer survival in 87 percent of the study cohort. The model was then refined to perform similarly using only 589 data points.

When the tool was refined even further, the researchers were able to identify blood-based proteins that acted as the best predictor of pancreatic cancer outcomes. Using this information, the research team developed a blood test and compared it to CA 19-9, the only FDA-approved blood test to predict pancreatic cancer survival.

The test significantly outperformed CA 19-9, and these findings were validated in independent datasets from Johns Hopkins University, Massachusetts General Hospital, and The Cancer Genome Atlas.

The researchers emphasized that the tool has significant potential to advance cancer care.

“Molecular Twin… can be used to study any tumor type, including pancreatic cancer, which is notoriously difficult to treat,” explained senior author Dan Theodorescu, MD, PhD, director of Cedars-Sinai Cancer and the PHASE ONE Foundation Distinguished Chair, in a press release. “Using our Molecular Twin technology, we anticipate creating tests that can be used even in locations that lack access to advanced resources and technology, pairing patients with the most effective therapies and expanding the availability of precision medicine.”

Biomarkers play a key role in guiding treatment for many conditions, including cancer, making the tool’s ability to identify features of blood and tissue crucial for outcome prediction.

“There's a huge unmet need for the development of biomarkers to guide our treatment of pancreatic cancer,” said first author Arsen Osipov, MD, assistant professor of Medicine and program lead in the Pancreatic Cancer Multidisciplinary Clinic and Precision Medicine Program at Cedars-Sinai Cancer. “We had already undertaken a comprehensive collection of blood and tissue samples from patients with pancreatic cancer, and this gave us a good opportunity to test the Molecular Twin platform. As we grow the platform with more patients, Molecular Twin will become an even more robust tool, not just in pancreatic cancer, but across all cancers.”

The study also underscored the major role of proteins as biomarkers in cancer care.

“Once a patient has cancer, proteins act as the body’s first responders, and their activity helps us determine how a patient’s body is reacting,” noted Jennifer Van Eyk, PhD, director of the Advanced Clinical Biosystems Institute in the Department of Biomedical Sciences at Cedars-Sinai. “Proteins turned out to be the main drivers of our pancreatic cancer models. And in future studies, proteins will also help us track how well a patient is responding to treatment.”

Moving forward, the research team aims to expand the Molecular Twin platform to include additional data types, such as wearable device feedback, samples from the tumor microenvironment and gut microbiome, and medical imaging data.

“A majority of our cancer patients are allowing us to include their clinical information and samples from blood, tumor, and other sources so that we can continue to build the Molecular Twin platform,” Theodorescu said. “This rich pool of data will help us discover biomarkers for additional cancer types, and eventually lead to the development of new treatments and the opportunity to identify at-risk patients before their cancer develops, so that we can prevent it entirely.”

Other institutions are also investigating how AI can advance precision oncology.

This month, researchers from Sylvester Comprehensive Cancer Center at the University of Miami Miller School of Medicine shared that they have developed an individual risk prediction model for multiple myeloma.

The tool was developed to improve upon the performance of current prognostic tools by leveraging insights into tumor biology, which could enhance prognosis and treatment.

Using clinical, treatment, and genetic data, the model flagged 90 “driver genes” — those with cancer cell mutations that appear to drive tumor growth – and accurately predicted each patient’s risk.