A machine learning model developed by researchers at the Johns Hopkins Kimmel Cancer Center filters out biological noise in liquid biopsy samples, helping clinicians better match therapies to their patients' tumors, according to a June 9 announcement.
The research, published May 1 in Clinical Cancer Research and funded in part by the National Institutes of Health, addresses challenges with liquid biopsies. Liquid biopsies analyze cell-free DNA fragments from tumors in blood samples to identify mutations and enable targeted therapy selection. However, these tests may also detect mutations that accumulate in white blood cells through an aging-related process called clonal hematopoiesis. Such mutations are common among older individuals and those who have undergone chemotherapy or radiation.
To address this issue, Canzoniero and colleagues developed a machine learning model called plasmaCHORD. The model uses characteristics of DNA fragments—including differences in how tumor-derived and white blood cell-derived DNA is fragmented—as well as patient age and gene mutation type to estimate whether a mutation originates from the tumor or white blood cells. The team trained plasmaCHORD on samples from 225 patients with various cancers and verified its accuracy using matched genetic sequencing of tumor cells and white blood cells.
They then tested plasmaCHORD on a separate set of 114 patients from another institution using a different sequencing platform. The model maintained similar performance levels, improving correct identification of clinically relevant mutations’ origins from about 50% to 83%. Researchers also demonstrated that plasmaCHORD’s predictions helped clinicians avoid selecting likely ineffective therapies during evaluations at the Johns Hopkins Molecular Tumor Board.
"About one-third of mutations detected in tumor-naive liquid biopsies can originate from white blood cells, and our ability to match targeted therapies to each patient's genomic profile depends on our ability to distinguish tumor mutation from biological noise," said Valsamo Anagnostou, M.D., Ph.D., senior study author and leader of the Johns Hopkins Molecular Tumor Board at the Johns Hopkins University School of Medicine. "An artificial intelligence model applied to standard liquid biopsy tests could be both clinically valuable and quickly scalable." Canzoniero said, "PlasmaCHORD can be used going forward for both research and potentially for clinical purposes to identify the origin of mutations in a liquid biopsy if you're not sure. We are thinking about working on a future version that would hopefully have even better performance."
