An artificial intelligence (AI)-guided approach for detecting tumor DNA in blood has shown remarkable sensitivity in predicting cancer recurrence, according to study results.
In an article recently published in Nature, the team, comprised of researchers from Weill Cornell Medicine, the NewYork-Presbyterian Hospital, the New York Genome Center, and the Memorial Sloan Kettering Cancer Center demonstrated that their machine-learning model could detect circulating tumor DNA based on DNA sequencing data from patient blood tests with very high sensitivity and accuracy in patients with lung cancer, melanoma, breast cancer, colorectal cancer, and precancerous colorectal polyps.
The model, which the researchers call MRD-EDGE, was trained to detect patterns in sequencing data, as well as to distinguish patterns associated with cancer from those suggestive of sequencing errors or other noise.
"We were able to achieve a remarkable signal-to-noise enhancement, and this enabled us, for example, to detect cancer recurrence months or even years before standard clinical methods did so," co-corresponding study author Dr. Dan Landau, a professor of medicine in the division of hematology and medical oncology at Weill Cornell Medicine and a faculty member at the New York Genome Center, said.
In previous research, the MRD-EDGE system team developed an approach using whole-genome sequencing of DNA in blood samples. This approach was able to gather much more signal instead of noise, enabling simpler and more sensitive tumor DNA detection; the team's method has been increasingly adopted by liquid biopsy developers.
With MRD-EDGE, the signal-to-noise enhancement was refined further, with the potential to enable earlier detection of recurrence and improved monitoring of tumor response to therapy, the study findings show.
