Ultra-sensitive liquid biopsy technology detects cancer before standard methods

A method that uses artificial intelligence to detect tumor DNA in the blood has demonstrated unprecedented sensitivity in predicting cancer recurrence, according to a study led by scientists from Weill Cornell Medical School, NewYork-Presbyterian, New York Genome Center (NYGC) and Memorial Sloan Kettering (MSK). The new technology has the potential to improve cancer treatment by detecting relapse very early and closely monitoring tumor response to therapy.

In a study published June 14 in the journal Nature Medicine, researchers showed that they were able to train a machine learning model, a type of artificial intelligence platform, to detect circulating tumor DNA (ctDNA) based on DNA sequencing data from patient blood tests with very high sensitivity and accuracy. They successfully demonstrated the technology in patients with lung cancer, melanoma, breast cancer, colon cancer and precancerous colon polyps.

“We were able to achieve significant improvements in signal-to-noise ratio, allowing us, for example, to detect cancer recurrence months or even years before standard clinical methods,” said study co-author Dr. Dan Landau, professor of medicine at Department of Hematology and Medical Oncology at Weill Cornell Medical School and a core member of the New York Genome Center.

The study's co-author and first author was Dr. Adam Widman, a postdoctoral fellow in Landau's lab who is also a breast oncologist at MSK. Other first authors were Minita Shah from NYGC, Dr. Amanda Friedendahl from Aarhus University and Daniel Halmos from NYGC and Weill Cornell Medical School.

Liquid biopsy technology has been unable to realize its great potential for a long time. Most existing approaches target relatively small sets of cancer-associated mutations that are often too rare in the blood to detect reliably, leading to underestimation of cancer recurrences.

Several years ago, Dr. Landau and his colleagues developed an alternative approach based on whole-genome sequencing of DNA in blood samples. They showed that much more “signal” could be collected in this way, allowing for more sensitive and logistically easier detection of tumor DNA. Since then, this approach has been increasingly adopted by liquid biopsy developers.

In the new study, the researchers took it one step further, using an advanced machine learning strategy (similar to that used in popular AI applications such as ChatGPT) to detect subtle patterns in sequencing data, specifically to distinguish patterns indicative of presence of cancer, from patterns indicating sequencing errors and other “noise.”

In one test, the researchers trained their system, which they called MRD-EDGE, to recognize patient-specific tumor mutations in 15 colon cancer patients. After surgery and chemotherapy, the system predicted, based on blood data, that nine of them still had cancer. In five of these patients, relapse was later detected several months later by less sensitive methods. However, there were no false negatives: none of the patients considered free of tumor DNA by MRD-EDGE experienced a relapse during the study period.

MRD-EDGE has demonstrated similar sensitivity in studies of patients with early-stage lung cancer and triple-negative breast cancer, detecting all but one relapse early and monitoring tumor status during treatment.

Researchers have demonstrated that MRD-EDGE can even detect mutant DNA from precancerous colon adenomas, the polyps from which colon cancers develop.

"It was not clear that these polyps could release detectable ctDNA, so this is a significant advance that may point to future strategies aimed at detecting precancerous changes," said Dr. Landau, who is also a member of the Sandra and Edward Meyer Cancer Center at Weill Cornell Medical School and as a hematologist-oncologist at NewYork-Presbyterian/Weill Cornell Medical Center.

Finally, the researchers showed that even without prior training on patient tumor sequencing data, MRD-EDGE can detect immunotherapy responses in melanoma and lung cancer patients weeks before detection using standard X-ray imaging.

"Overall, MRD-EDGE addresses a great need, and we are excited about its potential and are working with industry partners to try to bring it to patients," said Dr. Landau.