Advanced Liquid Biopsy Technology Detects Cancer Earlier Than Conventional Methods

Traditional methods have focused on a narrow range of cancer-associated mutations that are often present in such low quantities in the blood that they escape detection, leading to undetected cancer recurrences. Now, an artificial intelligence (AI)-powered technique for detecting tumor DNA in the bloodstream has demonstrated remarkable sensitivity in predicting cancer recurrence, promising to enhance cancer management through early detection of recurrences and close monitoring of tumor response during treatment.

Several years back, researchers at Weill Cornell Medicine (New York, NY, USA) pioneered a method that employs whole-genome sequencing of DNA from blood samples. This approach has proven to capture a greater "signal," facilitating a more sensitive and simpler means of detecting tumor DNA. This methodology has gained traction among liquid biopsy developers. In their latest research, the team employed a machine learning model, a form of AI, to identify circulating tumor DNA (ctDNA) using sequencing data from patient blood samples, achieving high levels of sensitivity and accuracy. They successfully applied this technology in patients with lung cancer, melanoma, breast cancer, colorectal cancer, and precancerous colorectal polyps.

In their latest study, which was published on June 14 in Nature Medicine, the researchers utilized an advanced machine learning strategy (similar to that used in ChatGPT and other popular AI tools) to detect subtle patterns in the sequencing data, particularly distinguishing cancerous patterns from sequencing errors and other "noise." In one instance, they trained their system, named MRD-EDGE, to identify specific tumor mutations in 15 colorectal cancer patients. Post-surgery and chemotherapy, the team used MRD-EDGE to analyze blood data to predict residual cancer in nine patients. Months later, using less sensitive techniques, five of these nine patients were confirmed to have experienced a recurrence of cancer. Notably, there were no false negatives; patients identified by MRD-EDGE as tumor DNA-free did not experience recurrences during the study period.

MRD-EDGE also demonstrated comparable sensitivity in studies involving patients with early-stage lung cancer and triple-negative breast cancer, accurately detecting nearly all recurrences and effectively monitoring tumor status throughout treatment. The system proved capable of detecting mutant DNA from precancerous colorectal adenomas, which are the polyps that can develop into colorectal tumors. Furthermore, the researchers found that MRD-EDGE could track responses to immunotherapy in melanoma and lung cancer patients, identifying changes weeks before they could be detected by traditional X-ray-based imaging, even without prior training on sequencing data from patients' tumors.

“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,” said Dr. Dan Landau, a professor of medicine in the division of hematology and medical oncology at Weill Cornell Medicine. “On the whole, MRD-EDGE addresses a big need, and we’re excited about its potential and working with industry partners to try to deliver it to patients.”

Related Links:
Weill Cornell Medicine
New York Genome Center