An AI-powered blood test was able to correctly detect the presence of lung cancer over 90% of the time in patient samples. The corresponding study was published in Nature Communications by researchers from multiple institutions.
Lung cancer is the most common cause of cancer death, with over 2 million people dying from the disease around the world each year. Despite this, however, less than 6% of Americans who are at risk of the illness undergo low-dose computed tomography screening. Some of the reasons for this include radiation exposure, fear of false-positive results, and fears around invasive procedures.
The researchers behind the current study thus sought to create a screening method for lung cancer that quells the above concerns and encourages more people at risk for the disease to undergo screening.
For their study, the researchers performed gene sequencing of cell-free DNA in blood samples taken from 365 people, the majority of which were at high risk for lung cancer and had smoking-related symptoms, including coughing and difficulty breathing.
Via a machine learning approach called DNA evaluation of fragments for early interception (DELFI), the researchers found that patients later diagnosed with cancer had widespread variation in their fragmentome (cell-free DNA) profiles, whereas those that did not develop the disease had consistent fragmentome profiles.
The researchers then validated the technology on a different population of 385 people without cancer and 46 with cancer. In the end, the approach was able to detect over 90% of patients with lung cancer, including those with early and advanced stages and different subtypes.
The DELFI approach is now undergoing a clinical trial in the US called DELFI-L101, involving 1,700 participants, including healthy patients, people with lung cancer, and those with other cancers.
"DNA fragmentation patterns provide a remarkable fingerprint for early detection of cancer that we believe could be the basis of a widely available liquid biopsy test for patients with lung cancer," said Rob Scharpf, Ph.D., author of the study.