Powered by artificial intelligence, a new lung cancer blood test developed at Johns Hopkins, combined with other measurements, correctly identified 94% of cancer cases in nearly 800 patients.
The lung cancer blood test, published in Nature Communications, looks for tiny fragments of DNA released by tumor cells. AI is looking for patterns in this broken DNA, rather than looking for specific pieces of cancerous DNA like other blood tests in development, New Atlas explained.
Lung cancer kills the most people in the world, the authors note, “largely due to the late stage of diagnosis where treatments are less effective than in earlier stages” – and rates of lung cancer are rising in the world. the whole world.
“We believe that a blood test, or ‘liquid biopsy’, for lung cancer could be a good way to improve screening efforts because it would be easy to do, widely accessible and cost effective,” said the first author of the study, Dimitrios Mathios. .
The DNA difference: Blood tests for cancer usually focus on looking for pieces of mutated tumor DNA.
Rather, Hopkins’ lung cancer blood test is based on the fact that cancer cells are much more chaotic than healthy cells when it comes to their DNA. Healthy cells pack their genetic code like a well-organized suitcase, the researchers said. When cancer cells die, it is as if their loosely tied suitcase has opened, scattering all the pieces in a different way than organized, healthy cells.
To take advantage of this botched signature, researchers developed a technique called DELFI.
DELFI uses machine learning to spot patterns of pieces of DNA associated with tumors, including their size and number, and score them based on their likelihood of indicating cancer.
“DNA fragmentation models provide a remarkable fingerprint for the early detection of cancer which we believe could be the basis of a widely available liquid biopsy test for lung cancer patients,” said Rob Scharpf, associate professor of oncology at Johns Hopkins.
Lung cancer signature search: The researchers put DELFI to the test using blood samples from 796 patients in the United States, Denmark and the Netherlands. When combined with clinical assessment of risk markers, use of a protein biomarker, and CT scans, the model correctly reported 94% of lung cancers. This dropped to 91% for early stage lung cancer and jumped to 96% for more advanced cancers.
But there is a trade-off in making sure to detect as many cancer cases as possible and find false positives. The test had a specificity rate of 80%, which means that 20% of people without lung cancer would also incorrectly test positive.
The test uses machine learning to spot sloppy patterns in scattered tumor DNA.
With such a high false-positive rate, screening everyone would result in far more false positives than true cancers – still requiring additional tests to sort it out. However, the model could be modified to raise the bar for a positive result, which would skip more real cancers, but also potentially make the result more useful.
As the researchers note, DELFI will need to be evaluated in a large-scale clinical trial before it is ready for wider use. They plan to test DELFI on more than 1,000 patients across the United States, including healthy patients, patients with lung cancer and other types of cancers.
We would love to hear from you! If you have a comment on this article or have a tip for a future Freethink story, please email us at [email protected]