A new universal cancer blood test can spot over 50 types of tumors and identify where they are in the body
Cancer is one of humanity’s leading killers, and the main reason for that is it’s often hard to detect until it’s too late. But that might be about to change. Researchers have developed a new type of AI-powered blood test that can accurately detect over 50 different types of cancer and even identify where it is in the body.
There are just so many types of cancer that it’s virtually impossible to keep an eye out for all of them through routine tests. Instead, the disease usually isn’t detected until doctors begin specifically looking for it, after a patient experiences symptoms. And in many cases, by then it can be too late.
Ideally, there would be a routine test patients can undergo that would flag any type of cancer that may be budding in the body, giving treatment the best shot of being successful. And that’s just what the new study is working towards.
The test uses a machine learning algorithm to search for specific chemical changes to DNA, called methylation patterns, that are associated with cancer. This is found in the form of cell-free DNA (cfDNA), which is shed into the bloodstream from many cells, including tumors.
The researchers started by training a machine learning algorithm on over 3,000 blood samples in the Circulating Cell-free Genome Atlas (CCGA). Half of these had cancer – one of 50 different types – while the other half didn’t. Once the algorithm had learned what methylation patterns to look for, it was put to work on classifying a further 1,200 samples, of which half had cancer.
And sure enough, the new test was largely successful, becoming more accurate for later stages of cancer. It was able to detect 18 percent of stage I tumors, 43 percent of stage II, 81 percent of stage III and 93 percent of stage IV. It was also able to pinpoint which tissue the cancer originated in with an accuracy of 93 percent, and importantly the false positive rate was just 0.7 percent.
“These data support the ability of this targeted methylation test to meet what we believe are the fundamental requirements for a multi-cancer early detection blood test that could be used for population-level screening: the ability to detect multiple deadly cancer types with a single test that has a very low false positive rate, and the ability to identify where in the body the cancer is located with high accuracy to help healthcare providers to direct next steps for diagnosis and care,” says Michael Seiden, senior author of the study.
The team says that the results should be generalizable to a larger population, but more tests will need to be done in larger groups. There are a few issues to iron out too – the researchers say that for some reason, cancers caused by human papillomavirus (HPV) made it more difficult for the system to figure out where the cancer was. Plus, the patients hadn’t all been followed for a whole year, so they can’t rule out the possibility that some of the “non-cancer” subjects actually do have the disease.
But the main drawback is that detections for early-stage cancers are still low. Improving this will no doubt be an important factor in future success for the method. Still, the general consensus seems to be that it’s a promising breakthrough.
“This is exciting work bringing together cutting edge laboratory techniques with AI,” says Kristina Warton of the University of New South Wales, who wasn’t involved in the study. “It highlights the potential of a test for cancer DNA in blood.
“One of the strengths of the study is the large number of samples from healthy people that were included. You need lots of samples from people without cancer to show that the test doesn’t give false positives, and this study had several thousand.
“Finally, where the challenge is, for this screening test and for all cancer screening tests, is to identify small, early-stage cancers. Advanced cancers are a lot easier to detect. I would say that detecting the small, early ones is still a work in progress.”
The work builds on previous advances in universal cancer blood tests that also look for cfDNA methylation patterns, but were only tested on 20 types of cancer. Others look for different signs of cancer in the blood, such as mutated genes, platelet RNA profiles, damaged white blood cells, elevated levels of certain proteins, and even DNA from microbes that are affected by tumors.
While all of these cancer screening tests are still a long way from clinical use, it’s encouraging to see such promising results in the area. And the newest appears to be one of the most accurate and wide-reaching.
The latest study was published in the journal Annals of Oncology.