One of humanity’s leading killers – cancer – is only so deadly because it’s hard to detect early on. Once symptoms emerge, it is often too late. Furthermore, there are so many different kinds of cancer that it’s nearly impossible to keep an eye out for all of them during routine checkups. That’s why the disease tends only to be spotted when the doctor is looking for it.
For this reason, researchers have been developing tests that can detect several different kinds of cancer in the bloodstream. Now, a team has created a new type of AI-powered blood test that can detect over 50 different types of cancer. It even identifies where the tumor is in the body. The work builds on previous progress in universal cancer blood tests, but those were only able to detect 20 types of cancer. The study has been published in Annals of Oncology.
The test employs a machine-learning algorithm to search for methylation patterns – specific chemical changes to DNA – that are associated with cancer. What it’s looking for is cell-free DNA (cfDNA), which is something tumors and cancerous cells shed into the bloodstream.
First, the researchers had to train the algorithm to be able to spot out the methylation patterns. They did this by feeding it data from over 3,000 blood samples in the Circulating Cell-free Genome Atlas (CCGA). Half of the samples had one of the 50 different types of cancer. Then, the team put the machine to work on classifying a further 1,200 samples, again half of which had cancer.
Of those second batch of samples, the new test was able to detect 93% of stage IV tumors, 81% of stage III, 43% of stage II, and 18% of stage I. The false-positive rate was 0.7%. Perhaps most impressively was the accuracy rate at which it could pinpoint which tissues the cancer originated in, which was 93%. It almost only had a difficult time figuring out where diseases caused by human papillomavirus were.
Michael Seiden, the senior author of the study, said:
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.
Unfortunately, the detections for early-stage cancers are still quite low. The team is now dedicated to improving this issue. Nevertheless, people are impressed with the outcome overall.
Kristina Warton, affiliated with the University of New South Wales but not involved in the study, said:
This is exciting work bringing together cutting-edge laboratory techniques with AI. It highlights the potential of a test for cancer DNA in the 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.
There are other types of universal cancer blood tests that look for different signs of cancer in the blood, such as mutated genes, damaged white blood cells, elevated levels of specific proteins, platelet RNA profiles, and even DNA from microbes that are affected by tumors. And while all these different cancer screening tests are not ready yet for clinical use, it is still encouraging to know their advancement has been providing promising results. When such tests are available, it would ideally be incorporated into a yearly checkup routine.
