Cancer is one of the leading causes of death in humans, and it is often difficult to detect until it is too late,media reported. But that could change. Researchers have developed a new type of AI blood test that can accurately detect more than 50 different types of cancer and even identify their location in the body.
Cancers are so diverse that it is almost impossible to keep a close eye on all cancers through routine tests. Instead, the disease is usually detected after the patient has developed symptoms and until the doctor begins to look for it specifically. By then, in many cases it may be too late. Ideally, the doctor will routinely examine the patient to find any type of cancer that may be occurring in the body, providing the best way to succeed in treatment. That’s where the new study is working.
The test uses machine learning algorithms to search for specific chemical changes in cancer-related DNA, called methylation patterns. It is found in the form of cell-free DNA (cfDNA), which falls off into the bloodstream from many cells, including tumors. The researchers first trained machine learning algorithms using more than 3,000 blood samples in the Circulating Cell-Free Genome Map (CCGA). Half of them have cancer – one of 50 different types of cancer, while the other half don’t. Once the algorithm understands the methylation pattern sought, another 1,200 samples can be classified.
To be sure, the new test has been largely successful, with more accurate future cancer tests. It can detect 18% of Stage I tumors, 43% stage II tumors, 81% stage III tumors and 93% of stage IV tumors. It was also able to ascertain with 93 per cent accuracy which tissue the cancer originated in, and, importantly, the false positive rate was only 0.7 per cent.
“These data support the ability of targeted methylation tests to meet the basic requirements of early blood testing for multiple cancers that we believe can be used for population-level screening: the ability to detect multiple types of deadly canceratdes at once.” The study had a very low false positive rate and was able to identify the location of the cancer in the body with high accuracy, which could help healthcare providers guide the next steps in diagnosis and care. “
The team says the results should be extended to larger populations, but more tests are needed in larger groups. There are also some issues that need to be addressed. For some reason, cancer caused by the human papillomavirus (HPV) makes it harder for the system to locate the cancer, the researchers said. In addition, patients were not followed for a full year, so they could not rule out the possibility that some “non-cancerous” subjects did have the disease. But the main drawback is that early cancer detection rates are still low. There is no doubt that improving this approach will be an important factor in the future success of this approach. Still, the general consensus seems to be a promising breakthrough.
Christina Wharton, of the University of New South Wales, who was not involved in the study, said: “This is an exciting effort to combine state-of-the-art laboratory technology with artificial intelligence. It highlights the potential to detect cancer DNA in the blood. One of the advantages of the study was that it included a large sample from healthy people. You need to get many samples from people who don’t have cancer to prove that the test doesn’t produce false positive results, and there are thousands of this study. Finally, the challenge for this screening test and all cancer screening tests is how to identify early small cancers. Advanced cancer is easier to detect. I would say that the discovery of smaller, early ones is still in progress. “
The work builds on previous advances in the general cancer blood test, which also looks for cfDNA methylation patterns, but only 20 cancers. Others are looking for different signs of cancer in the blood, such as genetic mutations, platelet RNA maps, damage to white blood cells, elevated levels of certain proteins, and even DNA from microbes affected by tumors.
While all of these cancer screening tests are still a long way from clinical application, it is encouraging to see such promising results in this area. And the latest seems to be one of the most accurate and influential.
The latest study is published in the journal Annals Oncolog of the