A game-changer in cancer detection might be just around the corner. A simple blood test could reveal cancer years before conventional tests — before symptoms even develop.
The quicker you detect any type of cancer, the better your chances to eliminate it without any dangerous complications. Survival rates have improved substantially in recent years, in part due to earlier detection. Unfortunately, detecting cancer early is often difficult — especially before any actual symptoms show up — and the tests themselves are often expensive and unpleasant. But that might soon change.
Using a blood test to detect cancer is not a new idea. It’s extremely challenging, but in the past few years, there have been a few encouraging results. In the latest effort, an international team reports developing a blood test that can detect some types of cancer before it’s even cancer.
“We demonstrated that five types of cancer can be detected through a DNA methylation-based blood test up to four years before conventional diagnosis,” the authors write in the journal Nature Communications.
The team collected blood samples before people had any signs of cancer. They recruited over 123,000 healthy individuals in Taizhou, China, and asked them to undergo annual health checks. This was a huge effort that ultimately required building a specialized warehouse to store the 1.6 million samples that were gathered over time. These volunteers were tracked for 10 years.
The researchers analyzed the blood samples, looking for chemical alterations of the genetic code (called DNA methylation) or telltale blood proteins. They then used machine learning algorithms (a type of artificial intelligence) to look for signs of cancer. They focused on the five most common types of cancer: stomach, esophageal, colorectal, lung, and liver malignancies.
Out of the volunteers, around 1,000 developed cancer. The algorithm successfully flagged cancer in 88% of the participants who had already been diagnosed — and even more impressively, in 95% who were not diagnosed with cancer, but did go on to develop the disease later. The test, called PanSeer, also correctly identified those without cancer 96% of the time, encouragingly calling false positives in very few cases.
Researchers stress that this test isn’t a predictor of cancer — it’s detecting cancerous growths that are asymptomatic.
“While we have demonstrated early detection of cancer four years before conventional diagnosis through use of a longitudinal cohort, we would like to emphasize that the PanSeer assay is likely not predicting patients that will later develop cancer. Instead, the assay is most likely identifying patients who already have cancerous growths but who remain asymptomatic to current detection methods and standard of care, as many cancers do not cause the appearance of symptoms until late in disease development,” the study reads.
The study has some limitations (the storage is imperfect and could have affected some of the samples, the sample size is not very large, and there are always inherent questions with AI). The tests will also always be unable to detect all cancers. Some cancers evolve very slowly, while others grow quickly and might not even be detected with annual screening — but the results are very promising.
It will still be a while before blood cancer tests start becoming a reality, but if the results are confirmed, it seems like a tangible possibility. Even if the results are imperfect, this could, at the very least, direct other diagnostic mechanisms to where they are most needed.
For instance, this sort of test could be used in at-risk populations or even be incorporated into routine checks. If the results are validated in larger trials, this could be a very useful tool in our fight against cancer.
The study “Non-invasive early detection of cancer four years before conventional diagnosis using a blood test” has been published in Nature Communications.