Swiss Cancer ResearchOur commitmentWe fund researchersFeatured scientific projectsImproving colorectal cancer screening

Improving colorectal cancer screening

Colorectal cancer is one of the three most deadly cancers in Switzerland because, unfortunately, it is often detected too late. In order to detect the precursors of the disease more reliably than before, researchers supported by Swiss Cancer Research are developing new examination methods.

The gastroenterologist Kaspar Truninger with the monitor for colonoscopy.

Every year, around 1700 people in Switzerland die of colorectal cancer. "That doesn't have to be the case," says gastroenterologist Kaspar Truninger. "Because at an early stage, the disease is curable." However, it does not cause any symptoms then - and therefore remains undetected in many cases. Truninger wants to change this: He analyzes the molecular traces that a developing tumor leaves in the genome long before it becomes apparent through symptoms of the disease, such as long-lasting, cramping abdominal pain.


Reading and understanding traces in the genome

The most important examination methods for the early detection of colorectal cancer are the blood-in-stool test and colonoscopy. If the blood-in-stool test is abnormal, a colonoscopy is indicated, during which the doctor examines the inside of the intestine - and can also remove any precancerous lesions, known as polyps. Truninger and his colleagues from Basel and Lugano began amassing a considerable collection of more than 1600 intestinal mucosa samples seven years ago. The samples came from people who had undergone such colonoscopies.

Since then, they have been studying the genetic material contained in the samples using the most modern methods of molecular biology. The traces that the team around Truninger wants to read and understand are  referred to as epigenetic methylation patterns. Methyl groups are small appendages that mark when which genes are active or inactive in a cell. Indeed, in the colon, numerous genes that protect against colorectal cancer become silenced with age.

Truninger refers to this as a methylation signature that changes throughout life. It is anything but trivial to distinguish which patterns belong to a healthy aging process and which, on the other hand, point to a worrying development of colorectal cancer - a precancerous lesion. Accordingly, Truninger's team also relies on machine learning to analyze the vast amount of data.

"We want to read and understand the traces that a developing tumor leaves in the genome long before it becomes apparent through signs of disease."

Polyps hidden in the folds of the intestine

Using these methods of artificial intelligence, the researchers can correctly deduce from the traces in the genetic material of the healthy intestinal mucosa in about three out of four cases whether or not a precancerous lesion is present in the corresponding section of the intestine, as Truninger writes in the final report of his project. The intestinal specialist hopes that with even more data, this discrimination accuracy can be increased to more than 90 percent. "From then on, it will become clinically relevant," Truninger says. That's because with such high values, the method could improve early detection of colorectal cancer. Today, an estimated five percent of larger polyps and up to 20 percent of smaller polyps go undetected during a colonoscopy because they hide in the folds of the intestine. "They get missed," Truninger says.

That can have nasty consequences, because if a colonoscopy is unremarkable, the next screening isn't due for another ten years. "In that interval, the missed precancerous lesions can develop into cancer," Truninger says. He envisions that in the future the medical profession will not only perform colonoscopies, but also have the genetic material of a tissue sample from the intestinal mucosa examined at the same time. "If the epigenetic signature indicates the presence of a lesion, then the interval until the next colonoscopy should be shortened," Truninger holds.

Project ID: KFS-4301-08-2017