Artificial intelligence to determine what cysts will be pancreatic cancer

A team of scientists, led by US researchers, has shown that a laboratory test using artificial intelligence tools has the potential to determine more precisely which people with pancreatic cysts will develop pancreatic cancer in the future.

The description of the tool, for now, a proof of concept, is published this Wednesday in the journal Science Translational Medicine and according to those responsible, although the results are "exciting, needs to be validated prospectively."

Called "CompCyst" (in English, for the complete analysis of cysts), this test incorporates measurements of molecular and clinical markers of the fluid of the cysts and "seems to be in the way of significantly improving conventional clinical and imaging tests," points out in a note the Johns Hopkins Medicine.

The test was developed from patient data, including those related to symptoms, computed tomography images, and molecular features such as DNA alterations within the fluid of the cyst.

Thus, the molecular profiles of a large number of pancreatic cysts (862) were evaluated, to which information from clinical and radiological data was added in a program that used artificial intelligence to classify patients into three different groups.

Those with cysts without the potential to become tumor and for which periodic monitoring would not be required; cysts that have a small risk of progressing to cancer and for which patients should receive regular monitoring; and cysts for which surgery is recommended due to the high probability of progression to cancer.

The scientists found that surgery was not necessary in 45% of the cases: this unnecessary surgery was performed because the doctors could not determine if the cysts were dangerous.

If CompCyst had been used, 60 to 74% of patients - depending on the type of cyst - could have been saved from unnecessary surgery, according to the same sources.

Pancreatic cysts - sacs of fluid in the pancreas - are universal; they are found in 4% of people in their 60s and 8% of those over 70. Some 800,000 people with pancreatic cyst are identified every year in the US alone, although, the researchers say, only a small part of the these develop cancer.

Although most of these cysts are not cancerous, the dilemma facing patients and doctors is the ability to distinguish between those who are precancerous and cysts that will not progress to the tumor, says Anne Marie Lennon, one of the authors of this work and doctor at the Johns Hopkins Kimmel Cancer Center.

And, the current clinical and imaging tests "often" do not, which makes it challenging to determine which patients will not require follow-up, which ones will need it in the long term and which patients should be treated with immediate removal surgery.

That is why adds Christopher Wolfgang, virtually all people diagnosed with pancreatic cyst are followed up in the long term, and surgeons have to make risk-benefit recommendations with limited information.

"We seldom lose cancer, but it is at the expense of operating that a posteriori may not have been necessary," says Wolfgang, who sums up: this study directly addresses these fundamental problems in the management of these cysts.