Lynch syndrome affects an estimated one in 300 people, making it the most common hereditary cause of colorectal cancer. Despite regular surveillance colonoscopies, the lifetime risk of colorectal cancer for many affected individuals remains around 50 percent. This makes the highest possible quality of colonoscopy all the more important.
“People with Lynch syndrome often undergo regular colonoscopies over the course of decades. For them, any improvement in surveillance is of great importance,” says co-first author Dr. Robert Hüneburg, senior physician at Medical Clinic I of the UKB, which is part of the Center for Integrated Oncology (CIO) in Bonn. “That’s why we wanted to know whether artificial intelligence could make an additional contribution to detecting precancerous lesions even more reliably.”
Largest Study of Its Kind
For the international CADLY2 study, 757 people with genetically confirmed Lynch syndrome were examined at nine specialized centers in Germany, the Netherlands, Spain, and Belgium. The researchers investigated whether an AI-supported assistance system detects more adenomas during colonoscopy than a high-quality standard colonoscopy without AI support. These changes in the intestinal mucosa are considered important precursors to colorectal cancer, and their early removal can prevent the development of cancer.
This makes CADLY2 the largest endoscopic study to date worldwide involving people with Lynch syndrome and, at the same time, the largest randomized trial on the use of artificial intelligence in this high-risk group. The study was initiated and coordinated at the University Hospital Bonn, which also recruited a significant proportion of the patients included in the study.
Experience Remains Crucial
The analysis showed that, in specialized centers, additional AI support did not lead to the detection of more relevant precancerous lesions compared with standard colonoscopy performed by experienced examiners. The high quality of the examinations left little room for further improvements through the technology used.
“The result may seem surprising at first,” says co-senior author Prof. Dr. Jacob Nattermann, head of the Hepatogastroenterology Section at Medical Clinic I of the UKB and a member of the Transdisciplinary Research Area (TRA) “Life & Health” at the University of Bonn. “Our findings show that modern AI systems do not automatically lead to better outcomes. In specialized centers with a high level of expertise, careful colonoscopy performed by experienced examiners remains the decisive factor in cancer prevention.”
According to the researchers, the results do not fundamentally rule out the use of artificial intelligence in endoscopy. Rather, they show that the benefits of new technologies depend heavily on the clinical setting, the examiners’ experience, and the initial quality of the examination.
“Our study helps to realistically assess the use of artificial intelligence in high-risk surveillance,” said Dr. Hüneburg. “At the same time, the results underscore the importance of specialized centers and structured surveillance programs for people with Lynch syndrome.”
“The study makes an important contribution to the evidence-based assessment of artificial intelligence in colorectal cancer surveillance,” emphasizes Prof. Dr. Christian P. Strassburg, Director of Medical Clinic I and author of the study. Future studies will now investigate in which areas and under what conditions AI systems can actually provide additional benefits for patients.