Francesc Balaguer - CGA-IGC Research Committee
Lynch syndrome (LS), an autosomal dominant disorder caused by pathogenic germline variants in DNA mismatch repair (MMR) genes, represents the most common hereditary colorectal cancer (CRC) syndrome. Although colonoscopy has been shown to reduce CRC incidence, a substantial proportion of LS carriers develop incident CRC despite regular surveillance. In the general population, one of the main causes of post-colonoscopy CRC are missed lesions on previous colonoscopies. In LS, high miss rates for colorectal neoplasia (12% – 74%) have been reported in several back-to-back colonoscopy studies. Adenomas in LS are usually small with a flat or non-polypoid morphology, which makes the detection of these lesions challenging.
Recently, artificial intelligence (AI) has been introduced in the endoscopy clinical setting, and computer-aided diagnosis (CAD) based on a deep-learning system is able to provide real-time output in lesion detection and characterization. The CAD system has been trained to process colonoscopy images and to superimpose them in real-time with green boxes over suspected lesions. A recent meta-analysis including 5 RCT with 4354 patients showed a significantly higher adenoma detection rate (ADR) in the CAD group with respect to control (RR=1.44).
Dr. Robert Hüneburg, from the University Hospital Bonn, National Center for Hereditary Tumor Syndromes in Germany, aimed to describe the role of AI in LS with a study entitled “REAL-TIME USE OF ARTIFICIAL INTELLIGENCE (CADEYE) IN COLORECTAL CANCER SURVEILLANCE OF PATIENTS WITH LYNCH SYNDROME – A RANDOMIZED PILOT TRIAL”. This study investigated AI-assisted colonoscopy in comparison to HD white-light endoscopy (HD-WLE). It included 96 individuals with LS (excluding PMS2 carriers) stratified by previous CRC and affected MMR gene with a 1:1 allocation ratio (AI-assisted with CAD-EYE; Fujifilm, Japan vs. HD-WLE). In the HD-WLE arm, adenomas were detected in 12/46 patients compared to 18/50 in the AI arm (26.1% [95% CI 14.3-41.1] vs. 36.0% [22.9-50.8]), however, this difference did not reach statistical significance (p=0.379). Interestingly, the increased ADR was due to identification of mostly flat adenomas, which were more frequently found with AI (p=0.018). The median withdrawal time was not statistically different between the HD-WLE and AI arms (14 vs. 15 min; p=0.170).
Dr. Hüneburg presented these findings at the Collaborative Group of the Americas on Inherited Gastrointestinal Cancers (CGA-IGC) Annual Meeting in Nashville, TN on November 12th 2022. He concluded: “This is the first study analyzing the clinical utility of AI in LS. Although the results show no significant increase in ADR with AI, we saw a significant increase in the detection of flat adenomas, which are the ones that we usually miss during colonoscopy. The effect on ADR could be explained by the small sample size. For this reason, we are conducting a multicenter prospective trial that will tackle this question”.
The full article describing this study, published in February 2023, can also be accessed at the United European Gastroenterology Journal.
If you missed this presentation, you can still register for an On-Demand subscription.
Comentários