ECCO Fellowship Study Synopsis: Brecht Hens
Brecht Hens, ECCO Grant Awardee
Circulating tumour DNA and artificial intelligence as screening tools for dysplasia in Inflammatory Bowel Disease
© Brecht Hens
Background & aim of research
Patients with colonic Inflammatory Bowel Disease (IBD) are at increased risk for the development of colorectal cancer. Surveillance colonoscopy is therefore advised, starting at 8 years after initial diagnosis of IBD and then repeated every 1–5 years based on the individual risk profile. However, screening based solely on colonoscopy is flawed as interval carcinomas still account for around 40%–50% of all colitis-associated carcinomas (CAC). In sporadic colorectal cancer, both liquid biopsies and artificial intelligence (AI) have proved to be feasible and to yield promising results. Patients with IBD were systemically excluded from these trials.
The aim of this research project is to improve the early detection of IBD-associated dysplasia by (1) developing non-invasive biomarkers using blood and/or stool samples to identify high-risk individuals and (2) developing a machine learning algorithm to aid in the detection of neoplastic lesions during colonoscopy.
Methodology/experiments that will be used
To detect circulating tumour DNA fragments in blood or stool samples of patients with IBD-associated dysplasia and colorectal cancer, we will leverage a recently developed technique, CyclomicsSeq, which is based on Oxford Nanopore MinION sequencing of concatenated copies of a single DNA molecule. To construct the computer-assisted detection system, we will collect and analyse still images and video recordings of dysplastic lesions encountered during colonoscopy in these patients.
Anticipated main impact
We believe that CAC-specific biomarkers could identify high-risk patients and guide the timing of surveillance colonoscopy, and that machine learning can improve detection of dysplastic lesions in real time during IBD surveillance colonoscopy.
The combination of AI and liquid biopsies may pave the way for an individually tailored surveillance programme, ultimately resulting in improved outcomes for patients with IBD and an opportunity for less frequent surveillance.
As of now, the first patients have been included. We anticipate that further inclusion and experimental optimisation will continue throughout 2023. A first interim analysis to evaluate the efficacy and accuracy of CyclomicsSeq will be performed after 4 months.