P267 Real-world artificial intelligence-aided colonoscopy does not improve adenoma detection rates in patients with Inflammatory Bowel Disease
Levartovsky, A.(1)*;Levy, I.(1);Bruckmayer, L.(2);Klang, E.(3);Ben-Horin, S.(1);Kopylov, U.(1);
(1)Sheba Medical center, Department of Gastroenterology, Tel Hashomer, Israel;(2)Sheba Medical center, Department of Internal Medicine D, Tel Hashomer, Israel;(3)Sheba Medical center, Department of Diagnostic Imaging, Tel Hashomer, Israel;
Background
Recently, the implementation of artificial intelligence-aided colonoscopy (AIAC) has resulted in improved performance of colorectal cancer (CRC) screening. The adenoma detection rate (ADR) is a key quality indicator in screening colonoscopies and achieving high ADRs is crucial for optimal prevention of CRC. Guidelines suggest that ADRs of 20% in women and 30% in males are indicative of adequate performance. Patients with inflammatory bowel disease (IBD) may be at increased risk for CRC based on disease extent and duration. However, ADR in patients with IBD can be generally lower than in average-risk patients, mainly due to an age gap and the presence of dysplasia associated lesions as opposed to sporadic adenomas. Indeed, there is no consensus for an accepted ADR target in these patients. In addition, the impact of AIAC on ADR in these patients has yet to be described. We aimed to explore the ADR of patients with IBD in a large-volume endoscopic center and evaluate the effect of AIAC on ADR.
Methods
This was a retrospective study conducted at a high-volume gastroenterology department where all endoscopy suites were equipped with an AIA device (GI Genius, Medtronic, Ireland) starting from July 1st 2021. Data was collected on all colonoscopies performed in the 11 months before the mentioned date and compared to a 15-month period afterwards. We excluded patients that performed colonoscopy due to evaluation of IBD severity, for known or suspected malignancy, therapeutic endoscopies, incomplete colonoscopies and colonoscopies with inadequate preparation. The primary outcome of the study was ADR.
Results
Our study included 996 colonoscopies (237 pre-AIAC, 759 AIAC). The groups were similar in age (median 43.8 years, interquartile ratio (IQR) 28.7-61.2 vs 44.5 years, IQR 30.7-59.1, p=0.76) and gender distribution (55.3% vs 54.3% males, p=0.82). In both groups, there were more Crohn’s disease (62.9% pre-AIAC, 57.2% AIAC, p=0.13) than ulcerative colitis patients. The ADR in the pre-AIAC group tended to be higher in comparison to AIAC (6.3% vs 4%, p=0.15); When limiting to experienced gastroenterologists only (≥ 5 years), ADR results were significant higher in the pre-AIAC group (7.6% vs 3.8%, p=0.035). In addition, total procedure time (considered the time from procedure start and end as recorded by the endoscopy nurse) was significantly shorter in the AIAC group (21 minutes, IQR 17-28 vs 25 minutes, IQR 19-37, p<0.0001).
Conclusion
In a large-volume tertiary center cohort, the introduction of AIAC to real-world colonoscopies did not improve ADR in patients with IBD, questioning the integration of AIAC in routine practice.