Artificial Intelligence for Leaving in Situ Colorectal Polyps: results from a Randomised Trial

Aims Artificial Intelligence (AI)-powered optical-diagnosis systems (CADx) hold the promise to permit real-time pathology prediction during colonoscopy and consequently reduce the many burdens of unnecessary polypectomies. However, no randomised trial exists to date on cost saving strategies based on optical diagnosis and on the potential impact of CADx on them. We aimed to evaluate the safety and effectiveness of a Leave-in-Situ strategy when using a CADx tool in a multicenter RCT, using ADR, the most important and validated metric in colonoscopy, as primary endpoint.

Methods Consecutive 18+years old subjects undergoing colonoscopy for CRC primary screening, surveillance or FIT+were enrolled in four centres. Colonoscopies were assisted by a regulatory approved CADe/CADx module and were performed by experienced operators. Patients were randomised to two different strategies: “Leave-In-Situ”, where diminutive (< 5mm) rectosigmoid polyps optically diagnosed as hyperplastic were left in place, or “Resect-All”, where all detected diminutive polyps were resected regardless of their optical diagnosis. Primary endpoint was non-inferiority of Adenoma Detection Rate (ADR). Secondary endpoints included mean number of polypectomies per patient (MPP), mean number of adenomas per patient (APC), True Histology Rate (THR, the number of polypectomies with histologically confirmed adenoma or CRC in any colon location divided by the total number of polypectomies) and withdrawal time (WT). Here, results based on per-protocol analysis (excluding poor bowel preparation) are reported.

Results Overall, 847 patients (480, 56.7% female, mean age, 60.9 years) were included in the final analysis, 417 and 430 randomised to Leave-In-Situ or Resect-All colonoscopy, respectively. ADR in the Leave-In-Situ arm (194/417, 46.5%) was non-inferior to the Resect-All arm (206/430, 47.9%): p for non-inferiority: 0.006, 95% CI for ADR difference: -1.4% (95%CI:-8.1-5.3%; RR, 0.98; 95%CI:0.85-1.12). Non-inferiority was maintained when results were stratified by colonic location, polyp size, morphology, multiplicity and histology. The MPP was lower in the Leave-In-Situ than Resect-All arm: 1.48 vs. 2.98; Incidence Rate Ratio (IRR): 0.50; 95% CI:0.42-0.59, while no differences were seen in APC: Leave-In-Situ vs. Resect-All arm, 1.01 vs. 0.96; IRR, 1.06; 95% CI:0.86 to 1.30. Per-polyp THR were 68.7% (423/616) and 32.2% (413/1283) in the Leave-In-Situ vs. Resect-All arm: ORs, 4.6; 95% CI:3.38 to 6.32. No difference in WT and other safety outcomes was observed [1] [2] [3].

Conclusions Our study is the first RCT performed on cost saving strategies based on optical diagnosis in colonoscopy. We show data supporting the safety and effectiveness of a Leave-In-Situ strategy when using real time CADe/CADx assistance, by showing a reduction of approximately 50% of unnecessary polypectomies without a loss in terms of adenoma detection rate. The use of CADx may increase cost-effectiveness of screening colonoscopy without jeopardizing safety, facilitating the standardization of the procedure.

Publikationsverlauf

Artikel online veröffentlicht:
27. März 2025

• References

• 1 Hassan C, Misawa M, Rizkala T. et al. Computer-Aided Diagnosis for Leaving Colorectal Polyps In Situ : A Systematic Review and Meta-analysis. Ann Intern Med 2024; 177 (07): 919-928
• 2 Hassan C, Balsamo G, Lorenzetti R, Zullo A, Antonelli G.. Artificial Intelligence Allows Leaving-In-Situ Colorectal Polyps. Clin Gastroenterol Hepatol 2022; 20 (11): 2505-2513 e4
• 3 Messmann H, Bisschops R, Antonelli G. et al. Expected value of artificial intelligence in gastrointestinal endoscopy: European Society of Gastrointestinal Endoscopy (ESGE) Position Statement. Endoscopy 2022; 54 (12): 1211-1231