AI-assisted colonoscopy does not improve detection of advanced neoplasias
While colonoscopies with artificial intelligence (AI) assistance yielded more adenomas compared with conventional colonoscopy, rates of advanced neoplasias per colonoscopy did not differ, a systematic review and meta-analysis found.
Colonoscopy with artificial intelligence assistance found more adenomas than traditional colonoscopy but did not appear to detect more advanced neoplasias, a recent study found.
Researchers performed a systematic review and meta-analysis of randomized clinical trials that compared computer-aided detection-enhanced (CADe) colonoscopy with conventional colonoscopy and were published through February 2024. The average number of adenomas per colonoscopy (APC) and advanced colorectal neoplasias (ACNs) per colonoscopy were the primary outcomes, and adenoma detection rate, adenoma miss rate (AMR), and advanced colorectal neoplasia (ACN) detection rate were the secondary outcomes. The results were published Oct. 22 by Annals of Internal Medicine.
A total of 44 randomized clinical trials involving 36,201 cases were included in the analysis. CADe colonoscopies were found to have higher average APC (0.98 vs. 0.78; incidence rate difference [IRD], 0.22 [95% CI, 0.16 to 0.28]) and ADR (44.7% vs. 36.7%; rate ratio [RR], 1.21 [95% CI, 1.15 to 1.28]) than conventional colonoscopy. While average ACN per colonoscopy was similar between the two methods (0.16 vs. 0.15; IRD, 0.01 [95% CI, −0.01 to 0.02]), the ACN detection rate was higher with CADe colonoscopy (12.7% vs. 11.5%; RR, 1.16 [95% CI, 1.02 to 1.32]). CADe colonoscopy led to resection of almost two additional nonneoplastic polyps per 10 colonoscopies and entailed a longer total withdrawal time.
The researchers noted that endoscopists could not be blinded to the interventions in the included studies and that there was statistically significant heterogeneity in study quality and sample size, among other limitations. They concluded that CADe colonoscopies have increased APC and detection rate compared with conventional colonoscopy, do not differ in ACN per colonoscopy, and result in a small increase in ACN detection rate and an increase in procedure time. “Future studies with focus on more important clinical outcomes, such as interval postcolonoscopy [colorectal cancer], and designed with randomization of physicians rather than patients may be considered to better evaluate the effect of these systems,” the authors wrote.