Studies show mixed results for artificial intelligence during colonoscopies

Two studies compared outcomes with the use of artificial intelligence, or computer-aided detection (CAD), during colonoscopy.

First, to quantify the benefits and harms of CAD, researchers conducted a systematic review and meta-analysis of randomized trials through February 2023 comparing CAD-assisted colonoscopy with standard colonoscopy for polyp and cancer detection. Adenoma detection rate, number of adenomas, advanced adenomas (≥10 mm with high-grade dysplasia and villous histology), number of serrated lesions, and adenoma miss rate were extracted as benefit outcomes. Number of polypectomies for nonneoplastic lesions and withdrawal time were extracted as harm outcomes. Twenty-one trials with 18,232 patients were included. The results were published Aug. 29 by Annals of Internal Medicine.

The adenoma detection rate was higher in the CAD group than in the standard colonoscopy group (44.0% vs. 35.9%; relative risk, 1.24 [95% CI, 1.16 to 1.33]; low-certainty evidence), corresponding to a 55% relative reduction in miss rate (risk ratio, 0.45 [95% CI, 0.35 to 0.58]). More nonneoplastic polyps were removed in the CAD group than in the standard group (0.52 vs. 0.34 per colonoscopy; mean difference, 0.18 polypectomy [95% CI, 0.11 to 0.26]; low-certainty evidence). Mean inspection time increased only marginally with CAD (mean difference, 0.47 min [95% CI, 0.23 to 0.72 min]; moderate-certainty evidence).

The second study, also published Aug. 29 by Annals of Internal Medicine, was a multicenter, parallel, randomized controlled trial of 3,213 Spanish patients with a positive fecal immunochemical test who were randomized to colonoscopy with or without CAD. The two comparison groups showed no significant difference in advanced colorectal neoplasia detection rate (34.8% with intervention vs. 34.6% for controls; adjusted risk ratio [aRR], 1.01 [95% CI, 0.92 to 1.10]) or mean number of advanced colorectal neoplasias detected per colonoscopy (0.54 vs. 0.52; adjusted rate ratio, 1.04 [99.9% CI, 0.88 to 1.22]). Adenoma detection rate also did not differ (64.2% vs. 62.0%; aRR, 1.06 [99.9% CI, 0.91 to 1.23]). CAD increased the mean number of nonpolypoid lesions (0.56 vs. 0.47; aRR, 1.19 [99.9% CI, 1.01 to 1.41]), proximal adenomas (0.94 vs. 0.81; aRR, 1.17 [99.9% CI, 1.03 to 1.33]), and lesions of 5 mm or smaller (polyps in general and adenomas and serrated lesions in particular) detected per colonoscopy.

An editorial called the results of the systematic review “promising” for CAD, or CADe, systems but noted that both studies suggest that such systems do not meaningfully improve detection of larger clinically significant polyps, defined as those 10 mm or larger. “This tempers enthusiasm for CADe but does not negate the clear performance benefit for detecting adenomas of all sizes,” the editorialist wrote. “Overall improved [adenoma detection rate] with widespread CADe system uptake could lead to lengthening of recommended surveillance intervals.”