University of Kansas
School of Medicine
Kansas City
The focus of this month’s Sharma’s Endoscopy Insights is improving the quality of colonoscopy. We’ll discuss three studies: one that examines adenoma detection with artificial intelligence, another that evaluates rates of repeat colonoscopy and a third that explores methods to measure polyp size more accurately.
The first study is a multicenter trial from Italy that investigated the role of AI, specifically computer-aided diagnosis, in evaluating the adenoma detection rate (ADR) and mean adenomas per colonoscopy (APC) in fecal immunochemical test (FIT)-positive patients undergoing colonoscopy. Both ADR and APC were significantly higher in colonoscopies performed with AI.
Second, a multicenter study from Europe showed there were considerable variations between endoscopy centers and endoscopists in whether they needed to perform a second or even a third colonoscopy before actual surveillance started. The main reasons for a second procedure were poor bowel preparation, incomplete colonoscopy or incomplete polypectomy.
At the time of the initial colonoscopy, we should endeavor to perform a high-quality procedure, so that in the presence of good bowel preparation, a good mucosal inspection of the colonic mucosa can be conducted and all polyps removed.
Finally, a Canadian study tested a new tool to improve the accuracy of polyp size estimates. Accuracy here is important because the estimated size of the polyps determines the methodology of polyp removal and future surveillance programs.
In this study, the use of a laser-based virtual scale endoscope (VSE) was tested during live endoscopy and found to have high relative accuracy for polyp size measurement compared with visual size estimation. Such tools will help us more precisely estimate polyp size.
Computer-Aided Detection of Adenomas
(Endoscopy 2022;54[12]:1171-1179)
In this multicenter, single-blind, randomized trial, researchers assessed whether using computer-aided detection (CADe) increases the ADR in a FIT-based colorectal cancer screening program.
The study was conducted in five open-access endoscopy centers in Italy. Endoscopists examined patients with high-definition white light (HDWL) colonoscopy with or without an additional real-time deep-learning CADe system for identifying polyps (CAD EYE, Fujifilm).
Of the 800 patients included in the study, 405 underwent CADe-assisted colonoscopy and 395 underwent HDWL colonoscopy alone.
The researchers found that both ADR and APC were significantly higher for patients examined with CADe compared with those who were examined only with HDWL. The ADR for the CADe group was 53.6% (95% CI, 48.6%-58.5%) compared with 45.3% (95% CI, 40.3%-50.45%) for the HDWL group. The APC was 1.13 (SD, 1.54) and 0.90 (SD, 1.32) (P=0.03) for the CADe and HDWL groups, respectively.
Meanwhile, there was no significant increase in advanced ADR. However, researchers noted that the study was limited by a lack of blinding and no preevaluation of the endoscopists.
The authors concluded that using CADe could potentially lead to an absolute increase in ADR that could then lead to a reduction of interval cancers in FIT-positive individuals.
Colonoscopy Repeat
(Gut 2022 Oct 28. doi:10.1136/gutjnl-2022-327696)
In Norway, researchers investigated variation in the number of colonoscopies needed before patients with low- and high-risk adenomas were enrolled in polyp surveillance.
The study included data from 15,581 patients enrolled in European Polyp Surveillance, a large, multinational trial that encompasses three clinical trials performed in eight European countries. The study included data from 38 endoscopy centers in five European countries.
Data were collected from April 2015 to March 2020 on patients aged 40 to 74 years who had colorectal polyps removed at participating endoscopy centers. Overall, the researchers reviewed data for 6,794 patients with low-risk adenomas and 8,797 with high-risk adenomas.
Before starting polyp surveillance, 961 patients underwent two or more colonoscopies (6.2%; 95% CI, 5.8%-6.6%), with 101 patients in the low-risk group (1.5%; 95% CI, 1.2%-1.8%) and 860 in the high-risk group (9.8%; 95% CI, 9.2%-10.4%) undergoing two or more procedures.
The main reasons for a second colonoscopy were incomplete colonoscopy or polypectomy (14.4%), poor bowel preparation (21.3%), or a planned second procedure (27.8%). On a second colonoscopy, the most common reasons for a third procedure was a piecemeal resection (26.5%) or reason not specified (23.9%).
The researchers concluded that their findings indicate the need for targeted quality improvement efforts to reduce the number of additional colonoscopies when possible.
Virtual Polyp Measurements
(Gut 2022 Nov 21. doi:10.1136/gutjnl-2022-328654)
Researchers in Canada launched a clinical pilot study to measure colorectal polyp size during live colonoscopies using a new VSE with a laser-based size measurement function.
The VSE (Scale Eye, Fujifilm) superimposes a virtual linear or circular ruler onto objects in the endoscopist’s field of view.
In this pilot study, polyps were first measured visually without any additional measurement tools to reflect common practice. Then the measurement was repeated using VSE. Finally, the physician resected and removed the polyps from the colon and the specimens were measured.
The analysis was performed on 36 of 72 total resected polyps up to 20 mm from 59 patients. Two endoscopists performed the procedures, a trainee and staff gastroenterologist, at the Centre Hospitalier de l’Université de Montréal.
In the results, the researchers found the relative accuracy of polyp size measurement was higher with VSE compared with visual size estimation alone (85.4% and 66.85%, respectively) (P<0.001). The findings also indicated that 33.3% of the measurements made by visual estimation and 86.1% of the measurements made by VSE were within 25% of the true polyp size (P<0.001).
The study was small, with a limited number of large polyps in the study sample. However, the findings suggest that relying on endoscopists’ subjective visual estimation of polyp size in routine practice may not be optimal.
—Compiled and written by Jillian Mock
Dr. Sharma is a member of the Gastroenterology & Endoscopy News editorial board.
This article is from the February 2023 print issue.