AI fails to improve detection of advanced neoplasms with colonoscopy
Colonoscopy is the most common procedure performed by Gastroenterologists and is critical for early detection and management of precursors to colorectal cancer (CRC). The screening of CRC involving Fecal immunochemical test (FIT) followed by diagnostic-therapeutic colonoscopy, as indicated, has improved the detection of at-risk premalignant neoplasms, thereby reducing CRC mortality. A key determinant of post-colonoscopy CRC occurrence is the detection rate of high-grade lesions during the procedure. Artificial intelligence has shown promise in improving the adenoma detection rate, but there needs to be more clarity on whether such improvements arise from detecting low-grade or high-grade lesions.
A study has recently demonstrated that Computer-aided detection did not improve colonoscopic identification of advanced colorectal neoplasias. This study published in Annals of Internal Medicine investigated the impact of Medtronic’s GI-Genius AI on detecting advanced colorectal neoplasia among Spanish patients who had a first positive FIT.
It is already known that data are scarce on the role of computer-aided detection in identifying advanced colorectal neoplasia. Considering this background, researchers evaluated the contribution of computer-aided detection to colonoscopic detection of advanced colorectal neoplasias and adenomas, serrated polyps, and nonpolypoid and right-sided lesions in the Multicenter, parallel, randomized controlled trial.
There were 3213 persons with a positive faecal immunochemical test. Researchers randomly assigned enrolled participants to colonoscopy with or without computer-aided detection. Advanced colorectal neoplasia was defined as advanced adenoma and advanced serrated polyp.
Key findings of this study are:
- The two groups had no significant difference in advanced colorectal neoplasia detection rate( 34.8 % for intervention and 34.6 % for control).
- The mean number of advanced colorectal neoplasias detected per colonoscopy with intervention and control was 0.54 and 0.52, respectively [aRR:1.04]
- There was no difference in the Adenoma detection rate for intervention and control, with 64.2% and 62.0%, respectively [aRR, 1.06].
- Computer-aided detection increased the mean number of nonpolypoid lesions, proximal adenomas, and lesions of 5 mm or smaller (polyps in general and adenomas and serrated lesions in particular) detected per colonoscopy.
Researchers said, “In our study, the high adenoma detection rate in the control group may limit the generalizability of the findings to endoscopists with low detection rates.”
They concluded that computer-aided detection did not improve colonoscopic identification of advanced colorectal neoplasias.
Medtronic funded the study, as acknowledged.
Reference:
Mangas-Sanjuan C, de-Castro L, Cubiella J, Díez-Redondo P, Suárez A, Pellisé M, Fernández N, Zarraquiños S, Núñez-Rodríguez H, Álvarez-García V, Ortiz O, Sala-Miquel N, Zapater P, Jover R; CADILLAC study investigators. Role of Artificial Intelligence in Colonoscopy Detection of Advanced Neoplasias : A Randomized Trial. Ann Intern Med. 2023 Sep;176(9):1145-1152. doi: 10.7326/M22-2619.
