Prospective Observational Study On Optical Detection Of Residual Colorectal Neoplasia After Endoscopic Mucosal Resection: Impact Of Endoclip Closure Of Emr Site

Application of clips after EMR may reduce post EMR bleeding but may also hamper the detection of recurrent adenoma at follow up due to nodular distortions of scar tissue. This could lead to an overtreatment/or snare resection of benign lesions. In this prospective trial, we aimed to evaluate the diagnostic accuracy of white light and NBI with near-focus in recurrent adenoma detection with or without clip artifacts. In this prospective trial consecutive patients undergoing follow-up colonoscopy after a previous EMR of a ≥20mm colorectal neoplasia were included. Endoscopists predicted the presence of recurrence on EMR scars with or without clip artifact and their level of confidence (high or low) based on visual assessment of the previous EMR site subsequently with the following four modalities 1) white light without and 2) with near-focus, and 3) NBI without and 4) with near-focus. Each EMR-site was biopsied as a reference standard, and histology was assessed by a pathologist blinded to the endoscopists’ visual assessment. Diagnostic accuracy was calculated for all diagnostic modalities for the real-time endoscopic assessment of the EMR scar tissue with or without clip artifacts. A total of, 230 patients (51% male, mean age: 65 years [± SD]) with 254 scar sites were included, with a median follow up of 7 (range: 1-70) months. At the original EMR, clips were used in 111/254 (44%) lesions. The rate of post-EMR scar clip artifact was 53% (59 of 111). Diagnostic accuracy for all EMR-scar sites with or without clip- artifact and confidence level in real time are summarized Table 1 and 2. The prevalence of residual neoplasia was 24% for all EMR scars. 21.6% of cases were over treated (snare resection) for benign disease when clips were used versus 7.6% of cases when clips were not used. Overall, clips reduced the proportion of high confidence diagnosis, and appear to lower the accuracy of scar assessment in low confidence diagnoses but not high confidence. These results suggest that clips used to close EMR defects may impair the ability to accurately assess the EMR scar at follow up. Clips reduced the proportion of high confidence diagnoses as well as the accuracy of scar assessment in low confidence diagnoses, and furthermore, increased the proportion of over treatment of benign clip artifact. Alternative methods to potentially reduce post-EMR bleeding are needed.Tabled 1Diagnostic value of optical detection of residual neoplasia with or without clip artifactImaging ModesDiagnostic accuracy with clip artifact, (n=59)Diagnostic accuracy without clip artifact, (n=195)NBI near focusHigh confidence95% [88-98]97% [91-99]Low confidence76% [66-83]80% [71-86]Overall90% [82-94]95% [88-98]White lightHigh confidence95% [88-98]97% [91-99]Low confidence75% [65-82]82% [73-88]Overall82% [73-88]95% [88-98]N, number of EMR-scar sites; NBI, narrow-band imaging; [.-.], 95%-confidence interval. Open table in a new tab Tabled 1Proportion of confidence level on optical detection of residual neoplasia with or without clip artifactImaging ModesLevel of confidence with clip artifact, (n=59)Level of confidence without clip artifact, (n=195)p-valueNBI near focus( high confidence diagnosis)83% (49)96% (181)0.01*White light (high confidence diagnosis)66% (39)88% (173)0.01*Proportion of overtreatment of benign disease with or without clip.Original EMR with clip use (n=111)Original EMR without clip use (n=143)p-valueOvertreatment21.6%[24]7.6% [11]0.01*N, number of EMR-scar sites; NBI, narrow-band imaging; [.-.], 95%-confidence interval, * chi squire test Open table in a new tab