Can white-light endoscopy or narrow-band imaging avoid biopsy of colorectal endoscopic mucosal resection scars? A multicenter randomized single-blind crossover trial.

BACKGROUND : Current guidelines suggest that routine biopsy of post-endoscopic mucosal resection (EMR) scars can be abandoned, provided that a standardized imaging protocol with virtual chromoendoscopy is used. However, few studies have examined the accuracy of advanced endoscopic imaging, such as narrow-band imaging (NBI) vs. white-light endoscopy (WLE) for prediction of histological recurrence. We aimed to assess whether NBI accuracy is superior to that of WLE and whether one or both techniques can replace biopsies. METHODS : The study was a multicenter, randomized, pathologist-blind, crossover trial, with consecutive patients undergoing first colonoscopy after EMR of lesions ≥ 20 mm. Computer-generated randomization and opaque envelope concealed allocation. Patients were randomly assigned to scar examination with NBI followed by WLE (NBI + WLE), or WLE followed by NBI (WLE + NBI). Histology was the reference method, with biopsies being performed for all tissues. RESULTS : The study included 203 scars (103 in the NBI + WLE group, 100 in the WLE + NBI group). Recurrence was confirmed histologically in 29.6 % of the scars. The diagnostic accuracy of NBI was not statistically different from that of WLE (95 % [95 %CI 92 %-98 %] vs. 94 % [95 %CI 90 %-97 %]; P = 0.48). The negative predictive values (NPVs) were 96 % (95 %CI 93 %-99 %) for NBI and 93 % (95 %CI 89 %-97 %) for WLE (P = 0.06). CONCLUSIONS : The accuracy of NBI for the diagnosis of recurrence was not superior to that of WLE. Endoscopic assessment of EMR scars with WLE and NBI achieved an NPV that would allow routine biopsy to be avoided in cases of negative optical diagnosis.