DOP69 The detection with targeted biopsy and characterisation of neoplastic lesions by magnifying chromoendoscopy and NBI in surveillance colonoscopy of patients with ulcerative colitis: a sub-analysis of the Navigator Study

We recently reported the UC surveillance colonoscopy (SC) pancolonic NBI observation was not inferior to panchromoendoscopy (PCE) for the detection of both neoplastic lesions (13.4% vs. 9.0%) and colitis-associated dysplasia or cancer (CC/D) (6.3% vs. 4.9%). Moreover, the total examination time with NBI (15.0 min) was significantly shorter than that with PCE (19.8 min) (p < 0.01). This sub-analysis aimed to evaluate the magnified chromoendoscopic and NBI findings of detected lesions in the preceding prospective multi-centre randomised controlled trial (Navigator Study), and to also investigate the utilities of existing endoscopic classification to characterise the neoplastic lesions. In total, 263 patients were randomised to either the PCE group (n = 130) or the NBI group (n = 133). SC in both groups was performed mainly with targeted biopsy. After either procedure detected a suspected neoplastic lesion, both procedures (Kudo’s classification by chromoendoscopy and J-NET classification by NBI) were allowed to characterise the lesion with magnified observation. The central pathological diagnoses were made by two expert pathologists based on examinations including immunohistochemical staining (p53, Ki-67). In total, 20 sporadic adenomas, 10 indefinite dysplasia lesions, 9 low-grade dysplasia (LDG) lesions, 7 high-grade dysplasia (HGD), or cancer lesions were detected. Only 2 indefinite dysplasia lesions were identified by random biopsy, the others (96.1%) were identified by targeted biopsy. There were no significant differences in locations between sporadic adenoma and CC/D (p = 0.12). Most sporadic adenomas showed the type 3L pit pattern in contrast to CC/D, which were distributed from the 3L to the 5 irregular pit pattern (p < 0.01). However, Kudo’s pit pattern was not useful for differentiating between LGD and HGD/cancer (p = 0.62). Most detected lesions (40/46, 87.0%) were brownish on NBI. The J-NET surface pattern and vessel pattern were also both useful for differentiating between sporadic adenoma and CC/D (p = 0.05, p = 0.02). Most sporadic adenomas showed type 2A of J-NET surface pattern or vessel pattern in contrast to CC/D, which were divided into type 2A and type 2B (p = 0.05, p = 0.02). However, J-NET surface pattern or vessel pattern also was not useful for differentiating between LGD and HGD/cancer (p = 0.52, p = 0.76). Identification of neoplastic lesions by SC that depends on targeted biopsy is a reasonable approach. Brownish colouring on NBI observation may provide improved detectability in pancolonic NBI SC. The existing endoscopic classifications has limitations in terms of characterising CC/D. A newly developed endoscopic classification to judge the indications for endoscopic resection is thus warranted.