Endoscopic tri-modal imaging for surveillance in ulcerative colitis: randomised comparison of high-resolution endoscopy and autofluorescence imaging for neoplasia detection; and evaluation of narrow-band imaging for classification of lesions

Abstract

Background: Endoscopic tri-modal imaging (ETMI) incorporates white light endoscopy (WLE), autofluorescence imaging (AFI) and narrow-band imaging (NBI).

Aims: To assess the value of ETMI for the detection and classification of neoplasia in patients with longstanding ulcerative colitis.

Design: Randomised comparative trial of tandem colonoscopies.

Setting: Academic Medical Centre Amsterdam, Netherlands.

Patients and methods: Fifty patients with ulcerative colitis underwent surveillance colonoscopy with ETMI. Each colonic segment was inspected twice, once with AFI and once with WLE, in random order. All detected lesions were inspected by NBI for Kudo pit pattern analysis and additional random biopsies were taken.

Main outcome measures: Neoplasia miss-rates of AFI and WLE, and accuracy of the Kudo classification by NBI.

Results: Among patients assigned to inspection with AFI first (n = 25), 10 neoplastic lesions were primarily detected. Subsequent WLE detected no additional neoplasia. Among patients examined with WLE first (n = 25), three neoplastic lesions were detected; subsequent inspection with AFI added three neoplastic lesions. Neoplasia miss-rates for AFI and WLE were 0% and 50% (p = 0.036). The Kudo classification by NBI had a sensitivity and specificity of 75% and 81%; however, all neoplasia was coloured purple on AFI (sensitivity 100%). No additional patients with neoplasia were detected by random biopsies.

Conclusion: Autofluorescence imaging improves the detection of neoplasia in patients with ulcerative colitis and decreases the yield of random biopsies. Pit pattern analysis by NBI has a moderate accuracy for the prediction of histology, whereas AFI colour appears valuable in excluding the presence of neoplasia.

Trial registration number: ISRCTN05272746.

Figure 1. (A) Images during high-resolution white-light endoscopy (WLE) (a); autofluorescence imaging (AFI) (b); and narrow-band imaging (NBI) (c), of mucosa with (d) no significant changes on histology. On AFI normal mucosa appears green; NBI shows a normal pit pattern (Kudo type I). (B). Images during WLE (a), AFI (b), and NBI (c), of a lesion revealing hyperplastic-like mucosal changes on histopathology (d). Tissue autofluorescence is disturbed leading to a purple (false positive) colour on AFI; during NBI a normal pit pattern is seen. (C) Images during WLE (a), AFI (b), and NBI (c), of an area showing inflammation on histopathology (d). On AFI, inflammation becomes purple (false positive), drawing the attention of the endoscopist. On NBI, an irregular pit pattern is seen, partly with elongated pits (Kudo type IIIL). (D) Images during WLE (a), AFI (b), and NBI (c), of a mass revealing low-grade intraepithelial neoplasia on histopathology (d). The neoplastic lesion appears deep purple on AFI and reveals Kudo pit pattern type IV on NBI.

Figure 2. Study design and flow chart of patients who gave informed consent during the study. The number of detected neoplastic lesions and number of patients with neoplasia are summarised per randomisation group. During white-light endoscopy (WLE), three neoplastic lesions (50%) were missed which were detected by autofluorescence imaging (AFI). Two random biopsies showed neoplasia after inspection with AFI and subsequent WLE; these were found in a patient in whom AFI already detected three areas of flat neoplasia.