Narrow Band Imaging Does Not Improve the Detection Rate of Neoplasia Among Patients with Longstanding Ulcerative Colitis When Compared to High Definition White Light Endoscopy

Abstract

Background: Methylene blue chromoendoscopy improves the detection of neoplasia in patients with longstanding ulcerative colitis (UC). However, this procedure is labor-intensive and operator dependent. Narrow band imaging (NBI) is a push-on-a-button imaging technique using optical filters that resembles chromoendoscopy. This randomized cross-over study compares NBI with high definition white light endoscopy (HD-WLE) for the detection of neoplasia in patients with UC. Patients and Methods: 49 patients with longstanding pancolitis have been planned to undergo 2 colonoscopies, once with NBI and once with HD-WLE; powered to demonstrate a 3-fold increase of neoplasia by NBI. The order of colonoscopies is randomized, and the time between the two procedures is set at 4-8 weeks. The examinations are performed by different endoscopists, blinded for each others findings. During each procedure, the Kudo pit pattern and vascular pattern of detected lesions are assessed with NBI before taking targeted biopsies. Random biopsies are taken during the second examination only. Outcomes: Neoplasia detected by NBI vs. HD-WLE (paired analysis); and accuracy of Kudo pit pattern and vascular pattern by NBI using histology as reference standard. Results: Thus far, 41 patients have given informed consent; 32 patients have completed both examinations (24 male, mean age 56 yrs), of whom 17 were randomized to HD-WLE first. Overall, in 11 patients neoplasia was detected by either NBI or HD-WLE. On a per-patient analysis, NBI detected 6 patients with neoplasia (55%; 8 lesions) vs. 11 patients (100%; 12 lesions) during HD-WLE (p=0.062). On a per-lesion analysis, 5 DALMs were detected (1 by NBI only) and 7 adenoma-like masses (5 by HD-WLE only). Two out of 1019 random biopsies contained neoplasia (0.2%), taken in 2 patients with DALMs that were already detected by NBI and HD-WLE. The Kudo pit pattern had a sensitivity, specificity and accuracy of 64%, 64% and 64% for differentiating neoplasia from non-neoplastic mucosa. Corresponding figures for vascular pattern were 77%, 68% and 69%. Conclusions: The interim results of this randomized cross-over trial in patient with longstanding UC suggest that HD-WLE has a higher detection rate of UC-associated neoplasia compared to NBI, although final results should be awaited. When using NBI or HD-WLE for surveillance, random biopsies do not detect additional neoplasia and therefore their use may be questioned. For differentiating neoplastic from non-neoplastic mucosa during surveillance of UC patients, the accuracy of the Kudo pit pattern and vascular pattern by NBI are disappointing.