Role of Autofluorescence Endoscopy in Surveillance of Barrett's Esophagus: A Prospective Multicenter Randomized Trial

Abstract

Background and Aims: The current reference method surveillance method for Barrett's esophagus (BE) is four quadrant biopsies (4-QB). Autofluorescence endoscopy (AF-E) claims the promise of visualization of otherwise unrecognized dysplasias. Hitherto, findings from small studies have been conflicting. We conducted a multicenter randomized controlled trial to elucidate whether AF-E in addition to the gold-standard improves diagnostic workups. Methods: We enrolled 200 patients with BE (73% male; mean age 67 yrs, BE mean segment length: 4.6 cm) from four tertiary referral centers. Of these, 16 had known dysplasia. The primary endpoints were detection/localization of early cancer (EC) and high-grade dysplasia (HGD). Patients were randomly allocated to either AF-E or white light endoscopy (WL-E) and a 3-month follow up using the reverse method. Examiners were blinded for the histology of the first endoscopy. A second endoscopy was obtained in 125 patients (25 requiring mucosectomy or surgery, 34 declining a second look, and 16 pending). For AF-E we used a Xenon lamp, a short-pass filter and a video camera attached to a fiber gastroscope with an observation long-pass filter (Storz, Germany). WL-E was done using video endoscopes. Visually suspicious sites were biopsied, unsuspicious sites were sampled according to the 4-QB protocol every 2 cm. Results: In 24 (12%) patients we found 45 pathologic lesions (12 EC and 33 HGD) and in 15 (7.5%) low-grade dysplasia (LGD). Ten EC/HGD lesions in 9 patients were only visualized by AF-E, 6 lesions by both methods and 5 (erosions, nodules) were seen with WL-E only. Unguided 4-quadrant biopsies from visually unsuspicious sites revealed further 25 sites of dysplastic lesions. The 325 endoscopies gave rise to 1012 biopsy sites (2 to 4 biopsies per site), of which 569 were obtained during AF-E. Of the 114 sites appearing suspicious under AF-E, 6 (5%) were EC, 7 (6%) yielded HGD, 12 (11%) showed LGD and 89 (78%) had no histological evidence of dysplasia (sensitivity of 21%). Specificity was 81%. The positive likelihood ratio for a suspicious area under AF being HGD or EC was 2.22. Conclusions: In this referral population, AF-E slightly enhanced visualization and localization of neoplastic lesions in Barrett's esophagus in addition to standard 4-QB endoscopy. For every 21 patients undergoing AF-E one additional neoplastic lesion was detected. AF-E appeared unsuited to replace standard 4-QB from suspicious BE.