Narrow-band imaging for the diagnosis of nonerosive reflux disease: an international, multicenter, randomized controlled trial.

ABSTRACT

BACKGROUND AND AIMS: We examined the accuracy of narrow-band imaging (NBI) findings in nonerosive reflux disease (NERD) patients compared with control subjects and the impact of proton pump inhibitor (PPI) therapy on these mucosal changes in a multicenter, double-blind, randomized controlled trial. METHODS: NERD patients (typical symptoms using a validated GERD questionnaire, absence of erosive esophagitis, and abnormal 48-hour pH study) and control subjects underwent high-definition white-light endoscopy followed by NBI and biopsy sampling of the distal esophagus. Then, NERD patients were randomized to esomeprazole 40 mg/day or placebo for 8 weeks, followed by repeat endoscopy. The presence of distal esophageal mucosal changes on NBI were recorded at baseline and after treatment intrapapillary capillary loops (IPCLs; number, dilation, and tortuosity), microerosions, increased vascularity, columnar islands, and ridge/villous pattern (RVP) above the squamocolumnar junction. RESULTS: Of 122 screened, 21 NERD and 21 control subjects were identified (mean age, 49.5 ± 14.6 years; 62% men; and 85% white). The combination of IPCL tortuosity, RVP, and microerosions (62% vs 19%, P < .05) had a high specificity (86%) and moderate sensitivity (60%) for NERD with an area under the curve of .74. In 10 NERD patients treated with PPIs, resolution of microerosions was most significant (P = .047) compared with placebo (n = 11). RVP resolved in all NERD patients after therapy (P = .02) and correlated with acid exposure time (P = .004). Papillary length (P = .02) and basal cell thickness (P = .02) significantly correlated with a combination of IPCL tortuosity, RVP, and microerosions. CONCLUSIONS: In this randomized controlled trial, RVP on NBI demonstrated a high specificity, correlated with acid exposure time, and improved with PPI therapy, suggesting that it could be used as a surrogate marker for diagnosis of NERD. (Clinical trial registration number NCT02081404.).