Third‐Generation Narrow‐Band Imaging Versus White‐Light Imaging for the Detection of Early Gastric Cancer: A Randomized Controlled Study

Objectives: The endoscopic system EVIS X1 with improved image quality has been introduced into clinical practice. We examined whether third-generation narrow-band imaging (3G-NBI) is more effective than white-light imaging (WLI) for detecting early gastric cancer (EGC). Methods: Our study, performed at a single center, had a parallel-group, open-label, two-arm, randomized, controlled design. Patients who had undergone endoscopic submucosal dissection for EGC were randomly assigned to a group undergoing 3G-NBI after initial WLI (initial WLI group) or a group undergoing WLI after initial 3G-NBI (initial 3G-NBI group). The primary endpoint was the EGC detection rate of the two methods. The secondary endpoints were as follows: proportions of EGC detected and missed lesions, positive predictive value (PPV) for EGC diagnosis, and observation time for WLI and 3G-NBI. Results: The EGC detection rate was 9.0% (17/188) in the initial WLI group and 8.5% (16/188) in the initial 3G-NBI group. The missed lesion rate was 5.6% (1/18) in the initial WLI group and 0% (0/18) in the initial 3G-NBI group. The PPV of the initial WLI group was 42.5% (17/40), whereas that of the secondary 3G-NBI was 25% (1/4). The PPV of the initial 3G-NBI group was 30.2% (16/53). No biopsies were performed during secondary WLI. The examination times were 274 ± 78.2 and 280 ± 82.9 s for WLI and 3G-NBI, respectively. Conclusions: 3G-NBI was not superior to WLI in detecting EGC. This finding is likely due to improved WLI image quality.