External validation of a classification for methylene blue magnification chromoendoscopy in premalignant gastric lesions

Background Conventional endoscopy has low sensitivity, specificity, and interobserver agreement for the diagnosis of gastric atrophy, intestinal metaplasia, and dysplasia. Magnification chromoendoscopy (ME) may optimize the evaluation of premalignant gastric lesions. Objective and Design As part of a multicenter trial, we aimed at validating a previously proposed classification for gastric methylene blue ME at a different center. Setting, Patients, and Interventions A sample of patients (n = 42) with previously diagnosed chronic atrophic gastritis with or without intestinal metaplasia underwent ME (Pentax EG-3430Z) with 1% methylene blue by 2 endoscopists. Main Outcome Measurements A simplified version of a previously published ME classification (group I, group II [further divided into subgroups IIE and IIF], and group III) was used for macroscopic lesions (n = 203) with Sydney-Houston and Vienna classifications being used for histologic analysis (n = 479 biopsy specimens). Results and Limitations Excellent reproducibility (wK = 0.92 [95% CI, 0.88-0.96]) was observed for classification in groups and substantial reproducibility (wK = 0.78 [95% CI, 0.72-0.84]) was found for classification in subgroups. Global validity was 82% (range 78%-86%), showing no false negatives (sensitivity of 100% [1/1 biopsy]) and a very low rate of false positives (specificity 99% [297/299 biopsies]) for dysplasia detection. Conclusions This classification for methylene blue ME was highly reproducible and valid for the diagnosis of premalignant gastric lesions when used in a center different from that involved in its conception. Despite requiring an unconventional endoscope and a longer procedure, these results could reinforce ME as a valuable technique in the surveillance of patients at risk for gastric cancer.