Experienced versus inexperienced confocal endoscopists in the diagnosis of gastric adenocarcinoma and intestinal metaplasia on confocal images

Background Confocal laser endomicroscopy (CLE) may be used to diagnose gastric cancer and intestinal metaplasia, but the impact of CLE experience on the accuracy of confocal diagnosis of gastric cancer and intestinal metaplasia is not clear. Objective To establish the sensitivity, specificity, and intragroup interobserver agreement of CLE image interpretation by 3 experienced (group 1) and 3 inexperienced (group 2) CLE endoscopists for diagnosing gastric intestinal metaplasia (GIM) and adenocarcinoma. Design Blinded review of CLE images for the diagnosis of gastric cancer or intestinal metaplasia. Setting Tertiary care hospital. Patients CLE images obtained ex vivo from gastrectomy specimens with proven gastric cancer and CLE images obtained in vivo from Chinese subjects older than 50 years of age by using matched biopsy specimens as reference standards. Main Outcome Measurements Sensitivity, specificity, and intragroup interobserver agreement of CLE image interpretation. Results Interpretation of in vivo images by group 1 was associated with higher sensitivity (95.2% vs 61.9%, P = .039) and higher specificity (93.3% vs 62.2%, P < .001) for GIM than interpretation by group 2. The agreement between interpretation by group 1 and histology for GIM was higher than that for group 2 (κ = 0.864 vs 0.217). The sensitivity (93.3% for group 1 vs 86.7% for group 2, P = 1.000) and specificity (87.7% for group 1 vs 80.7% for group 2, P = .344) of interpretation of ex vivo CLE images for the diagnosis of gastric adenocarcinoma was similar for groups 1 and 2. Limitations Single-center study. Conclusions Experience in CLE was associated with greater accuracy in the diagnosis of intestinal metaplasia.