Confocal endomicroscopy for in vivo prediction of completeness after endoscopic mucosal resection.

BACKGROUND: Endoscopic mucosal resection (EMR) is an alternative to surgery for removal of superficial gastric neoplastic lesions. Residual neoplastic tissue of the resection interface is difficult to detect by conventional endoscopy. The aim of this study is to assess the efficacy of confocal laser endomicroscopy (CLE) in predicting complete resection margins after EMR. METHODS: EMR was performed by using cap-assisted or "inject and cut" resection technique. Two weeks after EMR, the circumferential margins of the defect were inspected by using CLE, and completeness of excision was predicted from the CLE image. Additional EMR was performed if necessary. In vivo CLE diagnosis was validated against final histopathology. RESULTS: Twenty-seven lesions were removed by EMR in 27 patients. After excluding 3 patients for gastrectomy, a total of 24 patients underwent CLE assessment, of whom 9 with indefinite lateral margins underwent at least two consecutive CLE follow-ups. A total of 19 lesions were regarded as complete remission, and 5 lesions (21.7%) were incompletely excised according to final pathologic diagnosis. Accuracy of CLE in predicting incomplete resection for original lesions was 91.7%, with sensitivity and specificity of 100.0 and 89.5%, respectively. The residual lesions were treated by additional EMR guided by CLE. There was no recurrence on endoscopic biopsies at mean (range) follow-up of 8.3 (4-15) months. CONCLUSIONS: Confocal laser endomicroscopy has high accuracy for prediction of remnant tissue after EMR, and may lead to significant improvements in clinical surveillance after endoscopic resection.