ABSTRACT
BACKGROUND & AIMS: Accurate histopathology of endoscopic duodenal biopsy specimens is critical in the diagnosis of celiac disease (CD) but sampling error and poor quality specimens may generate a false-negative result. Confocal endomicroscopy (CEM) is a novel technology allowing real-time in vivo microscopy of the mucosa that may diagnose CD and evaluate its severity and response to treatment more accurately than histopathology. METHODS: Subjects with CD and controls prospectively underwent CEM. Features of villous atrophy and crypt hypertrophy were defined. A CEM score measuring CD severity was devised and validated against the diagnosis of CD and blinded histopathology. Receiver operator characteristics, sensitivity to change after treatment, and reliability of findings were assessed. RESULTS: From 31 patients (6 untreated CD, 11 treated CD, and 14 controls), 7019 CEM images paired with 326 biopsy specimens were obtained. The accuracy of CEM in diagnosing CD was excellent (receiver operator characteristics area under the curve, 0.946; sensitivity, 94%, specificity, 92%) and correlated well with the Marsh grading (R-squared, 0.756). CEM differentiated CD from controls (P < .0001) and was sensitive to change after treatment with gluten-free diet (1787 optical biopsies; P = .012). The intraclass correlation of reliability was high (0.759-0.916). Of the 17 cases with diagnosed CD, 16 (94%) were diagnosed correctly using CEM but only 13 (76%) had detectable histopathology changes. The procedure was safe and well-tolerated. CONCLUSIONS: CEM effectively diagnoses and evaluates CD severity in vivo. This promising technique has the potential to improve endoscopy efficiency.
