Diagnostic performance of two confocal endomicroscopy systems in detecting Barrett's dysplasia: a pilot study using a novel bioprobe in ex vivo tissue

Background: There are currently 2 existing confocal laser endomicroscopy (CLE) platforms: probe-based CLE (pCLE) and endoscope-based CLE (eCLE) systems, each with its own criteria for identifying dysplasia in Barrett's esophagus (BE). The diagnostic performance of these 2 systems has not been directly compared.

Design: Preclinical, feasibility study.

Objectives: We compared the interrater agreement and diagnostic performance of the pCLE and eCLE systems. In addition, we evaluated a new BE endomicroscopy criteria based on fluorescent glucose intensity uptake.

Patients: Thirteen patients with Barrett's esophagus and high-grade dysplasia or early cancer undergoing 16 EMR.

Intervention: CLE imaging was performed using two different probes with 2-[N-(7-nitrobenz-2-oxa-1,3-diaxol-4-yl)amino]-2-deoxyglucose, a fluorescent glucose analog with preferential uptake in dysplastic mucosa to supply contrast. Four quadrants were imaged per specimen with a total of 64 imaged mucosal sites presented to three gastroenterologists.

Main outcome measurements: Interobserver agreement and accuracy for dysplasia was assessed of images classified according to Miami criteria, stacked eCLE images classified using the Mainz criteria and a novel fluorescence intensity criteria.

Results: The interrater agreements were 0.17, 0.68, and 0.87 for the Miami, Mainz, and the fluorescence intensity criteria, respectively. Overall accuracy in detecting dysplasia was 37% (95% CI, 30.3-43.9), 44.3% (95% CI, 37.3-50.9), and 78.6% (95% CI, 72.2-83.3) for the Miami, Mainz, and the fluorescence intensity criteria, respectively.

Limitations: This imaging technique and proposed fluorescence intensity criteria using 2-[N-(7-nitrobenz-2-oxa-1,3-diaxol-4-yl)amino]-2-deoxyglucose in EMR tissue will require in vivo validation and cannot be directly used with the current eCLE and pCLE clinical applications.

Conclusions: In this preclinical feasibility study, the use of an eCLE system with a topical fluorescent contrast in ex vivo EMR tissue demonstrated higher interrater agreement and accuracy.