Autofluorescence-Directed Confocal Endomicroscopy in Combination With a Three-Biomarker Panel Can Inform Management Decisions in Barrett's Esophagus.
di Pietro M., Bird-Lieberman EL., Liu X., Nuckcheddy-Grant T., Bertani H., O'Donovan M., Fitzgerald RC.
OBJECTIVES: Barrett's esophagus (BE) surveillance with white-light endoscopy and quadrantic biopsies (Seattle protocol) is resource intensive and limited by sampling error. Previous work suggests that autofluorescence imaging (AFI) in combination with a molecular panel might reduce the number of biopsies, but this was not sufficiently sensitive for low-grade dysplasia, now a point for endoscopic intervention. Here we used AFI to direct narrow-field imaging tools for real-time optical assessment of dysplasia and biopsies for a biomarker panel. We compared the new diagnostic algorithm with the current standard. METHODS: A total of 55 patients with BE were recruited at a single tertiary referral center. Patients underwent high-resolution endoscopy followed by AFI. AFI-targeted areas (n=194) were examined in turn by narrow-band imaging with magnification (NBIz) and probe-based confocal laser endomicroscopy (pCLE). Biopsies were taken from AFI-targeted areas and tested using an established molecular panel comprising aneuploidy plus cyclin A and p53 immunohistochemistry. RESULTS: In the per-patient analysis the overall sensitivity and specificity of AFI-targeted pCLE were 100% and 53.6% for high-grade dysplasia/intramucosal cancer and 96.4% and 74.1% for any grade of dysplasia, respectively. NBIz had equal specificity for dysplasia detection (74.1%), but significantly lower sensitivity (57.1%) than pCLE. The time required to perform AFI-targeted pCLE was shorter that that taken by the Seattle protocol (P=0.0004). We found enrichment of molecular abnormalities in areas with optical dysplasia by pCLE (P<0.001), regardless of histologic dysplasia. The addition of the 3-biomarker panel reduced the false positive rate of pCLE by 50%, leading to sensitivity and specificity for any grade of dysplasia of 89.2% and 88.9%, respectively. CONCLUSIONS: The combination of pCLE on AFI-targeted areas and a 3-biomarker panel identifies patients with dysplasia.