Introduction Endoscopic surveillance in Barrett’s oesophagus (BO) has major limitations including sampling error due to inconspicuous dysplasia, need for multiple biopsies and subjectivity of pathological assessment of dysplasia. Probe-based confocal laser endomicroscopy (pCLE) allows real time histological assessment of the esophageal epithelium. However, in the absence of visible lesions, random pCLE is, like biopsies, subject to sampling error. We therefore used autofluorescence imaging (AFI) as a red-flag technique followed by a comparison of different advanced imaging technologies in these areas. The aims of the study were to evaluate the diagnostic accuracy for dysplasia of pCLE and to compare pCLE with narrow band imaging (NBI) for the prediction of the dysplasia.
Methods 27 patients with BO have been recruited at a single tertiary referral centre (non-dysplastic BO n = 9, indefinite for dysplasia n = 2, BO with low grade dysplasia n = 8, BO with high grade dysplasia (HGD) n = 3, BO with intramucosal cancer (IMC) n = 5). Mean BO length was 8cm. AFI positive (AFI+) areas were identified and assessed by NBI-zoom and pCLE, according to previously published criteria. Targeted biopsies were taken from AFI+ areas as well as from random quadrantic locations. Correlation between imaging patterns and histological outcome was analysed in a per lesion analysis (AFI+ areas individually) and per patient analysis (overall histology). Differences between pCLE and NBI were analysed with Fisher’s exact test
Results In the per lesion analysis, sensitivity and specificity of pCLE were 100% and 67%, respectively, for a diagnosis of HGD/IMC and 69% and 78% for a diagnosis of any grade of dysplasia. pCLE had a significantly higher sensitivity for dysplasia (P = 0.007) than NBI, but a lower specificity (p = 0.04), due to a slightly higher false positive rate. In the per patient analysis the combination of AFI and pCLE had a sensitivity and a specificity of 100% and 47%, respectively for a diagnosis of HGD/IMC, and 100% sensitivity and specificity for a diagnosis of any dysplasia (Table 1). AFI+pCLE did not change the overall patient outcome compared to the Seattle protocol, but allowed identification and correct characterization of 4 additional areas containing HGD/IMC
Conclusion pCLE can efficiently identify any grade of dysplasia. These preliminary data suggest that the combination of AFI and pCLE allows accurate in vivo diagnosis of dysplasia within BO. A larger study is warranted to determine whether this approach afford the potential to eliminate the necessity for multiple random biopsies during endoscopic surveillance
Disclosure of Interest None Declared