Classification of probe-based confocal laser endomicroscopy findings in pancreaticobiliary strictures

The accurate diagnosis of indeterminate pancreaticobiliary strictures presents a clinical dilemma. Probe-based confocal laser endomicroscopy (pCLE) offers real-time in vivo microscopic tissue examination that may increase sensitivity for the detection of malignancy. the objective of this study was to develop and validate a standard descriptive classification of pcle in the pancreaticobiliary system.A total of 102 patients undergoing endoscopic retrograde cholangiopancreatography (ERCP) with pCLE to assess indeterminate pancreaticobiliary strictures were enrolled in a multicenter registry; 89 of these patients were evaluable. Information and data on the following were collected prospectively: clinical, ERCP, tissue sampling, pCLE, and follow-up. A uniform classification of pCLE findings ("Miami Classification") was developed, consisting of a set of image interpretation criteria. Thereafter, these criteria were tested through blinded consensus review of 112 randomized pCLE videos from 47 patients, and inter-observer variability was assessed in 42 patients .A consensus definition of the specific criteria of biliary and pancreatic pCLE findings for indeterminate strictures was developed. Single-image interpretation criteria did not have a high enough sensitivity for predicting malignancy. However, combining two or more criteria significantly increased the sensitivity and predictive values. The characteristics most suggestive of malignancy included the following: thick white bands (>20 µm), or thick dark bands (>40 µm), or dark clumps or epithelial structures. These provided sensitivity, specificity, positive predictive value, and negative predictive value of 97%, 33%, 80%, and 80% compared with 48%, 100%, 100%, and 41% for standard tissue sampling methods. Inter-observer variability was moderate for most criteria.The Miami Classification enables a structured, uniform, and reproducible description of pancreaticobiliary pCLE. Combining individual characteristics improves the sensitivity for the detection of malignancy.