| Background&Aims:Ulcerative colitis(UC) is a chronic relapsing condition, hence,it is not only important to reach a correct diagnosis but also to assess disease activity. This is mainly done with the integration of clinical manifestation, serological parameters, endoscopic and histological assessment. Histological assessment is considered as the "gold standard". But mucosa biopsy easily causes time and resources consumption. Also, the assessment of disease activity by conventional colonoscopy is inaccurate in some cases, especially for those seemed to be in remission.Confocal laser endomicroscopy (CLE) has been recently proposed as a new technique that allows in vivo histologic assessment of mucosa during ongoing endoscopy, which is also reputed as "optical biopsy". This advanced endoscopic technique has a good application in the research of gastrointestinal disease. Three different classification systems for the evaluation of inflammation activity in UC with CLE has been proposed recently. Until now, no comparative assessment of these classification systems has been performed. We aimed to evaluate the accuracy and interobserver agreement of the three classification systems for the prediction of inflammation activity in ulcerative colitis.Methods:Patients previously diagnosed as UC for confocal endomicroscopy surveillance were recruited from the outpatient clinic of Qilu Hospital, Shandong University between March2011and December2011. The interested mucosa which seemed abnormal with white light mode was observed seriously by confocal endomicroscopy with the aid of contrast agent--fluorescein sodium. The serial confocal images were acquired which were also stored on laser discs for subsequent review. At last, targeted biopsies of the colon were obtained with a stand biopsy forceps.Only high quality images showing clear crypts or vascular architectures were selected by one expert endoscopist who was aware of the endoscopic and histological information. Four confocal images were selected from one suspicious area in each patient and stored in an absolute set. Each set was labelled with three different random orders, one for each classification system. Then, the images were observed by two groups of endoscopists with different levels of CLE experience in postprocedure. The prediction of inflammation activity according to three classification systems was compared with the histological assessment, respectively. Interobserver agreement was also analysed in the study.All calculations were carried out using SPSS v13.0. Differences in proportions were tested with the chi-squared test. P<0.05was considered to be statistically significant. Cohen’s κ coefficient was used to measure the degree of interobserver agreement.Results:In total, thirty-seven patients were enrolled in this study, so thirty-seven sets of confocal images were obtained. Histologically, nine were normal mucosa, twelve were non-active inflammation mucosa, and sixteen were active inflammation mucosa. Overall accuracy was87.8%for the Qilu system,82.9%for the Mainz system,82.4%for the Nagoya system, respectively. Sensitivity and specificity for the prediction of active inflammation were98.9%and95.2%for the Qilu,100%and93.2%for the Mainz,99.0%and99.9%for the Nagoya, respectively. Sensitivity and specificity for the prediction of non-active inflammation were65.3%and98.7%for the Qilu classification system,65.3%and94.0%for the Mainz,54.2%and96.0%for the Nagoya grading system, respectively. No significant difference related to the assessor’s level of expertise was observed. Interobserver agreement was substantial for the Nagoya, Mainz, and Qilu system(κ=0.714,0.788,0.797, respectively).Conclusion:1.The three classification systems suggesting good accuracy and feasibility could be used in the prediction of inflammation activity of UC. No obvious differences were found in these systems.2.The substantial interobserver agreement suggests a short learning curve for the endoscopists with less expertise in CLE. |
PDF Full Text Request