Lix Oliveira

Magnifying colonoscopy: interobserver agreement in the assessment of colonic pit patterns and its correlation with histopathological findings

Background and study aimsMagnifying colonoscopy (MC) is recognized as an aid to the differential diagnosis between neoplastic and nonneoplastic lesions. This study evaluated interobserver agreement of experienced endoscopists in the assessment of colonic pit patterns through the Kudo’s classification and correlated morphological aspects with histopathological findings.Materials and methodsA total of 213 magnification chromoendoscopic pictures of colonic lesions were collected from 161 consecutive patients and presented to three independent observers who expressed opinion about predominant pit pattern. All lesions were excised and sent for histopathological study.ResultsKappa statistics showed that the general agreement index with respect to the aspects of the pits was good among the three observers (0.561). Regarding prediction of histopathology according to the pit pattern diagnosis, overall accuracy was 84%, sensitivity was 91.4%, specificity was 67.2%, positive predictive value was 86.6%, and negative predictive value was 79.3%.ConclusionAlthough the interobserver reproducibility of the colonic pit pattern is good for experienced endoscopists, MC must not be used to replace the histopathological analysis, since it does not differentiate with the necessary safety neoplastic from nonneoplastic lesions.

Colonoscopia com magnificação de imagem: análise da variação interobservadores para os padrões de criptas e comparação das imagens endoscópicas com os achados histopatológicos

ABSTRACT: Colonoscopy with magnified view is the most reliable nonbiopsy method for distinguishing non-neoplastic fromneoplastic colorectal lesions. This study analysed lhe interobserver concordance in diagnosis of pit patem, with 3 experientsendoscopists using Kudo Classification and correlating the this aspects with histological ones. A total of 213 magnificated imagesinjuries (polipoides and plain) had been collected of 161 independent observing patients consecutive and presented tres that hadexpressed opinion on the predominant standard of criptas. All the injuries had been completely ressected and send for histopato logicostudy. The analysis esteem statistics of Kappa showed that the indice of general agreement for standards of criptas between the3 observers was 84%; sensitivity 91.4%; specificity 67.2% positive preditivo value 86.6%; negative preditivo value 79.3%; e ind exof Kappa of 0,61%. Although the interobserver reproduction of pit patern are good, the image for colonoscopia with magnificacao ,does not have to substitute the histopatologic study, therefore it does not differentiate with the necessary security neoplasticinjuries of non neoplastic.Key words: Magnification colonoscopy; neoplastic colorectal lesions; pit patem; Kudo classification.