Accuracy of narrow-band imaging in predicting the histology of anal intraepithelial lesions

Abstract

Background and aim: Anal intraepithelial neoplasia (AIN) is the precursor of anal squamous carcinoma, frequently underappreciated by most gastroenterologists. Purpose of the study is to assess the diagnostic accuracy of narrow-band imaging (NBI) in predicting the histology of anal lesions, considering a new NBI classification.

Methods: This is a retrospective analysis of consecutive patients with suspected anal squamous intraepithelial lesions (SILs) and superficially invasive squamous cell carcinoma (SISCCA) who underwent anorectal-sigmoidoscopy. Three expert endoscopists reviewed all the images collected during colonoscopies in white light and NBI in order to generate a practical classification of three main different NBI patterns.

Results: After the modified Delphi process, the final NBI classification, mainly based on intrapapillary capillary loops (IPCL) changes included three different NBI pattern. NBI pattern I: elongation of IPCL toward epithelial surface; NBI pattern 2: thickened and tortuous IPCL; NBI pattern 3: mosaic-like disposition of IPCL. We found that NBI pattern positively correlated to the histologic type of the lesions and tumor grade (Spearman's rho = 0.9671, P = 0.0000). Among 58 anal lesions included, 41/41 (100%) LSILs have been classified as NBI pattern I, 10/11 (90.91%) HSILs as NBI pattern II, 1/11 HSIL as NBI pattern I, 4/4 SISCCA as NBI pattern III and 2/2 invasive cancer as NBI pattern III. Diagnostic accuracy of NBI pattern II or III in predicting histology differentiating HSIL or SISCCA/IC versus LSIL showed a sensitivity of 100%, (95% CI: 92.3-100%) and a specificity of 94.1% (95% CI: 83.8-98.3%).

Conclusion: The accuracy of this new NBI score system in predicting the histology of anal lesions showed encouraging data in term of sensitivity and specificity in differentiating HSIL or SISCCA/IC versus LSIL, but the clinical usefulness and application of these findings should be evaluated in a prospective larger study.