Clinical and Sonographic Classification of Neurofibromas in Children with Neurofibromatosis Type 1 – A Cluster Analysis
Francisco Javier García‐Martínez, Fernando Alfageme, Anna Duat Rodríguez, Eva María Andrés Esteban, Á. Hernández‐Martín
Abstract
PURPOSE: High-frequency ultrasound allows the accurate identification of neurofibromas in neurofibromatosis type 1 (NF1). This study aimed to analyze the ultrasound features of neurofibromas in children with NF1, to establish a classification based on the clinical and sonographic patterns of the different types of neurofibromas, and to evaluate the interobserver correlation coefficient (κ) of this classification. MATERIALS AND METHODS: In this prospective, single referral center observational study, clinical and ultrasound findings of neurofibromas in children diagnosed with NF 1 were analyzed. To identify the ultrasound patterns, a cluster analysis allowing the inclusion of both clinical and ultrasound data was designed. The κ coefficient was calculated using 9 external evaluators. RESULTS: 265 ultrasound scans were performed on a total of 242 neurofibromas from 108 children diagnosed with NF1. Cluster analysis allowed the identification of 9 patterns (Snedecor's F, P < 0.001) classified as "classic" cutaneous neurofibroma, blue-red neurofibroma, pseudoatrophic neurofibroma, nodular subcutaneous neurofibroma, diffuse subcutaneous neurofibroma, congenital cutaneous neurofibroma, congenital plexiform neurofibroma, congenital diffuse and plexiform neurofibroma, and subfascial neurofibroma. The κ coefficient of the interobserver ratings was 0.82. CONCLUSION: Patterns identified in the cluster analysis allow neurofibromas to be classified with a very high interobserver correlation.