To examine correlations of the modified Bosniak categories assigned by radiologists to histological results and inter-rater reliability, focusing on intermediate-risk lesions.

The data of pediatric patients who underwent surgery for intermediate-risk complex renal cyst at a tertiary medical center in 2006-2019 were collected retrospectively. Four pediatric radiologists from 2 different medical centers reviewed the available imaging scans, and assigned each to one of the four modified Bosniak classification categories. Binary cohorts of the Bosniak categories (I-II vs III-IV) were compared to the histological results. Diagnostic accuracy (benign- vs intermediate-risk lesion) was calculated for each radiologist and for each imaging modality. Krippendorff's α test was used to measure inter-rater reliability.

The cohort included seven children, each with 1 complex cyst that was rated as intermediate-risk on pathological study. The median age was 1.5 years (IQR 1, 11.9). A correct classification was made in 41/56 imaging readings (sensitivity 73.2%). Applying Krippendorff's test to the binary Bosniak cohorts yielded poor inter-rater agreement (α = 0.08).

Implementation of the modified Bosniak classification in children caused a disconcerting underestimation of intermediate risk. There was a low inter-rater consistency for the categories intended to guide decisions regarding surgery or conservative management. The findings suggest that clinicians should be cautious using the modified Bosniak system for children.