Validation of the modified Bosniak classification system to risk stratify pediatric cystic renal masses: An international, multi-site study from the pediatric urologic oncology working group of the societies for pediatric urology

Background: Pediatric cystic renal lesions are challenging to manage as little is known about their natural course. A modified Bosniak (mBosniak) classification system has been proposed for risk stratification in pediatric patients that takes ultrasound (US) and/or computed tomogram (CT) characteristics into account. However, literature validating this system remains limited. Objective: To determine if the mBosniak classification system correlates with pathologic diagnoses. The hypothesis is that mBosniak classification can stratify the risk of malignancy in children with renal cysts. Study design: Patients treated for cystic renal masses with available imaging and pathology between 2000 and 2019 from five institutions were identified. Clinical characteristics and pathology were obtained retrospectively. Characteristics from the most recent US, CT, and/or magnetic resonance imaging (MRI) were recorded. Reviewers assigned a mBosniak classification to each scan. mBosniak scores 1/2 were considered low-risk and 3/4 high-risk. These groups were compared with pathology (classified as benign, intermediate, malignant). Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (+LR), and negative likelihood ratio (-LR) were calculated to assess this categorization as a screening tool to guide surgical intervention. Agreement between imaging modalities was also explored. Results: 99 patients were identified. High-risk imaging findings were correlated with malignant or intermediate pathology with a sensitivity of 88.3%, specificity of 84.6%, PPV of 89.8%, NPV of 82.5%, +LR of 5.7, and -LR of 0.14. The sensitivity for detecting malignant lesions only was 100%. There was substantial agreement between US/CT (n = 55; κ = 0.66) and moderate agreement between US/MRI (n = 20; κ = 0.52) and CT/MRI (n = 13; κ = 0.47). Discussion: The mBos classification system is a useful tool in predicting the likelihood of benign vs. intermediate or malignant pathology. The relatively high sensitivity and specificity of the system for prediction of high-risk lesions makes this classification applicable to clinical decision making. In addition, all malignant lesions were accurately identified as mBosniak 4 on imaging. This study adds substantial data to the relatively small body of literature validating the mBosniak system for risk stratifying pediatric cystic renal lesions. Conclusions: Pediatric cystic renal lesions assigned mBosniak class 1/2 are mostly benign, whereas class 3/4 lesions are likely intermediate or malignant pathology. We observed that the mBosniak system correctly identified pathology appropriate for surgical management in 88% of cases and did not miss malignant pathologies. There is substantial agreement between CT and US scans concerning mBos classification.