Evaluating the Feasibility of Standard Ultrasound in the Bosniak Classification of Cystic Renal Masses Compared with Computed Tomography/Magnetic Resonance Imaging: A Multicenter Study.

OBJECTIVE:To investigate the diagnostic performance of standard ultrasound (US) using a US-adapted Bosnian classification for characterizing cystic renal masses (CRMs) compared with computed tomography/magnetic resonance imaging (CT/MRI) as the current standard. METHODS:From July 2016 to June 2022, pathology-proven CRMs and CRMs with ≥3 y of follow-up from six centers were reviewed. Four readers with varying levels of expertise in US categorized CRMs using a US-based Bosnian classification, followed by a consensus reading by two high-level readers. For CRMs with concurrent CT/MRI, the CT/MRI Bosniak classification was performed again by a blinded radiologist. With a cut-off category of ≥III as malignancy, diagnostic performance was compared using receiver operating characteristic curve analysis. Inter-modality agreement and inter-observer variability were statistically evaluated. RESULTS:A total of 603 CRMs graded to US Bosniak I, II, IIF, III and IV were analyzed, including 239 pathology-proven lesions and 364 lesions with adequate follow-up. The area under curve was 0.78 for US and 0.82 for CT/MRI, with a sensitivity of 77% versus 86% and a specificity of 79% versus 78%, respectively, showing no significant differences (p = 0.17). US and CT/MRI classifications matched in 68% (substantial agreement, kappa [κ] = 0.75), with US leading to higher scores in 19.5%. Comparison with reader consensus was substantial (highest κ = 0.77) for low-level readers and excellent (highest κ = 0.89) for high-level readers. Inter-observer variability showed substantial agreement (κ = 0.66) in low-level readers and excellent agreement (κ = 0.88) in high-level readers. CONCLUSION:US-based Bosniak classification is reproducible and effective for identifying malignancy among CRMs, offering a lower-cost alternative to contrast-enhanced modalities, which is particularly important in resource-limited care settings.