Diagnostic Accuracy of Bosniak v2019 for Predicting Cancer and Unfavorable Histology in Class III-IV Cystic Renal Masses.

Background The current Bosniak classification system (Bosniak Classification version 2019 [Bosniak v2019]) helps predict cancer, but it is unknown whether it also helps predict unfavorable histologic features. Purpose To determine whether Bosniak v2019 and nodule size help predict cancer and unfavorable histologic features. Materials and Methods This retrospective study included Bosniak III or IV cystic renal masses across 10 institutions and three countries (January 2007 to February 2024). Pretreatment renal CT or MRI scans obtained within 1 year of resection were independently reviewed by two radiologists per site. The reference standard was surgical histopathologic findings. Generalized linear mixed models and multivariable analyses were used to estimate the probability of cancer and cancer with unfavorable histologic features by Bosniak class, subclass, and nodule size, adjusted for mass size. Reader agreement was assessed with the Gwet agreement coefficient (ie, AC1) and intraclass correlation coefficient. Results A total of 419 patients with 421 masses (264 male; median age, 59 years [IQR, 50-67 years]) were included. The probabilities of cancer were 74.8% (95% CI: 69.7, 79.4) and 91.0% (95% CI: 88.2, 93.1) for Bosniak classes III and IV, respectively. The probabilities of cancer with unfavorable histologic features were 10.3% (95% CI: 7.4, 14.2) and 32.3% (95% CI: 28.5, 36.4) for Bosniak classes III and IV, respectively. Bosniak IV subclasses were associated with a greater probability of cancer and unfavorable histologic features than were Bosniak III subclasses (P < .001 to P = .006). There was no evidence of a difference in the odds of cancer or unfavorable histologic features within the Bosniak III (P = .56-.995) or IV (P = .10-.50) subclasses. Nodule size was the strongest predictor of cancer (odds ratio [OR], 4.1 per 10 mm; P = .001) and unfavorable histologic features (OR, 1.9 per 10 mm; P < .001). Interrater agreement was substantial for Bosniak class (Gwet AC1, 0.70 [95% CI: 0.63, 0.77]) and good for nodule size (intraclass correlation coefficient, 0.89 [95% CI: 0.85, 0.91]). Conclusion Bosniak v2019 helps predict cancer and cancer with unfavorable histologic features. Nodule size was strongly associated with unfavorable histologic features. © RSNA, 2026 Supplemental material is available for this article.