Abstract
PURPOSE: To develop a preoperative prediction model using a computer-assisted volumetric assessment of potential spared parenchyma to estimate the probability of chronic kidney disease (CKD, estimated glomerular filtration rate [eGFR] <60 mL/min/1.73 m(2)) 6 months from extirpative renal surgery (nephron-sparing surgery [NSS] or radical nephrectomy [RN]). PATIENTS AND METHODS: Retrospective analysis of patients who underwent NSS or RN at our institution from January 2000 to June 2013 with a compatible CT scan 6-month renal function follow-up was performed. Primary outcome was defined as the accuracy of 6-month postoperative eGFR compared with actual postoperative eGFR based on root mean square error (RMSE). Models were constructed using renal volumes and externally validated. A clinical tool was developed on the best model after a given surgical procedure using area under the curve (AUC). RESULTS: We identified 130 (51 radical, 79 partial) patients with a median age of 58 years (interquartile range [IQR] 48-67) and preoperative eGFR of 82.1 (IQR 65.9-104.3); postoperative CKD (eGFR <60) developed in 42% (55/130). We performed various linear regression models to predict postoperative eGFR. The Quadratic model was the highest performing model, which relied only on preoperative GFR and the volumetric data for a RMSE of 15.3 on external validation corresponding to a clinical tool with an AUC of 0.89. CONCLUSION: Volumetric-based assessment provides information to predict postoperative eGFR. A tool based on this equation may assist surgical counseling regarding renal functional outcomes before renal tumor surgical procedures.