Preoperative multivariable prognostic models for prediction of survival and major complications following surgical resection of renal cell carcinoma with suprahepatic caval tumor thrombus

Surgical resection for renal cell carcinoma (RCC) with suprahepatic inferior vena cava tumor thrombus is associated with significant morbidity, yet there are currently no tools for preoperative prognostic evaluation. Our goal was to develop a preoperative multivariable model for prediction of survival and risk of major complications in patients with suprahepatic thrombi.We identified patients who underwent surgery for RCC with suprahepatic tumor thrombus extension from 2000 to 2013 at 4 tertiary centers. A Cox proportional hazard model was used for analysis of overall survival (OS) and logistic regression was used for major complications within 90 days of surgery (Clavien ≥ 3A). Nomograms were internally calibrated by bootstrap resampling method.A total of 49 patients with level III thrombus and 83 patients with level IV thrombus were identified. During median follow-up of 24.5 months, 80 patients (60.6%) died and 46 patients (34.8%) experienced major complication. Independent prognostic factors for OS included distant metastases at presentation (hazard ratio = 2.52, P = 0.002) and Eastern Cooperative Oncology Group (ECOG) performance status (hazard ratio = 1.84, P0.0001). Variables associated with increased risk of major complications on univariate analysis included preoperative systemic symptoms, level IV thrombus, and elevated preoperative alkaline phosphatase and aspartate transaminase levels; however, only systemic symptoms (odds ratio = 8.45, P0.0001) was an independent prognostic factor. Preoperative nomograms achieved a concordance index of 0.72 for OS and 0.83 for major complications.We have developed and internally validated multivariable preoperative models for the prediction of survival and major complications in patients with RCC who have a suprahepatic inferior vena cava thrombus. If externally validated, these tools may aid in patient selection for surgical intervention.