Introduction

Predictive models for recurrence in non-metastatic renal cell carcinoma (RCC) have been developed from general populations of RCC patients, including very few high risk patients. According, these models may not be ideal to stratify high risk patients for enrollment in adjuvant clinical trials such as patients with venous tumor thrombus. The purpose of this study was to compare predictive accuracy for a nomogram developed from non-metastatic RCC patients with venous tumor thrombus with 3 existing prognostic models for RCC recurrence following surgery.

Methods

A nomogram was developed from independent predictors after multivariable modeling of common clinical and pathologic variables. Receiver operator characteristic (ROC) curves were constructed to compare predictive accuracy of the nomogram with the University of California?Los Angeles Integrated Staging System(UISS), Stage Size Grade Necrosis (SSIGN) and Sorbellini models.

Results

A total of 669 consecutive non?metastatic RCC patients with tumor thrombus were treated surgically from 2000?2014 at 6 institutions. Variables independently associated with RCC recurrence were increased tumor diameter per centimeter, body mass index per point, preoperative hemoglobin < lower limit of normal, IVC thrombus level above hepatic vein, perinephric fat invasion, and non?clear cell histology. Estimated overall 5?year RFS was 49%. A nomogram was developed using weighted variables in a development cohort (n=465) and validated in a separate cohort (n=204). _x000D_ _x000D_ The AUC for the thrombus nomogram was 0.738, which was significantly higher than AUC for the Sorbellini model (0.623), UISS model (0.615) and SSIGN model (0.584)._x000D_ _x000D_

Conclusions

A nomogram developed specifically for non?metastatic RCC patients with venous tumor thrombus has significantly greater predictive accuracy than general prognostic models in this high risk population.

Funding

none