Risk of Recurrence After Nephrectomy: Comparison of Predictive Ability of Validated Risk Models - Beyond the Abstract

According to current European guidelines, nephrectomy (either partial or radical) is the treatment of choice for localized and locally-advanced renal cell carcinoma (RCC), whose incidence is increasing, mainly due to incidental detection of small renal masses.1 Nonetheless, up to 20% of non-metastatic patients at the time of diagnosis will develop recurrence after nephrectomy.2 Over time, several post-operative prognostic models based on the prognostic value of histopathologic variables have been developed to choose patient management, assess the need for adjuvant therapy, and define post-operative surveillance protocols.3

In our study,4 we tested the ability of validated risk models to predict recurrence free survival (RFS) within a “real-life” contemporary cohort of 358 non-metastatic surgically-treated patients, with predominantly low-stage and low-grade RCC. The predictive ability (expressed by C-index) was assessed for the SSIGN, the Leibovich, the ASSURE, and the UISS models, the AJCC/TNM staging system, and the GRANT score. Additionally, we also tested for the discriminatory ability of all models over time. According to our results, the predictive ability of the SSIGN models was the highest (c-index ranging from 0.89 at 6-month follow-up to 0.82 at 60-month follow-up), followed by the Leibovich models (c-index from 0.89 to 0.82), the AJCC/TNM staging system (c-index from 0.82 to 0.77), the ASSURE score (c-index from 0.81 to 0.76), the GRANT score (c-index from 0.83 to 0.73) and the UISS models (c-index from 0.76 to 0.72). The graphical depiction of the estimated C-indexes over time showed that each model reached its peak discriminatory ability before 12-month follow-up, followed by a steep decline within the first 24 months and then a steady decline over time for all models.

Our study confirms the reliability and clinical utility of the currently validated prognostic models to predict the risk of recurrence after surgical treatment for non-metastatic ccRCC and therefore they should be recommended to stratify surveillance protocols after nephrectomy. The SSIGN and Leibovich scores have the highest predictive accuracy. The GRANT and the ASSURE score are easy to calculate and have a good predictive performance. Despite their lower predictive accuracy compared to that of SSIGN and Leibovich score, taking into account age and thus competing risk of mortality may be of primary importance in patients with predominantly localized low risk RCC. Therefore, an age-based score model may be useful in planning follow-up schedule, in order to avoid unnecessary imaging. Additionally, the AJCC/TNM itself reached a good performance ability, which was even higher compared to some of the other scores. Nonetheless, the analysis of granular data of these scores showed that the differential delta of RFS estimates in each stratum of each model was higher compared to AJCC/TNM, where stage I and stage II showed very similar survival estimates throughout all follow-up. This may be of paramount importance not only for follow-up planning but also to provide useful prognostic information for patient counseling. Finally, it should be taken into account that the prediction ability decreases over time for all models with the peak discriminatory ability reached within the first 12 months after surgery.

Written by: Carlotta Palumbo, MD, Davide Perri, MD, and Alessandro Volpe, MD, Division of Urology, Department of Translational Medicine, University of Eastern, Maggiore della Carità Hospital, Novara, Italy. 


  1. Palumbo C, Pecoraro A, Knipper S, Rosiello G, Luzzago S, Deuker M, et al. Contemporary Age-adjusted Incidence and Mortality Rates of Renal Cell Carcinoma: Analysis According to Gender, Race, Stage, Grade, and Histology. Eur Urol Focus 2020:S2405456920301140. https://doi.org/10.1016/j.euf.2020.05.003.
  2. Brookman-May SD, May M, Shariat SF, Novara G, Zigeuner R, Cindolo L, et al. Time to recurrence is a significant predictor of cancer-specific survival after recurrence in patients with recurrent renal cell carcinoma - results from a comprehensive multi-centre database (CORONA/SATURN-Project): Time to recurrence as predictor for CSS after recurrence in patients with RCC. BJU Int 2013:n/a-n/a. https://doi.org/10.1111/bju.12246.
  3. Klatte T, Rossi SH, Stewart GD. Prognostic factors and prognostic models for renal cell carcinoma: a literature review. World J Urol 2018;36:1943–52. https://doi.org/10.1007/s00345-018-2309-4.
  4. Palumbo C, Perri D, Zacchero M, Bondonno G, Martino JD, D’Agate D, et al. Risk of recurrence after nephrectomy: Comparison of predictive ability of validated risk models. Urol Oncol 2022:S1078-1439(21)00542-1. https://doi.org/10.1016/j.urolonc.2021.11.025.

Read the Abstract
email news signup