Prognostic factors in renal cell carcinoma.

purposeRenal cell carcinoma (RCC) is a very heterogeneous disease with widely varying prognosis. An accurate knowledge of the individual risk of disease progression and mortality after treatment is essential to counsel patients, plan individualized surveillance protocols and select patients for adapted treatment schedules and new clinical trials.

methodsA systematic review of the literature on prognostic factors of localized and metastatic RCC was performed.

resultsPrognostic factors in RCC include anatomical (TNM classification, tumor size), histological (Fuhrman grade, histologic subtype), clinical (symptoms and performance status), and molecular features. All these features are not perfectly accurate when used alone. Therefore an increasing number of prognostic models or nomograms that include several combined prognostic features have been designed in order to improve predictive accuracy. UCLA Integrated Staging System (UISS) and the Mayo Clinic's SSIGN score are the two most used prognostic models for localized RCC. In the setting of metastatic RCC the classical anatomical and histological tumor features have little predictive value. However, accurate prognostic models have been designed to predict response to therapy, and progression-free and overall survival. The two most used tools to predict response to immunotherapy are the model designed by the French Group of Immunotherapy and the Motzer's model. The advent of tyrosine kinase inhibitors and antiangiogenic drugs have deeply changed the treatment of metastatic RCC. Predictive tools that are adapted to the modern targeted therapies are now needed.

conclusionThere is increasing knowledge on prognostic factors of localized and metastatic RCC. Several predictive models have been developed by combining different prognostic features and are valuable tools for patient counseling, treatment decision-making and trial design. Further research is needed to assess whether the combination of classical prognostic factors with molecular features and information from gene and protein expression profiling can increase the predictive accuracy of the current prognostic models.