Limited prognostic value of tumor necrosis in patients with renal cell carcinoma.

OBJECTIVES: To test whether tumor necrosis (TN) could improve the prognostic ability of the predictors of 2 established prognostic renal cell carcinoma (RCC) models. Presence of TN within the nephrectomy specimen is considered an important prognostic marker in patients with RCC. However, its added prognostic value along with established cancer-specific mortality (CSM) predictors has never been formally tested.

METHODS: We retrospectively analyzed data of 1526 patients with all stages of RCC, who were treated with radical or partial nephrectomy at 6 institutions between 1988 and 2004. Univariate and multivariate Cox-regression models tested the statistical significance of TN in CSM predictions. Covariates consisted of TNM stage, Fuhrman grade, tumor size, and symptom classification. The analyses first addressed the entire patient population (n=1526) and then repeated in patients with exclusive clear-cell histology (n=1320).

RESULTS: TN was present in 476 patients (31.2%). TN was a statistically significant predictor of CSM (hazard ratio: 2.73; P<.001) but not an independent predictor of CSM (adjusted hazard ratio: 0.88; P=.4). Accuracy of TN ranked sixth among 7 examined predictors and TN failed to improve the accuracy of other variables. The same results were recorded in patients with exclusive clear-cell histology.

CONCLUSIONS: TN does not improve the accuracy of established predictors of CSM that are used in 2 prognostic RCC models for patients with RCC of all stages.