FG has no added value in prediction of mortality after partial and radical nephrectomy for chromophobe renal cell carcinoma patients

Our objective was to test whether FG (FG) is applicable in the context of chromophobe renal cell carcinoma patients treated with partial and radical nephrectomy. Patients (n = 1862) with chromophobe renal cell carcinoma treated with partial and radical nephrectomy were identified within the Surveillance, Epidemiology, and End Results (1988-2008). Univariable and multivariable Cox regression analyses were fitted to predict cancer-specific mortality. Discriminant properties were assessed for the conventional four-tiered FG scheme. Additionally, discrimination of the three-tiered FG scheme (1-2 vs 3 vs 4) and the two-tiered FG scheme (1-2 vs 3-4) was also assessed. The statistical significance of the differences in accuracy estimates was compared using the Mantel-Haenszel test. A total of 65 of the 1862 died of the disease. The overall 5-year cancer-specific mortality-free survival rate was 94.8% (95% confidence interval: 93.5-96.2). In univariable analyses, none of the FG strata were significantly associated with cancer-specific mortality. Furthermore, FG was less informative (63%) than tumor size (72%) and tumor stage (69%), using measures of discrimination in univariable analyses. After accounting for all covariates, prediction of 5-year cancer-specific mortality was 79.0% vs 80.3% accurate, respectively, with vs without the consideration of FG (P=0.01). Similar discrimination estimates were obtained for the modified three-tiered FG scheme (78.5%; P=0.009) and the modified two-tiered FG scheme (79.5%; P=0.02). In conclusion, FG is not an informative predictor of prognosis, defined as cancer-specific mortality, after partial and radical nephrectomy for chromophobe renal cell carcinoma patients.