Development of a prognostic risk model for clear cell renal cell carcinoma by systematic evaluation of DNA methylation markers

Table 2 Comparison of model performance and fit

Models	Population-based series				Hospital-based series				TCGA series
Models	n	Df	AIC	C-statistic	n	Df	AIC	C-statistic	n	Df	AIC	C-statistic
Clinical model^a	219	9	681	0.65	42	5	63	0.86	227	7	470	0.75
Prognostic model^b	219^c	14	674	0.71	42	10	55	0.95	227	12	475	0.76

^a–cPerformance of both the clinical Cox proportional hazards model a (including age at diagnosis, sex, Fuhrman grade, tumor size and TNM stage) and the prognostic biomarker Cox proportional hazards model b (containing age at diagnosis, gender, Fuhrman grade, tumor size, TNM stage, methylation of NEFH, GREM1, GATA5, LAD1, and NEURL). Numbers in the table refer to the number of cases included in the analysis (n), degrees of freedom (Df), Akaike Information Criterion (AIC) and the Harrel’s C-statistic (C-statistic). c Lower number of patients due to missing data on methylation status of the included genes

ISSN: 1868-7083