Using tumor markers to predict the survival of patients with metastatic renal cell carcinoma Kim HL, Seligson D, Liu X, Janzen N, Bui MHT, Yu H, Shi T, Belldegrun AS, Horvath S, Figlin RA, Department of Urologic Oncology, Roswell Park Cancer Institut

摘要

PURPOSE: Approximately 30% of renal cell carcinomas (RCCs) present as metastatic disease. Molecular markers have the potential to characterize accurately the biological behavior of tumors and they may be useful for determining prognosis. MATERIALS AND METHODS: A custom tissue array was constructed using clear cell RCC from 150 patients with metastatic RCC who underwent nephrectomy prior to immunotherapy. The tissue array was stained for 8 molecular markers, namely Ki67, p53, gelsolin, carbonic anhydrase (CA)9, CA12, PTEN (phosphatase and tensin homologue deleted on chromosome 10), epithelial cell adhesion molecule and vimentin. Marker status and established clinical predictors of prognosis were considered when developing a prognostic model for disease specific survival. RESULTS: On univariate Cox regression analysis certain markers were statistically significant predictors of survival, namely CA9 (p < 0.00001), p53 (p = 0.0072), gelsolin (p = 0.030), Ki67 (p = 0.036) and CA12 (p = 0.043). On multivariate Cox regression analysis that included all markers and clinical variables CA9 (p = 0.00002), PTEN (p < 0.0001), vimentin (p = 0.0032), p53 (p = 0.028), T category (p = 0.0025) and performance status (p = 0.0013) were significant independent predictors of disease specific survival and they were used to construct a combined molecular and clinical prognostic model. The bias corrected concordance index (C-index) of this combined prognostic model was C = 0.68, which was significantly higher (p = 0.0033) than that of a multivariate clinical predictor model (C = 0.62) based on the UCLA Integrated Staging System (T category, histological grade and performance status). CONCLUSIONS: In patients with clear cell RCC a prognostic model for survival that includes molecular and clinical predictors is significantly more accurate than a standard clinical model using the combination of stage, histological grade and performance status.