Creation of a primary tumor tissue expression biomarker-augmented prognostic model for patients with metastatic renal cell carcinoma

Abstract Background Clinical and pathological factors alone have limited prognostic ability in patients with metastatic clear cell renal cell carcinoma (ccRCC). Bim, a downstream pro-apoptotic molecule in the PD-1 signaling pathway, has recently been associated with survival in other malignancies. We sought to determine if tissue biomarkers including Bim, added to a previously reported clinical metastases score, improved prediction of cancer-specific survival (CSS) for patients with metastatic ccRCC. Methods Patients with metastatic ccRCC who underwent nephrectomy between 1990 and 2004 were identified using our institutional registry. Sections from paraffin-embedded primary tumor tissue blocks were used for immunohistochemistry staining for Bim, PD-1, B7-H1 (PD-L1), B7-H3, CA-IX, IMP3, Ki67, and survivin. Biomarkers that were significantly associated with CSS after adjusting for the metastases score were used to develop a biomarker-specific multivariable model using a bootstrap resampling approach and forward selection. Predictive ability was summarized using a bootstrap-corrected c-index. Results The cohort included 602 patients: 192 (32%) with metastases at diagnosis and 410 (68%) who developed metastases after nephrectomy. Median follow-up was 9.6 years (IQR 4.2–12.8), during which 504 patients died of RCC. Bim, IMP3, Ki67, and survivin expression were significantly associated with CSS after adjusting for the metastases score, and were eligible for biomarker-specific model inclusion. After variable selection, high Bim (hazard ratio [HR] = 1.44; 95% confidence interval [CI] 1.16–1.78; P Conclusion We created a prognostic model combining the clinical metastases score and 2 primary tumor tissue expression biomarkers, Bim and survivin, for patients with metastatic renal cell carcinoma who underwent nephrectomy.