Abstract: PD52-11
Sources of Funding: R21CA176422, R01CA134466

Introduction

Multiple prognostic algorithms exist for patients with renal cell carcinoma (RCC). Models for localized disease, especially those that incorporate pathologic features, are highly accurate, while models for metastatic disease have inferior predictive ability. Thus, we investigated the gene expression profile in metastatic RCC to improve prognostication and provide a rationale for future biologic studies.

Methods

A custom Nanostring panel was used to evaluate 124 candidate genes on specimens from 111 patients with resected primary and matched metastatic clear cell RCC between 1990 and 2005., No patient received systemic therapy prior to metastasectomy. After initial candidate genes were identified, a multivariable gene expression model was built using the lasso method and a scoring algorithm to predict likelihood of death was developed using the coefficients. Multivariable Cox regression used to determine if the scoring algorithm was predictive after adjusting for our previously published algorithm for metastatic RCC.

Results

Nanostring assay was successful in 91 of 111 primary clear cell RCC tumor specimens. Median follow-up for survivors was 108.0 months, during which 79 patients died from RCC. In primary tumors, 18 of 124 genes interrogated were univariately associated with RCC-specific survival (false discovery rate <0.10) and five genes were retained in the multivariable model. After adjusting for clinical and pathological indices previously shown to be predictive of survival in metastatic clear cell RCC, the five gene scoring algorithm remained highly significant (p<0.0001). When expression levels were determined in metastatic tissue, rather than the primary tumor tissue, the five gene scoring algorithm remained significantly associated with survival.

Conclusions

We have identified a panel of genes that predict prognosis in patients with metastatic clear cell RCC and provides significant risk stratification after adjusting for existing models. These genes may provide insight into the biology of metastatic RCC and warrant further investigation.

Funding

R21CA176422, R01CA134466