Hypoxia-inducible factor pathway genes predict survival in metastatic clear cell renal cell carcinoma.

Abstract

Hypoxia inducible factor (HIF) pathway alterations drive progression of clear cell renal cell carcinoma (ccRCC). We aim to evaluate genes within the canonical and non-canonical HIF pathways as predictors of survival in metastatic ccRCC. Gene expression was determined from 324 archival pretreatment nephrectomy specimens from CALGB90206, a phase III trial of patients treated with interferon alpha (INF-α) vs. INF-α plus bevacizumab. TaqMan RT-qPCR was performed using RNA from tumors macrodissected based on review by genitourinary pathology. A total of 35 HIF-related genes were assessed by Cox regression analysis. After adjusting for sex and Memorial Sloan Kettering Cancer Center risk score (MSKCC-RS), 11 genes predicted OS: HIF2A (HR 1.059, P = 0.012), EGLN3 (HR 1.089, P = 0.012), VEGFC (HR 0.904, P = 0.039), VEGFD (HR 1.085, P = 0.016), FLT4 (HR 1.093, P = 0.038), CCND1 (HR 1.077, P = 0.026), TGFA (HR 1.127, P = 0.003), EGFR (HR 1.151, P = 0.028), VHL (HR 0.764, P = 0.002), HSP90AA1 (HR 0.845, P = 0.002), and PTEN (HR 1.163, P = 0.050); 7 genes predicted PFS: HIF2A (HR 1.060, P = 0.011), CCND1 (HR 1.082, P = 0.016), TGFA (HR 1.096, P = 0.026), EP300 (HR 1.171, P = 0.031), VHL (HR 0.775, P = 0.007), HSP90AA1 (HR 0.871, P = 0.015), and TP53 (HR 1.119, P = 0.050). Most of these genes validated as significant predictors of survival in the external, TCGA dataset. In multivariate analysis of all externally validated genes, VEGFC (HR 0.906, P = 0.043), TGFA (HR 1.122, P = 0.003), CITED2 (HR 1.113, P = 0.035) and EP300 (HR 1.136, P = 0.049) predicted OS; and HIF2A (HR 1.049, P = 0.036) and EP300 (HR 1.199, P = 0.010) predicted PFS. EGLN3 (HR 1.156, P = 0.045) and BNIP3 (HR 1.254, P = 0.049) significantly interacted with treatment status and predicted PFS in patients treated with IFN-α and IFN-α+bevacizumab, respectively. We identified specific gene isoforms in both the canonical and non-canonical HIF pathways associated with metastatic RCC survival. EGLN3 and BNIP3 showed significant interaction with treatment arm and may be predictive of treatment response. We have identified genes for future prospective investigation as predictive biomarkers and novel drug targets.