Impact of chromosomal copy number variation on outcome in metastatic clear cell renal cell carcinoma patients treated with antiangiogenic agents.

Abstract

393 Background: Antiangiogenic agents (AA) are used to treat metastatic clear cell renal cell carcinoma (ccRCC). Currently there are no biomarkers of therapeutic efficacy with these agents. We have previously shown that chromosome copy number variation (CNV) is associated with poor (14q loss) or more favorable outcomes (5q31-ter gain) in nonmetastatic patients. The purpose of the current study was the impact of CNV on response, progression free survival (PFS) and on overall survival (OS) after AA in patients with metastatic ccRCC. Methods: We obtained archival FFPE or frozen tumor specimens from patients with mRCC treated with sorafenib (after tumor removal), bevacizumab or sunitinib (neoadjuvant treatment). DNA was analyzed with Affymetrix 250K Nsp SNP microarrays. We identified the presence of genomic imbalances and loss of heterozygosity (LOH) to obtain “virtual karyotypes”. We then compared CNV to outcome using Wilcoxon-Gehan statistical analysis. Results: A total of 83 patients were used to analyze CNV after treatment with AA: 22 sorafenib, 29 sunitinib, 31 bevacizumab. Gain of 8q (p = 0.036) and loss of 16q (p = 0.0031), 20p (p = 0.038) or 20q (p = 0.022) were associated with a shorter OS, whereas gain of 1q (p = 0.037) and 5q (p = 0.019) were associated with longer OS in this patient cohort. When 14q loss was combined with 8q gain, median OS was further decreased (p 0.01). Surprisingly, when assessed as a group, no specific CNV was associated with PFS. However, when compared separately, 5q gain was predictive for better PFS in sorafenib or bevacizumab treated patients (p = 0.006), but no such effect was seen in the sunitinib treated cohort. Conclusions: Our results show that chromosomal imbalances are associated with divergent clinical outcome in metastatic ccRCC patients treated with AA, and are predominantly prognostic. Dissecting out driver tumor suppressor or tumor activating genes within these chromosomal regions will guide our understanding of what defines a particularly lethal phenotype. The possibility exists that specific CNV profiles predict for response to therapy, but a larger sample size is required to validate this possibility.