Abstract
Purpose: :
We recently identified a gene expression signature that accurately predicts metastatic death in uveal melanomas from enucleated eyes. Tumors with the class 1 signature have a low risk and those with the class 2 signature have a high risk for metastasis. However, most uveal melanomas are treated with globe–sparing techniques such as plaque radiotherapy. Therefore, we sought to develop a molecular predictive assay that could be performed on small biopsy samples rather than requiring a large piece of tumor tissue.
Methods: :
Matched whole tumor tissue and mock FNABs were obtained on 20 uveal melanomas at the time of enucleation. True FNABs were obtained in 24 patients at the time of radiotherapy. RNA was prepared and microarray gene expression analysis performed using Affymetrix U133A chips. Data were normalized using RMA and analyzed using Affymetrix, GeneCluster, Spotfire and Dchip software.
Results: :
Good quality chip data were obtained from all biopsy samples with a mean scaling factor of 15 (median, 10; range, 4.8–91.4). Using unsupervised and supervised techniques, the correlation between gene expression data obtained from biopsy samples and whole tumor samples was statistically significant (P=0.008). Unsupervised analysis of true FNAB samples revealed two tumor clusters, similar to our published dichotomous classification using whole tumor samples. Further, the gene list that discriminated these two clusters was highly similar to our previously published gene list that discriminated the class 1 and class 2 tumors (P<0.001), suggesting that the biopsy–based predictive test and the published gene expression profile identified the same low– and high–risk tumor groups. No more than 50 genes were needed to predict accurately the tumor classes using the FNAB samples. Interestingly, a subset of these genes predicted accurately the rate of tumor regression following plaque radiotherapy (P<0.001).
Conclusions: :
Tumor RNA obtained by fine needle biopsy of uveal melanomas is of sufficient quantity and quality to perform microarray gene expression profiling and to accurately predict radioresistance and patient survival. The small set of genes that predict these outcomes are being analyzed for functional themes and provide new insights into uveal melanoma pathogenesis. This biopsy–based molecular predictive test may allow for individualized clinical management of uveal melanoma patients.
Keywords: melanoma • gene microarray • gene/expression