Subscribe to RSS
DOI: 10.1055/s-0029-1222680
Exon-level gene expression analyses of primary neuroblastoma improves risk prediction and identifies JARID1C as a candidate target
Neuroblastoma (NB), the most common solid tumor of childhood, is characterized by a remarkable heterogeneity of patients' courses and survival rates of older patients with metastatic disease have remained poor. Numerous prognostic factors including amplification of the MYCN oncogene have been described. However, despite advanced clinical risk stratification, current trials still fail to determine the best treatment strategy for a substantial number of patients. On the route to clinical application of improved array-based classifiers to predict the risk profile of individual patients, significant progress has been achieved over the past years. We here present data on a cohort of 138 primary NB using a novel approach including information for all human coding exons described to date (Affymetrix ExonST Array). Using a classifier trained on 100 patient samples and then used to predict the outcome of the remaining 38 patients, we were able to achieve prediction accuracies >80% in the independent test set using support vector machine (SVM) learning algorithms, which is superior to the current clinical risk stratification. Interestingly, the histone demethylase JARID1C, which had not been linked to cancer formation previously, was up-regulated in relapsing NB and found to have prognostic value independent of MYCN amplification. JARID1C was also highly expressed in all NB cell lines investigated. Down-regulation of JARID1C using siRNA inhibited cell proliferation in vitro. Taken together, exon level analysis appears to be a highly promising tool both for prediction of outcome and for providing new insights into tumor biology.