Our results indicate that amplification of the MYCN transcription factor has significantly affected miRNA expression in neuroblastoma tumors, consistent with our earlier observations 
and the observations of other groups 
. Our present studies, based on a larger set of miRNAs and tumors than what was previously studied, has permitted us to identify many additional miRNAs whose expression levels are correlated with MYCN levels. In addition, we demonstrate that many of the recurrent large-scale chromosomal imbalances, including loss of 1p, 3p, 11q and 14q, along with gain of 1q and 17q, have also had a major impact upon miRNA expression. Our finding that approximately 15% of miRNA loci have expressional changes attributable to genomic imbalances is comparable to the results of Zhang et al 
showing that 15% of miRNAs are deregulated by such alterations in ovarian cancer. Interestingly, we identified several miRNAs whose expression levels were inversely correlated to the genomic imbalances, indicating that other factors can significantly negate the effects of DNA copy number alteration. For example, miR-29c is expressed at lower levels in MNA tumors in spite of the fact that it maps to a region on chromosome 1q that is commonly gained in neuroblastoma. In this instance, and in at least two additional instances (miR-30a and miR-25), over-expression of the MYCN
oncogene would appear to be the factor that is over-riding the effect of DNA copy number change.
Many of the miRNAs that are down or up-regulated by either MYCN or chromosomal imbalances are likely to play key roles in tumorigenesis. For example, miR-34a mapping to chromosome 1p is clearly down-regulated by 1p deletion because it targets important pro-proliferation genes 
, while the miR-17-5p cluster is directly up-regulated by MYCN and targets the p21 tumor suppressor. Wei et al 
showed that over-expression of the miR-17-5p-92 cluster host gene, MIRHG1
, is correlated with poor patient survival and that expression of miR17-5p-92 is positively correlated with host gene expression. Presumably over-expression of miR-17-5p-92 is also correlated with poor survival, although this was not directly reported on. Our results indicate a correlation of over-expression of the miR17-5p-92 cluster members with poor patient survival (p<0.009), although the p-value was not significant following a highly conservative correction for multiple comparisons. In this regard, the lack of a significant p-value associated with miR-17-5p over-expression and poor survival is probably a false negative result in our study, particularly in view of the fact that functional studies of this miRNA in neuroblastoma have indicated an oncogenic role in neuroblastoma 
Interestingly, miR-29a, which we show to be under-expressed in MNA tumors, was also recently shown to be down-regulated in neuroblastoma as it directly targets B7-H3, a cell surface immunomodulatory glycoprotein which contributes to the process of immune escape 
. It is equally important to realize that some of the dysregulated miRNAs might be passive events and have no effect on disease pathogenesis. For example, miR-30e was highly under-expressed in MNA tumors, but we have previously shown that ectopic over-expression of this miRNA in MNA cell lines has no detectable impact on cell proliferation 
We have also demonstrated that a signature based on the expression of 37 miRNAs that are differentially expressed in MNA tumors relative to non-MNA tumors permits the clustering of patients into two groups that differ significantly in survival. However, this signature yields P-values that are not as good as the use of MNA or 11q status by themselves, possibly because some of the miRNAs in the signature play no role in tumorigenisis (e.g. miR-30e). The inclusion of these passive miRNAs in the MYCN miRNA signature could decrease the predicting power of a miRNA expression signature based on miRNAs identified by their association with MYCN levels. We therefore developed a miRNA expression signature by directly identifying the miRNAs that were most significantly associated with overall survival. Interestingly, only miR-572 and miR-190 from the set of 15 miRNAs most closely associated with EFS and OS overlapped the set of 37 miRNAs that were associated with MYCN amplification. This signature classified patients into groups and subgroups associated with EFS and OS with p-values that were significantly better than those obtained with the MYCN associated miRNA signature. This 15 miRNA expression signature requires validation in an independent tumor set, and represents a method of patient stratification that is independent of MYCN amplification status.
Under-expression of miRNAs in MNA tumors appears to occur twice as frequently as over-expression. This observation is consistent with our studies on a smaller set of miRNAs (n
and with observations on a related family member, c-MYC 
. One of the notable miRNAs under-expressed in MNA tumors was the large polycistronic miRNA 379–656 cluster; a 44.7 kb polycistron containing 38 miRNAs mapping to chromosome 14. Four miRNAs from this cluster were expressed at significantly lower levels in MNA tumors relative to non-MNA tumors. Gene ontology analysis of predicted gene targets for miRNAs from this cluster revealed significant over-representation of biological processes such as neurogenesis, embryonic development and transcriptional regulation 
Treatment regimens for patients with neuroblastoma categorized as high risk involve intensive, multi-modal chemotherapy. Despite the severity of treatment, in many cases they have only short-term effects, emphasizing the fact that accurate identification of low and high risk is of great importance. The identification of clinical relevant miRNAs is an initial step towards identifying miRNAs as significant prognostic markers in neuroblastoma with potential novel therapeutic applications for this disease.