miRNA expression of human samples of hepatoblastoma (HB) and normal and fetal liver

Forty-nine hepatoblastoma (HB) tumor specimens were profiled on a microarray containing 250human miRNA, in order to identify miR classifiers that could discriminate HBs with diverse prognosis, and to get insight on the functional mechanisms in which miRNAs are involved. Hierarchical unsupervised clustering of HB samples according to their miRNA expression profile identifies 2 tumor subgroups associated with different degree of differentiation, proliferation rate and invasive phenotype. As subset of these tumors is used as training set to build a mir classifier. When the signature is applied to an independent test set, an independent series of HB is classified patients into two groups associated with the clinical features identified in the first set of patients.