kinase inhibitors tumors from both models also contained a high percentage of proliferating cells characterized by both Ki67 staining and a large number of mitotic cells and exhibited regions with substantial numbers of necrotic and apoptotic cells. The similarities between the mouse and human tumors are also highlighted by the gene expression microarray studies which found a 68.1% Pearson correlation in quartiled and grade adjusted gene expression. However, there are also several differences including: the presence of large nondividing ganglion-like cells in all the mouse tumors and larger, somewhat ganglion-like cells in only some of the human OSI-774 xenografts. In addition, the majority of the mouse tumors fail to macrometastasize beyond the local area. Eighteen genes were also differentially expressed in the human and mouse tumors. These differences may contribute to their tumorigenicity and/or metastatic abilities. For example the mouse tumors expressed higher levels of BRCA2, a known tumor suppressor gene in breast and ovarian cancer and involved in maintenance of genome stability, specifically the homologous recombination pathway for doublestrand DNA repair. We do not yet know if any of these 18 genes account for the differences in disease progression between the human and mouse neuroblastomas but our ultrasound guided engraftment procedure is suited for testing these hypotheses. Our ultrasound studies showed that 81% of initial tumors in the TH-MYCN mice are found surrounding or in the vicinity of the aorta in the paravertebral ganglia. The percentage of tumors in humans originating from the paraspinal ganglia is,60%. We speculate that this site in the mouse provides factors from the blood system and from the adjacent adrenal that can enhance tumor growth and progression. Surprisingly, none of the mouse NB tumors originated in the adrenal medulla itself as confirmed by ultrasonography, histology and electron microscopy. These results are in agreement with recent works observing initial tumor formation in the TH-MYCN mouse in early postnatal sympathetic ganglia and not in the adrenal. However, in a few cases we observed that the mouse tumors invaded the adrenal at later stages of tumor progression. This is in contrast to the human disease where approximately 40% of the patient tumors originate in the adrenal medulla. Recording the in vivo regression of a small group of TH-MYCN tumors by ultrasound and MRI opens the possibility of using this model to study this process and suggests these tumors may exhibit similarities to human stage 4S tumor which also spontaneously regress. Understanding the regression process in vivo could help identify the biological pathways that will facilitate tumor growth arrest and drive the tumor to complete regression.