Brain tumors are an important cause of cancer morbidity and mortality, particularly in children, where it will soon surpass leukemia as the leading cause of pediatric cancer-related deaths in the United States. Although the actual incidence of primary brain tumors is relatively low compared with the more common epithelial tumors, their significant contribution to the overall cancer mortality in this country is a testament to their lethality. Therapeutic advances for primary brain tumors have come painstakingly slowly as a result of both their intrinsic resistance to standard cytotoxic therapies as well as their anatomical location within the exquisitely sensitive tissue of the central nervous system (CNS), limiting the extent to which surgical resection and radiation therapy can be safely accomplished. Medulloblastomas, the most common malignant brain tumors in children, are embryonal tumors (formally described as primitive neuroectodermal tumors, or PNETs) of the cerebellum and account for approximately 20% of all childhood brain tumors. These tumors are thought to arise from either the neural stem cells of the subependymal zone of the fourth ventricle or from the granule cell progenitors (GCPs) destined to form the external granule layer of the cerebellum, where they will ultimately form the granule cells that help regulate the activity of cerebellar purkinje cells. These tumors are highly aggressive and, in contrast to the more common gliomas, have a high propensity for spreading along cerebral spinal fluid pathways and metastasizing widely both within the CNS and systemically. Although many patients who develop medulloblastoma can be cured of their disease with aggressive surgery, cranial spinal radiation, and chemotherapy, nearly half of all children afflicted with this tumor still die of their disease. Furthermore, subgroups of patients, such as the very young (< 2 years of age) and those who present with metastatic disease, have particularly poor prognoses. Finally, many of the children who are cured of their tumors by aggressive multimodality therapy are left with lifelong physical and neurocognitive deficits.

Understanding the molecular pathogenesis of a disease offers an opportunity for identifying novel signal transduction pathways, resulting in the discovery of novel therapeutic targets. Better yet, if the pathogenic mechanism (s) of an illness can be shown to be an environmental insult (ie, infection), one can envision potential preventative approaches to the disease. Thus, the study by Del Valle et al.(1) in this issue should be of interest for those interested in neuro-oncology as well as those interested in cancer pathogenesis. Del Valle et al. demonstrate the presence of JC virus (JCV) T antigen (Tag), and agnoprotein DNA in 65% and 69% of 20 and 16 medulloblastoma specimens, respectively. They further demonstrate the presence of agnoprotein and Tag in approximately half of these tumor specimens, although not always within the same tumor. The authors conclude that the finding of agnoprotein expression in the absence of Tag expres-