Metastatic medulloblastoma is a malignant tumor of the developing CNS with little hope for a permanent cure. A report by McDonald et al offers alternatives by identifying potential therapeutic targets using expression array analysis.1,2
The last three decades have seen impressive advances in the successful treatment of many common childhood malignancies such as acute lymphoblastic leukemia. This success has unfortunately not been paralleled for malignant brain tumors.3 The most common of these, the cerebellar medulloblastoma, exhibits a tremendous metastatic potential: up to 50% of patients present with metastatic disease at the time of diagnosis (see Figure 1).4 While the spread of malignant brain tumors is usually limited to the CNS, medulloblastomas can aggressively metastasize into extraneural tissues.5 Due to its prognostic relevance, the metastatic status is used for stratification of patients into risk groups.3
The treatment of patients with standard risk tumors, ie those without metastases has been rather successful with survival rates of up to 78%.4,6 In contrast, the cure of metastatic disease has until recently been limited to single cases.
In metastatic medulloblastoma standard chemotherapy regimens combined with neuraxis radiotherapy have shown only temporary response. In younger children radiotherapy is associated with unacceptable neurotoxicity and attempts at delaying the time to radiotherapy are often futile. Survival rates for this group of patients have been close to zero and any relapse has been associated with an inevitably fatal outcome.4
This grim scenario has to some extent changed with the introduction of high dose chemotherapy protocols supported by autologous stem cell rescue.7
Even though promising, all of the currently successful treatment options for high-risk medulloblastoma are associated with neural and neuroendocrine side effects. Many follow-up studies demonstrate a tremendous decline in quality of life among survivors of medulloblastoma.8
To avoid toxicity to the developing brain, therapeutic avenues targeting the underlying molecular lesions have to be explored. The search for the underlying genetic lesions in medulloblastoma has been long, however, knowledge has come only slowly.2 Genomic changes, such as the formation of an isochromosome 17q and a loss of heterozygosity distal to 17p13.1 (distal to the TP53 locus) have been known for over 10 years,9 however a proposed tumor suppressor gene in the region has not yet been identified. Table 1 recognizes some of the genes that have been identified in recent years as showing genetic differences between normal (cerebellar) and medulloblastoma tissue.2,9
Recently medulloblastomas have been extensively analyzed by novel genome-wide screening methodologies ranging from comparative genome hybridization over methylation scans to expression profiling.1,2,9,10 Expression profiling has stirred high hopes among clinicians and scientists alike as it can rapidly identify expression differences in multiple genes, which then may serve as potential therapeutic targets.
MacDonald et al compared expression arrays of 10 metastatic medulloblastomas with 13 non-metastatic tumors.1 A total of 85/1992 transcripts were differentially expressed in metastatic tumors, 59 of these up- and 26 downregulated. Most of the upregulated genes play fundamental roles in mechanisms important to metastasis, including angiogenesis, growth factor signaling, cell-adhesion and invasion. Two of the upregulated genes, platelet derived growth factor alpha (PDGFRA) and secreted protein acidic and rich in cysteine (SPARC) were further analyzed, since they implicated a novel pathway in metastasis of medulloblastomas. Surprisingly, differential protein expression in metastatic vs non-metastatic tumors could be established using immunohistochemistry and the authors found that blocking of the PDGFA signal or its downstream effector pathways in vitro inhibited cell adhesion and migration, two key components of metastatic spread.
This study is a milestone in medulloblastoma research in several ways. In contrast to previous approaches using unselected tumors comparing them to 'normal' tissue, MacDonald et al used specifically metastatic tumors and compared them to non-metastatic ones. Medulloblastoma is a rare tumor affecting up to 6/million children in the US each year. Thus as most studies in the field, this one also suffers from the rather low sample numbers. Furthermore these data represent a thus far unique resource for future pharmacogenetic research in medulloblastomas with many surprises still to be uncovered.
Intriguingly none of the previously described 'prognostic' genes MYC, TRKC and ERBB show any upregulation.2,3 This is in clear contrast to published data and deserves further validation in larger sample sets.
In the future it will be highly desirable to combine molecular analysis modalities. For instance several of the identified overexpressed genes are located on chromosome 7 (eg HOX1, RPA . . .), which is often present in three copies in medulloblastomas. A combination approach may help to determine a rank order of target genes, by separating primary from secondary lesions.
The finding of overexpression of PDGFRA and other downstream effectors of the RAS protooncogene pathway is a most remarkable point. PDGFRA, a receptor tyrosine kinase, has previously been implied in the adult counterpart of medulloblastomas, glioblastomas, and indeed oral treatment of mice with implanted glioblastoma cells using a tyrosine kinase inhibitor (STI 571) inhibited growth.11 Molecular therapeutics, like STI 571 or the ras-pathway inhibitors, have become fashionable due to their limited toxicity and a potentially high specificity.12 The expression of PDGFRA had previously been examined in medulloblastomas with no clear finding. Additionally overexpression does not reflect the functionality of the receptor and may not suffice as an indicator for in vivo treatment responses.13 Inhibition of the downstream pathway of PDGFRA (the MAPK) pathway might also be an oversimplistic approach, as an abundance of stimuli other than PDGFRA transmit their signals through this pathway.14
Nevertheless this study demonstrates that we are not far from individualized treatment protocols. Soon molecular lesion profiles will be established for individual tumors and clinicians may then adjust their therapeutics according to the profile. Unfortunately, metastatic medulloblastomas may take a little longer. For one, most of the molecular lesions against which the currently available drugs are targeted were elucidated 10-20 years ago, when virtually nothing was known on the molecular pathology of medulloblastoma. Secondly, it is rather likely that some of the key features of a medulloblastoma molecular lesion profile are undiscovered and will differ from adult tumors.
Despite the increasing research activities on metastatic medulloblastoma it will take years until solid molecular profiles will be available. Until then we welcome every potentially efficient therapeutic as an adjunct to current regimens.
| 
e 33, 48149 Münster, Germany. Tel: (0251) 83 47742 Fax: (0251) 83 47828 E-mail: 