For reprint orders, please contact reprints@expert-reviews.com

491www.expert-reviews.com ISSN 1473-7175© 2011 Expert Reviews Ltd

Key Paper Evaluation

10.1586/ERN.11.22

Medulloblastoma, the most common primary malignant brain tumor of childhood, has been an enigmatic disease for clinicians and biolo-gists alike [1,2]. It has a high propensity to spread along cerebrospinal fluid pathways, with lepto-meningeal dissemination observed in up to 30% of patients at initial diagnosis [1,3]. Incremental advances in neurosurgical techniques, improved radiotherapy planning and delivery, and new chemotherapeutic agents have vastly improved prognosis [4–6] for this disease that was hitherto considered largely incurable a few decades ago. However, such aggressive multimodal therapy results in considerable long-term morbidity almost universally in survivors, particularly in growing children [4,7,8].

Complex regulatory mechanisms involving core developmental pathways facilitate the pro-liferation and differentiation of the developing cerebellum. It is believed that dysregulation of such normal developmental processes are a key mechanism of tumorigenesis in at least a subset of medulloblastomas [2,9,10]. Several biological pathways have been implicated in the develop-ment of medulloblastoma; among these, the wingless (WNT) and sonic hedgehog (SHH) signaling pathways have been the most con-sistently and robustly identified [11]. Evidence from clinico-pathological and biological studies has increasingly validated the hypothesis that medulloblastoma is a heterogeneous disease with subtypes having different developmental

Tejpal Gupta1 and Rakesh Jalali†1

1Neuro-Oncology, Tata Memorial Centre, Dr Ernest Borges Road, Parel, Mumbai 400 012, India †Author for correspondence:Tel.: +91 222 417 7153 Fax: +91 222 414 6937 rjalali@tmc.gov.in

Evaluation of: Northcott PA, Korshunov A, Witt H et al. Medulloblastoma comprises four distinct molecular variants. J. Clin. Oncol. DOI: 10.1200/JCO.2009.27.4324 (2010) (Epub ahead of print).

Medulloblastoma, the most common primary pediatric malignant brain tumor, is a molecularly heterogeneous disease with different developmental origins, distinct phenotypes, diverse biological behavior and contrasting clinical outcomes. The current clinico-radiological risk classification fails to take account of this heterogeneity and resultant prognostic variability. It is widely accepted that dysregulation of normal developmental processes constitutes a key mechanism of tumorigenesis in at least a subset of medulloblastomas. Several attempts at biological classification have successfully identified distinct subgroups with subgroup-specific gene signatures, demographics, histologic subtypes and metastases; among these, tumors involving the wingless and sonic hedgehog signaling pathways have been the most reliably and consistently identified. However, such integrative genomic approaches have limited applicability in the clinic owing to the need for fresh frozen tissues and elaborate molecular biology tools. A novel four-antibody approach to subgroup medulloblastoma using immunohistochemistry on archival specimens as proposed by Northcott et al. appears extremely promising as it can be applied in any basic neuropathology laboratory across the globe. There is a compelling need to integrate assays of molecular biomarkers performed on archival specimens into stratification schemes for medulloblastoma alongside clinical and pathologic outcome indicators to refine risk stratification for individualizing therapy.

Keywords: antibody array • developmental pathways • genomics • heterogeneity • immunohistochemistry • medulloblastoma • risk stratification

Molecular biology of medulloblastoma: bridging the gap between research and practiceExpert Rev. Neurother. 11(4), 491–494 (2011)

THeMed ArTICLe y CNS neoplasms

Expert Rev. Neurother. 11(4), (2011)492

Key Paper Evaluation Gupta & Jalali

origins, distinct phenotypes, diverse biological behavior and con-trasting clinical outcomes [10–12]. Children >3 years of age with gross total resection and no evidence of leptomeningeal dissemi-nation have traditionally been classified as having average-risk disease, while the presence of residual tumor >1.5 × 1.5 cm2, an age of <3 years or leptomeningeal metastases are consid-ered high-risk features, either singly or in combination [12,13]. Intensification of therapy for patients with high-risk disease and reduced-intensity treatment for low- or average-risk patients is a widely accepted and logical strategy to optimize therapeutic index. Although generally simple and useful, this clinical risk-stratification system does not adequately account for hetero-geneity in the biological behavior of the disease and existent prognostic variability. Developing an improved classification system for guiding risk-adapted therapy and assigning prognosis has been an active area of medulloblastoma research [12].

Methods & resultsThe authors integrated genome-wide DNA copy number and mRNA expression profiles from 103 primary medulloblastoma samples to generate a novel molecular classification system using a robust biostatistics and bioinformatics approach [14]. Unsupervised hierarchical clustering of medulloblastoma gene-expression data identified four unique clusters. Subgroup-specific signature genes were identified by a multivariate permutation test restricted on the proportion of false discoveries and one-way ana lysis of vari-ance. Bootstrap support ana lysis of clustering data revealed high confidence (>96%) for these four unique subgroups: WNT, SHH, group C and group D. Separation among the four subgroups was further assessed via principal component ana lysis. WNT and SHH tumors showed clear separation from each other, as well as group C and D tumors, while group C and D tumors shared few similarities. Supervised analyses of these four subgroups revealed significant subgroup-specific demographics, histology, metastatic status and DNA copy number aberrations. Desmoplastic medul-loblastomas were predominantly found in the SHH subgroup, although they were also seen in group C and D. Large-cell/ana-plastic histology was seen in SHH, group C and group D tumors. Known targets of the WNT and SHH pathways were significantly differentially expressed in their respective subgroups. Elevated MYC family expression was lacking in group D. Monosomy 6 and 9q deletion was confined to WNT and SHH, respectively, while isochromosome 17q was limited to groups C and D. SHH-driven tumors were commonly seen either in infants (<3 years) or adults (>16 years). Group C tumors demonstrated a peak in early childhood (3–10 years) and were virtually never seen in older children. WNT and group D tumors were evenly distributed throughout all age groups. Approximately 70% of males belonged to groups C and D, compared with 42% females. Conversely, 47% of females had SHH-driven tumors compared with 24% of males. The incidence of WNT tumors was almost double in females (11%) compared with males (6%).

Prediction ana lysis of microarrays was performed by applying an 84-gene classifier of medulloblastoma subgroup-signature genes derived from the current data set (training set) to a published

data set of 62 medulloblastomas from Kool et al. [15] (test set) dividing them into the same four subgroups. Subgroups C and D of the Kool data set corresponded with subgroup D of the current data set. Comparison of the 20 most highly differen-tially expressed genes discriminating between subgroups C and D in both data sets demonstrated that 38 out of 40 genes were concordantly differentially expressed. Non-negative matrix factorization on the expression data set strongly supported the existence of four medulloblastoma subgroups. WNT and SHH tumors showed over-representation of genes involved in axonal guidance, while groups C (phototransduction and glutamate sig-naling) and D (semaphoring, cAMP and b-adrenergic receptor signaling) tumors were characterized by over-representation of pathways involved in neuronal development. The authors further selected highly expressed, subgroup-specific signature genes based on microarray data and commercial availability of good-quality antibodies. Staining two separate medulloblastoma tissue micro-arrays using these antibodies, as well as b-catenin and GLI1, demonstrated robust subgroup-specific staining. Approximately 98% of 294 tumor samples stained positive for a single marker protein. Immunohistochemistry (IHC) for DKK1 (WNT), SFRP1 (SHH), NPR3 (group C) and KCNA1 (group D) reli-ably and uniquely classified formalin-fixed paraffin-embedded medullo blastomas. Demographics for the tissue microarray patients validated results from expression ana lysis. Metastases and large-cell/anaplastic histology was significantly associated with group C tumors. Survival ana lysis from 287 patients dem-onstrated that group C tumors had the worst prognosis regardless of metastases. A multivariate ana lysis evaluating age, M stage, his-tology, extent of resection and molecular subgroup revealed that large-cell/anaplastic histology and group C subgroup were inde-pendently associated with poor outcome, with group C anplastic tumors faring the worst.

Discussion & significanceIndividualization of therapy for children with brain tumors in general and medulloblastoma in particular has been a major challenge for the pediatric neuro-oncology community in the current era of personalized medicine. There have been several previous attempts at a molecular genetic classification of medullo-blastoma [16,17] with potential for refined risk stratification. While the number of subgroups have generally varied from four to six, all of the authors have identified WNT and SHH subgroups reliably and consistently. Thompson and colleagues were the first to report five distinct molecular subgroups (A to E) using gene-expression profiling in a set of 46 medulloblastoma samples [18]. Subgroup-specific abnormalities included mutations in WNT signaling and chromosome 6 deletion (subgroup B) and SHH pathway (sub-group D). Reverse-transcriptase PCR and IHC confirmed selec-tive upregulation of genes within the WNT and SHH pathways in group B and D tumors, respectively.

Kool et al. studied mRNA expression profiles in a set of 62 sam-ples to gain deeper insights into the molecular biology of medul-loblastoma [15]. Genomic abnormalities were further characterized in a subset of these using comparative genomic hybridization

www.expert-reviews.com 493

Key Paper EvaluationMolecular biology of medulloblastoma

arrays. Five distinct molecular subtypes with characteristic gene signatures, specific clinico-pathologic features and distinctive genetic aberrations were identified, characterized by WNT sig-naling (group A), SHH signaling (group B), expression of neu-ronal differentiation genes (groups C and D) or photo receptor genes (groups D and E). Bootstrap analyses supported differ-ences between all five clusters and was highly robust for clusters A (WNT) and B (SHH) but weaker for clusters C, D and E.

Using an integrative genomics approach, Pomeroy and colleagues profiled the mRNA transcriptome of 194 tumor samples and performed high-density SNP array and miRNA ana lysis [19]. An unsupervised clustering algorithm based on non-negative matrix factorization identified six molecular subgroups of medulloblas-toma (c1 to c6). Silhoutte width ana lysis suggested that samples within each subgroup were highly coherent. Gene set enrichment ana lysis showed marked enrichment of MYC and related transcrip-tional/ribosomal genes in subgroup c1. Subgroups c2 and c4 were characterized by neuronal differentiation markers, while the c5 subgroup was enriched for photoreceptor transcriptional programs. The c3 subgroup demonstrates enrichment of gene sets associated with SHH signaling, while subgroup c6 was enriched with TGF-b, WNT/b-catenin and epithelial–mesenchymal transition gene sets. The current study is ample testimony and lends further credence to the existence of molecular heterogeneity in medulloblastoma.

Expert commentary & five-year viewWith our ever-increasing understanding of molecular biology of medulloblastoma demonstrating significant genetic heterogeneity and prognostic variability, it is believed that the current clinical risk stratification with its inherent limitations will be replaced by

a robust integrated clinico-patho-biological classification with the potential to guide therapeutic decision-making and predict out-comes. The main drawback of previous approaches has been the need for fresh-frozen tumor tissue for genomic analyses using com-plex molecular biology tools mandating highly skilled personnel, a high level of experience and extensive standardization, which may not be routinely available in large parts of the world. This novel four-antibody approach by Northcott et al. to subgroup medullo-blastoma using IHC on formalin-fixed paraffin-embedded blocks is extremely promising as it can be applied across the globe in basic neuropathology laboratories without requiring elaborate and expensive molecular biology techniques [14]. However, stand-alone molecular classification systems completely disregarding clinical or pathological features are unlikely to be used routinely in the clinic. There is a compelling need to integrate assays of molecular biomarkers performed on archival specimens into stratification schemes for medulloblastoma alongside clinical and pathologic outcome indicators to refine risk stratification for individualizing therapy [20]. The suggested classification would need validation in appropriate prospective clinical trials to bridge the gap between biological research and clinical practice that can help translate novel scientific insights to inform evidence-based decision-making.

Financial & competing interests disclosureThe authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

Key issues

• Medulloblastoma is the most common primary malignant brain tumor of childhood, with a high propensity for leptomeningeal dissemination.

• Dysregulation of normal developmental processes constitutes a key mechanism in the tumorigenesis of at least a subset of medulloblastomas, with aberrations in WNT and SHH signaling identified consistently and robustly.

• Evidence from clinico-pathological and biological studies has validated the hypothesis that medulloblastoma is a heterogenous disease.

• The study under evaluation by Northcott et al. uses an integrative genomics approach to classify medulloblastomas into four distinct molecular subtypes with subgroup-specific demographics, histology, metastasis and DNA copy number aberrations. More importantly, the authors demonstrate a pragmatic immunohistochemistry-based ana lysis of archival specimens using four antibodies to uniquely assign medulloblastoma to respective subgroups, making it widely applicable in the clinic.

• There is a need to validate and integrate assays of molecular biomarkers into stratification schemes alongside clinico-pathologic outcome indicators to refine risk stratification in medulloblastoma for individualizing therapy.

ReferencesPapers of special note have been highlighted as:• of interest

1 Bailey PC. Medulloblastoma cerebelli: a common type of midcerebellar glioma of childhood. Arch. Neurol. Psychiatry 14, 192–224 (1925).

2 Pizer B, Clifford S. Medulloblastoma: new insights into biology and treatment. Arch. Dis. Child Educ. Pract. Ed. 93(5), 137–144 (2008).

3 Koeller KK, Rushing EJ. From the archives of the AFIP: medulloblastoma: a comprehensive review with radiologic-pathologic correlation. Radiographics 23(6), 1613–1637 (2003).

4 Gottardo NG, Gajjar A. Current therapy for medulloblastoma. Curr. Treat. Options Neurol. 8(4), 319–334 (2006).

5 Crawford JR, MacDonald TJ, Packer RJ. Medulloblastoma in childhood: new biological advances. Lancet Neurol. 6(12), 1073–1085 (2007).

6 Rutkowski S, von Hoff K, Emser A et al. Survival and prognostic factors of early childhood medulloblastoma: an international meta-analysis. J. Clin. Oncol. 28(33), 4961–4968 (2010).

7 Mulhern RK, Reddick WE, Palmer SL et al. Neurocognitive deficits in medulloblastoma survivors and white matter loss. Ann. Neurol. 46(6), 834–841 (1999).

8 Fossati P, Ricardi U, Orecchia R. Pediatric medulloblastoma: toxicity of current

Expert Rev. Neurother. 11(4), (2011)494

Key Paper Evaluation Gupta & Jalali

treatment and potential role of protontherapy. Cancer Treat. Rev. 35(1), 79–96 (2009).

9 Srivastava VK, Nalbantoglu J. The cellular and developmental biology of medulloblastoma: current perspectives on experimental therapeutics. Cancer Biol. Ther. 9(11), 843–852 (2010).

10 Parsons DW, Li M, Zhang X et al. The genetic landscape of the childhood cancer medulloblastoma. Science 331(6016), 435–439 (2011).

11 Gibson P, Tong Y, Robinson G et al. Subtypes of medulloblastoma have distinct developmental origins. Nature 468(7327), 1095–1099 (2010).

• Firstevidencethatdifferentsubtypesofmedulloblastomahavedistinctcellularorigins.Tumorscontainingactivatingmutationsinthewinglesspathwayariseoutsidethecerebellumfromcellsofthedorsalbrainstem,whilesonichedgehog-driventumorsoriginatefromgranuleneuronprecursorcellsandarelocatedwithinthecerebellarhemispheres.

12 Polkinghorn WR, Tarbell NJ. Medulloblastoma: tumorigenesis, current clinical paradigm, and efforts to improve risk stratification. Nat. Clin. Pract. Oncol. 4(5), 295–304 (2007).

13 Gilbertson R, Wickramasinghe C, Hernan R et al. Clinical and molecular stratification of disease risk in medulloblastoma. Br. J. Cancer 85(5), 705–712 (2001).

14 Northcott PA, Korshunov A, Witt H et al. Medulloblastoma comprises four distinct molecular variants. J. Clin. Oncol. DOI: 10.1200/JCO.2009.27.4324 (2010) (Epub ahead of print).

15 Kool M, Koster J, Bunt J et al. Integrated genomics identifies five medulloblastoma subtypes with distinct genetic profiles, pathway signatures and clinicopathological features. PLoS One 3(8), e3088 (2008).

• Providedarobustvalidationofthehypothesisthatmedulloblastomaisamolecularlyheterogeneousdisease.IdentifiedfivesubgroupscharacterizedbyWNTsignaling(groupA),SHHsignaling(groupB),expressionofneuronaldifferentiationgenes(groupsCandD)orphotoreceptorgenes(groupdDandE).

16 Pomeroy SL, Tamayo P, Gaasenbeek M et al. Prediction of central nervous system embryonal tumour outcome based on gene expression. Nature 415(6870), 436–442 (2002).

• Firstgrouptoclassifyembryonalbraintumorsanddemonstratemolecularheterogeneityofmedulloblastomabygene-expressionprofiling.Alsodemonstratedthatclinicaloutcomeofchildrenwithmedulloblastomacouldbeoptimallypredictedbyaneight-genemodel.

17 Northcott PA, Rutka JT, Taylor MD. Genomics of medulloblastoma: from Giemsa-banding to next-generation sequencing in 20 years. Neurosurg. Focus 28(1), E6 (2010).

18 Thompson MC, Fuller C, Hogg TL et al. Genomics identifies medulloblastoma subgroups that are enriched for specific genetic alterations. J. Clin. Oncol. 24(12), 1924–1931 (2006).

• Usedlarge-scalegenome-wideexpressionprofilingtodistinguishsubgroupsofmedulloblastomathatareenrichedforspecificgeneticalterations.

19 Cho YJ, Tsherniak A, Tamayo P et al. Integrative genomic analysis of medulloblastoma identifies a molecular subgroup That drives poor clinical outcome. J. Clin. Oncol. DOI: 10.1200/JCO.2010.28.5148 (2010) (Epub ahead of print).

• ProfilesthemRNAtranscriptomeandidentifiessixmolecularsubgroupswithuniquecombinationsofnumericalandstructuralchromosomalalterations,includingapreviouslyunrecognizedpoor-prognosismolecularsubgroupcharacterizedgeneticallybyc-MYCcopynumbergainsandtranscriptionallybyenrichmentofphotoreceptorpathways.

20 Ellison DW, Kocak M, Dalton J et al. Definition of disease-risk stratification groups in childhood medulloblastoma using combined clinical, pathologic, and molecular variables. J. Clin. Oncol. DOI: 10.1200/JCO.2010.30.2810 (2010) (Epub ahead of print).

• Clinical,pathologicalandmolecularvariableswerecombinedtodefineanovelriskstratificationsysteminchildhoodmedulloblastomawithsignificantlydifferentoutcomes.Low-riskdiseasewasdefinedasb-cateninnucleopositivetumorswithoutmetastasis.High-riskmedulloblastomawasdefinedastumorswithmetastaticdisease,largecell/anaplasticphenotypeorMYCamplification.