Genome-wide molecular characterization of central nervous system primitive neuroectodermal tumor and pineoblastoma

Abstract

Central nervous system primitive neuroectodermal tumor (CNS PNET) and pineoblastoma are highly malignant embryonal brain tumors with poor prognoses. Current therapies are based on the treatment of pediatric medulloblastoma, even though these tumors are distinct at both the anatomical and molecular level. CNS PNET and pineoblastomahave aworse clinicaloutcomethanmedulloblastoma; thus, improved therapies based on an understanding of the underlying biologyofCNSPNETandpineoblastoma are needed. To this end, we characterized the genomic alterations of 36 pediatric CNS PNET s and 8 pineoblastomas using Affymetrix single nucleotide polymorphism arrays. Overall, the majority of CNS PNET s contained a greater degree of genomic imbalance than pineoblastomas, with gain of 19 p (8 [27.6%] of 29), 2 p (7 [24.1%] of 29), and 1 q (6 [20.7%] of 29) common events in primary CNS PNETs. Novel gene copy number alterations were identified and corroborated by Genomic Identification of Significant Targets In Cancer (GISTIC) analysis: gain of PCDHGA3, 5 q31.3 in 62.1% of primary CNS PNET s and all primary pineoblastomas and FAM129A, 1 q25 in 55.2% of primary CNS PNET s and 50% of primary pineoblastomas. Comparison of our GISTIC data with publically available data for medulloblastoma confirmed these CNS PNET-specific copy number alterations. With use of the collection of 5 primary and recurrent CNS PNET pairs, we found that gain of 2 p21 was maintained at relapse in 80% of cases. Novel gene copy number losses included OR4C12, 11p11.12 in 48.2% of primary CNS PNETs and 50% of primary pineoblastomas. Loss of CDKN2A/B (9p21.3) was identified in 14% of primary CNS PNET s and was significantly associated with older age among children (P 5=05). CADPS, 3 p14.2 was lost in 27.6% of primary CNS PNET s and was associated with poor prognosis (P 5=043). This genome-wide analysis revealed the marked molecular heterogeneity of CNS PNET s and enabled the identification of novel genes and clinical associations potentially involved in the pathogenesis of these tumors. © 2011 The Author(s).