Background: Embryonal cancers show multiple chromosomal rearrangements, suggesting a complicated series of acquired alterations during malignant evolution. This is best exemplified in neuroblastoma, a common and often lethal cancer of early childhood, in which tumor genomics correlate with disease phenotype. Current data indicate that neuroblastoma represents a spectrum of diseases and distinct patient subsets exist based on tumor biological features. There are multiple regions of the genome that are commonly and recurrently aberrant in neuroblastoma, and we hypothesize that epigenetic events are very likely to play a key role in functional activation or inactivation of candidate genes at these loci. Methods: We quantified DNA methylation status at 1505 CpG sites from 807 genes in a representative set of 88 (76 stage 4, 12 stage 1) diagnostic neuroblastoma primary tumor samples and 9 control tissues (fetal brain and fetal adrenal tissues) using a SNP-based genotyping system of bisulfite-converted genomic DNA on universal bead arrays, for which genome-wide copy number data were available. We used a matrix of ß-values (quantitation of the methylation level at each CpG site) to identify CpG sites that show differential methylation. For each of the 807 genes on the chip, we performed a t-test to assess differential methylation based on binary DNA copy number status (gain vs no gain/loss vs no loss) . Results: A total of 221 of the 1505 CpG sites showed significant differential methylation in neuroblastoma compared to control tissues after Bonferroni correction (p<3.3 x 10-5), and 194 showed significant differential methylation when the Stage 4 samples were compared to the Stage 1 samples. We focused on regions of the genome showing large regions of hemizygous deletion in order to discover tumor suppressor genes. The most highly differentially methylated genes in regions of recurrent copy number aberration were APOA1 on 11q23 (p=0.0222), XPC (p= 1.44 x 10-5) on 3p25, and RASSF1 (p=1.33 x 10-4) on 3p21.3. Of these, both APOA1 and XPC exhibited lowered gene expression as determined by the Illumina Human-6 mRNA expression array in samples with a deleted copy of the gene (p=0.008 and p=8.21 x 10-5, respectively), while RASSF1 showed a trend towards significance (p=0.0878). Conclusions: Genome-wide methylation profiling can identify regional candidate tumor suppressor genes when integrated with other genome-scale experiments. APOA1 and XPC are novel candidate neuroblastoma suppressor genes that are currently undergoing further evaluation. DNA copy number alterations are markers for the location of hypo- and possibly hyper-methylation events, and that these will identify the critical oncogenes and tumor suppressors that can be exploited therapeutically.

Citation Information: In: Proc Am Assoc Cancer Res; 2009 Apr 18-22; Denver, CO. Philadelphia (PA): AACR; 2009. Abstract nr 5178.

100th AACR Annual Meeting-- Apr 18-22, 2009; Denver, CO