An integrated data analysis workflow for the characterization of genetic events in Glioblastoma multiforme using Affymetrix GeneChip® U133 plus 2 expression arrays, Mapping 100K arrays and RPCI BAC arrays
Glioblastoma multiforme (GBM) is the most common adult brain tumor, and despite treatment including surgical resection, radiation and chemotherapy, the median survival time remains less than 12 months. While evidence for a genetic progression of the disease has been shown, relatively few specific genes have been identified to date. Here we attempted to characterize genetic events in GBMs by the complementary use of gene expression profiling using the Affymetrix U133 plus 2 arrays, and LOH/Copy Number prediction using the Affymetrix Mapping 100K SNP array, and array comparative genome hybridization (aCGH) using RPCI BAC arrays. Large contiguous segments of losses and gains were defined at high resolution using aCGH and we found concordance between the prediction from SNP and BAC arrays. Regions showing well-characterized copy number predictions include 1q32.1 (MDM4), 4q12 (PDGFRA), 7p11.2 (EGFR), 9p21.3 (CDKN2A), 10q23.31 (PTEN) and 12q13.3 (CDK4). Due to the yet higher resolution provided by the SNP array, copy number prediction algorithm from the SNP array identified additional regions of micro-deletions that were not seen in the BAC arrays. Using the SNP array, previously unreported LOH events were identified. These LOH events include 11p, 12q, 14q, 15q and 19q, which can be seen as a single copy loss on the BAC array. In addition, the SNP array showed LOH events that were not associated with copy number reduction including regions of 3p, 3q, 9p, 17p and 20p. Using the Affymetrix U133 Plus 2 expression array we identified differentially expressed genes which are involved in GBM related pathways and regions of copy number change. We developed a data analysis workflow for the integration of different types of microarray datasets such as expression and Mapping arrays which are processed on the same tumor samples. The statistical analysis results from each type of microarray are combined to identify the dependency of gene expression, LOH and copy number change.