Abstract 117: jackTom, an algorithm to integrate aCGH and gene expression microarrays to predict activation of oncogenes by DNA amplification in individual tumors

Abstract

Abstract Oncogene over-expression due to DNA amplification is a frequent type of oncogene activation event that is responsible, in part, for tumorigenesis. The determination of which oncogenes in a tumor are genetically activated by DNA amplification can be predicted by the integration of aCGH and gene expression microarray data. Several methods have been previously implemented that combine aCGH and gene expression microarray data, but all of these methods focus on the gene expression changes in genomic regions with the most frequent DNA amplifications in a set of tumor samples. The implementation of personalized cancer therapy will require methods that predict which oncogenes are activated in a specific tumor sample by analyzing genes in all amplified regions. Here, we describe a method and R package for the Joint Analysis of Chromosomes and Karyotypes with Transcription On Microarrays (jackTom) that provides predictions on which genes are over-expressed due to DNA amplification in individual tumors by analyzing all amplicons. The analysis requires a collection of aCGH and gene expression arrays from a set of tumors. First, regions of genetic alteration are identified from aCGH microarrays with circular binary segmentation (DNAcopy package) using a copy number amplification minimum of 1.75. Additionally, regions lacking genetic amplification or genetic loss are also mapped using the same procedure. The expression of amplified genes in a specific tumor are compared to tumors without genetic alterations in the amplified region. The significance of the gene expression alterations in amplified regions is determined using either a Z-score or a modified ANOVA (limma package). The jackTom package was used to analyze the 8p11-p12 region in 10 human breast cancer (HBC) cell lines and 19 HBC tumor samples. This genetic region is amplified in approximately 15% of HBC. Results from jackTom were subsequently compared with a multiple regression analysis of the genes in this amplicon (performed by globalTest). In the 29 HBC samples analyzed, 9 samples exhibited chromosome 8p11-p12 amplification of at least 1.75 fold. Sixty one genes that were amplified in at least two samples spanned bases 31.01 Mb to 48.81 Mb. GlobalTest analysis demonstrated that 20 genes had expression changes that were significantly altered by the amplification event. Nevertheless, jackTom analysis of each sample demonstrated that MYST3 is significantly over-expressed in 3/8 tumors harboring an amplification (p less than or equal to 0.024, 0.010 and 0.0053, respectively) and DKK4 is significantly over-expressed in 1/6 tumors harboring an amplification (p less than or equal to 0.0074). Thus, jackTom analysis demonstrates that frequently amplified regions select for genetic activation of different candidate oncogenes in individual tumors. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 117.