Prediction of chromosomal aneuploidy from gene expression data

Chromosomal aneuploidy is commonly observed in neoplastic diseases and is an important prognostic marker. Here we examine how gene expression profiles reflect aneuploidy and whether these profiles can be used to detect changes in chromosome copy number. We developed two methods for detecting such changes in the gene expression profile of a single sample. The first method, fold-change analysis, relies on the availability of gene expression data from a large cohort of patients with the same disease. The expression profile of the sample is compared with that of the dataset. The second method, chromosomal relative expression analysis, is more general and requires the expression data from the tested sample only. We found that the relative expression values are stable among different chromosomes and exhibit little variation between different normal tissues. We exploited this novel finding to establish the set of reference values needed to detect changes in the copy number of chromosomes in a single sample on the basis of gene expression levels. We measured the accuracy of the performance of each method by applying them to two independent leukemia datasets. The second method was also applied to two solid tumor datasets. We conclude that chromosomal aneuploidy can be detected and predicted by analysis of gene expression profiles. This article contains Supplementary Material available at http://www.interscience.wiley.com/jpages/1045-2257/suppmat.