Meta-analysis of global gene-expression profiles identify molecular signatures for histological subtypes of sarcomas

Abstract

SUMMARY Sarcomas are rare mesenchymal malignancies and comprise over 50 histological subtypes. Sarcomas are not well studied because the number of cases of individual sarcoma is low. The utilization of public data, such as gene expression data, may allow for improvement in the novel discovery of sarcoma. In this study, to gain insight into histological subtypes of sarcoma from a public database, we performed a meta-analysis of the gene-expression profiles by survey-ing the data deposited in the Gene Expression Omnibus database from 2001 to 2014. The gene-expression data for 10 sarcoma subtypes and the gene-expression profiles for 1002 cases were selected for comparative analysis. Genes with histology-oriented molecular signatures were identified, and the results were verified by functional validation using gene oncology analysis. Pathway analysis suggested the existence of differential biological processes among sarcoma subtypes. Furthermore, as an application of the sarcoma gene expression datasets used in this study, we investigated the gene expression patterns of the targets of pazopanib to predict the response of sarcoma to pazopanib. We found that the gene expression distribution patterns of targets of pazopanib were without distinction among 10 subtypes of sarcoma. Taken together, we identified the tissue-specific genes of 10 subtypes of sarcoma by bioinformatics analysis; our results demonstrated the utility of sarcoma datasets in public databases and provide valuable information for future rare cancer research.