Molecular pathogenesis of uterine smooth muscle tumors from transcriptional profiling

Abstract

Uterine smooth muscle tumors range from the very common benign leiomyoma to the uncommon, but frequently lethal, leiomyosarcoma. Morphological and clinical differences between these tumors are presumed to result from differences in gene expression. To test this hypothesis, RNAs from four normal uterine myometria, seven uterine leiomyomas, and nine uterine leiomyosarcomas were profiled using microarrays of oligonucleotides representing about 7,000 unique probe sets. RNAs whose levels distinguished any of the three sample types were selected by analysis of variance (ANOVA). The 153 (2.2% of the total) probe sets representing 146 unique genes with the highest test statistic selected for further analysis met minimum ratio and range thresholds between groups. Cluster analysis distinguished benign and malignant samples at the first node, and myometrium and leiomyoma were resolved in a secondary node. Downregulation of specific genes in uterine leiomyosarcoma was the most common pattern of differential gene expression selected by the three-way ANOVA. Four extrauterine leiomyosarcomas had profiles most similar to that of the uterine leiomyosarcomas. Functional analysis of the 146 genes did not reveal any strong biological theme. These genes were distributed throughout the genome, but there was slight overrepresentation of genes on 1p and 2q. These genes define a tumor signature for uterine smooth muscle neoplasia, and they suggest that the molecular pathways in leiomyoma and leiomyosarcoma are distinct.