Gene expression profiling in human endometrial cancer tissue samples: utility and diagnostic value

OBJECTIVE: Recently, gene expression profiling techniques have been used on several human cancers to classify tumor subgroups with a specific biological behavior, which were previously undetected by the conventional histopathologic staging systems. In the current study, the clinical usefulness and prognostic value of gene expression profiling in human endometrial carcinomas were studied. METHODS: A macro cDNA array, containing cDNAs of 588 genes selected from different areas of cancer research, was used to generate gene expression profiles of tumor tissue samples. The gene expression profiles of 12 endometrial cancers, 3 benign (e.g. noncancer) endometrial tissue samples and 3 myometrial tissue samples, taken from human surgical specimen, were compared. RESULTS: The efficacy to generate a gene expression profile of these tissue samples was 77%. The RNA samples could be randomly taken from the tissue samples and were highly reproducible. Cluster analysis of gene expression profiles of the different samples showed that the benign endometrial and the myometrial samples clustered separately from the tumor samples, indicating that the gene expression profiles were tissue specific and not patient specific. Cluster analysis of the tumor samples revealed two distinct tumor clusters. Ranking of the tumors in the two clusters showed high similarity with the histopathologic classification [International Federation of Gynecology and Obstetrics (FIGO) grading]. CONCLUSION: Classification of endometrial tumors on basis of their gene expression profiles showed similarity with the FIGO grading system.