Objective: Endometrioid endometrial carcinoma is caused by a combination of mutational events and hormonal factors. We used large-scale messenger RNA expression analysis to discover genes that distinguish neoplastic transformation and examine the patterns of tumor expression of those genes which are normally regulated during the menstrual cycle.
Methods: Expression of approximately 6000 unique genes was quantified in 4 normal (2 proliferative, 2 secretory) and 10 malignant endometria using Affymetrix Hu6800 GeneChip probe arrays. Expression differences between normal and malignant tissue groups were measured by a test of statistical significance comparing the individual t statistic for each gene to the distribution of maximum t statistics among all genes following 1001 permutations of the tissue group assignments (Permax test). Hormonally responsive genes, selected by comparison of proliferative and secretory subsets of normal endometria using a combination of filters applied to the group means and t test rankings, were then examined in the tumors.
Results: Fifty genes with a Permax <0.50 provided excellent discrimination between normal and malignant groups and were predominantly characterized by diminished expression levels in the cancers. We found that 100 genes which are hormonally regulated in normal tissues are expressed in a disordered and heterogeneous fashion in cancers, with tumors resembling proliferative more than secretory endometrium.
Conclusion: Neoplastic transformation is accompanied by predominant loss of activity of many genes constitutively expressed in normal source tissues and absence of expression profiles which characterize the antitumorigenic progestin response.
Copyright 2001 Academic Press.