Frontiers in Oncology (May 2019)
A Novel Set of WNT Pathway Effectors as a Predictive Marker of Uterine Corpus Endometrial Carcinoma–Study Based on Weighted Co-expression Matrices
Abstract
Uterine corpus endometrial carcinomas (UCEC) are clinically divided into two subgroups—endometrioid endometrial carcinoma (EEC) or non-endometrioid endometrial carcinoma (NEEC). The first group shows relatively better prognosis. However, the discrimination seems to be insufficient due to the fact that in the mildest EEC are patients with poor treatment response and bad prognosis. Our aim was to examine the molecular background of such phenomenon and whether gene expression patterns might be of importance for the clinic. We focused our analysis on WNT pathway target genes since it is one of the main regulators of endometrial proliferation and differentiation. In silico analysis of TCGA data, including Weighted Co-expression Network Analysis, Principle Component Analysis, and Multiple Factor Analysis, allows to select 28 genes that serve as a predictive markers for UCEC patients. Our study revealed that there is a subgroup of the endometrioid cases that molecularly resembles mixed/serous groups. This may explain the reason for existence of subgroup of patients, that although clinically diagnosed with the mildest endometrioid UCEC type, yet present failure in treatment and aggressive course of the disease. Our study suggests that worse outcome in these patients may be based on a disruption of proper WNT signalling pathway resulting in deregulation of its effector genes. Moreover, we showed that mixed group consisting of tumours containing both endometrioid and serous types of cells, has serous expression profile of WNT targets. The proposed gene set allows to predict progression of the disease trough dividing patients into groups of low or high grade with 70.8% sensitivity and 88.6% specificity (AUC = 0.837) as well as could predict patient prognosis associated with UCEC subtype with 70.1% sensitivity and 86.2% specificity (AUC = 0.855). Relatively small number of implicated genes makes it highly applicable and possibly clinically simple and useful tool.
Keywords
