Meta-analysis identifies candidate key genes in endometrium as predictive biomarkers for clinical pregnancy in IVF

Abstract

Genetic factors in endometrium are likely to be involved in the embryo implantation failure (IF), one of the major limiting factors in the success of in vitro fertilization (IVF). In this study, we aimed to identify critical genes from the transcriptional profile for the establishment of the endometrial receptivity which supporting the normal pregnancy. Three GEO datasets, including 12 samples of IF and 12 samples of controls, were used for the meta-analysis. We identified 182 different expression genes (DEGs) by comparing IF with controls and present here the successful clustering according to sample type, not by the origin. The gene ontology (GO) enriched analysis demonstrated the significant downregulation in activation and regulation of inflammatory and immune response in IF patients. Furthermore, network analysis of down-regulated genes identified the significant hub genes containing GADD45A (growth arrest and DNA damage inducible alpha, Degree = 77), GZMB (granzyme B, Degree = 38) and NLRP2 (NLR family pyrin domain containing 2, Degree = 37). The lower expression of NLRP2, related to inflammatory responses with the most degree in the network, was validatied by other GEO data. Besides, it was confirmed that the NLRP2 could act as a predictor for pregnancy after IVF (AUC = 87.93%; sensitivity, 60.00%; specificity, 91.30% ). Our meta-analysis will help us to better understand the molecular regulation of endometrial receptivity, and guiding further line of treatment for IF during IVF.

Keywords: IVF; endometrial receptivity; implantation failure; meta-analysis; microarray.

Figure 1. Flowchart of the selected process of microarray datasets for the meta-analysis

Figure 2. Genes differentially expressed in endometrium between IF and control patients across three datasets

(A) Heat map representation of the DEGs between control and IF patients across different microarrays identified from the meta-analysis. Each color above represents a single dataset. The heat map was rescaled to prevent domination by study-specific effects. (B) Unsupervised clustering of the transcriptome of the DEGs in the three datasets.

Figure 3. GO enrichment analysis for the up- and down-regulated genes

Figure 4. The network analysis of down-regulated genes in endometrium of IF patients

(A) Network including 434 nodes and 484 edges. Red: down-regulated genes. Purple: interaction genes. (B) The network genes are enriched in T cell receptor signaling pathway. Red rectangles represent the genes in the network. Purple is the color of Kyoto Encyclopedia of Genes and Genomes (KEGG) database.

Figure 5. Evaluation of NLRP2 as predictive biomarker for clinical pregnancy in IVF

(A) Relative abundance of NLRP2 in endometrium of IF patients (red circles) compared with control (green circles) in samples from the datasets of GSE58144. A Student T test (two-tailed) was used to estimate the significance between IF and control patients. (B) ROC curves analysis for clinical pregnancy prediction by NLRP2 expression in endometrium.