Placenta inflammation is closely associated with gestational diabetes mellitus

Abstract

Objective: To investigate the potential role of placenta inflammation in gestational diabetes mellitus (GDM) and construct a model for the diagnosis of GDM.

Methods: In this study, transcriptome-wide profiling datasets, GSE70493 and GSE128381 were downloaded from Gene Expression Omnibus (GEO) database. Significant immune-related genes were identified separately to be the biomarkers for the diagnosis of GDM by using random forest model (RF), support vector machine model (SVM), and generalized linear model (GLM).

Results: RF was the best model and was used to select the four key immune-related genes (FABP4, DKK1, CXCL10, and IL1RL1) to diagnose GDM. A nomogram model was constructed to predict GDM based on the four key immune-related genes by using "rms" package. The relative proportion of 22 immune cell types were calculated by using CIBERSORT algorithm. Higher M1 macrophage ratio and lower M2 macrophage ratio in GDM placenta compared to normal patients were observed.

Conclusions: This study provides clues that inflammation was correlated with GDM and suggests inflammation may be the cause and also the potential targets of GDM.

Keywords: GDM; consensus clustering; immune-related genes; inflammation; nomogram; placenta.

Figure 1

Workflow of the research. Abbreviations are defined as follows, Gene Expression Omnibus database (GEO), differentially expressed gene (DEG), immune-related gene (IRG), protein-protein interaction (PPI), gestational diabetes mellitus (GDM), decision curve analysis (DCA).

Figure 2

Differentially expression analysis and protein-protein interaction analysis. A. The intersection of DEGs in GSE70493 dataset and IRGs downloaded from ImmPort contains 79 DIRGs. B. The PPI network analysis of the 76 DIRGs.

Figure 3

Gene Ontology and KEGG pathway enrichment analysis. A. Biological process (BP, up), cellular component (CC, middle), and molecular function (MF, low) analysis results of 76 DIRGs. B. Result of KEGG pathway enrichment analysis of the 76 DIRGs. Proportion of DIRGs are exhibited in the X-axis and different categories are shown in the Y-axis. The number of genes enriched in particular category are manifested by the size of the circle. The color of the circle denotes different properties.

Figure 4

Construction and assessment of RF, GLM and SVM model. A. Cumulative residual distribution map of the sample. B. Boxplots of the residuals of the sample. Red dot stands for root mean square of residuals. C. The importance of the variables in RF, GLM and SVM model.

Figure 5

Relative expression level of CXCL10, FABP2, DKK1, IL1RL1. A. Heat map of the expression pattern of CXCL10, FABP4, DKK1 and IL1RL1. B. The chromosomal locations of DKK1, ILRL1, FABP4 and CXCL10. C. The relative expression level of CXCL10, FABP4, DKK1 and IL1RL1 between GDM and non-GDM from GSE70493 dataset. D. Principal component analysis shows that the four genes aforementioned can clearly distinguished GDM and non-GDM.

Figure 6

Correlation among selected immune-related genes. A. The correlation among DKK1, ILRL1, CXCL9, HLA-DQA2, CXCL10 and FABP4. B. The correlation among CXCL10, FABP4, DKK1 and IL1RL1.

Figure 7

Construction and validation of a nomogram model for GDM diagnosis based on the training dataset GSE70493. A. Nomogram to predict the occurrence of GDM. B. Calibration curve to assess the predictive power of the nomogram model. C. DCA curve to evaluate the clinical value of the nomogram model. D. Clinical impact curve based on the DCA curve to assess the nomogram model.

Figure 8

Distribution of the immune cells in placenta. A. Characteristics of infiltrated immune cells. B. Differences of the infiltrate immune cells between the GDM group and non-GDM group.