Background/Purpose: Rheumatoid arthritis (RA), while still an incurable disease, can often improve naturally during pregnancy. The specific mechanism(s) underlying this pregnancy-induced improvement, however, are not known. Our goal was to use our prospective pregnancy cohort (a) to identify genes whose expression are modulated during pregnancy in healthy women and in RA, as the disease improves, and (b) to investigate their functional enrichment using co-expression network analysis.

Methods: Data and samples were collected at pre-pregnancy (T0) and each trimester (T1, T2, T3) from women with RA who improved during pregnancy (n=14) and healthy women (n=15) enrolled in our prospective pregnancy cohort. Total RNA samples from whole blood were sequenced (RNA-seq). Following alignment (HISAT2) and transcript assembly (StringTie), gene-level expression was quantified (featureCounts) and normalized (edgeR). Cell type proportions were inferred through deconvolution (CIBERSORTx). Generalized Estimating Equation (GEE) models were used to assess modulation of gene expression across pregnancy in each group of women, adjusting for changes in cell type proportions across time points. Thresholds of FDR< 0.05 and fold-change >=2 (vs T0) were used to determine significance. Functional enrichment analysis (Webgestalt, Enrichr) and co-expression network analysis (WGCNA) were also performed.

Results: Temporal changes were observed in the proportions of neutrophils, resting NK cells, and resting CD4 memory T cells among healthy women only, and of monocytes among the women with RA. Numerous genes (334 protein-coding and 115 non-coding) were significantly up- or down-regulated during pregnancy among the women with RA, some demonstrating significant changes starting from early pregnancy (T1: n=26) or mid-pregnancy (T2: n=262) or during late-pregnancy (T3: n=161). Of those, 116 demonstrated similar temporal expression patterns among healthy women; another 112 genes were modulated during pregnancy only in the healthy group. The genes whose expression were modulated during pregnancy as RA improved were enriched in specific co-expression network modules related to neutrophil activation (upregulated in early pregnancy), binding of erythroid transcription factors (upregulated in mid-pregnancy), immunoglobulin constant and variable chain genes (downregulated in mid-pregnancy), and translation (mostly ribosomal genes, upregulated in late pregnancy). Similar enrichment in co-expression modules were observed in healthy pregnancy, except for the module related to binding of erythroid transcription factors which seemed to be specific to the RA group. Additionally, a module related to interferon signaling was enriched in genes modulated only in healthy pregnancy.

Conclusion: In our pregnancy cohort, multiple co-expression modules overlapped between the RA and healthy pregnancies. However, there were also differences including a specific module related to binding of erythroid transcription factors that was unique to RA, and a module related to interferon signaling that was unique to healthy pregnancy. It remains to be investigated how those differences may influence improvement of the disease during pregnancy.

« Back to ACR Convergence 2021

ACR Meeting Abstracts - https://acrabstracts.org/abstract/investigating-temporal-changes-in-pregnancy-among-women-with-rheumatoid-arthritis-and-healthy-women-using-co-expression-network-analysis/