Predicting the Risk of Preterm Birth Throughout Pregnancy Based on a Novel Transcriptomic Signature

Abstract Objective This study focused on the prediction of preterm birth (PTB). It aimed to identify the transcriptomic signature essential for the occurrence of PTB and evaluate its predictive value in early, mid, and late pregnancy and in women with threatened preterm labor (TPTL). Methods Blood transcriptome data of pregnant women were obtained from the Gene Expression Omnibus database. The activity of biological signatures was assessed using gene set enrichment analysis and single-sample gene set enrichment analysis. The correlation among molecules in the interleukin 6 (IL6) signature and between IL6 signaling activity and the gestational week of delivery and latent period were evaluated by Pearson correlation analysis. The effects of molecules associated with the IL6 signature were fitted using logistic regression analysis; the predictive value of both the IL6 signature and IL6 alone were evaluated using receiver operating characteristic curves and pregnancy maintenance probability was assessed using Kaplan-Meier analysis. Differential analysis was performed using the DEseq2 and limma algorithms. Results Circulatory IL6 signaling activity increased significantly in cases with preterm labor than in those with term pregnancies (normalized enrichment score (NES) = 1.857, P = 0.001). The IL6 signature (on which IL6 signaling is based) was subsequently considered as the candidate biomarker for PTB. The area under the curve (AUC) values for PTB prediction (using the IL6 signature) in early, mid, and late pregnancy were 0.810, 0.695, and 0.779, respectively; these values were considerably higher than those for IL6 alone. In addition, the pregnancy curves of women with abnormal IL6 signature differed significantly from those with normal signature. In pregnant women who eventually had preterm deliveries, circulatory IL6 signaling activity was lower in early pregnancy (NES = −1.420, P = 0.031) and higher than normal in mid (NES = 1.671, P = 0.002) and late pregnancy (NES = 2.350, P < 0.001). In women with TPTL, the AUC values for PTB prediction (or PTB within 7 days and 48 hours) using the IL6 signature were 0.761, 0.829, and 0.836, respectively; the up-regulation of IL6 signaling activity and its correlation with the gestational week of delivery ( r = −0.260, P = 0.001) and latency period ( r = −0.203, P = 0.012) were more significant than in other women. Conclusion Our findings suggest that the IL6 signature may predict PTB, even in early pregnancy (although the predictive power is relatively weak in mid pregnancy) and is particularly effective in symptomatic women. These findings may contribute to the development of an effective predictive and monitoring system for PTB, thereby reducing maternal and fetal risk.