Prediction of vaginal birth after induction of labor with maternal circulating RNA transcripts.

Abstract

Induction of labor is commonly performed in pregnancies for clinical benefits, but 25% to 30% of inductions eventually fail and require cesarean delivery. So far, there is no biomarker for the prediction of successful vaginal birth after induction of labor, and clinical prediction remains unsatisfactory. This study aimed at identifying circulating RNA transcripts of which the level changed before and after induction of labor and to assess their predictability of successful vaginal birth after induction. This observational study comprised 3 phases: I: a longitudinal discovery study; II: a longitudinal verification study; and III: a case-control study. Women with singleton pregnancies undergoing labor induction at term were recruited for preinduction and postinduction peripheral blood sampling. In Phase I, circulating transcripts with differential expression during induction were identified using transcriptome analysis in 18 subjects. In Phase II, the identified circulating transcripts were verified with quantitative polymerase chain reaction in another 27 subjects. In Phase III, the potential of these selected transcripts in predicting the outcomes after induction of labor was assessed in a case-control study comparing 45 subjects with vaginal birth and 45 matched subjects who had cesarean birth. In Phase I and II, a panel of 10 circulating RNA transcripts with differential expressions before and after labor induction were identified and verified. GMEB2, IGLV1-36, PDCD11, TRBV15, and UPF2 were downregulated with log₂ fold changes ranging from -4.36 to 0.50 (P<.05), while CA4, DHRS13, PADI4, SH3BP5, and TRAPPC11 were upregulated with log₂ fold changes ranging from 0.55 to 7.09 (P<.01). Kaplan-Meier analyses indicated that women whose preinduction of labor levels of the 5 downregulated transcripts (GMEB2, IGLV1-36, PDCD11, TRBV15, and UPF2) were low had a 10.8-hour to 16.6-hour shorter time from induction to birth than their respective high-level counterparts. Similarly, those women with a high preinduction of labor level of the SH3BP5 (an upregulated transcript) progressed about 5.8 hours earlier to birth than their respective low-level counterparts. In Phase III, we identified that GMEB2, IGLV1-36, PDCD11, and UPF2 were most predictive for cesarean birth. Their combined predictive accuracy was superior to Bishop score (area under the curve=0.714 vs 0.620, 95% confidence interval: 0.606-0.821; P=.031) with a sensitivity of 0.578 and a false positive rate of 0.222. A panel of 10 circulating birth-associated RNA transcripts were identified. Their preinduction levels may help to estimate the chance of successful vaginal birth and facilitate clinical decision-making.