Characteristics of vaginal microbiota in pregnant women with premature rupture of membranes and establishment of prediction model / 中华微生物学和免疫学杂志

ABSTRACT

Objective:To study the characteristics of vaginal microbiota in pregnant women with premature rupture of membranes (PROM) and to establish prediction models for PROM.Methods:This study involved 35 women with preterm premature rupture of membranes (PPROM), 180 with term premature rupture of membranes (TPROM) and 255 term birth cases without premature rupture of membranes (TBWPROM, control group). The V3-V4 hypervariable region sequences in the vaginal samples collected at 16-28 weeks of gestation were detected by 16S rRNA gene next-generation sequencing. The differences in Alpha and Beta diversity, and the attributes and metabolic function prediction of each recognized species among the three groups were analyzed. Subsequently, a random forest model was used to establish the prediction models for PROM using vaginal microbiota species and environmental risk factors.Results:Compared with the control group, the Alpha diversity of the PPROM group was higher (Observed features, P=0.022; Faith_pd index, P=0.024) and Beta diversity was also significantly different (Unweighted UniFrac, P=0.010; Jaccard index, P=0.008). In PPROM cases, Megasphaera genomosp. typeⅠ was significantly increased ( P=0.017) and Lactobacillus mulieris was significantly decreased ( P=0.003). In the patients with TPROM, Megasphaera was significantly increased ( P=0.009) and Lactobacillus mulieris was significantly decreased ( P=0.002). In terms of functional pathways, sulfur oxidation ( P=0.021), methanogenesis from acetate ( P=0.036), L-histidine biosynthesis ( P=0.009), adenosylcobalamin biosynthesis ( P=0.041) and fucose degradation ( P=0.001) were significantly increased in patients with PPROM; L-histidine biosynthesis ( P<0.001) and fucose degradation ( P=0.030) were significantly increased in patients with TPROM. The prediction models were established using the random forest model with vaginal microbiota species and environmental risk factors and the prediction model for PPROM performed well [AUC 0.739 (95%CI 0.609-0.869), sensitivity 0.928, specificity 0.659, positive predictive value 0.750, negative predictive value 0.906], which had a certain reference value. Conclusions:Vaginal microbiota might be related to the development and progression of PROM. Studying the differences in vaginal microbiota might provide a new idea for the prevention and treatment of PROM. Functional prediction provided a direction for further research on the mechanism of PROM. The established prediction model could prevent the occurrence of PPROM and promote maternal and infant health.