Prospective observational study of vaginal microbiota pre- and post-rescue cervical cerclage

Abstract

Objective: To investigate the relation between vaginal microbiota composition and outcome of rescue cervical cerclage.

Design: Prospective observational study.

Setting: Queen Charlotte's and Chelsea Hospital, London.

Population: Twenty singleton pregnancies undergoing a rescue cervical cerclage.

Methods: Vaginal microbiota composition was analysed in women presenting with a dilated cervix and exposed fetal membranes before and 10 days following rescue cervical cerclage and was correlated with clinical outcomes.

Main outcome measures: Composition of vaginal bacteria was characterised by culture-independent next generation sequencing. Successful cerclage was defined as that resulting in the birth of a neonate discharged from hospital without morbidity. Unsuccessful cerclage was defined as procedures culminating in miscarriage, intrauterine death, neonatal death or significant neonatal morbidity.

Results: Reduced Lactobacillus spp. relative abundance was observed in 40% of cases prior to rescue cerclage compared with 10% of gestation age-matched controls (8/20, 40% versus 3/30, 10%, P = 0.017). Gardnerella vaginalis was over-represented in women presenting with symptoms (3/7, 43% versus 0/13, 0%, P = 0.03, linear discriminant analysis, LDA (log 10) and cases culminating in miscarriage (3/6, 50% versus 0/14, 0%, P = 0.017). In the majority of cases (10/14, 71%) bacterial composition was unchanged following cerclage insertion and perioperative interventions.

Conclusions: Reduced relative abundance of Lactobacillus spp. is associated with premature cervical dilation, whereas high levels of G. vaginalis are associated with unsuccessful rescue cerclage cases. The insertion of a rescue cerclage does not affect the underlying bacterial composition in the majority of cases.

Tweetable abstract: Preterm cervical dilatation associates with reduced Lactobacillus spp. Presence of Gardnerella vaginalis predicts rescue cerclage failure.

Keywords: Infection; preterm birth; rescue cerclage; vaginal microbiome.

Figure 1

Vaginal microbial community structure prior to rescue cerclage compared with gestation age‐matched controls. Relative abundance of bacterial genera and Lactobacillus spp. (shades of green) within the vaginal microbiota, inverse Simpson index (diversity), and classification by Lactobacillus spp. dominance; dominant (blue) >75%, intermediate (pink) 50–75%, and deplete (red) <50%. Women with premature cervical dilation compared with gestation age‐matched controls (closed cervix subsequent term delivery). The proportion of women with a vaginal microbiome with reduced (intermediate or deplete) Lactobacillus spp. abundance was significantly higher (P = 0.01) in women prior to rescue cerclage.

Figure 2

(A) Vaginal microbial community structure prior to rescue cerclage in women with and without symptoms. Relative abundance of bacterial genera and Lactobacillus spp. (shades of green) within the vaginal microbiome, inverse Simpson index (diversity), and classification by Lactobacillus spp. dominance; dominant (blue) >75%, intermediate (pink) 50–75%, and deplete (red) <50%. There were no significant differences in vaginal community structure between asymptomatic women and those presenting with symptoms. A difference at bacterial genera level within these communities was apparent with an overrepresentation of Lactobacillus crispatus (dark green) and Bifidobacterium (blue) in the asymptomatic cohort and Gardnerella vaginalis (red) in the symptomatic group. (B) Gardnerella vaginalis and L. crispatus are differentially expressed between women with and without symptoms prior to emergency cerclage. Cladogram describing differentially abundant vaginal microbial clades and nodes observed between asymptomatic women and those with symptoms as identified using LEfSe analysis. The effect size for each differentially abundant species was estimated using LDA. Vaginal microbiome of symptomatic women was enriched with Gardnerella vaginalis, Prevotella disiens, and Streptococcus pseudopneumoniae, whereas those without symptoms were comparatively enriched with Lactobacillus crispatus.

Figure 3

Vaginal microbial communities dominated by Gardnerella vaginalis were associated with poor outcomes. Relative abundance of bacterial genera and Lactobacillus spp. (shades of green) within the vaginal microbiome, inverse Simpson index (diversity), and classification by Lactobacillus spp. dominance; dominant (blue) >75%, intermediate (pink) 50–75%, and deplete (red) <50%. There were no significant differences in the proportion of women with a Lactobacillus spp.‐deplete microbiome between successful and unsuccessful cerclages or between cases delivering before and after 28, 32, and 37 weeks. The presence of Gardnerella vaginalis (red) was only seen prior to cerclages with unsuccessful outcomes.

Figure 4

Rescue cerclage and postoperative interventions did not perturb the vaginal microbiome in the majority of cases. (A) Relative abundance of bacterial genera and Lactobacillus spp. (shades of green) within the vaginal microbiome before (top) and 10 days after (bottom) emergency cervical cerclage for cases where Lactobacillus spp. abundance remained stable (left, black), increased (middle, dark grey) and decreased (right, light grey). In 70% (10/14) of cases there was minimal change or increased Lactobacillus spp. abundance following rescue cervical cerclage. In the minority of cases (4/14, 30%) Lactobacillus spp. abundance fell and Gardnerella vaginalis emerged. Overall, no significant differences were observed in richness (B), diversity (C) or total Lactobacillus spp. abundance (D).