Understanding the impact of the vaginal microbiota on miscarriage: Are we there yet?

Miscarriage affects approximately 20% of recognised pregnancies1 and can cause severe psychological distress. A history of miscarriage is associated with complications in subsequent pregnancies and other adverse health outcomes.2 As with many pregnancy outcomes, few established risk factors have been identified,3 and preventative treatments are limited. In understanding the aetiological pathways that result in idiopathic miscarriage, researchers may discover affordable and accessible treatments to intervene upon specific biological pathways. Factors that disrupt the maternal immune system are promising targets for novel therapeutic interventions. The maternal immune system is highly dynamic (not immunosuppressive as previously suggested) and protects against ascending infections while maintaining a receptive environment that ensures decidualisation, implantation and placentation.4 Failures in implantation and placentation have been linked to miscarriage,5 and emerging evidence indicates that reproductive tract microbiota plays an essential role in maternal immune regulation and modulation in addition to preventing the overgrowth of harmful bacteria.4 Scientific understanding of the impact of microbial communities on human health has increased dramatically with technological advances such as 16S ribosomal RNA, DNA hybridisation and next-generation sequencing, which have overcome the limitations of culture-based methods. The vaginal microbiota has been extensively studied. A 'normal' vaginal microbiota is determined by the dominance of Lactobacillus spp., which controls the proliferation of harmful bacterial species through the production of lactic acid, hydrogen peroxide and bacteriocins.6 Variations in the stability of some bacterial communities across menstrual cycles7 and by patient characteristics and sexual behaviours pose a challenge to epidemiologic investigations. Studies of the vaginal microbiota suggest that dysbiosis is associated with gynaecologic conditions, sexually transmitted infection acquisition, endometritis and pregnancy complications such as preterm delivery.8 However, the association between the vaginal microbiota and the risk of miscarriage is inconsistent across studies. Much of this variation is due to differences in study designs and analyses, the timing of vaginal sampling and the use of various molecular-based methods that yield different information on microbial compositions (i.e. genus vs. species and strains). Of particular concern is the inability to obtain vaginal samples before or near key reproductive processes such as implantation. Indeed, many studies lack a biological rationale for their sampling timing, and because microbial communities can change markedly with hormonal fluctuations, studies with specimens sampled later in gestation may not represent microbial communities present in early pregnancy. In this issue of Paediatric and Perinatal Epidemiology, McClelland and colleagues9 address many of the limitations of prior investigations. The authors nested this investigation within the Kenyan Microbiota and Pre-term Birth (MPTB) study, a preconception cohort of participants who had recently discontinued contraception, to measure the vaginal microbiota at periconception (<28 days of pregnancy) and in the first trimester. Periconception was selected to be as close to implantation as possible. Their measurement of the vaginal microbiota was robust. It included an evaluation of the relative abundance of bacteria using n16s rRNA PCR with next-generation sequencing, which informed their targets for quantitative PCR (qPCR), a method the investigators have used previously. The results of the next-generation sequencing revealed that relative abundances of some periconception and first-trimester bacteria were associated with miscarriage in unadjusted models. However, their primary analysis revealed little association between greater concentrations of Megasphaera hutchinsoni, Mageeibacillus indolicus, Mobiluncus mulieris and Sneathia sanguinegens/vaginalis and the risk of miscarriage. Their exploratory crude model found that first-trimester detection of Mageeibacillus indolicus and Mobiluncus Mulieris was associated with miscarriage; this warrants further investigation. Most epidemiological studies of miscarriage have been retrospective, confined to pregnant participants and prone to outcome misclassification, left truncation and recall and selection biases. The MPTB study overcomes these limitations with its prospective design, preconception enrollment, early detection of miscarriage and biomarker measurement of the vaginal microbiome. Participants would have been unaware of their microbiome status or future pregnancy chances at the time of cohort enumeration, thereby reducing the potential for selection bias. Given that preconception and early pregnancy exposures can affect miscarriage risk, MPTB is better able to ascertain exposures in aetiologically relevant periods. The study has some limitations. The sample size included only 45 cases of miscarriage and 159 controls, which reduced precision. However, the difficulties in obtaining periconceptional samples should be considered. Lack of access to early pregnancy samples has plagued prior investigations. Thus, the small sample size should be weighed against access to a well-designed preconception cohort study. While using qPCR was a strength of the investigation because it overcomes limitations of next-generation sequencing, including misinterpretation of microbial communities, the study was limited to measuring absolute quantities of only four selected bacteria, a weakness appropriately acknowledged by the authors. While measurement of the vaginal microbiota at periconception and in the first trimester is a strength, the preconception samples from this cohort were not utilised. The process of decidualisation is essential for implantation as well as placentation. As the vaginal microbiota also fluctuates with hormonal changes, it would have been interesting to know if there were any changes to the microbial communities from the preconception through the periconception periods and into the first trimester. This may be a target for future research. Vaginal samples are more accessible, but data are unclear regarding the extent to which vaginal microbiota reflect the upper reproductive tract. Potential mechanisms underlying an association between the vaginal microbiota and miscarriage are not fully understood, but investigators should make these hypotheses clear. One pathway may be through the increased risk for ascension of harmful bacteria (including sexually transmitted infections) to the upper genital tract, which can result in excessive inflammation through pattern recognition receptor signalling. McClelland et al. were not able to determine bacterial ascension to the upper genital tract. Sexually transmitted infections were measured, but there were few positive cases and no evidence that those with a pregnancy loss were more or less likely to have Chlamydia trachomatis (4.4% vs. 6.9%), Trichomonas vaginalis (0% vs. 1.4%) or bacterial vaginosis (33.3% vs. 29.9%) than those who had pregnancies continue past 20 weeks gestation. On the other hand, the endometrium and fallopian tubes are reported to have unique microbial communities and may not be as sterile as previously thought.7 Mouse models suggest that endometrial receptivity may be mediated by baseline type I interferon responses controlled by the local microbiota.4 Chen et al.7 have suggested that vaginal or cervical microbiota may be useful samples to identify pathologies of the upper reproductive tract. Others have found some discordance between vaginal and endometrial microbiota, although low biomass and contamination remain concerns for endometrial microbiota investigations.8 Access to reproductive tissues can be challenging and almost impossible during pregnancy. Mechanistic studies may improve our understanding of the impact of the microbiota on reproduction, which could guide clinical and epidemiologic investigations. However, animal models and in vitro studies using single-cell systems do not fully reflect the human experience. The growing use of organ-on-chip models to replicate the female reproductive tract and pregnancy is promising, but incorporating the vaginal microbiota into these models remains a challenge.10 In summary, McClelland et al. did not find strong evidence that the vaginal microbiota was associated with miscarriage, but these findings should not discourage future research in this area. This study highlights the challenges of this area of investigation and identifies areas for future research. Brandie DePaoli Taylor is a professor and perinatal epidemiologist at the University of Texas Medical Branch, Galveston, TX. Her research examines how reproductive tract infections and maternal innate immunity impact reproductive and pregnancy success. Dr. DePaoli Taylor collaborates closely with basic scientists to incorporate mechanistic studies with epidemiologic investigations of infection, immunology and pregnancy complications. Lauren A. Wise is a professor and reproductive epidemiologist at Boston University School of Public Health. She is the Principal Investigator of Pregnancy Study Online (PRESTO), a preconception cohort study of North American couples. Her research focuses on preconception environmental, dietary and lifestyle determinants of reproductive and pregnancy outcomes, including miscarriage. Dr. Wise serves as an associate editor for Paediatric and Perinatal Epidemiology. BDT took the lead on drafting the commentary. Both authors contributed substantially to the conception of the work and revised it critically for intellectual content. Both authors approve the final version to be published and agree to be accountable for all aspects of the work. None. Dr. Taylor is supported by grants from the National Institutes of Health (R01AI141501-01A1 and R01AI143653-01A1). Dr. Wise is supported by grants from the National Institutes of Health (R01HD086742, R01HD105863, R01ES029951 and R01HD115096) and the National Science Foundation (NSF 1914792). Dr. Wise serves as a paid consultant for AbbVie, Inc. and the Gates Foundation and receives in-kind donations for primary data collection in Pregnancy Study Online (PRESTO) from Swiss Precision Diagnostics (home pregnancy tests) and Kindara.com (fertility-tracking apps). There was no data involved in this commentary.