The placental microbiome is characterized metagenomically as a low-biomass, sparse, reliably detected collection of genetic material that is vastly outnumbered by source host DNA. This complicates its molecular detection, leading some research teams to achieve detection while others attribute detection to contamination. Contributing to this misclassification is the failure to appreciate microbial:host nucleic acid stoichiometry in amplicon generation via polymerase chain reaction (PCR). We hypothesized that germ-free pregnant mice inoculated with specific microbes during gestation will have placentas with detectable microbial signatures via targeted quantitative PCR (qPCR) but not untargeted standard PCR methods. Germ-free mice were mated and divided into three groups: n=3 remained germ-free (GF), n=4 were orally gavaged with Fusobacterium nucleatum mixed with Lactobacillus reuteri (Oral), and n=3 were orally gavaged with F. nucleatum and L. reuteri following treatment with an oral inflammatory agent (TNBS). Sterile Cesarean was performed, and the placentas were subject to DNA extraction with contaminant controls. Microbial DNA was targeted with the V4 region of the 16S rRNA gene, F. nucleatum FadA gene, and L. reuteri gene-specific primers via standard PCR and TaqMan® qPCR. No differences are observed between GF, Oral, and TNBS groups via standard PCR of any target region, yet the sensitivity of detection of microbial rRNA/DNA in vitro is dependent on the relative ratio of murine host DNA present (Fig 1). By qPCR, increasing the amount of murine host DNA with a set amount of L. reuteri DNA reveals a sensitivity in the detection of microbial rRNA/DNA that is dependent on the relative amount of murine host DNA (Fig 2). We observe a relationship between low-abundance microbial rRNA/DNA and high-abundance host DNA in PCR-based techniques. We caution against concluding that a microbial signal is a contaminant without assuring appropriate stoichiometric balance. Our data advocates for careful interpretation of microbial DNA signatures in the presence of abundant host DNA.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

Background: Lactobacillus was described as a keystone bacterial taxon in the human vagina >100 years ago. Using metagenomics, we and others have characterized lactobacilli and other vaginal taxa across health and disease states, including pregnancy. While shifts in community membership have been resolved at the genus/species level, strain dynamics remain poorly characterized. Methods: We performed a metagenomic analysis of the complex ecology of the vaginal econiche during and after pregnancy in a large US based longitudinal cohort of women who were initially sampled in the third trimester of pregnancy, and then validated key findings in a second cohort of women initially sampled in the second trimester of pregnancy. Findings: First, we resolved microbial species and strains, interrogated their co-occurrence patterns, and probed the relationship between keystone species and preterm birth (PTB) outcomes. Second, to determine the role of human heredity in shaping vaginal microbial ecology in relation to PTB, we performed a mtDNA-bacterial species association analysis. Finally, we explored the clinical utility of metagenomics in detection and co-occurrence patterns for the pathobiont group B Streptococcus (causative bacterium of invasive neonatal sepsis). Conclusions: Our highly refined resolutions of the vaginal ecology during and post-pregnancy not only provide insights into structural and functional community dynamics but also highlight the capacity of metagenomics to reveal the finer aspects of the vaginal microbial ecologic

The maternal immune response is thought to protect against fetal pathogenic infections, including Zika Virus (ZIKV) via interferon (IFN) signaling. Part of the trigger for maternal immunity is the microbiome, but it is not known if an established "conventional" (CONV) microbiome mediates this protection or if recently-acquired "conventionalized" (CONVZ) microbes can confer the same benefit. We hypothesize that CONVZ mice have altered type 1 IFN signaling compared to CONV mice. CONV and CONVZ Swiss Webster (SW) mice were mated then inoculated with ZIKV. Separately, germ-free (GF) SW mice were mated then inoculated with Fusobacterium nucleatum mixed with Lactobacillus reuteri, with or without an oral inflammatory agent (TNBS; FNLR group). Both groups included GF controls. Sterile Cesarean was performed, then measures of IFN signaling and microbial transmission. In the ZIKV group (Fig 1), CONVZ mice had increased IFN-α in the spleen (7.8 vs 4.5 fold change) and uterus (0.6 vs 0.1 fold change p=0.03) versus CONV mice, and increased uterine IFN-β transcripts (2.3 vs 0.0 fold change p< 0.001). Placental ZIKV detection was increased in CONVZ mice compared to both CONV mice (7.1 vs 2.1 log viral copies p< 0.0001) and GF controls (7.1 vs 2.4 log viral copies p< 0.0001). Complementarily, in the FNLR group (Fig 2), CONVZ mice had increased IFN-α transcripts in the spleen (6.5 vs 1.1 fold change) and placenta (4.5 vs 2.1 fold change p=0.022) versus GF mice. TNBS treatment further increased placental IFN transcripts (IFN-α 10.2 vs 2.1 fold change p< 0.0001; IFN-β 3.9 vs. 2.0 fold change p=0.026) compared to GF. We detected the inoculum more frequently in the placenta of TNBS treated CONVZ mice versus GF (83.3% vs 45.1% p=0.0077). We tested the hypothesis that conventionalization of the maternal microbiome in previously GF mice alters type I IFN signaling, resulting in increased ZIKV susceptibility. Aberrant IFN signaling, rather than a simplistic "too high" or "too low" mentality, may be associated with increased fetal susceptibility to pathogen infection.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

BACKGROUND:Reliable prediction of spontaneous preterm birth remains limited, particularly for nulliparous and multiparous women without a personal history of preterm birth. Although previous preterm birth is a risk factor for recurrent preterm birth, most spontaneous preterm births occur in women with no previous history of preterm birth.OBJECTIVE:This study aimed to determine whether patients' self-reported maternal family history of preterm births among siblings and across 3 generations was an independent risk factor for spontaneous preterm births after controlling for potential confounders.STUDY DESIGN:This was a retrospective analysis of a prospectively acquired cohort using a comprehensive single, academic center database of deliveries from August 2011 to July 2017. The objective of the current analysis was to evaluate the risk of preterm birth among women with and without a family history of preterm birth. All subjects in the database were directly queried regarding familial history across 3 generations, inclusive of obstetrical morbidities. Index subjects with probable indicated preterm birth (eg, concurrent diagnosis of preeclampsia; hemolysis, elevated liver enzymes, and low platelet count; or placenta previa or placenta accreta) were excluded, as were nonsingleton pregnancies. Univariate and multivariate analyses with logistic regression were used to determine significance and adjusted relative risk.RESULTS:In this study, 23,816 deliveries were included, with 2345 (9.9%) born prematurely (<37 weeks' gestation). Across all subjects, preterm birth was significantly associated with a maternal family history of preterm birth by any definition (adjusted relative risk, 1.44; P<.001), and the fraction of preterm birth occurring in women with a positive family history increased with decreasing gestational age at which the index subjects of preterm birth occurred. For nulliparous women, a history in the subject's sister posed the greatest risk (adjusted relative risk, 2.25; P=.003), whereas for multiparous women with no previous preterm birth, overall family history was most informative (P=.003). Interestingly, a personal history of the index subject herself being born preterm presented the greatest individual risk factor (adjusted relative risk, 1.94; P=.004).CONCLUSION:Spontaneous preterm birth in the current pregnancy was significantly associated with a maternal family history of preterm birth among female relatives within 3 generations and notably sisters. The risk persisted among gravidae without a previous preterm birth, demonstrating the capacity for familial history to independently predict risk of spontaneous preterm birth even in the context of a negative personal history. This study provides evidence that self-reported maternal family history is relevant in a US population cohort and across more distant generations than has previously been reported.

Part of the maternal immune response that protects against fetal pathogenic infections, like Zika Virus (ZIKV), is an established commensal microbiome. Germ-free mice infected with ZIKV during pregnancy have higher rates of fetal growth restriction, intrauterine demise, and ZIKV transmission compared to mice with an established "conventional" (CONV) microbiome (Seferovic, SMFM 2019). However, it is not known if only a naturally CONV microbiome mediates this protection or if postnatally-acquired "conventionalized" (CONVZ) microbes can confer the same benefit. We hypothesized that CONVZ versus CONV mice would have impaired resistance against ZIKV infection in association with an inability to truly normalize a germ-free mouse with postnatal CONVZ. CONV and CONVZ Swiss Webster mice were mated then infected with ZIKV on embryonic days 6-9 (CONV-ZIKV, CONVZ-ZIKV). Additional CONV mice received a Mock infection (CONV-Mock) instead. Sterile Cesarean was performed, then we measured fetal growth restriction, intrauterine demise, viral transmission, and stool bacteria. CONVZ-ZIKV mice had similar fetal weight and intrauterine fetal demise compared to CONV-ZIKV mice. However, increased ZIKV viral load was detected in CONVZ-ZIKV versus CONV-ZIKV spleen (7.7 vs 2.2 log viral copies), placenta (7.1 vs 2.1 log viral copies, p< 0.0001), and uterus (4.9 vs 0.8 log viral copies, p=0.0494). These findings correlate with differences seen in the maternal stool microbiome, whereby CONVZ-ZIKV separates from both CONV-ZIKV and CONV-Mock via PCA (p=0.025), indicating differences in taxonomical structure of the microbial communities (Figure). We tested the hypothesis that conventionalization of the maternal microbiome would not restore resistance against ZIKV infection due to persistent gut microbiome abnormalities. However, ZIKV resistance is preserved in CONVZ mice, but rather we observe an increased molecular susceptibility to ZIKV in all tissues.

It is postulated that the vaginal microbiome lends to the risk/prediction of preterm birth (PTB), although data demonstrating this is inconsistent. As microbiome function results from production of strain-specific metabolites, it is important to adequately characterize vaginal ecology at the strain level. Reliable strain detection requires deep metagenomic sequencing, which cannot be achieved with 16S approaches. Here we tested the hypothesis that keystone vaginal bacterial species are present as dissimilar strains that might differentially associate with PTB. In this prospective study we performed metagenomic sequencing (Illumina HiSeq) on microbial DNA extracted from vaginal samples (n=201, 60 gravidae) collected during 3rd trimester, delivery, and 4-6 weeks postpartum. Strain-level classification of bacteria was performed using PanPhlAn. Lactobacillus spp. are present as one of two strains, whereas Gardnerella vaginalis comprises five strains ( Fig1A). When we then examined interactions at the strain level, we found strain modeling mirrored species level co- occurrence models. However, we found L. crispatus and L. jensenii strains negatively co-occurred with Gardnerella strains; at the species level L. iners negatively and positively co-occurred with Gv2b and Gv3 strains, respectively (Fig1B). As exclusionary interactions between these taxa are thought to lead to PTB, we tested whether differences in strain frequencies associated with PTB. There was no significant difference in the frequency of multiple Gardnerella strains during pregnancy on PTB. When we examined the frequency of Gardnerella and lactobacilli strains, we also failed to identify clear associations with PTB (p>0.05) (Fig1C). Metagenomic sequencing, although costlier and more computationally intensive compared to 16S metabarcoding, provides a highly sensitive method for detecting the dominant members of the vaginal microbiome, enabling strain-level differentiation. The current findings are noteworthy, as they demonstrate for the first time the degree to which strain-level differences among dominant vaginal microbial species occur. The failure to see an association with PTB and strains likely explains the evident disparity among others PTB studies. Given these strain-level differences, we speculate larger studies will be required to definitively determine whether an association between the vaginal microbiome and PTB exists.

Long considered to be a sterile organ, emerging data using culture-independent methods have provided contrasting results suggesting the presence of placental microbiota. Here we sought to rigorously test whether removing sequences detected from DNA extraction kits (kit contaminants) from placental samples rejects the placental microbiome hypothesis, as well as evaluate the impact of sequencing methodology on detection of kit contaminants. Placental samples (n=63) with matched kit negative controls (n=8) were extracted and sequenced in parallel using the V4 region of the 16S rDNA gene. DADA2 was used to generate amplicon sequence variants (ASVs) and profile bacterial taxonomy, and decontam was then used to perform prevalence-based filtering of contaminant ASVs. Additionally, parallel sequencing of 30 blank DNA extractions by the V1V2, V4, V5V6, V7V8 regions of the 16S rDNA gene were performed to determine if different variable regions detect different contaminant taxa. Comparison of taxonomic results of kit negative samples reveals different variable regions detect different contaminant taxa (A,B, p = 0.001). Jackknifed principal coordinate analysis (PCoA) of ASVs from placenta vs matched kit negative controls reveals placenta samples cluster distinctly from kit negative controls, even prior to contaminant filtering (C, p = 0.026). The isNotContaminant function of decontam for low microbial biomass samples determined that 266 out of 1475 total inferred ASVs have sufficient statistical evidence to indicate they are truly present in the placental samples (D,E). After filtering contaminant ASVs from placenta samples and matched kit negative controls, we again found distinct clustering (F, p = 0.024). Beta diversity analysis demonstrates distinct clustering of the placenta microbiota from kit contaminant microbiota, and prevalence-based removal of sequences present in kit negative controls does not fully deplete placenta microbial sequences. Together, this provides evidence that microbial DNA detected in placenta samples are not solely derived from kit contamination. Additionally, identification of contaminant taxa is dependent upon the variable region chosen for sequencing, and should be considered when performing analysis of low abundance microbial communities.

When accompanied by stagnant and contaminated water, major flooding can lead to an increase in incidence or severity of infectious morbidities. Our objective was to evaluate the rate of maternal and neonatal infections prior to and after Hurricane Harvey. Data was abstracted from our Perinatal research database (PeriBank). All singleton pregnancies with recorded delivery date were included. Deliveries prior to August 25, 2017 were included in the "Pre-Harvey" group, and compared with those occurring up to 308 days (44 weeks) comprised the "Post-Harvey" group. The diagnosis of "infection" was abstracted from antenatal, intrapartum and postnatal data. Specifically, infection was defined as a documented occurrence of a urinary tract infection, herpetic lesion, chlamydia, gonorrhea, trichomoniasis, Hepatitis B/C, chorioamnionitis, skin infections, and pneumonia. In addition, there is allowance for provider stipulated "other infection" by free-text. Odds ratios were corrected for possible confounders as identified by univariate analysis. Neonatal outcomes were adjusted for infant sex and GA at delivery. To evaluate changes in the rate of these outcomes over time, the cohort was grouped into mutually exclusive 3-month intervals and the time- trend was graphed in a moving window analysis. "Pre-Harvey" subjects were more likely to be nulliparous, have lower BMI, have a high school education, have higher household income and private insurance as method of payment but were less likely to be married or be Hispanic or African American. The overall rate of infections that were diagnosed antenatally, intrapartum and postnatally remained similar before and after the Hurricane (table). Even after correcting for possible confounders, the rate of skin infections and URI infections diagnosed during the patient's intrapartum course did increase significantly after Hurricane Harvey. Time-trend analysis revealed that the rate of skin infections diagnosed intrapartum was not predicted by fluctuations over time (R2 = 0.0771, p = 0.16); however URI infections diagnosed intrapartum were correlated with variation over time but significantly increased coincident with Hurricane Harvey (R2 = 0.2167, p = 0.01) (Figure). Most rates of the infectious morbidities evaluated were not modified because of Hurricane Harvey. However, there was a clinically meaningful and statistically significant increased occurrence of URI and skin infections.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

OBJECTIVE: To evaluate the association between Hurricane Harvey landfall with maternal and neonatal morbidity. METHODS: Using an institutional perinatal database from two hospitals in Houston, Texas, women with nonanomalous singletons delivering after 24 weeks of gestation between August 2011 and June 2018 were included. To evaluate the possible association of hurricane landfall with pregnancy outcomes, gravid women delivering within 280 days (40 weeks of gestation) on or after August 25, 2017 (the day of hurricane landfall) were categorized as exposed, and women who delivered before August 25, 2017, were the reference group. Composite maternal morbidity included any of the following: hypertensive disorders of pregnancy, chorioamnionitis, endometritis, blood transfusion, peripartum hysterectomy, maternal critical care admission, pulmonary edema, or maternal death. Composite neonatal morbidity included any of the following: 5-minute Apgar score 3 or less, respiratory distress syndrome, use of ventilator or continuous positive airway pressure, suspected newborn sepsis, seizure, stillbirth, or neonatal death. Adjusted odds ratios (aORs) were calculated after correcting for possible confounders identified on univariate analysis. Disruption in outcome trends were measured in time series analyses. RESULTS: Of 40,502 deliveries in our database, 29,179 (72%) met the inclusion criteria, with 3,842 (13.2%) delivering within 280 days of Hurricane Harvey landfall. Women delivering after Hurricane Harvey were on average less likely to be obese and more likely to be older, Caucasian, married, have a household income higher than $75,000, a high school education, and private insurance. However, compared with the cohort of gravid patients who delivered before Hurricane Harvey, composite maternal morbidity increased by 27% (11.5% vs 14.7%, aOR 1.27, 95% CI 1.14-1.42) after the storm. Composite neonatal morbidity increased by 50% (7.8% vs 11.9%, aOR 1.52, 95% CI 1.34-1.71). In time series analyses, we observed a significant shift in composite maternal morbidity specific to women of low socioeconomic status (estimate 2.87, P=.028). CONCLUSION: Despite having fewer at-risk baseline characteristics, gravid patients delivering after landfall by Hurricane Harvey had a significantly higher likelihood of adverse outcomes as did their neonates.

The human body is cohabitated with trillions of commensal bacteria that are essential for our health. However, certain bacteria can also cause diseases in the human host. Before the microbiome can be attributed to disease risk and pathogenesis, normal acquisition and development of the microbiome must be understood. Here, we explore the evidence surrounding in utero microbial exposures and the significant of this exposure in the proper development of the fetal and neonatal microbiome. We further explore the development of the fetal and neonatal microbiome and its relationship to preterm birth, feeding practices, and mode of delivery, and maternal diet.

OBJECTIVE To assess whether recent anti-immigration rhetoric is significantly associated with inadequate prenatal care. METHODS This was a population-based cohort study (2011-2017). In their native language, patients were consented and queried regarding country of origin and time in the United States. Additional variables were collected or abstracted from the medical record, including documentation and timing of prenatal visits. Based on relevance and prevalence during the study period, publicly available Google search trends were mined for the terms "Make America Great Again," "Mexico Wall," and "Deportation" by geographic region. The time of first deviation from the mode Google search popularity value for each term was ascertained (mode inflection date). Perinatal data was averaged over 15 days moving windows, and the Adequacy of Prenatal Care Utilization Index was used to categorically define inadequate prenatal care by validated standards. RESULTS Twenty-four thousand nine hundred thirty-three deliveries occurred during the study period. A mode inflection date was extrapolated from Google trend analytics and used to define the period before change in trends use pre (before rhetoric) and post (after rhetoric). Coincident to the rhetoric change, there was a significant increase in days until the first prenatal visit, fewer prenatal visits, and a decreased trend of mean hemoglobin nadir among U.S. non-native Hispanic women (P<.001). Immigrant status was an independent predictor of inadequate prenatal care as defined by the Adequacy of Prenatal Care Utilization Index standard, with increased adjusted odds among Hispanic women (adjusted odds ratio 1.581, 95% CI 1.407-1.777 [1.4-1.8]) coincident with anti-immigration rhetoric. CONCLUSION Our findings are of likely significant public health importance and suggest that recent anti-immigrant rhetoric is associated with adequate, timely, and regular access to prenatal care among nearly 25,000 deliveries in Houston, Texas.

We and others have previously shown that the vaginal microbiome signature in pregnancy is significant for its reduced diversity and richness. However, the underlying mechanisms driving this loss of microbial diversity were previously unknown. Here, we aimed to study transitions and community dynamics within the vaginal microbiota during the third trimester, delivery, and postpartum. We performed a prospective study, using whole genome shotgun (WGS) metagenomic sequencing using the Illumina HiSeq platform of microbial DNA extracted from vaginal samples (n=201) collected from 60 gravidae during pregnancy (third trimester and delivery) and 4-6 weeks postpartum. Metagenomics demonstrates the vaginal microbiome clusters according to the predominant species – L. iners, L. crispatus, and G. vaginalis, a fourth cluster dominated by L. jensenii/L. iners, and a fifth mixed species cluster (Fig1A). Modeling the temporal dynamics of the community though discrete time Markov chains revealed differences in cluster membership within the 3rd trimester to delivery (Fig1B) and delivery to postpartum intervals ( Fig1C). To understand the patterns of species association, we used probabilistic modeling to determine significant positive and negative co-occurrences. We found a majority of significant species co-occurrences to be positive (85% of significant co-occurrences). When samples were stratified to pregnancy or postpartum, we found that the pattern of significant co-occurrences during pregnancy comprised a network of positive and negative co- occurrences (i.e., a signature microbiome of pregnancy)(Fig1D), which then increased in the number of positive co- occurrences and decreased in the number of negative co-occurrences at postpartum sampling (Fisher’s exact test, p=0.0011)(Fig1E). The vaginal microbiome has complex community dynamics that comprise a signature profile during pregnancy resulting from exclusionary co-occurrences. Taken together, our highly-refined resolutions of vaginal ecology provide both unique structural and functional signature profiles, and highlight the benefits of metagenomics when considered in a microbial ecologic framework.

BACKGROUND: Numerous reports have documented bacteria in the placental membranes and basal plate decidua in the absence of immunopathology using histologic techniques. Similarly, independent metagenomic characterizations have identified an altered taxonomic makeup in association with spontaneous preterm birth. Here we sought to corroborate these findings by localizing presumptive intact bacteria using molecular histology within the placental microanatomy. OBJECTIVE: Here we examined for microbes in term and preterm gestations using a signal-amplified 16S universal in situ hybridization probe set for bacterial rRNA, alongside traditional histologic methods of Warthin-Starry and Gram stains, as well as clinical culture methodologies. We further sought to differentiate accompanying 16S gene sequencing taxonomic profiles from germ-free (gnotobiotic) mouse and extraction and amplicon contamination controls. STUDY DESIGN: Placentas were collected from a total of 53 subjects, composed of term labored (n = 4) and unlabored cesarean deliveries (n = 22) and preterm vaginal (n = 18) and cesarean deliveries (n = 8); a placenta from a single subject with clinical and histologic evident choriomanionitis was employed as a positive control (n = 1). The preterm cohort included spontaneous preterm birth with (n = 6) and without (n = 10) preterm premature rupture of membranes, as well as medically indicated preterm births (n = 10). Placental microbes were visualized using an in situ hybridization probe set designed against highly conserved regions of the bacterial 16S ribosome, which produces an amplified stable signal using branched DNA probes. Extracted bacterial nucleic acids from these same samples were subjected to 16S rRNA metagenomic sequencing (Illumina, V4) for course taxonomic analysis, alongside environmental and kit contaminant controls. A subset of unlabored, cesarean-delivered term pregnancies were also assessed with clinical culture for readily cultivatable pathogenic microbes. RESULTS: Molecular in situ hybridization of bacterial rRNA enabled visualization and localization of low-abundance microbes after systematic high-power scanning. Despite the absence of clinical or histologic chorioamnionitis in 52 of 53 subjects, instances of 16S rRNA signal were confidently observed in 13 of 16 spontaneous preterm birth placentas, which was not significantly different from term unlabored cesarean specimens (18 of 22; P > .05). 16S rRNA signal was largely localized to the villous parenchyma and/or syncytiotrophoblast, and less commonly the chorion and the maternal intervillous space. In all term and unlabored cesarean deliveries, visualization of evident placental microbes by in situ hybridization occurred in the absence of clinical or histologic detection using conventional clinical cultivation, hematoxylin-eosin, and Gram staining. In 1 subject, appreciable villous bacteria localized to an infarction, where 16S microbial detection was confirmed by Warthin-Starry stain. In all instances, parallel sample principle coordinate analysis using Bray-Cutis distances of 16S rRNA gene sequencing data demonstrated consistent taxonomic distinction from all negative or potential contamination controls (P = .024, PERMANOVA). Classification from contaminant filtered data identified a distinct taxonomic makeup among term and preterm cohorts when compared with contaminant controls (false discovery rate <0.05). CONCLUSION: Presumptively intact placental microbes are visualized as low-abundance, low-biomass and sparse populations within the placenta regardless of gestational age and mode of delivery. Their taxonomic makeup is distinct from contamination controls. These findings further support several previously published findings, including our own, which have used metagenomics to characterize low-abundance and low-biomass microbial communities in the placenta.

Hurricane Harvey displaced over 60,000 of Houston's nearly 4.5 million residents. Natural disasters have been linked to posttraumatic stress disorders. Our objective was to evaluate the impact on perinatal outcomes of self-perceived stress among pregnant women in the weeks to months following Hurricane Harvey. Data was abstracted from our perinatal research database (PeriBank). An abbreviated version of the PTSD checklist – civilian version (PCL-C), a validated tool for evaluating both pregnant women and development of PTSD, was prospectively administered to women delivering within our network. Severe stress was defined as either a score >6 in her overall stress score (1-10) or a positive response for symptomatic stress on PCL-C. Composite maternal morbidity (CMM) included: preeclampsia/GHTN, chorioamnionitis, endometritis, blood transfusion, peripartum hysterectomy, maternal ICU admission, pulmonary edema or maternal death. Composite neonatal morbidity (CNM) included: 5-minute Apgar score ≤ 3 at 5 minutes, respiratory distress syndrome (RDS), use of ventilator/CPAP, newborn sepsis, seizure, stillbirth, or neonatal death. Odds ratios were corrected for possible confounders revealed during univariate analysis. Neonatal outcomes were corrected for gender and GA delivery. Of 2,121, 456 (22%) pregnant women delivering after Hurricane Harvey responded to our questionnaire with most suffering the storm effects during their second trimester (84%). Severe stress was diagnosed in 226 (49.6%) with only 3.1% seeking professional mental health services. Women reporting self-perceived severe stress were older and more likely to be married. These women were more likely to suffer sleeplessness (2.1 vs 12.4%, p<0.01), anxiety (9.2 vs. 18.1%, p<0.01), panic symptoms (2.2 vs. 7.5%, p<0.01) and hopelessness (2.2 vs. 7.5%, p<0.01) during the 4 weeks after the hurricane made landfall. They were also more likely to report having financial difficulty after the natural disaster (30.4% vs. 40.7%, p=0.03). Interestingly, the rates of perinatal outcomes were similar between both groups (Table) and did not vary by trimester of pregnancy in which stress occurred. Positive responses to individual PCL-C questions were unable to identify women at risk for adverse perinatal outcomes (figure). Women reporting levels of severe stress following Hurricane Harvey had similar perinatal outcomes compared to women not reporting severe stress.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

Hurricane Harvey significantly impacted the greater Houston area during August and early September of 2017, displacing over 60,000 people as a result of major flooding and destruction. Additionally, it limited access to medical care and severely damaged existing medical infrastructure. Our objective was to evaluate the rate of adverse outcomes prior to and after the storm impaired the city of Houston. Prospectively collected data was abstracted from our perinatal research database (PeriBank). All pregnancies with a recorded delivery date were included. "Pre-Harvey" subjects were defined as deliveries occurring prior to storm landfall on August 25, 2017. All deliveries occurring up to 308 days (44 weeks) after this date comprised the "Post-Harvey" subject subgroup. Composite maternal morbidity (CMM) was defined as: preeclampsia/GHTN, chorioamnionitis, endometritis, blood transfusion, peripartum hysterectomy, maternal ICU admission, pulmonary edema or maternal death. Composite neonatal morbidity (CNM) was defined as: 5-minute Apgar score ≤ 3 at 5 minutes, respiratory distress syndrome (RDS), use of ventilator/CPAP, suspected newborn sepsis, seizure, stillbirth, or neonatal death. Odds ratio were corrected for possible confounders including maternal age, education, race, income, nulliparity, BMI and marital status. Neonatal outcomes were adjusted for infant gender and GA at delivery. To evaluate changes in rate of these outcomes over time, the cohort was grouped into mutually exclusive 3-month intervals and time-trend analysis performed. "Pre-Harvey" subjects were more likely to be nulliparous, have a lower BMI, have a high school education, have a higher household income and private insurance as method of payment but were less likely to be married or be Hispanic or African American. The rate of preeclampsia/GHTN, CMM, CNM, newborn sepsis and rate of Cesarean delivery significantly increased in among post-Harvey subjects (table). Time-trend analysis demonstrated a significant spike in the rate of CNM and preeclampsia/GHTN that was not predicted by fluctuations over time, however CMM and newborn sepsis were correlated (Figure). Major natural disasters rendering displacement of a large proportion of the populations are associated with a significantly increased rate of adverse perinatal outcomes that may restitute over time.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

Given the inability to consistently replicate or independently validate studies linking G. vaginalis and Lactobacillus spp. to preterm birth, differences in the ethnic or racial makeup of risk-disparate cohorts may be masking underlying true associations. As we and others have published on associations between genetic polymorphisms of the host mitochondrial genome and the microbiome, including the gut and vagina, we sought to evaluate the association of the vaginal microbiome with mitochondrial DNA (mtDNA) single nucleotide polymorphisms (SNPs) as risk-modifiers of preterm birth (PTB). WGS paired-end reads identified as host were aligned to the human mitochondrial reference genome (NC_012920.1) using BWA and variant calls were generated using samtools. Only SNPs were considered for subsequent analysis. Species/mtSNP associations were performed using PLINK with the variants considered a haploid genotype and taxa a quantitative trait. For associations between variants, species and PTB, we utilized the quantitative trait interaction algorithm using term and PTB as the covariate groups. Resultant p-values were corrected for False Discovery Rates (FDR). Although a number of significant taxa-SNP associations were identified in WGS (n=1588) (Figure1A), these associations were all in relatively minor taxa and did not include the major landmark species driving the vaginal community, including L. crispatus, L. iners, L. jensenii or G. vaginalis. With respect to PTB, five SNP-species associations identified by WGS metagenomics were significantly different between term and PTB subjects (Figure1B). However, post-hoc comparisons revealed these to be minor taxa present at low abundance and frequency – Propionibacterium acnes, Haemophilus haemolytica, Veillonella atypica, Veillonella parvum, and Lactobacillus mucosa (Figure1C). Given the significance in disparity of preterm birth between racial and ethnic groups, it is logical to consider the role of mitochondrial genetics and the microbiome. However, within this study, when we rigorously accounted for race and ethnicity using mitochondrial DNA sequencing we failed to observe an association. This was not due to underpowering, since; we identified taxa-by-mtDNA SNP associations among rare species in the vaginal microbiome. However, they were weakly predictive of PTB.

Reliable prediction of PTB remains limited, particularly for nulliparous women, as a history of PTB persists as an independent risk factor for sPTB. Because PTB is thought to be inherited genetically, we sought to determine if a family history of PTB among siblings and across multiple generations was an independent risk factor for sPTB after controlling for potential confounders. This was a retrospective cohort study utilizing a comprehensive single, academic center database of deliveries from August 2011 to July 2017 to evaluate the risk of PTB among women with and without a family history of PTB. All subjects are directly queried regarding familial history across three generations, inclusive of obstetrical and gynecologic morbidities. Subjects with probable indicated PTB (e.g., preeclampsia, HELLP, or placenta previa/accrete) were excluded. Univariate and multivariate analysis with logistic regression were used to determine significance and adjusted relative risk (aRR). 23816 deliveries were included, with 2345 (9.9%) born prematurely (<37 weeks gestational age). Across all subjects, PTB was significantly associated with a family history by any definition (aRR: 1.44; p<0.001), and the fraction of PTB linked to a family history increased with decreasing gestational age at delivery (Fig. 1). For nulliparous women, a history in the subject’s sister posed the greatest risk (aRR:2.25; p=0.003) while for multiparous women with no prior PTB, overall family history was most informative (p=0.003) and a personal history of a subject herself being born preterm was the greatest individual risk factor (aRR: 1.94; p=0.004). PTB in the current pregnancy was strongly associated with a family history of PTB among female relatives within two generations, and notably sisters. The risk persisted among gravidae without a prior PTB, demonstrating the capacity for familial history to independently predict risk of sPTB even in the context of a negative personal history.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

In intrahepatic cholestasis of pregnancy (IHCP), primary bile acids (BA) and their microbial derived secondary BAs accumulate in the placenta. Since microbes are intolerant to BAs, we sought to determine if the taxonomical makeup of the placental microbiome is altered in association with exogenous placental bile acid accumulation in IHCP. We thus determined the molecular bile acid profile of placentas in IHCP with and without ursodeoxycholic acid treatment and compared to taxonomic changes using a metagenomic sequencing approach. Placentas were collected from IHCP with (n 30) and without (n 10) ursodexoycholic acid treatment and control (n 10) subjects. BAs were quantified by MRM mass spectrometry. The placenta microbiome of a subset of IHCP (n 9) and control (n 5) subjects was profiled by whole genome shotgun (WGS) sequencing to quantify bacterial taxa, and correlation analysis was used to determine associations between metabolic and microbial profiles. Of BAs detected, 9/11 placental and 4/7 serum were increased in IHCP (p<0.05). Hierarchical clustering revealed an association with time since treatment initiation and severity of BA number (A). BA Group 4 structured around increased secondary BAs associated with the treatment. These bile acids were highest in association with diagnosis/treatment (36 days, p=0.01) (A). Beta diversity analysis of the placental microbiome revealed clustering by virtue of treatment with UDCA (B, p = 0.052) with enrichment of numerous taxa including Actinomyces oris with UDCA treatment (D, p < 0.05). Trending microbial beta diversity clustering was detected by virtue of BA Group (C, p = 0.11), with significant correlations between the PC2 axis (microbiome beta diversity) and bile acids distinctive of BA Group 4 (C, p < 0.002). We present evidence of taxonomic changes in the placenta in association with bacterial toxic exogenous bile acid treatment (e.g., Ursodiol). The significance of interaction suggests (1) that ursodeoxycholic acid has effects beyond the management of pruritus, and (2) provides the strong evidence of intact live placental microbiota that shift with medications other than antibiotics.