RESEARCH Open AccessIncreased richness and diversity of thevaginal microbiota and spontaneouspreterm birthAline C. Freitas1, Alan Bocking2,3, Janet E. Hill1, Deborah M. Money4,5* and the VOGUE Research GroupAbstractBackground: The bacterial community present in the female lower genital tract plays an important role in maternaland neonatal health. Imbalances in this microbiota have been associated with negative reproductive outcomes,such as spontaneous preterm birth (sPTB), but the mechanisms underlying the association between a disturbedmicrobiota and sPTB remain poorly understood. An intrauterine infection ascending from the vagina is thought tobe an important contributor to the onset of preterm labour. Our objective was to characterize the vaginal microbiota ofpregnant women who had sPTB (n = 46) and compare to those of pregnant women who delivered at term (n= 170).Vaginal swabs were collected from women at 11–16 weeks of gestational age. Microbiota profiles were created by PCRamplification and pyrosequencing of the cpn60 universal target region.Results: Profiles clustered into seven community state types: I (Lactobacillus crispatus dominated), II (Lactobacillus gasseridominated), III (Lactobacillus iners dominated), IVA (Gardnerella vaginalis subgroup B or mix of species), IVC (G. vaginalissubgroup A dominated), IVD (G. vaginalis subgroup C dominated) and V (Lactobacillus jensenii dominated). The microbiotaof women who experienced preterm birth (< 37 weeks gestation) had higher richness and diversity and higher Mollicutesprevalence when compared to those of women who delivered at term. The two groups did not cluster according to CST,likely because CST assignment is driven in most cases by the dominance of one particular species, overwhelming thecontributions of more rare taxa. In conclusion, we did not identify a specific microbial community structure that predictssPTB, but differences in microbiota richness, diversity and Mollicutes prevalence were observed between groups.Conclusions: Although a causal relationship remains to be determined, our results confirm previous reports of anassociation between Mollicutes and sPTB and further suggest that a more diverse microbiome may be importantin the pathogenesis of some cases.Keywords: Microbiome, Vagina, Lactobacillus, CST, Diversity, Richness, Mollicutes, Preterm birth, Pregnancy, InfectionBackgroundPreterm birth is defined as delivery before 37 completedweeks of gestational age [1] and can be furthersub-categorized in extremely preterm (≤ 27+6 weeks+days),very preterm (28 to 31+6) and late preterm (32 to 36+6)[2]. Preterm birth comprises 11% of all livebirths world-wide, and its complications are estimated to cause 35% ofworld’s neonatal deaths, which represents 3.1 milliondeaths annually [3]. Children who are born prematurelyalso have higher rates of cardiovascular disorders, respira-tory distress syndrome, neurodevelopmental disabilitiesand learning difficulties compared with children born atterm [4].Preterm birth is a complex multi-factorial conditionwith several known risk factors, such as low and highmaternal ages [5–7], low BMI [8], black ethnicity [9], to-bacco use, heavy alcohol intake, illicit drug use [4], closetemporal proximity to a previous delivery [10], and mul-tiple gestation [11]. Although studied extensively, somepreterm cases remain unexplained for women with noknown risk factors. Intrauterine infection with organismsascending from the vagina has been hypothesized as an* Correspondence: deborah.money@ubc.ca4Department of Obstetrics and Gynaecology, University of British Columbia,Vancouver, BC V6T 1Z4, Canada5Women’s Health Research Institute, BC Women’s Hospital & Health Centre,Vancouver, BC V6H 3N1, CanadaFull list of author information is available at the end of the article© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Freitas et al. Microbiome (2018) 6:117 https://doi.org/10.1186/s40168-018-0502-8important contributor to preterm birth since many or-ganisms isolated from the amniotic fluid/membranes ofwomen who experienced preterm birth are also found inthe lower genital tract of pregnant women [12–15]. Alarge number of studies support this hypothesis basedon the strong association between intra-amniotic bacter-ial infection and preterm birth [12, 14–20].The microbiological diagnosis of a ‘normal’ or dis-turbed vaginal microbiota has historically been based onthe Nugent score, the current gold standard diagnosticmethod that relies on Gram stain of vaginal smears [21].The ‘normal’ vaginal microbiota in non-pregnant repro-ductive aged women is understood to be dominated byLactobacillus species, while an abnormal microbiota(defined as bacterial vaginosis) is characterized by lowabundance of lactobacilli and an overgrowth of anaer-obic bacteria, such as Gardnerella vaginalis, Prevotellaspp., Bacteroides spp., Mobiluncus spp. and Mycoplasmahominis [22]. In low-risk pregnant women, it has beenshown that the vaginal microbiota has reduced richnessand diversity and increased abundance of lactobacillicompared to those of non-pregnant women [23–27]. Anabnormal microbiota has been previously associatedwith preterm birth [28], but only a few in depthculture-independent studies of the vaginal microbiota ofwomen who had preterm birth have been published,with inconsistent conclusions [29–32].The objective of this study was to assess whether thereare differences in the vaginal microbiota composition,early in gestation, of women who had spontaneous pre-term birth (sPTB) and term delivery that could be fur-ther investigated as diagnostic indicators of pretermbirth risk. Microbiome profiling was based on sequen-cing of the cpn60 universal target, which provides higherresolution than 16S rRNA variable regions [33] and al-lows the resolution of Gardnerella vaginalis subgroups,a hallmark bacteria in the disturbed microbiota [34].MethodsStudy population and samplingThis retrospective cohort study analysed the vaginalmicrobiota of women who experienced spontaneous pre-term birth (sPTB) and compared the resulting microbialprofiles to those of pregnant women who delivered atterm. The bacterial profiles of pregnant Canadian womenat low risk of sPTB who had term deliveries (n = 170) werepreviously generated by our research group [24]. The vagi-nal microbial profiles of Canadian women who had pre-term birth originated from samples of this previous study(n = 7) [24] and from the Ontario Birth Study (n = 39),resulting in 46 samples. The Ontario Birth Study(ontariobirthstudy.com) is an open longitudinal pregnancycohort at Mount Sinai Hospital, Toronto, Canada. It is aplatform for studies of both pregnancy complications aswell as Developmental Origins of Health and Disease re-lated research. The PTB rates for the low-risk cohort andOBS cohorts were 4 and 6.2%, respectively. All biospeci-mens, including maternal vaginal swabs and maternal andinfant blood, are collected concurrently with routine clin-ical specimens to reduce the burden on study participants.Detailed demographic and lifestyle characteristics are ob-tained from women during pregnancy and postpartum,and clinical information is extracted from the health re-cords. For the purposes of this report, self-administeredvaginal swabs were taken at 16 weeks gestation and placedin dry tubes prior to being placed in − 80 °C for storage inthe Lunenfeld Tanenbaum Research Institute BiospecimenStorage and Processing Laboratory. Specimens from allcohorts were processed similarly in terms of sample col-lection, storage, DNA extraction, library preparation andsequencing.Clinical and behavioural questionnaire data (pregnancyhistory, family and personal medical history, psycho-social health, demographic factors and other lifestyle andenvironmental exposures) were transferred to theResearch Electronic Data Capture (REDCap) databaseprotected by a secure server [35]. For the PTB group,eligible participants for this study were women who hadundergone preterm delivery at greater than 20 weeks butless than 37 weeks gestational age, where onset of labouroccurred spontaneously or in association with cervicalincompetence or preterm premature rupture of mem-branes (PPROM). Vaginal swabs collected from pregnantwomen (both PTB and term groups) at 11–16 weeks ofgestational age were used for bacterial genomic analysis.Total nucleic acid was extracted from swabs using theMagMAX™ Total Nucleic Acid Isolation Kit (Life Tech-nologies, Burlington, ON, Canada) as per manufacturer’sinstructions. Kit reagents are aliquoted to eliminate re-peated accessing of open reagents, and samples are proc-essed in small batches using filter tips to preventcross-contamination. Pipettes and other lab surfaces areregularly treated with DNA surface decontaminant(DNA Away, Thermo Fisher Scientific, Waltham, MA).Samples from both cohorts were processed in exactlythe same way in terms of swab type, storage temperature(no stabilizer was used), DNA extraction, library prepar-ation and sequencing.Total bacterial DNA (qPCR) and detection of Mollicutes (PCR)Quantitative PCR (qPCR)Total bacterial DNA quantity in each sample was esti-mated using a SYBR Green assay based on amplificationof the V3 region of the 16S rRNA gene. Primer se-quences were as follows: SRV3-1 (5′-CGGYCCAGACTCCTAC-3′), SRV3-2 (5′-TTACCGCGGCTGCTGGCAC-3′) [36]. Reactions were run on a MyiQ ther-mocycler using the following cycling parameters: 95 °CFreitas et al. Microbiome (2018) 6:117 Page 2 of 15for 3 min, followed by 30 cycles of 95 °C for 15 s, 62 °Cfor 15 s and 72 °C for 15 s, with a final extension at 72 °Cfor 5 min [37].Conventional PCRSome Mollicutes (Mycoplasma and Ureaplasma) specieslack a cpn60 gene [38]. Thus, we performed a family-specific semi-nested PCR targeting the 16S rRNA geneto detect Mollicutes [39], and a PCR targeting themultiple-banded antigen gene to detect Ureaplasma spp.PCR products from U. parvum and U. urealyticum canbe differentiated by size [40].cpn60 universal target (UT) PCR and pyrosequencingUniversal primer PCR targeting the 549–567 bp cpn60UT region was performed using a mixture of cpn60primers consisting of a 1:3 M ratio of primers H279/H280:H1612/H1613, as described previously [41–43]. Toallow multiplexing of samples in a single sequencingrun, primers were modified at the 5′ end with one of 24unique decamer multiplexing identification (MID) se-quences, as per the manufacturer’s recommendations(Roche, Brandford, CT, USA). Amplicons were pooled inequimolar amounts for sequencing on the Roche GSJunior sequencing platform. The sequencing librarieswere prepared using the GS DNA library preparationkit, and emulsion PCR (emPCR) was performed with aGS emPCR kit (Roche Diagnostics, Laval, Canada).Samples were handled in small batches to avoidcross-contamination, and experimental controls were in-cluded at several steps in the study. Regular monitoring ofDNA extraction controls in our lab by universal PCR con-firms that these procedures are sufficient to eliminate de-tectable template contamination of study samples. A notemplate control was also included in each set of PCR re-action as negative controls. Experimental controls werenot sequenced as they did not yield any amplification.Analysis of operational taxonomic units (OTU)Raw sequence data was processed by using the defaulton-rig procedures from 454/Roche. Filter-passing readswere used in the subsequent analyses for each of the py-rosequencing libraries. MID-partitioned sequences weremapped using Bowtie 2 (http://bowtie-bio.sourceforge.net/bowtie2/) on to a manually curated reference set of1561 OTU sequences representing the human vaginalmicrobiota. Bowtie 2 was run using the defaultend-to-end alignment mode.The OTU reference set was generated originally by denovo assembly of cpn60 sequence reads from 546 vaginalmicrobiomes using the microbial Profiling Using Meta-genomic Assembly pipeline (mPUMA, http://mpuma.sourceforge.net) [44] with Trinity as the assembly tool[45] (Additional file 1). OTU were labeled according totheir nearest reference sequence determined bywatered-Blast comparison [46] to the cpn60 referencedatabase, cpnDB_nr (downloaded from http://www.cpndb.ca [38]). This reference assembly approach al-lows us to compare the microbial profiles from various co-horts under investigation, including the 46 pregnantwomen who had sPTB described in this study.The result of mapping is an OTU frequency table(Additional file 2) that was used for microbiome data ana-lysis. Some analyses were also performed at species level,i.e. combined OTU that have the same nearest neighbour.Statistical analysisComparisons of socio-demographic characteristics of co-horts and participants were based on analysis of variance(ANOVA), t test and chi-square, performed in IBM SPSS(Statistical Package for the Social Sciences, version 21)at 5% level of significance. For analysis of associationsbetween socio-demographic characteristics and micro-biota profiles (CST), a false discovery rate (FDR) correc-tion for multiple comparisons was applied [47].Alpha (Shannon diversity and Chao1 estimated speciesrichness) and beta diversity (jackknifed Bray–Curtis dis-similarity matrices) were calculated as the mean of 100subsamplings of 1000 reads (or all reads available whenless than 1000) in QIIME (Quantitative Insights IntoMicrobial Ecology) [48]. Plots of alpha diversity mea-sures against bootstrap sample number were generatedin R and visually inspected to ensure that an adequatesampling depth for each sample was achieved.For community state type (CST) analysis, a Jensen–Shannon distance matrix was calculated using the ‘veg-dist’ function in the vegan package [49] with a customdistance function that calculates the square root of theJensen–Shannon divergence [50]. This distance matrixwas used for hierarchical clustering using the ‘hclust’function in R, with Ward linkage.The function aldex.clr from the ALDEx2 package in Rwas used to compare the differential relative abundanceof individual taxa in term and preterm groups [51]. Sig-nificant differences were determined based on the falsediscovery rate (FDR), which is the result of a Benjamini–Hochberg corrected p value from a Welch’s t test calcu-lated within ALDEx2.ResultsDescription of the study population and pregnancy outcomesSocio-demographic characteristics of women who hadspontaneous preterm birth (n = 46) and women who hadterm deliveries (n = 170) are summarized in Table 1.There were no significant differences in maternal age,BMI, ethnicity, smoking status, consumption of alcoholor use of probiotics between term and preterm groups(all p > 0.05). Average maternal age was 33 forFreitas et al. Microbiome (2018) 6:117 Page 3 of 15participants in both cohorts. Average body mass index(BMI) was 22.9 and 24.2 for women in the term and pre-term groups, respectively. Most women in both cohortsidentified themselves as white ethnicity, followed by EastAsian and South/Southeast Asian (Table 1). Consump-tion of tobacco (term 2.3%; preterm 0%), alcohol (term5.9%; preterm 4.3%) or probiotic supplements (term4.1%; preterm 6.5%) was low among women in bothgroups (chi-square, all p > 0.05).Most women in the preterm group had a Bachelor/graduate degree (29/46) and an average house incomehigher than CAD 100,000 per year (25/46). A minorityof women who had preterm birth (5/46) reported con-sumption of substances without prescription prior preg-nancy, of which 3/46 women consumed marijuana/hashish, 1/46 woman consumed tranquilizers/nerve pillsand 1/46 woman consumed cocaine/crack. Approxi-mately 74% of the participants in the preterm group re-ported a pre-existing condition. A total of 12/46 womenhad some condition related to mental health, such de-pression or anxiety. Seventeen percent (8/46) had aneurological condition, including migraine headaches,Table 1 Socio-demographic and microbiological characteristics of subjectsCharacteristics Term pregnancies (n = 170) Preterm pregnancies(n = 46)p valueAge (mean ± SD, range)1 33.6 ± 4.2 (21–45) 33.65 ± 4.1 (25–45) 0.94821–25 5 (2.9%) 1 (2.1%)26–35 114 (67.1%) 32 (69.5%)36–45 51 (30.0%) 13 (28.2%)BMI (mean ± SD, range)1 22.9 ± 3.8 (17–40) 24.2 ± 5.6 (19–43) 0.125Underweight (< 18.50) 7 (4.1%) 0 (0%)Normal weight (18.51–24.9) 131 (77.0%) 33 (73.3%)Overweight (25.0–29.9) 25 (14.7%) 8 (17.7%)Obese (> 30) 7 (4.1%) 4 (9.0%)MD3 0 (0%) 1 (2.2%)Ethnicity2 0.261White 108 (63.5%) 22 (47.8%)East Asian 26 (15.3%) 6 (13.0%)South/Southeast Asian 15 (8.8%) 4 (8.7%)Latin America/Hispanic 8 (4.7%) 3 (6.5%)Black 3 (1.8%) 2 (4.4%)Other/mixed ethnicity 10 (5.9%) 6 (13.0%)MD 0 (0%) 3 (6.5%)Community state type (CST)2 0.361I 56 (32.9%) 17 (37%)II 9 (5.3%) 5 (10.9%)III 28 (16.5%) 8 (17.4%)IVA 31 (18.2%) 6 (13%)IVC 19 (11.2%) 2 (4.3%)IVD 11 (6.5%) 1 (2.2%)V 16 (9.4%) 7 (15.2%)Estimated bacterial load (log copies of 16S rRNA gene)/swab (mean ± SD, range)1 7.78 ± 0.93 (4.89–10.67) 8.07 ± 0.71 (6.32–10.33) 0.049Presence of Mollicutes2 68 (40%) 28 (60.8%) 0.012Presence of Ureaplasma2 40 (23.4%) 14 (30.4%) 0.337U. parvum 37 (21.7%) 14 (30.4%)U. urealyticum 3 (1.7%) 0 (0%)Shannon diversity (mean ± SD, range)1 1.28 ± 0.86 (0.13–4.52) 1.81 ± 1.13 (0.34–5.16) 0.004Chao1 richness (mean ± SD, range)1 36.22 ± 14.80 (14.39–115.74) 46.38 ± 24.19 (20.20–126.01) 0.0091t test; 2chi-square; 3MD = missing dataFreitas et al. Microbiome (2018) 6:117 Page 4 of 15and 24% (11/46) had a genitourinary condition (corpusluteal cyst, bicornuate uterus, cervical polyp, cervicaldysplasia (2), uterine polyp, ovarian cyst, polycystic ovar-ian syndrome, urinary tract infections with and withoutkidney stones (3).Characteristics regarding pregnancy and neonatal out-comes are described in Table 2. Pregnancy outcome infor-mation was not available for one woman in the pretermgroup as she was lost to follow-up. There were no signifi-cant differences in gestational age at enrolment, mode ofconception or fetal sex between groups (all p > 0.05). Aver-age gestational age at delivery was 39+3 weeks for thewomen who delivered at term and 34+2 weeks for womenwho had preterm birth, most of which were consideredlate preterm, i.e. delivery between 32 and 36+6 weeks ofgestational age. Women in the preterm group were morelikely to have experienced preterm birth or miscarriage intheir previous pregnancy (chi-square, p < 0.001). They alsohad higher percentage of caesarean sections than womenwho delivered at term. Number of previous gestations alsodiffered between groups; women who had preterm birthwere more likely to be primigravida (22/46) in comparisonwith women who had term deliveries (45/170). There wasa significant difference between term and preterm groupsregarding birth weight and number of infants admitted tolevel 3 neonatal intensive care unit (NICU) (Table 2).Apgar score at 1 (term 8.75 ± 0.6; preterm 8.38 ± 1.1) and5 min (term 8.97 ± 0.17; preterm 8.76 ± 0.7) betweengroups also differed (t test, all p < 0.001). One preterminfant (1/46) died shortly after birth (20 weeks of gesta-tional age).Among women who delivered preterm, 63% (29/46)had premature rupture of membranes (PPROM), 10.8%(5/46) had gestational diabetes and 4.3% (2/46) hadanemia unresponsive to therapy. Twenty-four percent ofwomen (11/46) presented one of the following condi-tions: maternal elevated liver enzymes, short cervix andincompetent cervix; fetal ascites, fetal distress and largefoetus for gestational age; and placental findings of mar-ginal cord insertion, two-vessel umbilical cord, placentaprevia and low-lying placenta.Sequencing results and OTU analysisRaw sequence data files for the samples described in thisstudy were deposited to the NCBI Sequence Read Arch-ive (Accession SRP073152, BioProject PRJNA317763;Table 2 Gestation characteristics, pregnancy and neonatal outcomesCharacteristics Term pregnancies (n = 170) Preterm pregnancies (n = 46) p valueGestational age in weeks+dayAt enrolment (mean ± SD, range)1 13+ 2 ± 1+1 (11+1–16+6) 13+3 ± 1+0 (11+6–16+0) 0.641At delivery (mean ± SD, range) 1 39+3 ± 0+6 (39+3–41+2) 34+2 ± 2+6 (20+0–36+6) < 0.0001Late preterm (32 to 36+6) NA 39 (84.8%)Very preterm (28 to 31+6) NA 5 (10.8%)Extremely early (≤ 27+6) NA 1 (2.1%)Previous pregnancy history (excludes women in first pregnancy)2 (n = 125) (n = 24) < 0.0001Livebirth, term 89 (71.2%) 9 (37.5%)Livebirth, preterm 0 (0%) 3 (12.5%) < 0.0001Spontaneous abortion 20 (16%) 8 (33.3%) 0.046Pregnancy termination 16 (12.8%) 3 (12.5%) 0.968Ectopic pregnancy 0 (0%) 1 (4.1%)Mode of delivery2 0.042Vaginal delivery 128 (75.3%) 27 (58.7%)C-section 42 (24.7%) 18 (39.1%)Parity2 0.0050 55 (32.3%) 28 (60.8%)1 94 (55.3%) 12 (26%)2–4 21 (12.3%) 6 (13%)Assisted conception2 17 (10%) 6 (13%) 0.591Fetal sex (% male/% female) 2 (48.8%)/(51.1%) (41.3%)/(56.5%) 0.430Birth weight (g) (mean ± SD, range)1 3398 ± 459 (1970–5200) 2550 ± 559 (1300–3595) < 0.0001Infant in NICU (n, %)2 1 (0.6%) 24 (52.2%) < 0.00011t test; 2chi-squareFreitas et al. Microbiome (2018) 6:117 Page 5 of 15BioProject PRJNA403856). Total dataset contained1,635,072 cpn60 reads; median and average read countper sample were 4936 and 7569 (range 402–37,378),respectively. Average read length was 424 bp. Results ofBowtie2 mapping showed that these reads correspondedto 728 OTUs from the reference assembly(Additional file 1).Microbiota profilesMicrobiota profiles were created by PCR amplificationand pyrosequencing of the cpn60 universal target region.Hierarchical clustering of vaginal microbiota profilesresulted in seven community state types (CST): I(Lactobacillus crispatus dominated), II (Lactobacillusgasseri dominated), III (Lactobacillus iners dominated),IVA (Gardnerella vaginalis subgroup B or mix of differ-ent species), IVC (G. vaginalis subgroup A dominated),IVD (G. vaginalis subgroup C dominated) and V(Lactobacillus jensenii dominated) (Fig. 1). Each CST isdefined by the dominance of one species of Lactobacillus (I,II, III, V), Gardnerella vaginalis (IVC, IVD) or a mixture ofbacteria species (IVA), as previously described [52, 53].Overall microbiota profiles did not cluster togetherbased on gestational age at delivery (Figs. 1 and 2). Mostmicrobial profiles from the preterm group (80.5%) wereassigned to Latobacillus-dominated CST: CST I (37% ofprofiles), CST III (17.4%), CST V (15.2%) and CST II(10.9%). The remaining profiles (19.5%) were assigned toCST IVA, IVC or IVD (Table 1). The CST IVA was themost heterogeneous group, represented by the domin-ance of Lactobacillus delbrueckii, Bifidobacterium den-tium, Bifidobacterium infantis, Atopobium vaginae,Bifidobacterium breve or a mixture of different bacteriaspecies. The CST IVC was dominated by G. vaginalissubgroup A and Megasphaera spp., and CST IVD wasdominated by G. vaginalis subgroup C (Fig. 1).Ecological analysis and total bacterial loadAssessment of alpha diversity revealed that microbiomesof women who delivered preterm were richer (Chao1richness 46.3 ± 24.1) and more diverse (Shannon diver-sity index 1.8 ± 1.1) when compared to those of womenin the term group (36.2 ± 14.8; 1.2 ± 0.8) (t test, p < 0.01)(Table 1). Total bacterial load was estimated based onqPCR targeting the 16S rRNA gene, and it was expressedas log 16S rRNA gene copy number per swab. Higherbacterial loads were detected in samples from thepreterm group (7.7 ± 0.9) compared to term group(8.0 ± 0.7) (t test, p = 0.049) (Table 1).Bacteria species relative abundance and prevalenceTo investigate whether there was an association betweenindividual taxa and sPTB, the abundance and prevalenceof each species was evaluated. The ALDEx2 analysisassessed the relative abundance of each taxa (at theOTU and species level) in term and preterm groups.Eight OTU/species were more abundant in the termgroup in comparison with preterm, all of which wereconsidered rare members of the bacterial community(Fig. 3). L. acidophilus represented 1% of the total readsin the dataset and had a low relative abundance averageof 1.98% (range 0–69%) and 0.18% (range 0–0.87%) insamples from term and preterm groups respectively. Allthe other seven bacteria together represented only 0.4%of the total reads in the dataset.Bacteria prevalence (presence/absence) was alsoassessed (only species with at least 10 total reads wereincluded). A total of 60 taxa had significant differencesin prevalence between term and preterm groups; 11 spe-cies had greater prevalence in the term cohort and 49species were more prevalent in the preterm cohort(Table 3). Bifidobacterium infantis, for example, was twotimes more prevalent in the term group in comparisonwith preterm, and Prevotella timonensis was 1.58 timesmore prevalent in the preterm group (Table 3). SeveralPrevotella spp. were associated with both term and pre-term. Prevotella amnii and P. tannerae had greaterprevalence in the term cohort, whereas P. timonensis, P.bivia, P. corporis and P. bucalis were more prevalent inthe preterm group (Table 3). It is important to note thatread depth distribution did not differ between term andpreterm cohorts (t test, p > 0.05); therefore, the differ-ences observed here in bacteria prevalence were unlikelyto be driven by cohort sequencing bias.Mollicutes (Mycoplasma and/or Ureaplasma) were de-tected by family-specific conventional PCR in 28/46(60%) of pregnant women who delivered preterm(Table 1). Ureaplasma species were detected bygenus-specific PCR in samples of 14/46 (30%) womenwho had PTB, with all women testing positive for U.parvum and none for U. urealyticum. Women who de-livered at term were less likely to be PCR positive forMollicutes compared to women who had PTB (Table 1).No significant differences were observed in Ureaplasmaprevalence between the two groups (Table 1). An associ-ation between Mollicutes/Ureaplasma detection and thecomposition of the vaginal microbiota, represented asCST, was also investigated. Detection of Mollicutes andUreaplasma was not associated with any CST in particu-lar when investigated in the term cohort, preterm cohortor both groups together (chi-square, p > 0.05).Relationships between microbiological and socio-demographiccharacteristics within the preterm groupThe association between CST (I, II, III, IVC, IVD, V)from profiles of women who delivered preterm and sev-eral microbiologic-socio-demographic characteristicswas investigated. Only two associations were significant:Freitas et al. Microbiome (2018) 6:117 Page 6 of 15ABCDEFig. 1 (See legend on next page.)Freitas et al. Microbiome (2018) 6:117 Page 7 of 15CST and microbiota richness, and CST and microbiotadiversity (ANOVA, p < 0.001). There was no significantassociation between CST and the remaining followingmetadata: microbiota richness and diversity (continuousvariable), presence of Mollicutes and Ureaplasma (yes/no), log 16S rRNA gene copies (continuous variable),maternal age (continuous variable; 18–25, 26–35,36–45), BMI category (underweight, normal, overweight,obese; < 25, ≥ 25), ethnicity (White, East Asian, SouthAsian, Black, Hispanic, Other), natural conception (yes/no),parity (0, > 1), gestational age (continuous variable), modeof delivery (vaginal, C-section), pre-existing condition(yes/no), folic acid intake before or during pregnancy(yes/no), drinking alcohol (yes/no), neonate in highlevel care (yes/no), birth weight (continuous variable)and Apgar score at 1 and 5 min (1–9).DiscussionIn this study, we determined the composition of the va-ginal microbiota of women who had spontaneous pre-term birth and compared these profiles to those ofwomen who delivered at term, previously reported byour research group [24]. The availability of foundationaldata on women who delivered at term and the infeasibil-ity of collecting large numbers of samples at 11–16 weeksgestation from women who would go on to deliverpre-term, our study design included comparison ofsamples collected in a previously published study [24]and from the OBS. To minimize any batch effects, wewere rigorous in implementation of consistent sampleprocessing and did extensive analysis of the clinical anddemographic characteristics to ensure they were wellmatched (Table 1). The cohorts were comparable interms of maternal age, BMI, ethnicity, consumption oftobacco, alcohol and probiotics, which is of interestgiven that several of these characteristics have been pre-viously associated with preterm delivery. In particular,previous described factors included low and high mater-nal ages [5–7], low BMI [8], black ethnicity [9], highlevels of tobacco, alcohol and illicit drugs consumption[4], close temporal proximity to a previous delivery [10]and multiple gestation [11]. This cohort is unique in thatit did offer the opportunity to have gestational age at de-livery as the main characteristic distinguishing these twogroups recognizing that the majority of preterm birthsoccurred beyond 32 weeks gestation.A difference in number of previous gestations was ob-served between groups, with women who experiencedpreterm birth more likely to be primigravida in compari-son with women who had term deliveries. It has been re-cently demonstrated that women with a priorconception, regardless of whether or not this proceededto a birth, have a decrease in the relative abundance ofL. crispatus and a concomitant increase in the(See figure on previous page.)Fig. 1 Vaginal microbiota profiles of women who had sPTB and term deliveries. a Hierarchical clustering of Jensen–Shannon distance matriceswith Ward linkage on the relative proportions of reads for each OTU within individual vaginal samples. b Community state type (CST). c Gestationalage at delivery. d Heatmap of relative abundances of bacterial species within each vaginal microbiota. Each column represents a woman’s vaginalmicrobiota profile, and each row represents a bacteria species. Only species that are at least 1% abundant in at least one sample are shown.e Shannon diversity indices calculated for each samplePC2 (14.60%)PC3 (10.65%)PC1 (29.56%)PC2 (14.60%)Gestational age at delivery Community State Type (CST)Term Preterm I II III IVA IVC IVD VPC3 (10.65%)PC1 (29.56%)Fig. 2 Vaginal microbiota profiles coloured by gestational age at delivery or CST. Jackknifed principal coordinates analysis (PCoA) of Bray–Curtisdistance matrices of microbial profiles from all participants in the studyFreitas et al. Microbiome (2018) 6:117 Page 8 of 15abundance of other Lactobacillus species as well asGardnerella [54].Other known risk factors for sPTB include maternalmedical disorders like hypertension, asthma, diabetesand thyroid disease [4]. Although some women in bothcohorts reported these conditions, there were notenough participants to stratify the data based on the in-dividual disorder and therefore was not possible to in-vestigate the interaction between those medicalconditions and gestation outcome. We were, however,able to confirm previous reports of history of prematur-ity as a risk factor for preterm birth [55].Since many organisms isolated from the amniotic cav-ity of women who experienced preterm birth are alsofound in the genital tract [12–15], an intrauterine infec-tion ascending from the vagina is one of the currentlyhypothesized triggers of PTB [56]. In this study, how-ever, we did not identify a signature microbiota compos-ition (CST) associated with preterm birth. Thisobservation is consistent with the results presented byothers [29, 30]. CST assignments are largely driven bythe dominance of a single species, which may mask dif-ferences in rare taxa that would differentiate term andpreterm groups, and indeed, further analysis revealedthat the vaginal microbiota of women who experiencedpreterm birth was richer and more diverse than those ofwomen who delivered at term. Also, most women(84.8%) in our study were considered late preterm andalthough we cannot address this question, it is possiblethat sPTB driven by an ascending infection would bemore evident in a high-risk cohort or extreme pretermcases. A recent study of a high-risk pregnant cohort hasreported that L. iners was strongly associated with shortcervix and preterm birth, as L. crispatus was associatedwith term deliveries [57]. Those differences in study out-comes indicate that the pathogenesis of sPTB in low-and high-risk groups might be different. Identifying dif-ferences in the causes of early and late sPTB and therole of the vaginal microbiota in those processes will re-quire further study.One controversy that challenges the current hypoth-esis of preterm caused by an ascending infection is thatantibiotic administration to pregnant women with a dis-turbed vaginal microbiota does not improve outcome inmost cases, as demonstrated by study trials [58, 59] andsystematic reviews [60–62]. One explanation for the in-efficacy of antibiotic treatment in the prevention of pre-term birth relies is the high rates of antibiotic resistanceamong bacterial-vaginosis-associated bacteria [63, 64]. Inthis case, antibiotics not only do not kill the targetedbacteria, but might also reduce the vaginal Lactobacilluspopulation leading to an even more disturbed micro-biota, as recently demonstrated [65].In addition to differences in richness and diversity, dif-ferences in the microbiota between the two cohorts re-garding bacterial abundance and prevalence were also4 605Difference in abundance between condition (Median log 2)−505−5OTUNearest neighbour20Difference in abundance within condition (Median log2)8Lactobacillus acidophilusStreptococcus parasanguinisSphingobium yanoikuyae0 2 4 6 8 10Eubacterium siraeumPrevotella tanneraeMassilia timonaeClostridium innocuumPseudovibrio sp.CLR relative abundanceTermPretermSpeciesA BFig. 3 Bacteria relative abundance differences between term and preterm groups represented by ALDEx2. a ALDEx2 between- and within-differencevalues for individual organisms across gestational age category. Organisms (at OTU and nearest neighbour species level) with significant p values areshown as pink circles (Welch’s t statistical test). b Violin plots showing the bacteria relative abundance (centre log transformed, CLR) in term andpreterm groups. Only the eight bacteria with significant relative abundance differences between term and preterm groups are shown. In the violinplots, the white dot represents the median value, the black bar is the interquartile range, and the vertical width of the plot shows the density of thedata along the X-axisFreitas et al. Microbiome (2018) 6:117 Page 9 of 15Table 3 Bacteria prevalence in the vaginal microbiomes of women who delivered preterm and at termSpecies TotalreadsPreterm (n = 46) Term (n = 170) PrevalenceratioFDR*Reads Prevalence (%) Reads Prevalence (%)A term/pretermBifidobacterium infantis 56,246 409 19.57 55,837 40.00 2.04 0.033Lactobacillus delbrueckii subsp. lactis 48,063 9 4.35 48,054 28.82 6.63 0.005Lactobacillus acidophilus 17,440 569 89.13 16,871 97.65 1.10 0.033Prevotella amnii 13,339 68 4.35 13,271 21.76 5.01 0.023Pseudovibrio sp. 1634 52 15.22 1582 91.76 6.03 0.000Streptococcus parasanguinis 1199 63 19.57 1136 88.82 4.54 0.000Sphingobium yanoikuyae 1174 16 13.04 1158 80.00 6.13 0.000Prevotella tannerae 930 58 15.22 872 88.24 5.80 0.000Clostridium innocuum 814 42 13.04 772 81.18 6.22 0.000Eubacterium siraeum 641 43 17.39 598 78.24 4.50 0.000Massilia timonae 426 6 8.70 420 70.00 8.05 0.000B preterm/termPrevotella timonensis 11,450 4211 71.74 7239 45.29 1.58 0.005Dialister micraerophilus 7381 2850 67.39 4531 44.12 1.53 0.021Prevotella sp. 2216 626 47.83 1590 14.71 3.25 0.000Bacteroides coagulans 1283 1212 41.30 71 7.06 5.85 0.000Corynebacterium accolens 1083 999 52.17 84 13.53 3.86 0.000Porphyromonas uenonis 1040 795 28.26 245 12.94 2.18 0.038Actinomyces neuii subsp. anitratus 877 738 43.48 139 10.00 4.35 0.000Peptoniphilus harei 875 811 45.65 64 7.65 5.97 0.000Lactobacillus fermentum 729 664 17.39 65 5.29 3.29 0.026Facklamia hominis 711 709 15.22 2 1.18 12.93 0.000Prevotella bivia 638 228 26.09 410 8.24 3.17 0.005Prevotella corporis 535 463 23.91 72 5.88 4.07 0.000Corynebacterium timonense 503 355 34.78 148 10.00 3.48 0.000Corynebacterium genitalium 372 323 34.78 49 7.06 4.93 0.000Tepidanaerobacter sp. 329 303 23.91 26 4.12 5.81 0.000Corynebacterium amycolatum 264 228 32.61 36 4.71 6.93 0.000Parvimonas micra 193 183 10.87 10 1.18 9.24 0.005Mobiluncus curtsii subsp. curtsii 183 171 21.74 12 2.35 9.24 0.000Finegoldia magna 155 149 13.04 6 3.53 3.70 0.038Coprococcus eutactus 154 68 47.83 86 27.06 1.77 0.026Brevibacterium linens 147 142 10.87 5 0.59 18.48 0.000Rothia dentocariosa 144 120 30.43 24 5.29 5.75 0.000Streptococcus thermophilus 132 126 19.57 6 2.35 8.32 0.000Magnetospirillum magnetotacticum 116 40 43.48 76 24.12 1.80 0.033Dethiobacter alkaliphilus 101 97 10.87 4 1.76 6.16 0.017Brevibacterium massiliense 99 86 15.22 13 1.18 12.93 0.000Eremococcus coleocola 90 82 10.87 8 1.76 6.16 0.017Anaeromusa acidaminophila 86 85 15.22 1 0.59 25.87 0.000Arthrobacter globiformis 66 48 10.87 18 1.76 6.16 0.017Staphylococcus epidermidis 54 47 13.04 7 2.35 5.54 0.009Freitas et al. Microbiome (2018) 6:117 Page 10 of 15identified. The ALDEx2 analysis indicated that eight raretaxa were more abundant in the term group, which doesnot necessarily mean they are associated with a ‘health-ier’ state or implicated in preventing sPTB. Since thesebacteria are detected at very low abundance within themicrobiota profiles, their biological significance in thevaginal microbiome is questionable. Differences in theprevalence of several other taxa between groups werealso observed. For example, more women in the termgroup had Prevotella amnii and P. tannerae detected intheir vaginal samples, whereas P. timonensis, P. bivia, P.corporis and P. bucalis were more frequently detected insamples from women in the preterm group (Table 3).Prevotella spp. have been previously associated with bac-terial vaginosis and preterm labour [22, 66, 67], and ourresults indicate that different Prevotella species mighthave different roles in sPTB. Several of the taxa thatwere significantly different in their prevalence amongwomen in the two groups also had low sequence readcounts (Table 3). Further investigation would be re-quired to determine if these rare members of the micro-bial community play a yet unknown role in sPTB.It is also important to note that the number of bacterialspecies with greater prevalence in the preterm (49/60) washigher than in the term (11/60) cohort (Table 3), which isconsistent with our results of increased microbial richnessand diversity in the samples from women who experi-enced preterm birth. This might indicate that increasedrichness, rather than the presence of specific taxa, mightbe associated with sPTB. Those differences might also bean indicator of physiological/biochemical dissimilarities inthe vaginal microbiomes of women who deliver at term orpreterm. In other words, the physiological state that leadsto sPTB might also create an environment that supports aricher/more diverse microbiota.Our results also confirmed previous reports of an as-sociation between Mycoplasma and preterm birth [68].Mollicutes were detected significantly more often inwomen in the preterm group compared to women in theterm group, but no differences were observed inUreaplasma prevalence between groups indicating thatthe difference in Mollicutes prevalence is primarilydriven by the presence of Mycoplasma spp. Although in-dividual Mycoplasma species could not be discernedbased on assay used in our study, both Mycoplasmagenitalium [69–71] and Mycoplasma hominis [72–75]have been previously associated with negative reproduct-ive outcomes including PTB.Collectively, our overall findings were similar to othertwo studies, which provided us the opportunity toTable 3 Bacteria prevalence in the vaginal microbiomes of women who delivered preterm and at term (Continued)Species TotalreadsPreterm (n = 46) Term (n = 170) PrevalenceratioFDR*Reads Prevalence (%) Reads Prevalence (%)Corynebacterium simulans 52 41 26.09 11 1.76 14.78 0.000Anaeroglobus geminatus 52 49 15.22 3 1.18 12.93 0.000Cellvibrio gilvus 42 26 15.22 16 1.76 8.62 0.000Prosthecochloris vibrioformis 41 40 6.52 1 0.59 11.09 0.028Prevotella buccalis 33 33 15.22 0 0.00 – 0.000Acidaminococcus fermentans 31 29 13.04 2 1.18 11.09 0.000Brevibacterium casei 25 24 8.70 1 0.59 14.78 0.005Streptococcus sanguinis 23 18 13.04 5 2.35 5.54 0.009Atopobium parvulum 23 20 8.70 3 1.18 7.39 0.023Peptoniphilus duerdenii 22 20 10.87 2 0.59 18.48 0.000Pelobacter propionicus 21 18 6.52 3 0.59 11.09 0.028Staphylococcus hominis 21 17 15.22 4 1.76 8.62 0.000Atopostipes suicloacalis 18 8 10.87 10 1.18 9.24 0.005Corynebacterium pyruviciproducens 18 13 13.04 5 1.76 7.39 0.005Corynebacterium jeikeium 14 11 13.04 3 0.59 22.17 0.000Sporichthya polymorpha 13 11 8.70 2 1.18 7.39 0.023Rhodococcus jostii 13 7 8.70 6 1.18 7.39 0.023Nitrospina gracilis 10 5 8.70 5 0.59 14.78 0.005Megasphaera sp. BV3C16-1 10 9 6.52 1 0.59 11.09 0.028Panel A: species with greater prevalence in the term group (ratio term/preterm)Panel B: species with greater prevalence in the preterm group (ratio preterm/term)*FDR (false discovery rate) represents the corrected p value for multiple comparisonsFreitas et al. Microbiome (2018) 6:117 Page 11 of 15compare different study designs (based on different co-horts and barcode gene) that addressed the same re-search question. Hyman and colleagues [30] describedthe vaginal microbiota of 83 pregnant women (term n =66, preterm n = 17) based on Sanger sequencing ofcloned 16S rRNA genes. Samples were collected at eachtrimester and preterm was defined as delivery before37 weeks of gestation. There was no correlation betweenpreterm and absence/low abundance of Lactobacillus inthe microbiota; in other words, preterm outcome couldnot be predicted based on CST. Similar to our results,they found an association between increased microbiotadiversity and preterm delivery among women of whiteethnicity (n = 40) (data from women of others ethnicitieswas not included in the analysis because of small samplesizes). Although there was no association between CSTand ethnicity, it is important to note that most womenenrolled in this study described themselves as beingwhite, and it is possible that an increased sample size ofparticipants of other ethnicities could result in a differ-ent conclusion.Romero and colleagues [29] also investigated the vaginalmicrobiota of pregnant women who experienced preterm,defined as delivery before 34 weeks of gestation (termn = 72, preterm n = 18). The profiles were created by16S rRNA amplicon sequencing, and samples were col-lected every 4 weeks until 24 weeks of gestation and thenevery 2 weeks. They found no differences in the frequencyof different CST between women who had term and pre-term deliveries. Likewise, no differences in bacteria rela-tive abundance were observed between the two cohorts,although only bacteria that were present in at least 25% ofsamples were included in the analysis. These results areconsistent with our findings of bacterial abundance basedon the ALDEx analysis since we only found significant dif-ferences in relative abundance for eight rare bacteria. Un-like Hyman et al. [30] and our results, Romero et al. [29]did not find differences in microbiota diversity betweenwomen who delivered preterm and at term. One possibleexplanation for this contradictory result might be relatedto differences in participant ethnicity among these studies.While most women in our study and the Hyman et al.study described themselves as white, the majority of par-ticipants in the Romero et al. study described themselvesas African American. It has been reported that the com-position of the vaginal microbiota is strongly associatedwith a woman’s ethnicity [52, 76]. Other studies have alsodemonstrated that black ethnicity is associated with an in-creased microbiota diversity in comparison with whiteethnicity [77], which could have masked differences inbacterial diversity between term and preterm cohorts inthe Romero study.Contrary to our overall findings, DiGiulio and col-leagues [31] found a strong association between thenon-Lactobacillus-dominated CST IV and preterm birthin a case–control study based on the 16S rRNA ampli-con sequencing. Pregnant women (preterm n = 34, termn = 15), mostly of white ethnicity, were sampled weeklythroughout gestation. Interestingly, the authors pointedout that if samples had been collected less frequently,short-term ‘excursions’ to CST IV would have beenmissed and probably the association between CST IVand preterm birth would have been less obvious. The de-tection of a temporary microbiota disturbance repre-sented by a change from a Lactobacillus-dominated CSTto CST IV may have been missed in our study sincesamples were not collected longitudinally. Moreover, arecent study has demonstrated that PTB–microbiota as-sociations are population-dependent [32]; lowerLactobacillus and higher Gardnerella abundances wereassociated with PTB in a low-risk predominantlyCaucasian cohort, but not in a high-risk predominantlyAfrican American cohort. These population-dependentassociations might contribute to explain contradictoryconclusions among different studies and emphasize theimportance of investigating the vaginal microbiota of dif-ferent populations with varying ethnic backgrounds andfrom different geographical locations.Most samples in the preterm group were dominated byLactobacillus, yet, they collectively had higher richnessand diversity compared to samples from the term group.The increased microbiota richness/diversity might indi-cate a transient state between Lactobacillus-dominatedCST and non-Lactobacillus-dominated, i.e., CST IV (A, Cor D). In other words, the increased richness and diversitywe observed might be a remnant characteristic of the pre-vious disturbed microbiota. In summary, although we didnot “detect” a specific microbial community structure thatis associated with preterm birth, the increased microbiotarichness/diversity was associated with preterm birth. Inaddition, the association with differences in Prevotella spe-cies and Mycoplasma presence may point to signaturespecies associated with preterm birth.ConclusionsTaken together, our results suggest that the differencesin the microbiota of women who had preterm deliveries,such as increased microbiota richness and diversity andgreater prevalence of Mollicutes and other bacteria, mayhave a role in sPTB. Other differences between cohortsmight have been masked by the presence of highly dom-inant bacteria like Lactobacillus. At the overall level, wedid not identify a specific vaginal microbial communitystructure at 11–16 weeks gestation age that predictssPTB. Also, differences in relative abundance of bacterialspecies between term and preterm groups were only sig-nificant for a few low abundance species. Although acausal relationship remains to be determined, our resultsFreitas et al. Microbiome (2018) 6:117 Page 12 of 15confirm previous reports of an association betweenMollicutes and preterm birth, and further suggest that adiverse bacterial community may contribute to themicrobiome’s role in sPTB. Alternatively, the more richand diverse microbiotas of the preterm group may re-flect physiological differences between the groups thataffect selection of bacteria. This study provides valuableevidence of subtle alterations in the microbiome associ-ated with preterm birth that requires further study utiliz-ing sequencing methodology. In addition, future studyshould include evaluation of the microbial metaboliteproduction and host response to further elucidate factorsleading to sPTB and identify women at risk early inpregnancy.Additional filesAdditional file 1: cpn60 OTU sequences. Multiple fasta file containing728 OTU sequences. (TXT 336 kb)Additional file 2: Summary of OTU analysed in this study. OTU ID,percentage of identity, length, cpnDB name, species, and abundance ineach library are shown. (XLSX 500 kb)AcknowledgementsThe authors are grateful to the women who participated in the study andacknowledge the contribution and support of Ontario Birth Study Teammembers and the participants. The VOGUE Research Group is Deborah Money,Alan Bocking, Sean Hemmingsen, Janet Hill, Gregor Reid, Tim Dumonceaux,Gregory Gloor, Matthew Links, Kieran O’Doherty, Patrick Tang, Julianne vanSchalkwyk and Mark Yudin.FundingFinancial support was provided by a joint Canadian Institutes of Health Research(CIHR) Emerging Team Grant and a Genome British Columbia (GBC) grantawarded to DMM, AB and JEH (grant reference #108030) as well as a CIHRgrant MOP-82799 to AB. ACF was supported by a University of Saskatchewangraduate scholarship. Funding for the Ontario Birth Study was provided byMount Sinai Hospital Foundation, Lunenfeld-Tanenbaum Research Institute andthe Department of Obstetrics and Gynecology at Mount Sinai Hospital.Availability of data and materialsThe dataset supporting the results of this article is available in the NCBI SRArepository (Accession SRP073152, BioProject PRJNA317763; BioProjectPRJNA403856).Authors’ contributionsDM, AB, JEH and the other members of the VOGUE Research Group conceivedthe study. DM, AB and JEH oversaw and contributed to data collection andparticipated in writing and manuscript review. ACF generated the Mollicutes PCR,16S rRNA qPCR and the cpn60 microbiome profiles data, performed data analysisand wrote the paper. All authors read and approved the final manuscript.Ethics approval and consent to participateThis study received ethical approval from the University of British ColumbiaChildren’s and Women’s Research Ethics Board (Approval Number H14-01954),and Mount Sinai Hospital Research Ethics Board (Approval Number 15-0184-E).All participants provided written consent at enrolment.Consent for publicationNot applicable.Competing interestsThe authors declare that they have no competing interests.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims in publishedmaps and institutional affiliations.Author details1Department of Veterinary Microbiology, University of Saskatchewan,Saskatoon, SK S7N 5B4, Canada. 2Departments of Obstetrics andGynaecology and Physiology, University of Toronto, Toronto, ON M5G 1L4,Canada. 3Lunenfeld-Tanenbaum Research Institute, M5T1X5, Toronto, ON,Canada. 4Department of Obstetrics and Gynaecology, University of BritishColumbia, Vancouver, BC V6T 1Z4, Canada. 5Women’s Health ResearchInstitute, BC Women’s Hospital & Health Centre, Vancouver, BC V6H 3N1,Canada.Received: 19 October 2017 Accepted: 18 May 2018References1. WHO, March of Dimes, PMNCH, Save the Children. In: Howson CP, KinneyMV, Lawn JE, editors. Born too soon: the global action report on pretermbirth. Geneva: World Health Organization; 2012.2. Quinn J, Munoz FM, Gonik B, Frau L, Cutland C, Mallett-Moore T, et al.Preterm birth: case definition & guidelines for data collection, analysis, andpresentation of immunisation safety data. Vaccine. 2016;34:6047–56.3. Blencowe H, Cousens S, Oestergaard MZ, Chou D, Moller AB, Narwal R, et al.National, regional, and worldwide estimates of preterm birth rates in theyear 2010 with time trends since 1990 for selected countries: a systematicanalysis and implications. Lancet. Elsevier Ltd. 2012;379:2162–72.4. Behrman RE, Butler AS. In: Behrman RE, Butler AS, editors. Preterm birth:causes, consequences, and prevention. Washington, D.C.: NationalAcademies Press; 2007.5. Fraser AM, Brockert JE, Ward RH. Association of young maternal age withadverse reproductive outcomes. N Engl J Med. 1995;332:1113–8.6. Jacobsson B, Ladfors L, Milsom I. Advanced maternal age and adverseperinatal outcome. Obstet Gynecol. 2004;104:727–33.7. Schempf AH, Branum AM, Lukacs SL, Schoendorf KC. Maternal age andparity-associated risks of preterm birth: differences by race/ethnicity.Paediatr Perinat Epidemiol. 2007;21:34–43.8. Han Z, Mulla S, Beyene J, Liao G, McDonald SD. Maternal underweight andthe risk of preterm birth and low birth weight: a systematic review andmeta-analyses. Int J Epidemiol. 2011;40:65–101.9. Kistka ZAF, Palomar L, Lee KA, Boslaugh SE, Wangler MF, Cole FS, et al.Racial disparity in the frequency of recurrence of preterm birth. Am J ObstetGynecol. 2007;196:1–6.10. Conde-Agudelo A, Rosas-Bermúdez A, Kafury-Goeta AC. Birth spacing andrisk of adverse perinatal outcomes. JAMA. 2006;295:1809–23.11. Stock S, Norman J. Preterm and term labour in multiple pregnancies. SeminFetal Neonatal Med Elsevier Ltd. 2010;15:336–41.12. Gardella C, Riley DE, Hitti J, Agnew K, Krieger JN, Eschenbach DA.Identification and sequencing of bacterial rDNAs in culture-negativeamniotic fluid from women in premature labor. Am J Perinatol. 2004;21:319–23.13. Krohn MA, Hillier SL, Nugent RP, Cotch MF, Carey JC, Gibbs RS, et al. Thegenital flora of women with intraamniotic infection. J Infect Dis. 1995;171:1475–80.14. Hillier SL, Martius J, Krohn M, Kiviat N, Holmes KK, Eschenbach DA. A case–control study of chorioamnionic infection and histologic chorioamnionitis inprematurity. N Engl J Med. 1988;319:972–8.15. Romero R, Sirtori M, Oyarzun E, Avila C, Mazor M, Callahan R, et al. Infectionand labor V. Prevalence, microbiology, and clinical significance ofintraamniotic infection in women with preterm labor and intactmembranes. Am J Obstet Gynecol. 1989;161:817–24.16. Watts HD, Krohn MA, Hillier SL, Eschenbach DA. The association of occultamniotic fluid infection with gestational age and neonatal outcome amongwomen in preterm labor. Obstet Gynecol. 1992;79:351–7.17. Yoon BH, Romero R, Bin MJ, Shim S-S, Kim M, Kim G, et al. Clinicalsignificance of intra-amniotic inflammation in patients with preterm laborand intact membranes. Am J Obstet Gynecol. 2001;185:1130–6.18. DiGiulio DB, Romero R, Amogan HP, Kusanovic JP, Bik EM, Gotsch F, et al.Microbial prevalence, diversity and abundance in amniotic fluid during pretermlabor: a molecular and culture-based investigation. PLoS One. 2008;3:e3056.Freitas et al. Microbiome (2018) 6:117 Page 13 of 1519. Hillier SL, Witkin SS, Krohn MA, Watts DH, Kiviat NB, Eschenbach DA. Therelationship of amniotic fluid cytokines and preterm delivery, amniotic fluidinfection, histologic chorioamnionitis, and chorioamnion infection. ObstetGynecol. 1993;81:941–8.20. Jacobsson B, Mattsby-Baltzer I, Hagberg H. Interleukin-6 and interleukin-8 incervical and amniotic fluid: relationship to microbial invasion of thechorioamniotic membranes. BJOG An Int. J. Obstet. Gynaecol. 2005;112:719–24.21. Nugent RP, Krohn MA, Hillier SL. Reability of diagnosing bacterial vaginosisis improved by standardized method of gram stain interpretation. J ClinMicrobiol. 1991;29:297–301.22. Hill GB. The microbiology of bacterial vaginosis. Am J Obstet Gynecol. 1993;169:450–4.23. Romero R, Hassan SS, Gajer P, Tarca AL, Fadrosh DW, Nikita L, et al. Thecomposition and stability of the vaginal microbiota of normal pregnant womenis different from that of non-pregnant women. Microbiome. 2014;2:1–19.24. Freitas AC, Chaban B, Bocking A, Rocco M, Yang S, Hill JE, et al. The vaginalmicrobiome of pregnant women is less rich and diverse, with lower prevalenceof Mollicutes, compared to non-pregnant women. Sci Rep. 2017;7:1–16.25. Aagaard KM, Riehle K, Ma J, Segata N, Mistretta TA, Coarfa C, et al. Ametagenomic approach to characterization of the vaginal microbiomesignature in pregnancy. PLoS One. 2012;7:e36466.26. Walther-António MRS, Jeraldo P, Berg Miller ME, Yeoman CJ, Nelson KE,Wilson BA, et al. Pregnancy’s stronghold on the vaginal microbiome. PLoSOne. 2014;9:e98514.27. MacIntyre DA, Chandiramani M, Lee YS, Kindinger L, Smith A, AngelopoulosN, et al. The vaginal microbiome during pregnancy and the postpartumperiod in a European population. Sci Rep. 2015;5:1–9.28. Leitich H, Bodner-Adler B, Brunbauer M, Kaider A, Egarter C, Husslein P.Bacterial vaginosis as a risk factor for preterm delivery: a meta-analysis. Am JObstet Gynecol. 2003;189:139–47.29. Romero R, Hassan SS, Gajer P, Tarca AL, Fadrosh DW, Bieda J, et al. Thevaginal microbiota of pregnant women who subsequently havespontaneous preterm labor and delivery and those with a normal deliveryat term. Microbiome. 2014;2:1–15.30. Hyman RW, Fukushima M, Jiang H, Fung E, Rand L, Johnson B, et al.Diversity of the vaginal microbiome correlates with preterm birth. ReprodSci. 2014;21:32–40.31. DiGiulio DB, Callahan BJ, Mcmurdie PJ, Costello EK, Lyell DJ, Robaczewska A,et al. Temporal and spatial variation of the human microbiota duringpregnancy. Proc Natl Acad Sci U S A. 2015;112:11060–5.32. Callahan BJ, DiGiulio DB, Goltsman DSA, Sun CL, Costello EK, Jeganathan P,et al. Replication and refinement of a vaginal microbial signature of pretermbirth in two racially distinct cohorts of US women. Proc Natl Acad Sci U S A.2017;114:9966–71.33. Links MG, Dumonceaux TJ, Hemmingsen SM, Hill JE. The chaperonin-60universal target is a barcode for bacteria that enables de novo assembly ofmetagenomic sequence data. PLoS One. 2012;7:e49755.34. Paramel Jayaprakash T, Schellenberg JJ, Hill JE. Resolution andcharacterization of distinct cpn60-based subgroups of Gardnerella vaginalisin the vaginal microbiota. PLoS One. 2012;7:e43009.35. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Researchelectronic data capture (REDCap)—a metadata-driven methodology andworkflow process for providing translational research informatics support. JBiomed Inform Elsevier Inc. 2009;42:377–81.36. Lee DH, Zo YG, Kim SJ. Nonradioactive method to study genetic profiles ofnatural bacterial communities by PCR—single-strand-conformationpolymorphism. Appl Environ Microbiol. 1996;62:3112–20.37. Chaban B, Albert AYK, Links MG, Gardy J, Tang P, Hill JE. Characterization ofthe upper respiratory tract microbiomes of patients with pandemic H1N1influenza. PLoS One. 2013;8:e69559.38. Hill JE, Penny SL, Crowell KG, Goh SH, Hemmingsen SM. cpnDB: achaperonin sequence database. Genome Res. 2004;14:1669–75.39. van Kuppeveld FJM, van der Logt JTM, Angulo AF, van Zoest MJ, Quint WGV,Niesters HGM, et al. Genus- and species-specific identification of mycoplasmasby 16S rRNA amplification. Appl Environ Microbiol. 1992;58:2606–15.40. Watson HL, Blalock DK, Cassell GH. Variable antigens of Ureaplasmaurealyticum containing both serovar-specific and serovar-cross-reactiveepitopes. Infect Immun. 1990;58:3679–88.41. Chaban B, Links MG, Jayaprakash TP, Wagner EC, Bourque DK, Lohn Z, et al.Characterization of the vaginal microbiota of healthy Canadian womenthrough the menstrual cycle. Microbiome. 2014;2:1–12.42. Hill JE, Town JR, Hemmingsen SM. Improved template representation incpn60 polymerase chain reaction (PCR) product libraries generated fromcomplex templates by application of a specific mixture of PCR primers.Environ Microbiol. 2006;8:741–6.43. Schellenberg JJ, Links MG, Hill JE, Dumonceaux TJ, Kimani J, Jaoko W, et al.Molecular definition of vaginal microbiota in East African commercial sexworkers. Appl Environ Microbiol. 2011;77:4066–74.44. Links MG, Chaban B, Hemmingsen SM, Muirhead K, Hill JE. mPUMA: acomputational approach to microbiota analysis by de novo assembly ofoperational taxonomic units based on protein-coding barcode sequences.Microbiome. 2013;1:1–7.45. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. Full-length transcriptome assembly from RNA-Seq data without a referencegenome. Nat Biotechnol. 2011;29:644–52.46. Schellenberg JJ, Links MG, Hill JE, Dumonceaux TJ, Peters GA, Tyler SD, et al.Pyrosequencing of the chaperonin-60 universal target as a tool fordetermining microbial community composition. Appl Environ Microbiol.2009;75:2889–98.47. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practicaland powerful approach to multiple testing. J R Stat Soc. 1995;57:289–300.48. Caporaso JG, Kuczynski J, Stombaugh JI, Bittinger K, Bushman FD, CostelloEK, et al. QIIME allows analysis of high-throughput community sequencingdata. Nat Methods. 2010;7:335–6.49. Oksanen AJ, Blanchet FG, Kindt R, Legendre P, Minchin PR, Hara RBO, et al.Community Ecology Package “vegan”. R version 2.3-1. Available: https://cran.r-project.org/web/packages/vegan/. 2015.50. Endres DM, Schindelin JE. A new metric for probability distributions. IEEETrans Inf Theory. 2003;49:1858–60.51. Fernandes AD, Macklaim JM, Linn TG, Reid G, Gloor GB. ANOVA-likedifferential expression (ALDEx) analysis for mixed population RNA-Seq. PLoSOne. 2013;8:e67019.52. Ravel J, Gajer P, Abdo Z, Schneider GM, Koenig SSK, McCulle SL, et al.Vaginal microbiome of reproductive-age women. Proc Natl Acad Sci U S A.2011;108:4680–7.53. Albert AYK, Chaban B, Wagner EC, Schellenberg JJ, Links MG, van SchalkwykJ, et al. A study of the vaginal microbiome in healthy Canadian womenutilizing cpn60-based molecular profiling reveals distinct Gardnerellasubgroup community state types. PLoS One. 2015;10:e0135620.54. Nasioudis D, Forney LJ, Schneider GM, Gliniewicz K, France M, Boester A,et al. Influence of pregnancy history on the vaginal microbiome ofpregnant women in their first trimester. Sci. Rep. Springer US; 2017;7:1–6.55. Mercer BM, Goldenberg RL, Moawad AH, Meis PJ, Iams JD, Das AF, et al. Thepreterm prediction study: effect of gestational age and cause of preterm birthon subsequent obstetric outcome. Am J Obstet Gynecol. 1999;181:1216–21.56. Goldenberg RL, Hauth JC, Andrews WW. Intrauterine infection and pretermdelivery. N Engl J Med. 2000;342:1500–7.57. Kindinger LM, Bennett PR, Lee YS, Marchesi JR, Smith A, Cacciatore S, et al.The interaction between vaginal microbiota, cervical length, and vaginalprogesterone treatment for preterm birth risk. Microbiome. 2017;5:1–14.58. Kekki M, Kurki T, Pelkonen J, Kukinen-Raty M, Cacciatore B, Paavonen J.Vaginal clindamycin in preventing preterm birth and peripartal infections inasymptomatic women with bacterial vaginosis: a randomized, controlledtrial. Obstet Gynecol. 2001;97:643–8.59. Andrews WW, Goldenberg RL, Hauth JC, Cliver SP, Copper R, Conner M.Interconceptional antibiotics to prevent spontaneous preterm birth: arandomized clinical trial. Am J Obstet Gynecol. 2006;194:617–23.60. Morency A, Bujold E. The effect of second-trimester antibiotic therapy onthe rate of preterm birth. J Obstet Gynaecol Canada. 2007;29:35–44.61. Okun N, Gronau KA, Hannah ME. Antibiotics for bacterial vaginosis orTrichomonas vaginalis in pregnancy: a systematic review. Obstet Gynecol.2005;105:857–68.62. Thinkhamrop J, Hofmeyr GJ, Adetoro O, Lumbiganon P, Ota E. Antibioticprophylaxis during the second and third trimester to reduce adversepregnancy outcomes and morbidity. Cochrane Database Syst Rev. 2015;1:CD002250.63. Beigi RH, Austin MN, Meyn LA, Krohn MA, Hillier SL. Antimicrobial resistanceassociated with the treatment of bacterial vaginosis. Am J Obstet Gynecol.2004;191:1124–9.64. Budd W, Bostwick DG, Woody J, Hunt C. Antimicrobial resistance genes andmodelling of treatment failure in bacterial vaginosis: clinical study of 289symptomatic women. J Med Microbiol. 2016;65:377–86.Freitas et al. Microbiome (2018) 6:117 Page 14 of 1565. Brown RG, Marchesi JR, Lee YS, Smith A, Lehne B, Kindinger LM, Terzidou V,Holmes E, Nicholson JK, Bennett PR, et al. Vaginal dysbiosis increases risk ofpreterm fetal membrane rupture, neonatal sepsis and is exacerbated byerythromycin. BMC Med. 2018;16(1):9.66. Paramel Jayaprakash T, Wagner EC, van Schalkwyk J, Albert AYK, Hill JE,Money DM. High diversity and variability in the vaginal microbiome inwomen following preterm premature rupture of membranes (PPROM): aprospective cohort study. PLoS One. 2016;11:e0166794.67. Ling Z, Kong J, Liu F, Zhu H, Chen X, Wang Y, et al. Molecular analysis ofthe diversity of vaginal microbiota associated with bacterial vaginosis. BMCGenomics. 2010;11:1–16.68. Foxman B, Wen A, Srinivasan U, Goldberg D, Marrs CF, Owen J, et al.Mycoplasma, bacterial vaginosis-associated bacteria BVAB3, race, and risk ofpreterm birth in a high-risk cohort. Am J Obstet Gynecol. 2014;210:226.e1–7.69. Hitti J, Garcia P, Totten P, Paul K, Astete S, Holmes KK. Correlates of cervicalMycoplasma genitalium and risk of preterm birth among Peruvian women.Sex Transm Dis. 2010;37:81–5.70. Edwards RK, Ferguson RJ, Reyes L, Brown M, Theriaque DW, Duff P.Assessing the relationship between preterm delivery and variousmicroorganisms recovered from the lower genital tract. J. Matern. NeonatalMed. 2006;19:357–63.71. Lis R, Rowhani-Rahbar A, Manhart LE. Mycoplasma genitalium infection andfemale reproductive tract disease: a meta-analysis. Clin Infect Dis. 2015;61:418–26.72. Donders GG, Van Calsteren K, Bellen G, Reybrouck R, Van Den Bosch T,Riphagen I, et al. Predictive value for preterm birth of abnormal vaginalflora, bacterial vaginosis and aerobic vaginitis during the first trimester ofpregnancy. BJOG An Int J Obstet Gynaecol. 2009;116:1315–24.73. Kwak D-W, Hwang H-S, Kwon J-Y, Park Y-W, Kim Y-H. Co-infection withvaginal Ureaplasma urealyticum and Mycoplasma hominis increases adversepregnancy outcomes in patients with preterm labor or preterm prematurerupture of membranes. J Matern Neonatal Med. 2014;27:333–7.74. Hillier SL, Nugent RP, Eschenbach DA, Krohn MA, Gibbs RS, Martin DH, et al.Association between bacterial vaginosis and preterm delivery of a low-birth-weight infant. N Engl J Med. 1995;333:1737–42.75. Lamont RF, Taylor-Robinson D, Wigglesworth JS, Furr PM, Evans RT, ElderMG. The role of mycoplasmas, ureaplasmas and chlamydiae in the genitaltract of women presenting in spontaneous early preterm labour. J MedMicrobiol. 1987;24:253–7.76. Zhou X, Brown CJ, Abdo Z, Davis CC, Hansmann MA, Joyce P, et al.Differences in the composition of vaginal microbial communities found inhealthy Caucasian and black women. ISME J. 2007;112:121–33.77. Fettweis JM, Brooks JP, Serrano MG, Sheth NU, Girerd PH, Edwards DJ, et al.Differences in vaginal microbiome in African American women versuswomen of European ancestry. Microbiology. 2014;160:2272–82.Freitas et al. Microbiome (2018) 6:117 Page 15 of 15