Clinical Infectious Diseases MAJOR ARTICLE

Transcription

1 Clinical Infectious Diseases MAJOR ARTICLE The Cervicovaginal Microbiota in Women Notified for Chlamydia trachomatis Infection: A Case-Control Study at the Sexually Transmitted Infection Outpatient Clinic in Amsterdam, The Netherlands Charlotte van der Veer, 1 Sylvia M. Bruisten, 1 Jannie J. van der Helm, 2 Henry J. C. de Vries, 2,3 and Robin van Houdt 4 1 Public Health Laboratory, and 2 Sexually Transmitted Infections Outpatient Clinic, Public Health Service Amsterdam, 3 Department of Dermatology, Amsterdam Medical Centre, University of Amsterdam, and 4 Department of Medical Microbiology and Infection Prevention, VU University Medical Centre, Amsterdam, The Netherlands (See the Editorial Commentary by DiGiulio on pages 32 3.) Background. Increasing evidence suggests that the cervicovaginal microbiota (CVM) plays an important role in acquiring sexually transmitted infections (STIs). Here we study the CVM in a population of women notified by a sex partner for Chlamydia trachomatis infection. Methods. We included 98 women who were contact-traced by C. trachomatis positive sex partners at the STI outpatient clinic in Amsterdam, the Netherlands, and analyzed their cervicovaginal samples and clinical data. CVMs were characterized by sequencing the V3/V4 region of the 16S ribosomal RNA gene and by hierarchical clustering. Characteristics associating with C. trachomatis infection were examined using bivariable and multivariable logistic regression analysis. Results. The CVM was characterized for 93 women, of whom 52 tested C. trachomatis positive and 41 C. trachomatis negative. We identified 3 major CVM clusters. Clustered CVM predominantly comprised either diverse anaerobic bacteria (n = 39 [42%]), Lactobacillus iners (n = 32 [34%]), or Lactobacillus crispatus (n = 22 [24%]). In multivariable analysis, we found that CVM was significantly associated with C. trachomatis infection (odds ratio [OR], 4.2 [95% confidence interval {CI}, ] for women with diverse anaerobic CVM and OR, 4.4 [95% CI, ], for women with L. iners dominated CVM, compared with women with L. crispatus dominated CVM), as was younger age (OR, 3.1 [95% CI, ] for those 21 years old) and reporting a steady sex partner (OR, 3.6 [95% CI, ]). Conclusions. Women who tested positive for Chlamydia trachomatis infection after having been contact-traced by a chlamydiapositive partner were more likely to have CVM dominated by L. iners or by diverse anaerobic bacteria, than by L. crispatus. Keywords. Chlamydia trachomatis; cervicovaginal microbiota; contact tracing. Chlamydia trachomatis is the most common bacterial sexually transmitted infection (STI), affecting >100 million people worldwide, with young women (<25 years old) being most at risk [1, 2]. Epidemiological studies, using microscopy, have demonstrated that a depletion of lactobacilli in the cervicovaginal microbiota (CVM) was associated with increased risk for STIs, including human immunodeficiency virus (HIV) [3 5]. Molecular epidemiological studies, using mostly 16S ribosomal RNA (rrna) gene sequencing, have identified several common CVM compositions, of which 4 are dominated by a single Lactobacillus species: L. iners, L. crispatus, L. jensenii, orl. gasseri [6, 7]. The other CVM compositions are not dominated by single Received 9 March 2016; accepted 6 July 2016; published online 27 August Correspondence: C. van der Veer, Public Health Laboratory, Amsterdam Public Health Service, Nieuwe Achtergracht 100, 1018 WT, Amsterdam, The Netherlands (cvdveer@ggd.amsterdam.nl) Clinical Infectious Diseases 2017;64(1):24 31 The Author Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, journals.permissions@oup.com. DOI: /cid/ciw586 taxa but rather consist of a mixture of anaerobic bacteria such as Gardnerella vaginalis, Atopobium vaginae, Megasphaera, Prevotella, and BVAB 1-3[6, 7]. High numbers of resident lactobacilli are thought to protect the urogenital tract from pathogenic microorganisms by competing for nutrients, reducing the ph (<4.5) through secretion of lactic acid, producing antimicrobials, and by modulating the immune system. But not all Lactobacillus species are associated with vaginal health: CVMs dominated by L. iners are more often associated with adverse outcomes, such as bacterial vaginosis (BV) and/or STIs, than CVMs dominated by other Lactobacillus species [8 12]. Molecular studies looking into the association between C. trachomatis infection and the CVM are scarce [12], and studies using microscopy have included low numbers of C. trachomatis cases [13, 14]. In the present study, we aimed to study the association between the CVM and C. trachomatis infection in women who were notified by a C. trachomatis positive sex partner and who were therefore presumably recently exposed to C. trachomatis. Clients of the STI clinic in Amsterdam, the 24 CID 2017:64 (1 January) van der Veer et al

2 Netherlands, who test positive for an STI are encouraged to notify their sex partners and are provided with an official contact-tracing slip to give to their sex partners. Here we collected cervicovaginal samples and demographic and clinical data from women presenting with such a C. trachomatis contact-tracing slip and compared the CVM of women testing positive for C. trachomatis to the CVM of women testing negative for C. trachomatis. METHODS Study Population and Sample Collection Our study was conducted at the STI clinic in Amsterdam, the Netherlands. Clients of the STI clinic are notified that remainders of their samples may be used for scientific research after anonymization. If clients object, their samples and data are discarded. This procedure was approved by the Amsterdam University Medical ethical committee (reference number W15_159, number ). For this study, we retrieved processed samples and data from women who had presented with a C. trachomatis contact-tracing slip at the STI clinic. During the timeframe of sample collection for this study, no contact-traced women objected to their data being used. As part of routine management, nurse-collected cervical and/or vaginal swabs were used for C. trachomatis testing, using the single Chlamydia trachomatis TMA assay up to May 2014, and the combo Chlamydia trachomatis/neisseria gonorrhoeae TMA assay (APTIMA, Hologic, Bedford, Massachusetts) assays thereafter. All samples retrieved for this study were stored at 80 C until further processing. Vaginal smears, when available, were Gram stained and assessed by the Nugent score [15]. A data manager collected and anonymized demographic and clinical data from electronic patient files. DNA Extraction Samples were thawed on ice and vortexed vigorously. A 500-µL aliquot was transferred to a sterile 2-mL Lysating Matrix Tube B (MPBio, Leicester, United Kingdom,). Samples were lysed with lysozyme (10 mg/ml; Sigma-Aldrich, St Louis, Missouri), mutanolysin (>4 U/µL; Sigma-Aldrich), lysostaphin (1 mg/ml; AMBI,Lawrence,NewYork),proteinaseK(20mg/mL;Thermo Fisher, Waltham, Massachusetts), and Rnase A (10 mg/ml; Thermo Fisher). A mechanical bead beating step followed, at 5 m/s for 40 seconds (MagnaPure, Roche Diagnostics, Basel, Switzerland). DNA was extracted from the crude lysates using the ChemaGen extraction robot (PerkinElmer, Baesweiler, Germany,). DNA templates were measured using the Quant-iT PicoGreen dsdna assay (Thermo Fisher) and normalized to contain a maximum DNA concentration of 5 ng/µl. DNA templates were stored at 80 C until further processing. 16S rrna Gene Amplification and Illumina MiSeq Sequencing Dual indexed universal primers 319F and 806R were used for polymerase chain reaction (PCR) amplification of the V3 V4 hypervariable regions of the 16S rrna genes, as previously described by Fadrosh et al [16]. PCR reactions were performed using a Gene Amp PCR System 9700 (Applied Biosystem, Carlsbad, California) using Phusion Mix (Thermo Fisher) and 5 25 ng of extracted DNA in a total volume of 25 µl. PCR products were quantified using the Quant-iT PicoGreen dsdna assay (Thermo Fisher) and pooled, with each sample contributing a maximum of 125 ng amplified DNA to the pool. The pooled sample was purified with Agencourt AMPure XP magnetic beads (Beckman Coulter, Brea, California). Raw Data Cleaning Unaligned sequences underwent quality control; only keeping sequences with >99.9% base call accuracy (Trimmomatic tool; [17]) and sequences were aligned and truncated (keeping sequences of >400 bp) using PandaSeq [18]. Primers were trimmed off the aligned sequences using a sliding door approach with tagcleaner [19]. Aligned sequences were mapped to their corresponding barcodes using the de-multiplex tool in QIIME, version 1.9 [20]. Operational taxonomic units (OTUs) were picked using the usearch tool in QIIME at 97% identity. A representative sequence for each OTU was taken forward and aligned to a template of sequences from the vaginal 16S reference package developed by Srinivasan et al [11], using the PyNAST alignment tool in QIIME with uclust clustering [21]. Taxonomic Classification Taxonomic classification of the representative OTUs was done by heuristic phylogenetic placement of query sequences, using PPLACER [22]. We used the previously published vaginal 16S reference package for our analysis [11]. Common OTUs in our dataset that could not be identified to a species level using the vaginal reference package were searched using BLAST and assigned to the deepest taxonomical level based on available published data. Relative abundances per taxonomy were calculated per sample. Samples with <100 reads were excluded from CVM analysis. Cluster analysis was performed in R (version 3.2.1) [23]. A dendrogram based on the maximum distance in matrix dissimilarity between relative abundances per sample was created using the ggplot2 package in R [24]. Statistical Analysis All statistical analyses were performed in SPSS Statistics software version 21 (IBM). Differences in study population characteristics were tested using the χ 2, Fisher exact (when the expected value was <5), or Mann Whitney U test, accordingly. To examine whether the CVM was associated with C. trachomatis infection, we performed a bivariable and multivariable logistic regression analysis. We used a directed acyclic graph to identify other variables that could be associated with C. trachomatis infection [25]. In univariable analysis, we assessed the effect of the CVM, age ( 21 or >21 years), ethnicity, Nugent score, clinical diagnosis of bacterial vaginosis, and determinants of sexual behavior (number of sex partners Vaginal Microbiota and Chlamydia CID 2017:64 (1 January) 25

3 in the preceding 6 months, condom use at last sex act, and type of last sex partner [steady or nonsteady]). Variables that did not display collinearity and that were associated with C. trachomatis infection at P.1 in univariable analysis were entered into a multivariable model. In the multivariable model we considered P <.05 as statistically significant. Additionally, to evaluate which bacterial taxa were most likely to explain the difference in CVM with respect to C. trachomatis infection, we performed a linear discriminant analysis effect size analysis with α set to.05 [26]. RESULTS Study Population From September 2013 to December 2014, we included samples and data from 98 women who had presented themselves at the STI clinic with a contact-tracing slip indicating that a sex partner had tested positive for C. trachomatis. Five women testing positive for C. trachomatis were excluded from statistical analyses as their CVM could not be determined. Of the 93 remaining women, 41 tested C. trachomatis negative and 52 tested C. trachomatis positive. Nugent scores were performed for 47 women; 29 of these women were C. trachomatis negative and 18 women were C. trachomatis positive. The C. trachomatis positive and C. trachomatis negative groups were similar in most demographic and behavioral characteristics (Table 1). Differences in the following 2 characteristics were of note: (1) The C. trachomatis positive group was younger (median age, 20.7 [interquartile range {IQR}, ] years) than the C. trachomatis negative group (median age, 22.9 [IQR, ] years; P =.010); and (2) the C. trachomatis positive women were more likely to report that their last sex act was with a steady partner, as opposed to a nonsteady partner (P =.005). Cervicovaginal Microbiota Clusters The 93 included women provided 46 vaginal and 82 cervical samples (Figure 1). An average of reads were generated per sample, with cervical samples producing fewer reads than vaginal samples (median, [IQR, ] and [IQR, ], respectively). In total, 230 OTUs were generated and allocated to 101 different taxonomies with a minimum likelihood >0.8. We identified 3 major CVM clusters (Figure 2). These CVMs either predominantly consisted of diverse anaerobic bacteria (n = 39 [42%]), L. iners (n = 32 [34%]), or L. crispatus (n = 22 [24%]). Two women had discordant cervical/vaginal cluster allocations; in that case the cervical cluster was used for statistical analysis (Figure 1). Characteristics Associated With C. trachomatis Infection In univariable analysis, having a CVM dominated by L. iners or diverse anaerobic bacteria was associated with having a C. trachomatis infection (OR, 4.2 [95% confidence interval {CI}, ] and 4.4 [95% CI, ], respectively), compared with having an L. crispatus dominated CVM (P =.013) Table 1. Characteristics of Chlamydia trachomatis Contact-Traced Women Attending the Sexually Transmitted Infection Outpatient Clinic in Amsterdam, The Netherlands, September 2013 December 2014, Stratified by C. trachomatis Infection Status Characteristic C. trachomatis Negative (n = 41), No. (%) C. trachomatis Positive (n = 52), No. (%) P Value Demographic characteristics Age y 10 (24.4) 27 (51.9) >21 y 31 (75.6) 25 (48.1) Ethnicity.502 European 30 (73.2) 34 (65.4) Non-European a 11 (26.8) 18 (34.6) Behavioral characteristics No. of sex partners in the.053 preceding 6 mo 1 10 (24.4) 23 (44.2) >1 31 (75.6) 29 (55.8) Type of last sex partner.005 (n = 89) b Steady 16 (40.0) 35 (71.4) Nonsteady 24 (60.0) 14 (28.6) Condom use at last sex act.606 (n = 90) b Yes 10 (24.4) 9 (18.4) No 31 (75.6) 43 (81.6) Urogenital symptoms.082 Yes 10 (24.4) 22 (42.3) No 31 (75.6) 30 (57.7) Contraceptive use.289 No contraception or 18 (43.9) 17 (32.7) nonhormonal (IUD) Hormonal (Pill/injection/ 23 (56.1) 35 (67.3) Implanon/Evra/Nuva) Use of antibiotics Preceding 3 mo 8 (19.5) 8 (15.4).783 Preceding 7 d 0 (0) 2 (3.8).502 Laboratory characteristics Nugent score (n = 47) b (No bacterial 23 (79.3) 8 (44.4) vaginosis) 4 6 (Intermediate) 4 (13.8) 9 (50) 7 10 (Bacterial vaginosis) 2 (6.9) 1 (5.6) Clinical bacterial vaginosis c 6 (14.6) 11 (21.2).590 Candida 0 (0) 7 (13.5).016 Condylomata acuminate 0 (0) 2 (3.8).502 Neisseria gonorrhoeae 0 (0) 4 (7.7).127 Trichomonas vaginalis 1 (2.4) 0 (0).441 HIV status (n = 90) b.581 Negative 39 (100) 51 (100) Positive 0 (0) 0 (0) Cervicovaginal microbiota.008 clusters Lactobacillus crispatus 16 (39.0) 6 (11.5) Lactobacillus iners 11 (26.8) 21 (40.4) Diverse anaerobic bacteria 14 (34.2) 25 (48.1) Values in boldface are statistically significant at P <.05. Abbreviations: C. trachomatis, Chlamydia trachomatis; HIV, human immunodeficiency virus; IUD, intrauterine device. a Non-European ethnicity was as follows: n = 9 (30%) African; n = 12 (43.3%) Dutch Antilles; n = 5 (16.7%) Asian; n = 2 (6.7%) Central or South American; and n = 1 (3.3%) other ethnicity. b Data missing on condom use (n = 3), type of last sex partner (n = 4), HIV status (n = 3), and Nugent score (n = 46). c Clinical bacterial vaginosis was diagnosed in the presence of vaginal discharge, clue cells, and a positive whiff test. 26 CID 2017:64 (1 January) van der Veer et al

4 Figure 1. Flowchart showing study population and sample composition of women notified for Chlamydia trachomatis attending the outpatient sexually transmitted infection clinic, Amsterdam, the Netherlands, September 2013 December All samples were included for cervicovaginal microbiome (CVM) cluster analysis. If cervical and vaginal samples coming from the same woman were allocated to different cluster types, the cluster to which the cervical sample belonged was included in the statistical analysis. (Table 2). In univariable analysis, ethnicity was not associated with C. trachomatis infection (P =.502), nor was condom use (P =.606), or clinical diagnosis of BV (P =.422). Nugent score was associated with C. trachomatis infection (OR, 6.5 [95% CI, ] for intermediate score and OR, 1.4 [95% CI, ] for high score, both compared to low Nugent score; P =.037). Younger age ( 21 years: OR, 3.0 [95% CI, ]; P =.015), reporting 1 sex partner in the preceding 6 months (compared with >1 partner: OR, 2.5 [95% CI, ]; P =.050),andreporting that the last sex act was with a steady sex partner (compared with a nonsteady partner: OR, 3.9 [95% CI, ]; P =.002) were all associated with C. trachomatis infection. The latter 2 variables were associated with each other (P <.001) and most likely reflect a steady partnership; women reporting 1 sex partner in the preceding 6 months also often reported that their last sex act was with a steady partner. Of these 2 variables, type of last partner was most strongly associated with C. trachomatis infection and was considered in multivariable analysis. The Nugent score displayed collinearity with the CVM (P <.001) and was, therefore, excluded from multivariable analysis. In multivariable analysis, C. trachomatis infection was associated with CVM: OR, 4.2 [95% CI, ] for L. iners dominated CVMs and OR, 4.4 [95% CI, ] for diverse anaerobic CVMs, compared with L. crispatus dominated CVMs (P =.043); younger age ( 21 years: OR, 3.1 [95% CI, ]; P =.026 compared with >21 years) and reporting a Vaginal Microbiota and Chlamydia CID 2017:64 (1 January) 27

5 Figure 2. Hierarchical clustering of 128 cervicovaginal samples from 93 women notified for Chlamydia trachomatis attending the outpatient sexually transmitted infection clinic, Amsterdam, the Netherlands, September 2013 December The dendrogram was generated using complete linkage clustering with Euclidean distance, based on the relative abundance of taxa in each sample. Relative abundance is illustrated by the color key. Sidebars above the heatmap depict cervicovaginal microbiome (CVM) clusters (top bar), Nugent score (middle bar; white spaces represent missing data), and C. trachomatis status (bottom bar). steady sex partner (OR, 3.6 [95% CI, ]; P =.009,compared with a nonsteady partner). In the linear discriminant analysis effect size analysis (Figure 3), L. iners, L. crispatus, Lactobacillus vaginalis, Lachnospiraceae, Gemella, BVAB3, and Actinomyces urogenitalis were the species that showed statistically and biologically consistent differences between women who had C. trachomatis infection and women who were not infected. However, apart from L. crispatus and L. iners, these species were rare (<1% abundance) within this population. The results indicate that L. iners was associated with having C. trachomatis infection, whereas L. crispatus was not associated with C. trachomatis infection. DISCUSSION All women included in this study were assumed to be exposed recently to C. trachomatis by a sexual partner with proven C. trachomatis infection. We compared the CVM of women with a chlamydial infection to those without. We found that women testing positive for C. trachomatis infection were more likely to have a CVM dominated by L. iners or diverse anaerobic bacteria than by L. crispatus. Furthermore, C. trachomatis positivity was more likely to occur in young women ( 21 vs >21 years old) and women reporting a steady sexual partner, as opposed to a casual partner. Presumably, most individuals in our study were exposed to C. trachomatis in just a few days or weeks prior to sampling, as more than half of contact-traced individuals came to the STI clinic within 3 days and >95% within 3 weeks according to a study performed in 2010 at the STI clinic in Amsterdam (unpublished data). Studies looking into spontaneous clearance of C. trachomatis are scarce, but the available data suggest that clearance can take from several weeks to several years [27, 28]. Therefore, it is unlikely that a spontaneous clearance of C. trachomatis infection could have occurred in the C. trachomatis negative group. 28 CID 2017:64 (1 January) van der Veer et al

6 Table 2. Univariable and Multivariable Logistic Regression Analyses of Characteristics Associated with Chlamydia trachomatis Infection Among C. trachomatis Contact-Traced Women Attending the Sexually Transmitted Infection Outpatient Clinic in Amsterdam, The Netherlands, September 2013 December 2014 Univariable (n = 93) Multivariable (n = 89) Characteristic OR (95% CI) P Value Adjusted OR (95% CI) P Value Demographic characteristics Age y 3.0 ( ) 3.2 ( ) >21 y 1 1 Ethnicity.422 European 1 Non-European a 1.4 (.6 3.5) Behavioral characteristics No. of sex partners in the last 6 mo ( ) >1 1 Type of last sex partner (n = 89) b Steady 3.9 ( ) 3.6 ( ) Nonsteady 1 1 Condom use at last sex act (n = 90) b.354 Yes 1 No 1.6 (.6 4.4) Laboratory characteristics Nugent score (n = 47) b (No BV) (Intermediate) 6.5 ( ) 7 10 (BV) 1.4 ( ) Clinical bacterial vaginosis c.422 No BV 1 BV 1.6 (.5 4.7) Cervicovaginal microbiota clusters Lactobacillus crispatus 1 1 Lactobacillus iners 5.1 ( ) 4.2 ( ) Diverse anaerobic bacteria 4.8 ( ) 4.4 ( ) Values in boldface are statistically significant at P <.05. Abbreviations: BV, bacterial vaginosis; CI, confidence interval; OR, odds ratio. a Non-European ethnicity was as follows: n = 9 (30%) African; n = 12 (43.3%) Dutch Antilles; n = 5 (16.7%) Asian; n = 2 (6.7%) Central or South American; and n = 1 (3.3%) other ethnicity. b Data missing on condom use (n = 3), steady sex partner (n = 4), and Nugent score (n = 46). c Clinical BV was diagnosed in the presence of vaginal discharge, clue cells, and a positive whiff test. Figure 3. Bacterial taxa that showed statistically and biologically consistent differences between women with and without Chlamydia trachomatis infection. Bacterial taxa were compared using a linear discriminant analysis (LDA) effect size analysis with α set to.05. Bar plots in black show LDA scores of taxa more prevalent in C. trachomatis positive samples, and bar plots in white show LDA scores of taxa more prevalent in C. trachomatis negative samples. Vaginal Microbiota and Chlamydia CID 2017:64 (1 January) 29

7 Only 1 other study examined the CVM in relation to C. trachomatis infection in C. trachomatis contact-traced women [5]. As in our study, they found that women with C. trachomatis infection were >3 times more likely to have vaginal dysbiosis, compared with having lactobacilli-dominated CVM [5]. Furthermore, a large prospective cohort study of vaginal microbiota in relation to STI (n = 3620 women followed quarterly for 1 year), using the Nugent score, found that intermediate and BV-like Nugent scores were associated with a 1.5- to 2-fold increased risk for incident Trichomonas vaginalis, N. gonorrhoeae, orc. trachomatis infection [3]. Similarly, Gallo et al found that the odds for C. trachomatis infection increased by 1.6-fold in women who had BV based on Amsel criteria [29]. Other longitudinal studies have found comparable, albeit statistically nonsignificant, point estimates for increased chlamydial infection incidence among women with BV [14, 30]. We also performed a longitudinal study in which the composition of the CVM of women 1 year before they were diagnosed with C. trachomatis was assessed using 16S rrna sequencing [12]. These data showed, similar to our current findings, that L. crispatus dominance was not associated with C. trachomatis infection, whereas L. iners was associated. Findings based on BV diagnoses by Nugent scores or Amsel criteria alone, may therefore give conservative estimates for association between the CVM and C. trachomatis infection as women with CVM dominated by L. iners,by definition, have healthy Nugent scores and do not fulfill the Amsel criteria for BV. Observational studies, using molecular tools, have also found that L. iners dominance was associated with STI prevalence [11, 31]. The link between lactobacilli and vaginal health is often attributed to the production of lactic acid. This acidifies the vagina (ph < 4.5) and restricts growth of other (pathogenic) organisms. Besides acidification, Witkin and colleagues propose that an L. crispatus induced low L/D-lactic acid ratio may also help protect against STIs by reducing host gene expression of matrix metalloproteinase and so keeping the endocervical barrier intact [32]. Also, others have found that L. crispatus inhibits adhesion and infectivity of C. trachomatis in human epithelial cells and that it reduces severe inflammation by stimulating the production of the anti-inflammatory cytokine interleukin (IL) 10 and reduces the production of the proinflammatory cytokines IL-6, IL-8, and tumor necrosis factor α [33, 34]. Lactobacillus iners, on the other hand, does not seem to be able to produce D-lactic acid, as in vitro and genome analyses have indicated [32]. Rather, L. iners likely metabolizes its carbohydrate source (mostly glycogen) into succinate and short-chain fatty acids, resulting in a ph > 4.5 [32, 35, 36]. Given the relatively small genome of L. iners (1.32 Mb, compared with Mb for other common vaginal Lactobacillus species), it is plausible that symbiotic relationships with other vaginal bacteria exist that allow L. iners to survive under differing vaginal conditions, possibly even differentially expressing approximately 10% of its genome during healthy or dysbiotic states [32, 36]. If so, L. iners could possibly reflect a transitional phase from healthy to BV-like state, or vice-versa. A strength of our study is that we have included a substantial number of well-characterized C. trachomatis cases and controls who had all been warned for chlamydial infection. However, the cross-sectional nature of our data limits any inference of causality that is, we do not know the CVM composition prior to exposure to C. trachomatis and to what extent the C. trachomatis infection is driving the CVM composition. Drivers of the CVM composition in reproductive-age women have not yet been fully described, but it has been suggested that factors such as hormonal contraception, recent sexual intercourse, vaginal douching, menses, and also STIs could impact the CVM composition and possibly cause fluctuations in CVM composition over time [29, 35, 37 39]. As we do not have information on all of the aforementioned variables, it is possible that residual confounding exists in our data. Regarding sexual risk behavior, we found that C. trachomatis positive women were more likely to report a steady sexual partner, as opposed to a casual partner. This association with C. trachomatis infection can possibly be explained by more frequent sex with a steady partner (transmission per partnership) than with a casual partner (transmission per sex act). Our study is also limited by its single-center design in that the majority of the included women (approximately 70%) were of Dutch ethnic origin and we have recently shown that women of Dutch ethnic origin are less likely to have vaginal dysbiosis compared with women of other ethnicities residing in Amsterdam, the Netherlands [40]. As the prevalence of different CVM compositions may differ between ethnic groups (also described by Ravel et al [6]), the association between the CVM and C. trachomatis infection should be assessed in different ethnic cohorts. Furthermore, prospective cohort studies, using molecular tools, are needed so that the moment of STI exposure can be estimated and CVMs prior to and after infection can be compared. To conclude, in a population of women who were notified for chlamydial infection, we found that those who tested positive for C. trachomatis infection were more likely to be younger, have a steady partnership, and were more likely to have CVMs dominated by L. iners or diverse anaerobic bacteria than by L. crispatus. Our findings warrant further epidemiological and fundamental research to guide intervention strategies for women at risk for contracting chlamydia. Notes Acknowledgments. The authors thank Sandra van den Broek for anonymizing the study participants data and Jeanette Gillebaart for her expert advice and assistance in performing the Nugent scores. Financial support. This work was supported by the Public Health Laboratory in Amsterdam, the Netherlands. Potential conflicts of interest. All authors: No potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of 30 CID 2017:64 (1 January) van der Veer et al

8 Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed. References 1. World Health Organization. Global incidence and prevalence of selected curable sexually transmitted infections Geneva, Switzerland: WHO, van Aar F, de Moraes M, Morre SA, et al. Chlamydia trachomatis IgG seroprevalence in the general population of the Netherlands in 1996 and in 2007: differential changes by gender and age. Sex Transm Infect 2014; 90: Brotman RM, Klebanoff MA, Nansel TR, et al. Bacterial vaginosis assessed by gram stain and diminished colonization resistance to incident gonococcal, chlamydial, and trichomonal genital infection. J Infect Dis 2010; 202: Martin HL, Richardson BA, Nyange PM, et al. Vaginal lactobacilli, microbial flora, and risk of human immunodeficiency virus type 1 and sexually transmitted disease acquisition. J Infect Dis 1999; 180: Wiesenfeld HC, Hillier SL, Krohn MA, Landers DV, Sweet RL. Bacterial vaginosis is a strong predictor of Neisseria gonorrhoeae and Chlamydia trachomatis infection. Clin Infect Dis 2003; 36: Ravel J, Gajer P, Abdo Z, et al. Vaginal microbiome of reproductive-age women. Proc Natl Acad Sci U S A 2011; 108(suppl 1): van de Wijgert JH, Borgdorff H, Verhelst R, et al. The vaginal microbiota: what have we learned after a decade of molecular characterization? PLoS One 2014; 9:e Borgdorff H, Tsivtsivadze E, Verhelst R, et al. Lactobacillus-dominated cervicovaginal microbiota associated with reduced HIV/STI prevalence and genital HIV viral load in African women. ISME J 2014; 8: Gautam R, Borgdorff H, Jespers V, et al. Correlates of the molecular vaginal microbiota composition of African women. BMC Infect Dis 2015; 15: Ravel J, Brotman RM, Gajer P, et al. Daily temporal dynamics of vaginal microbiota before, during and after episodes of bacterial vaginosis. Microbiome 2013; 1: Srinivasan S, Hoffman NG, Morgan MT, et al. Bacterial communities in women with bacterial vaginosis: high resolution phylogenetic analyses reveal relationships of microbiota to clinical criteria. PLoS One 2012; 7:e van Houdt R, Ma B, Speksnijder A, van der Helm JJ, de Vries HJC, Ravel J. The vaginal microbiome as a predictor for acquiring Chlamydia trachomatis. International Human Microbiome Consortium, Abstract #:Evi Masese L, Baeten JM, Richardson BA, et al. Incidence and correlates of Chlamydia trachomatis infection in a high-risk cohort of Kenyan women. Sex Transm Dis 2013; 40: Ness RB, Kip KE, Soper DE, et al. Bacterial vaginosis (BV) and the risk of incident gonococcal or chlamydial genital infection in a predominantly black population. Sex Transm Dis 2005; 32: Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol 1991; 29: Fadrosh DW, Ma B, Gajer P, et al. An improved dual-indexing approach for multiplexed 16S rrna gene sequencing on the Illumina MiSeq platform. Microbiome 2014; 2: Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 2014; 30: Masella AP, Bartram AK, Truszkowski JM, Brown DG, Neufeld JD. PANDAseq: paired-end assembler for Illumina sequences. BMC Bioinformatics 2012; 13: Schmieder R, Lim YW, Rohwer F, Edwards R. TagCleaner: identification and removal of tag sequences from genomic and metagenomic datasets. BMC Bioinformatics 2010; 11: Caporaso JG, Kuczynski J, Stombaugh J, et al. QIIME allows analysis of highthroughput community sequencing data. Nat Methods 2010; 7: Caporaso JG, Bittinger K, Bushman FD, DeSantis TZ, Andersen GL, Knight R. PyNAST: a flexible tool for aligning sequences to a template alignment. Bioinformatics 2010; 26: Matsen FA, Kodner RB, Armbrust EV. PPLACER: linear time maximumlikelihood and Bayesian phylogenetic placement of sequences onto a fixed reference tree. BMC Bioinformatics 2010; 11: R Foundation for Statistical Computing. R: a language and environment for statistical computing. Vienna, Austria: R Foundation, Wickham H. Ggplot2: elegant graphics for data analysis. New York: Springer, Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol 2008; 8: Segata N, Izard J, Waldron L, et al. Metagenomic biomarker discovery and explanation. Genome Biol 2011; 12:R Molano M, Meijer CJ, Weiderpass E, et al. The natural course of Chlamydia trachomatis infection in asymptomatic Colombian women: a 5-year follow-up study. J Infect Dis 2005; 191: Morre SA, van den Brule AJ, Rozendaal L, et al. The natural course of asymptomatic Chlamydia trachomatis infections: 45% clearance and no development of clinical PID after one-year follow-up. Int J STD AIDS 2002; 13(suppl 2): Gallo MF, Macaluso M, Warner L, et al. Bacterial vaginosis, gonorrhea, and chlamydial infection among women attending a sexually transmitted disease clinic: a longitudinal analysis of possible causal links. Ann Epidemiol 2012; 22: Abbai NS, Reddy T, Ramjee G. Prevalent bacterial vaginosis infection a risk factor for incident sexually transmitted infections in women in Durban, South Africa [manuscript published online ahead of print]. Int J STD AIDS Borgdorff H, Gautam R, Armstrong SD, et al. Cervicovaginal microbiome dysbiosis is associated with proteome changes related to alterations of the cervicovaginal mucosal barrier. Mucosal Immunol 2015; 9: Witkin SS, Mendes-Soares H, Linhares IM, Jayaram A, Ledger WJ, Forney LJ. Influence of vaginal bacteria and D- and L-lactic acid isomers on vaginal extracellular matrix metalloproteinase inducer: implications for protection against upper genital tract infections. MBio 2013; 4:e Kyongo JK, Crucitti T, Menten J, et al. Cross-sectional analysis of selected genital tract immunological markers and molecular vaginal microbiota in sub-saharan African women, with relevance to HIV risk and prevention. Clin Vaccine Immunol 2015; 22: Rizzo A, Fiorentino M, Buommino E, Donnarumma G, Losacco A, Bevilacqua N. Lactobacillus crispatus mediates anti-inflammatory cytokine interleukin-10 induction in response to Chlamydia trachomatis infection in vitro. Int J Med Microbiol 2015; 305: Gajer P, Brotman RM, Bai G, et al. Temporal dynamics of the human vaginal microbiota. Sci Transl Med 2012; 4: Macklaim JM, Fernandes AD, Di Bella JM, Hammond JA, Reid G, Gloor GB. Comparative meta-rna-seq of the vaginal microbiota and differential expression by Lactobacillus iners in health and dysbiosis. Microbiome 2013; 1: Borgdorff H, Verwijs MC, Wit FW, et al. The impact of hormonal contraception and pregnancy on sexually transmitted infections and on cervicovaginal microbiota in African sex workers. Sex Transm Dis 2015; 42: Brotman RM, Klebanoff MA, Nansel TR, et al. A longitudinal study of vaginal douching and bacterial vaginosis a marginal structural modeling analysis. Am J Epidemiol 2008; 168: Jespers V, van de Wijgert J, Cools P, et al. The significance of Lactobacillus crispatus and L. vaginalis for vaginal health and the negative effect of recent sex: a cross-sectional descriptive study across groups of African women. BMC Infect Dis 2015; 15: Borgdorff H, van der Veer C, van Houdt R, et al. Women of Dutch ethnic origin have lower prevalence of vaginal microbiota dysbiosis than women of other ethnic origins residing in Amsterdam the HELIUS study. In: European Congress of Clinical Microbiology and Infectious Diseases, Amsterdam, the Netherlands, 9 12 April Abstract O474. Vaginal Microbiota and Chlamydia CID 2017:64 (1 January) 31