Mother-to-child transmission (MTCT) of HIV remains a significant risk in the developing world. Without medical intervention, ∼25% of infants of HIV-infected mothers become infected before weaning.1,2 Short-course antiretroviral regimens are effective,3–6 but only half of HIV-infected mothers in resource-limited countries receive antiretroviral therapy during pregnancy.2,7 Thus, additional modalities for prevention and treatment are urgently needed to reduce the incidence of MTCT. Moreover, determining why ∼75% of infants of HIV-infected mothers are not infected, despite viral exposure in utero, at birth, and during breastfeeding may direct development of novel interventions.
Vaginal microbial communities influence the rates of both horizontal HIV acquisition8–12 and subsequent transmission.13 Bacterial vaginosis, characterized by imbalances in microbial community composition (dysbiosis) within the vagina, is significantly associated with HIV acquisition [eg, odds ratio (OR) 2.0].9 The mechanism(s) by which the defining microbiological features of bacterial vaginosis, loss of Lactobacillus spp. and gain of gram-negative facultative anaerobes (eg, Gardnerella vaginalis), enhance the infectivity of HIV remain unknown. Lactobacilli may create a physical environment that is inhospitable to viral propagation through production of lactic acid or H2O2.14 Alternatively, lactobacilli could induce an antiinflammatory state within the vaginal mucosa that lessens the abundance of cellular targets of HIV (dendritic cells and activated CD4+ T lymphocytes). In contrast, microorganisms enriched in bacterial vaginosis could either directly disrupt the vaginal mucosal barrier,15 thus increasing local HIV loads, or induce adverse sequelae such as premature rupture of membranes (PROM) or preterm labor16 that compromise the integrity and sterility of the amniotic cavity.
Analogous to sexual transmission of HIV,8,9,11,12,17 we hypothesized that particular microbial communities may also correlate with the frequency of MTCT. Mechanistically, vaginal microbes could modify HIV transmission rates either in utero18 or during delivery.19 To test this hypothesis, we retrospectively characterized the indigenous vaginal microbial communities of mothers with HIV infection in Burkina Faso, West Africa, who either did or did not transmit HIV to their infants. Microbial communities were surveyed by broad-range amplification of bacterial 16S rRNA genes and pyrosequencing. We found that the characteristics of the vaginal microbiota among African women with HIV infection was comparable with that described from women in other settings and geographical venues. Of note, altered vaginal microbial communities were associated with an increased risk for perinatal MTCT, consistent with results with horizontal transmission of HIV.
MATERIALS AND METHODS
This is a case-cohort study nested within the DITRAME prospective cohort of HIV-1–infected pregnant women and their live-born children from the ANRS 049a and 049b trials. The DITRAME ANRS 049a and 049b phase 2 multicenter, double-blind, randomized placebo-controlled trials evaluated the tolerance and prophylactic efficacy of azidothymidine (AZT; ANRS 049a3,20) and of the topical antiseptic benzalkonium chloride (CdB; ANRS 049b21,22) on MTCT among HIV-seropositive women. The DITRAME ANRS 049a trial was conducted in 2 large cities of West Africa (Abidjan, Côte d'Ivoire, and Bobo-Dioulasso, Burkina Faso) from September 1995 to May 1997.3 Briefly, eligible HIV-1–infected pregnant women were randomized at 36–38 weeks' gestation to receive either oral zidovudine (250–300 mg twice a day) or a matching placebo, until the beginning of labor; then a single oral dose of 500/600 mg until delivery; and a 7-day postpartum treatment of 500/600 mg per day. No treatment was given to the neonate. In the DITRAME ANRS 049b trial,21 women testing positive for HIV-1 infection in prenatal care units in Abidjan (Côte d'Ivoire) and Bobo-Dioulasso (Burkina Faso) from November 1996 to April 1997 were eligible, with their informed consent. Women self-administered daily a vaginal suppository of 1% CdB or matched placebo from 36 weeks of pregnancy, plus a single dose during labor. Only vaginal samples from Burkina Faso are analyzed in this report.
A single cervicovaginal lavage (CVL; 3 mL of sterile saline) fluid specimen was obtained from each study participant by gentle aspiration at 36–38 weeks of gestation. After low-speed centrifugation, the fluid supernatants were frozen at –80°C on the same day and shipped in liquid nitrogen dry shippers. All women for whom CVL specimens were available (n = 70) were included in this study, although 6 were excluded due to failure to produce polymerase chain reaction (PCR) products (see following). The protocols were approved by the Centre Muraz Ethical Committee and Ministry of Health of Burkina Faso, and the institutional review boards at the University of Minnesota and the University of Colorado (protocol 10-0708). Written informed consent was obtained from all participants.
HIV transmission was evaluated by measuring HIV RNA in infant plasma with Amplicor v1.5 Roche HIV RNA kits (Hoffman-La Roche Ltd, Basel, Switzerland) within 72 hours after birth and at day 45. Infants with positive results by 72 hours were defined as cases of antepartum MTCT, those with initial negative test for HIV (<72 hours) followed by a positive test by day 45 as intrapartum MTCT, and those who were negative at day 45 were classified as controls.23
16S rRNA Gene Pyrosequencing
One hundred microliters of CVL supernatant in 500 μL of Buffer B24 was heated 10 minutes at 70°C to inactivate viral particles. Five hundred microliters of buffer-saturated phenol (pH 8; Sigma-Aldrich, St Louis, MO) and 100 mg of 0.1-mm zirconium beads (Biospec Products Inc, Bartlesville, OK) were added and specimens agitated 3 minutes in a Mini-Beadbeater-8 (Biospec Products Inc). Samples were chloroform-extracted and DNA-precipitated after addition of 0.5 volume 7.5 M NH4OAc and 1 volume 100% isopropanol. Pellets were rinsed in 80% ethanol, dried, resuspended in 20 μL TE (10 mM Tris–HCl, 1 mM EDTA, pH 8.0), and stored at –80°C. Bacterial rRNA gene concentrations were determined by quantitative PCR (QPCR).25 Aliquots of genomic DNAs were diluted to 250,000 rRNA templates per microliter.
Multiplexed broad-range 16S amplicon libraries were constructed for pyrosequencing on a 454 Life Sciences GS FLX instrument (Branford, CT) using barcoded primers.25 Samples were amplified through 30 PCR cycles (early log phase for 250,000 templates/μL), and if necessary, additional cycles were performed to produce sufficient amplicons for sequencing (33, 36, or 38 cycles maximum). Bacterial 16S rRNA DNA was successfully amplified from 64 of 70 (91%) subjects (10 transmitters, 54 nontransmitters); these 64 subjects comprise the study population. No statistical differences in clinical parameters were observed between the 64 PCR-amplified and the 6 nonamplified specimens. Eighteen specimens were subjected to replicate DNA extraction (2–3 per specimen) and replicate PCR reactions (2–6 per specimen). Median within-subject Morisita–Horn community similarity values (CMH) of 0.99 [interquartile range (IQR) 0.94–1.0] for extraction replicates and 0.99 (IQR 0.99–1.0) for PCR replicates indicate that nearly identical distributions of operational taxonomic units (OTUs) were observed in replicate samples from the same subject.
Pyrosequences were polished, deconvoluted,26 and excluded by the following criteria: (a) trimmed length <150 nucleotides, (b) >0 ambiguous base(s), (c) absence of barcode, and/or (d) SINA27 quality score <75. Taxonomies were assigned by BLAST search of a database extracted from the All-Species Living Tree Project (version LTP_S95)28 and corroborated using the Ribosomal Database Project Naïve Bayesian classifier29 as described.25 On the basis of high BLAST percent identity scores (median 99%, IQR 98%–99%), most of the vaginal microorganisms could be identified to the species level (96% of sequences were ≥97% identical to LTP_S95 sequences). Biodiversity indices30,31 were estimated through 10,000 bootstrap replicates.32
Enumeration of Selected Microorganisms
Quantification of total bacteria, the 2 most prominent sequence types (Gardnerella vaginalis and Lactobacillus spp.) was performed by QPCR using standard curves derived from cloned 16S genes,24 using the following primers: (a) total bacteria33: 8F (5′ AGAGTTTGATCCTGGCTCAG) and 338R (5′ CTGCTGCCTCCCGTAGGAGT), (b) G. vaginalis34: GardnF (5′ GACTGAGATACGGCCCAGAC) and GardnR (5′ ATTCGAAAGGTACACTCACC), and (c) Lactobacillus34: LactoF (5′ TGGAAACAGRTGCTAATACCC) and LactoR (5′ GYCCATTGTGGAAGATTCCC). The cycling protocol was as follows: (a) denaturation at 95°C (10 minutes) and (b) 40 cycles of 95°C (15 seconds), 56°C (15 seconds), and 60°C (30 seconds followed by fluorescence plate read).
Analyses used the R statistical software package (v.2.8.1).35 Differences in mean values of continuous variables (eg, age) between pairs of categorical variables (eg, treatment group) were assessed by Student t test. Percent abundances of bacterial groups, estimated by 16S rRNA sequences, were logit transformed into continuous variables using the “car” package of R.36 The association between HIV transmission and genus-level clustering of subjects were assessed by Fisher exact test. Associations between the main outcome variables (eg, HIV transmission, treatment arm) and the abundances of individual species- and phylum-level taxa were evaluated by logistic regression using the glm function in R with a quasibinomial distribution (to accommodate overdispersion) and the logit link function.35 Because univariate analyses indicated potential associations between vaginal microbiotas and age, CD4+ T-cell count and CD8+ T-cell count, results were adjusted for these covariates [no associations were observed between microbiota and mode of delivery, low birth weight (<2.5 kg birth weight), PROM, or prematurity (<37 weeks' gestation37)]. Adding treatment as a covariate in the analysis of associations between microbiome and MTCT did not appreciably change the statistical results, so treatment arm was not included in the reported analysis of MTCT. Because of the exploratory nature of this study, P values were not adjusted for multiple comparisons. Rather, results are interpreted as indicating candidate microbes potentially associated with HIV transmission. Morisita–Horn community similarity indices (CMH) were calculated using the vegdist function of the R package “vegan”38 and compared between categories by Wilcoxon rank-sum test. Heatmaps were generated with the “heatmap.2” function of the R package “gplots.”39 Complete-linkage hierarchical clustering of subjects was performed using the default options of the “hclust” and “dist” functions of the R package “stats.”35 Euclidean dissimilarity matrices were calculated using logit-transformed OTU abundance data.
DNA Sequence Accession Numbers
Sequences were deposited in GenBank with the accession numbers JF461543–JF487783.
All CVL specimens available from women enrolled in the ANRS 049a or 049b study in Burkina Faso were included in this substudy (N = 70). Six subjects were excluded because the CVL specimens failed to produce bacterial 16S PCR products. Of the 64 HIV-infected mothers remaining, 10 transmitted HIV to their infants [4 (6%) antepartum and 6 (9.3%) intrapartum; Table 1]. Transmitters were characterized by significantly lower circulating CD4+ T-cell numbers compared with nontransmitters (P < 0.024; Table 1), as previously reported,3 whereas neither age nor CD8+ T-cell counts differed significantly between cases and controls. No significant differences in age, CD4+ T-cell counts, or CD8+ T-cell counts were observed between treatment arms (Table 1). HIV RNA levels were not routinely available in infant or maternal plasma or CVL in this cohort.
Characterization of Vaginal Microbial Ecology
High-throughput pyrosequencing of 16S rRNA genes amplified by PCR with pan-bacterial primers generated a data set (Table 2) consisting of 49,301 polished sequences, with a median length of 245 nucleotides (IQR 224–257 nucleotides) and a median of 350 sequences per specimen (IQR 255–600 reads/subject). Good's coverage indices for species-level OTUs ranged from 94.2% to 100% per subject (median 99.5%), indicating that this depth of sequencing reflected the true biodiversity of each specimen.
Analysis of vaginal rRNA sequences revealed that the study population as a whole contained 198 species-level bacterial OTUs from 111 genera and 9 phyla (Table 2 and Fig. 1). The majority of the sequences belonged to only 2 phyla, the Firmicutes (low G + C gram positives; 66.2% of sequences) and the Actinobacteria (high G + C gram positives; 17.3% of sequences). Members of the phyla Proteobacteria (7.9%), Fusobacteria (3.7%), Bacteroidetes (2.7%), and Tenericutes (eg, mycoplasmas and ureaplasmas; 2.1%) also were detected in CVL specimens at lower abundances (Table 2 and Fig. 1).
Substantial between-individual differences were observed in vaginal microbiotas (Fig. 1). Hierarchical clustering of subjects based on similarity of vaginal microbiotas defined 4 broad groups of subjects. The microbiotas of 30 of 64 subjects, of whom 4 were intrapartum HIV transmitters, were dominated by the genus Lactobacillus (Fig. 1, cluster 1), which included the species L. iners (77% of sequences in cluster 1), L. crispatus (11%), L. fornicalis (3.9%), L. gasseri (3.2%), and L. vaginalis (0.5%). A second cluster of 8 individuals (Fig. 1, cluster 2) was dominated by coagulase-negative staphylococci (eg, S. haemolyticus and S. epidermidis), with lesser quantities of lactobacilli, and did not include any HIV transmitters. The vaginal communities of 24 individuals in cluster 3 (Fig. 1) contained more complex mixtures of a variety of genera, dominated by Gardnerella spp. (eg, G. vaginalis, 36.8% of sequences in cluster 3) and Lactobacillus spp. (L. iners, 18.7% abundance). Other genera in this cluster of individuals included Sneathia (9.5%), Atopobium (6.0%), Prevotella (6.3%), Anaerococcus (4.4%), and Mycoplasma (4.3%), all of which have been associated with bacterial vaginosis.40,41 All 4 antepartum and 2 of 6 intrapartum HIV transmitters belonged to cluster 3, suggesting a weak association between membership in this group and antepartum MTCT (P = 0.09). The fourth cluster consisted of only 2 nontransmitting subjects characterized by low Lactobacillus spp. and diverse anaerobic groups (eg, Prevotella, Sneathia, Peptostreptococcus).
Vaginal Microbiota and MTCT
HIV transmission, either antepartum or intrapartum, was significantly associated with the abundances of several individual bacterial phyla and species-level OTUs (Table 2). Most notably, the relative abundances of G. vaginalis [36.7% vs 15.1% of sequences/subject; OR 1.7; 95% confidence interval (CI) 1.2 to 2.4; P = 0.004] and its phylum Actinobacteria (38.1% vs 17.3% of sequences/subject; OR 1.6; 95% CI 1.1 to 2.3; P = 0.009) were elevated by >2-fold in antepartum cases compared with nontransmitters. The OR values were adjusted for multiple covariates, including CD4+ and CD8+ T-cell counts and maternal age (univariate analyses suggested no associations between microbiota and mode of delivery, low birth weight, PROM, premature birth, or treatment, so these variables were not included in the analysis). Although the treatment group did not significantly affect vaginal microbial communities, G. vaginalis sequences tended to be more abundant in treatment groups (23.3% for AZT and 18.5% for CdB) compared with controls (7.8%; P = 0.32 and P = 0.14, respectively), as were actinobacterial sequences (28.2% and 20.8% vs 9.5%; P = 0.24 and P = 0.26, respectively). Several Prevotella spp. also were significantly enriched in antepartum cases compared with controls, although the increase in the phylum Bacteroidetes was not significant. In contrast, the phylum Firmicutes, of which the lactobacilli are prominent vaginal constituents, was reduced in relative sequence abundance in antepartum cases, compared with those in nontransmitting control mothers (52.8% vs 72.3%; OR 0.69; 95% CI 0.48 to 0.99; P = 0.05). Although the species L. iners, typically among the most common vaginal species,42 was also present in antepartum cases at somewhat lesser frequencies than controls (26.7% vs 42.0% sequence abundance), the difference was not statistically significant (P = 0.37). Only 2 bacterial species-level OTUs, Brevibacterium casei and L. gasseri, were significantly associated with intrapartum transmission (Table 2). Thus, MTCT is associated with differential abundances of several bacterial groups and in the diversity of these groups.
Quantification of Specific Microbial Groups
We sought to independently corroborate pyrosequencing results by targeted quantification of candidate microbial groups. Because of their relatively high abundances and suggested association with HIV transmission, both G. vaginalis and lactobacilli were enumerated in CVL specimens by QPCR. To normalize results between specimens, total bacterial PCR using pan-bacterial 16S rRNA primers was used to enumerate total bacteria in each specimen. Therefore, results are presented as logit-transformed percentages of total bacterial populations occupied by the particular bacterial groups (Figs. 2A, B).
Specific QPCR (Fig. 2A) confirmed that antepartum HIV transmitters carried significantly higher relative abundances of G. vaginalis compared with nontransmitters (32.4% vs 8.8% of total bacterial load; P = 0.009), but differences between antepartum and intrapartum transmitters were not significant. In contrast, QPCR of lactobacilli did not differ significantly in abundance between transmitters or controls (Fig. 2A). In general, the QPCR results matched those of pyrosequencing with Pearson correlation coefficients of 0.82 and 0.76 for G. vaginalis and lactobacilli, respectively.
We characterized for the first time the vaginal microbial communities of HIV-infected expectant mothers in sub-Saharan Africa. Consistent results for the vaginal bacteria identified by 16S rRNA pyrosequencing confirm that although 198 different species were identified in individual mothers, Lactobacillus spp. predominate, accounting for 41% of all bacteria identified. These results were similar to those reported in North America,40,41,43,44 South America,45 Africa,46 Asia,47 and Europe,48 confirming the existence of a core vaginal microbiota, despite geographic separation and ethnic and immunologic variations.
To our knowledge, these are also the first in-depth, culture-independent, high-throughput molecular analyses of the vaginal microbiological correlates of MTCT. After adjusting for maternal age, CD4+ and CD8+ T-cell counts, and treatment arm (AZT vs CdB vs placebo), several vaginal microbial groups were positively and significantly correlated with perinatal MTCT. The most striking association observed was between antepartum HIV transmission and increased relative abundance of G. vaginalis (OR 1.7; no significant difference in prevalence was observed between transmitters and nontransmitters). Although a prominent member of the vaginal microbiota of many study subjects, G. vaginalis was doubled in relative abundance in the broad-range 16S rRNA sequence libraries of antepartum transmitters (37% of sequences), relative to nontransmitters (15% of sequences). Species-specific PCR quantification of G. vaginalis in CVL specimens, normalized to total bacteria, confirmed this result.
Based on similarities in vaginal microbiota, 6 of 10 (60%) MTCT cases and 20 of 54 (37%) nontransmitter controls formed a microbiological cluster (cluster 3; Fig. 1) that was distinguished by increased abundances and/or prevalences of multiple genera, including Gardnerella, Sneathia, Atopobium, Prevotella, and Mycoplasma. Furthermore, lactobacilli, the primary constituents of the “normal” vaginal microbiota,40 typically were present at reduced abundances in these subjects in cluster 3. Although bacterial vaginosis was not evaluated or diagnosed during the ANRS 049 trial, these vaginal microbiotas in cluster 3 are similar to those reported for cases of vaginosis in developed countries.40,41 These results are consistent with a trend toward higher risk for MTCT in women with the microbiological hallmarks of bacterial vaginosis. Because of the complex relationships between bacterial vaginosis, herpes simplex virus 2 infection, and HIV acquisition, it is possible that bacterial vaginosis is a mediator between herpes simplex virus 2 and HIV infections.49 Neither treatment with AZT nor with CdB was associated with altered vaginal microbiotas (Figs. 1, 2 and data not shown).
Limitations of this case–control study include reduced statistical power due to a small number of cases, absence of clinical diagnosis of vaginal disorders (such as bacterial vaginosis and other sexually transmitted infections), and lack of data on HIV RNA viral load in CVL. The similarity of microbial communities reported here to those described previously40,41,43,44 indicates that long-term storage at –80°C since sampling did not compromise the specimens analyzed. Because of the study design, the causal relationships between vaginal HIV loads, alterations in the vaginal microbiota, and MTCT could not be addressed. For instance, independent of its effects on MTCT risk, compromised mucosal immunity in the reproductive tract may have indirectly altered the vaginal microbiota. Nonetheless, the statistically significant positive correlations between microbial groups such as G. vaginalis and MTCT justify prospective follow-up investigations to establish the generalizability of the observed associations between vaginal microbiota and risk for MTCT in the context of biologically relevant cofactors.
Determining the biological factors that naturally protect some neonates from perinatal HIV transmission may lead to novel strategies for preventing infections. Few in-depth culture-independent studies of the human vaginal microbiota have been reported among individuals with HIV infection,44 particularly in the context of clinical outcomes. Nonetheless, vaginal dysbiosis and/or acquisition of specific pathogens (eg, Neisseria gonorrhoeae, Trichomonas vaginalis) are well-established risk factors for sexual transmission of HIV8,9,11,12,17 and have been associated with higher HIV RNA burden in the genital tract.50,51 The results of this study suggest that vaginal microbial communities also may influence the risk for prenatal MTCT. The loss of normally protective function(s) provided by commensal vaginal microbes could disrupt cervicovaginal mucosal barrier integrity and thereby lead to increased local infiltration of immune effector cells, including HIV-infected leucocytes along with free virus, into cervicovaginal tissues. Under this model, the effects of increased HIV virion loads in the genital mucosa could raise the risk for both in utero and intrapartum transmission.19 Indeed, vaginal dysbiosis is associated with an increased risk for premature delivery. Because the risk for in utero MTCT is also significantly correlated with the occurrence of premature rupture of membranes,18,52 the impact of vaginal microbes on the upper genital tract also could contribute indirectly to the risk for both prenatal and perinatal MTCT.16 Alternatively, such local infection and inflammation may promote increased local HIV replication or enhanced transfer of virus from the systemic compartment. Indeed, ongoing studies are directed to confirm these results and to establish a more direct link between cervicovaginal microbial ecology, including the presence or absence of H2O2-producing lactobacilli, local HIV RNA and DNA levels, mucosal inflammation and epithelial integrity, and the rate of MTCT of HIV.
The authors thank Jana Palaia for assistance with specimen management and Prof. Helene Marchandin (University of Montpellier) for careful review of the manuscript, and also the patients and staff at Centre Muraz, Bobo-Dioulasso, Burkina Faso, for their investment of time and commitment.
1. Rollins NC, Dedicoat M, Danaviah S, et al.. Prevalence, incidence, and mother-to-child transmission of HIV-1 in rural South Africa. Lancet. 2002;360:389.
2. Fowler MG, Lampe MA, Jamieson DJ, et al.. Reducing the risk of mother-to-child human immunodeficiency virus transmission: past successes, current progress and challenges, and future directions. Am J Obstet Gynecol. 2007;197(3 suppl):S3–S9.
3. Dabis F, Msellati P, Meda N, et al.. 6-month efficacy, tolerance, and acceptability of a short regimen of oral zidovudine to reduce vertical transmission of HIV in breastfed children in Cote d'Ivoire and Burkina Faso: a double-blind placebo-controlled multicentre trial. DITRAME Study Group. DIminution de la Transmission Mère-Enfant. Lancet. 1999;353:786–792.
4. Guay LA, Musoke P, Fleming T, et al.. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet. 1999;354:795–802.
5. Chasela CS, Hudgens MG, Jamieson DJ, et al.. Maternal or infant antiretroviral drugs to reduce HIV-1 transmission. N Engl J Med. 2010;362:2271–2281.
6. Shapiro RL, Hughes MD, Ogwu A, et al.. Antiretroviral regimens in pregnancy and breast-feeding in Botswana. N Engl J Med. 2010;362:2282–2294.
8. Atashili J, Poole C, Ndumbe PM, et al.. Bacterial vaginosis and HIV acquisition: a meta-analysis of published studies. AIDS. 2008;22:1493–1501.
9. Myer L, Denny L, Telerant R, et al.. Bacterial vaginosis and susceptibility to HIV infection in South African women: a nested case-control study. J Infect Dis. 2005;192:1372–1380.
10. Taha TE, Kumwenda NI, Kafulafula G, et al.. Intermittent intravaginal antibiotic treatment of bacterial vaginosis in HIV-uninfected and -infected women: a randomized clinical trial. PLoS Clin Trials. 2007;2:e10.
11. van de Wijgert JH, Morrison CS, Brown J, et al.. Disentangling contributions of reproductive tract infections to HIV acquisition in African Women. Sex Transm Dis. 2009;36:357–364.
12. van de Wijgert JH, Morrison CS, Cornelisse PG, et al.. Bacterial vaginosis and vaginal yeast, but not vaginal cleansing, increase HIV-1 acquisition in African women. J Acquir Immune Defic Syndr. 2008;48:203–210.
13. Farquhar C, Mbori-Ngacha D, Overbaugh J, et al.. Illness during pregnancy and bacterial vaginosis are associated with in-utero HIV-1 transmission. AIDS. 2010;24:153–155.
14. Cherpes TL, Hillier SL, Meyn LA, et al.. A delicate balance: risk factors for acquisition of bacterial vaginosis include sexual activity, absence of hydrogen peroxide-producing lactobacilli, black race, and positive herpes simplex virus type 2 serology. Sex Transm Dis. 2008;35:78–83.
15. Patterson JL, Stull-Lane A, Girerd PH, et al.. Analysis of adherence, biofilm formation and cytotoxicity suggests a greater virulence potential of Gardnerella vaginalis relative to other bacterial-vaginosis-associated anaerobes. Microbiology. 2010;156(pt 2):392–399.
16. DiGiulio DB, Romero R, Amogan HP, et al.. Microbial prevalence, diversity and abundance in amniotic fluid during preterm labor: a molecular and culture-based investigation. PLoS One. 2008;3:e3056.
17. Taha TE, Hoover DR, Dallabetta GA, et al.. Bacterial vaginosis and disturbances of vaginal flora: association with increased acquisition of HIV. AIDS. 1998;12:1699–1706.
18. Group TIPH. Duration of ruptured membranes and vertical transmission of HIV-1: a meta-analysis from 15 prospective cohort studies. AIDS. 2001;15:357–368.
19. European Mode of Delivery Collaboration. Elective caesarean-section versus vaginal delivery in prevention of vertical HIV-1 transmission: a randomised clinical trial. Lancet. 1999;353:1035–1039.
20. Msellati P, Meda N, Welffens-Ekra C, et al.. Zidovudine and reduction of vertical transmission of HIV in Africa. ANRS 049 Trial Group. Am J Public Health. 1999;89:947–948.
21. Msellati P, Meda N, Leroy V, et al.. Safety and acceptability of vaginal disinfection with benzalkonium chloride in HIV infected pregnant women in west Africa: ANRS 049b phase II randomized, double blinded placebo controlled trial. DITRAME Study Group. Sex Transm Infect. 1999;75:420–425.
22. Mandelbrot L, Msellati P, Meda N, et al.. 15 Month follow up of African children following vaginal cleansing with benzalkonium chloride of their HIV infected mothers during late pregnancy and delivery. Sex Transm Infect. 2002;78:267–270.
23. Newell ML. Mechanisms and timing of mother-to-child transmission of HIV-1. AIDS. 1998;12:831–837.
24. Frank DN, St Amand AL, Feldman RA, et al.. Molecular-phylogenetic characterization of microbial community imbalances in human inflammatory bowel diseases. Proc Natl Acad Sci U S A. 2007;104:13780–13785.
25. Frank DN, Feazel LM, Bessesen MT, et al.. The human nasal microbiota and Staphylococcus aureus carriage. PLoS One. 2010;5:e10598.
26. Frank DN. BARCRAWL and BARTAB: software tools for the design and implementation of barcoded primers for highly multiplexed DNA sequencing. BMC Bioinformatics. 2009;10:362.
27. Pruesse E, Quast C, Knittel K, et al.. SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 2007;35:7188–7196.
28. Yarza P, Richter M, Peplies J, et al.. The All-Species Living Tree project: a 16S rRNA-based phylogenetic tree of all sequenced type strains. Syst Appl Microbiol. 2008;31:241–250.
29. Wang Q, Garrity GM, Tiedje JM, et al.. Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol. 2007;73:5261–5267.
30. Good IJ. The population frequencies of species and the estimation of population parameters. Biometrika. 1953;40:237–264.
31. Chao A. Nonparametric estimation of the number of classes in a population. Scand J Stat. 1984;11:265–270.
32. Frank DN. XplorSeq: a software environment for integrated management and phylogenetic analysis of metagenomic sequence data. BMC Bioinformatics. 2008;9:420.
33. Lane DJ. 16S/23S rRNA sequencing. In: Stackebrandt E, Goodfellow M eds. Nucleic Acid Techniques in Bacterial Systematics. New York, NY: Wiley; 1991:115–175.
34. Frank JA, Reich CI, Sharma S, et al.. Critical evaluation of two primers commonly used for amplification of bacterial 16S rRNA genes. Appl Environ Microbiol. 2008;74:2461–2470.
35. R, Team DC. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2005.
36. Fox J. An R and S-Plus Companion to Applied Regression. Thousand Oaks, CA: Sage; 2002.
37. Latis GO, Simionato L, Ferraris G. Clinical assessment of gestational age in the newborn infant. Comparison of two methods. Early Hum Dev. 1981;5:29–37.
39. Gregory R. Warnes. Includes R source code and/or documentation contributed by: Ben Bolker, Lodewijk Bonebakker, Robert Gentleman, Wolfgang Huber Andy Liaw, Thomas Lumley, Martin Maechler, Arni Magnusson, Steffen Moeller, Marc Schwartz and Bill Venables (2011). gplots: Various R programming tools for plotting data. R package version 2.10.1. Available at: http://CRAN.R-project.org/package=gplots
. Accessed February 28, 2012.
40. Fredricks DN, Fiedler TL, Marrazzo JM. Molecular identification of bacteria associated with bacterial vaginosis. N Engl J Med. 2005;353:1899–1911.
41. Oakley BB, Fiedler TL, Marrazzo JM, et al.. Diversity of human vaginal bacterial communities and associations with clinically defined bacterial vaginosis. Appl Environ Microbiol. 2008;74:4898–4909.
42. 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):4680–4687.
43. Zhou X, Brown CJ, Abdo Z, et al.. Differences in the composition of vaginal microbial communities found in healthy Caucasian and black women. ISME J. 2007;1:121–133.
44. Spear GT, Sikaroodi M, Zariffard MR, et al.. Comparison of the diversity of the vaginal microbiota in HIV-infected and HIV-uninfected women with or without bacterial vaginosis. J Infect Dis. 2008;198:1131–1140.
45. Dumonceaux TJ, Schellenberg J, Goleski V, et al.. Multiplex detection of bacteria associated with normal microbiota and with bacterial vaginosis in vaginal swabs by use of oligonucleotide-coupled fluorescent microspheres. J Clin Microbiol. 2009;47:4067–4077.
46. Anukam KC, Reid G. Organisms associated with bacterial vaginosis in Nigerian women as determined by PCR-DGGE and 16S rRNA gene sequence. Afr Health Sci. 2007;7:68–72.
47. Zhou X, Hansmann MA, Davis CC, et al.. The vaginal bacterial communities of Japanese women resemble those of women in other racial groups. FEMS Immunol Med Microbiol. 2010;58:169–181.
48. Kiss H, Kogler B, Petricevic L, et al.. Vaginal Lactobacillus microbiota of healthy women in the late first trimester of pregnancy. BJOG. 2007;114:1402–1407.
49. Nagot N, Ouedraogo A, Defer MC, et al.. Association between bacterial vaginosis and herpes simplex virus type-2 infection: implications for HIV acquisition studies. Sex Transm Infect. 2007;83:365–368.
50. Sha BE, Zariffard MR, Wang QJ, et al.. Female genital-tract HIV load correlates inversely with Lactobacillus species but positively with bacterial vaginosis and Mycoplasma hominis. J Infect Dis. 2005;191:25–32.
51. Coleman JS, Hitti J, Bukusi EA, et al.. Infectious correlates of HIV-1 shedding in the female upper and lower genital tracts. AIDS. 2007;21:755–759.
52. Burns DN, Landesman S, Muenz LR, et al.. Cigarette smoking, premature rupture of membranes, and vertical transmission of HIV-1 among women with low CD4+ levels. J Acquir Immune Defic Syndr. 1994;7:718–726.