Vaginal Microbiota and Susceptibility to HIV

Abstract

Bacterial vaginosis, characterized by the replacement of the Lactobacillus-dominant microbiota with anaerobic bacteria and facultative Gram-negative rods, has been associated with adverse reproductive health outcomes including HIV acquisition. With the advent of newer molecular techniques, the vaginal microbiota can be investigated in more detail and the association with HIV examined more thoroughly. This review examines recent evidence suggesting that vaginal dysbiosis with increased microbial diversity, specific vaginal bacterial communities, and the presence and concentrations of some individual bacterial species, may increase HIV susceptibility. Potential mechanisms through which vaginal microbiota could impact HIV susceptibility are discussed. Based on the available data, this review finds that there is a modest, but growing, body of evidence linking vaginal microbiota to HIV susceptibility in women. The evidence could be strengthened through two main pathways. First, laboratory studies such as ex vivo or animal experiments are needed to move from plausible mechanisms towards proven mechanisms that explain an effect of the vaginal microbiota on HIV susceptibility. Second, experimental evidence could directly test the hypothesis that sustaining optimal microbiota reduces HIV risk, though there are important obstacles to conducting such studies. Finally, this review examines strong evidence from a recent publication suggesting that deviations from an optimal vaginal microbiome, and particularly the presence of some bacterial communities with high relative abundance of Gardnerella vaginalis, reduces the efficacy of vaginal tenofovir-based microbicides.

Introduction

Women in sub-Saharan Africa continue to bear a greater burden of HIV compared to men [1]. Disruptions of the vaginal microbiota could play a key role in mediating HIV susceptibility in African women.

In the early 1900s, Doderlein bacillus, later classified as Lactobacillus, was linked with vaginal health in married white women [2]. In this population, deviation from a microbiota dominated by Lactobacillus species was associated with vaginal discharge. As the vaginal microbiota has been characterized in a wider range of populations, it is evident that non-Lactobacillus-dominant vaginal microbial communities are common [3], can occur with and without symptoms [2], and may be associated with a range of adverse reproductive health outcomes including HIV acquisition [4–9].

Heterosexual transmission of HIV is inefficient [10]. Co-factors including sexually transmitted infections (STIs) and vaginal dysbiosis likely contribute to increased HIV transmission efficiency [4, 11–23]. Bacterial vaginosis (BV), the most common type of vaginal dysbiosis, is a condition in which Lactobacillus-dominant microbiota is replaced by complex bacterial communities with anaerobic bacteria and facultative Gram-negative rods [2]. Bacterial vaginosis has been associated with HIV acquisition in women [4–9]. Because BV often persists [24, 25], and frequently recurs even when treated [26, 27], this condition may contribute substantially to the population attributable risk (PAR) of HIV infection. Two studies suggest that BV contributes substantially more to HIV PAR than any genital condition other than herpes simplex virus-2 (HSV-2) [28, 29].

Beginning in 2005, advances in molecular evaluation of the vaginal microbiota [30], have enabled examination of vaginal microbial communities in much finer detail. This paper summarizes recent literature using molecular characterization to explore the possible role of vaginal microbiota in mediating women’s susceptibility to HIV infection.

Laboratory methods for characterizing vaginal microbiota

There are two commonly used criteria to diagnose BV. Amsel criteria includes three or more of four clinical signs including clue cells on wet mount microscopy, “fishy” amine odor, vaginal pH>4.5, and thin, homogenous vaginal discharge [31]. The criteria developed by Nugent and Hillier (Nugent criteria) define BV based on Gram stain enumeration of bacterial morphotypes [32].

In addition to the appearance of bacteria on Gram stain, the vaginal microbiota has been characterized using culture-based methods. One advantage is that culture can identify some minority species more easily than through newer molecular techniques [33]. Culture also allows for antimicrobial sensitivity testing. However, many key bacterial taxa associated with BV are difficult to cultivate [34].

During the past 15 years, advances in molecular microbiology have contributed substantially to our understanding of the vaginal microbiota by providing a complementary approach [35]. Many of these novel methods begin with broad-range polymerase chain reaction (PCR) amplification targeting highly conserved 16S rRNA gene sequences. Additional steps are then employed to identify bacteria. These techniques include denaturing gradient gel electrophoresis (DGGE) [36], terminal restriction fragment length polymorphism analysis (T-RFLP) [37], cloning and Sanger sequencing [38], amplified ribosomal DNA restriction analysis (ARDRA) [39], and pyrosequencing [40, 41]. Taxon-directed quantitative PCR (qPCR) can been used to measure quantities of individual bacterial taxa [42, 43]. In addition, fluorescent in situ hybridization (FISH) is a non-amplified method that uses fluorescently-labeled 16S rRNA probes to detect bacterial taxa, characterize their morphology, and localize species into microniches [30]. In combination, newer methods have allowed for a simplified approach to identifying bacteria, including cultivation-resistant bacteria. Bacterial quantities can be evaluated in terms of their relative abundance (e.g. pyrosequencing data) and concentration (e.g. qPCR). Relative abundance data have been used to identify vaginal bacterial community types that provide more differentiation than the traditional Amsel and Nugent criteria [44].

Application of molecular methods has considerably advanced our understanding of both healthy and disrupted vaginal microbiota. Women with BV have more diverse vaginal microbiota compared to women without BV [41, 45, 46]. Diversity has been defined in a variety of ways including the Shannon Diversity Index, operational taxonomic units, and number of positive probes [41, 45, 46]. Multiple bacterial taxa have been associated with BV, including bacterial vaginosis-associated bacteria 1 (BVAB1), BVAB2, Mageeibacillus indolicus (previously BVAB3), Gardnerella vaginalis, Atopobium vaginae, Eggerthella-like uncultured bacteria, Leptotrichia, Megasphaera, Prevotella, Mycoplasma hominis, Bifidobacterium, and Dialister [30]. In contrast, Lactobacillus crispatus has been associated with a healthier vaginal microbiome [47–49].

Ethnic and geographic variations in vaginal microbiota

Since the 1990s, studies using Amsel or Nugent criteria have detected differences in BV prevalence by race and ethnicity [3, 50–54]. Non-white women generally have higher rates of BV compared to white women. In 2011, Ravel et al used broad-range PCR amplification of 16S rRNA genes with pyrosequencing to explore this question [44]. Compared to Hispanic and black women in North America, Asian and white women were more likely to have vaginal communities dominated by Lactobacillus species. Vaginal pH was lower in Asian and white women compared to black and Hispanic women, an effect the authors hypothesized was related to lactic acid production by Lactobacillus. Understanding the diversity of the vaginal microbiome across different populations is critical, because associations between the vaginal microbiota and susceptibility to HIV infection may vary by race, ethnicity, and geography.

Studies of BV as a risk for HIV infection in women

Bacterial vaginosis, diagnosed by Amsel or Nugent criteria, has been associated with HIV prevalence and incidence in numerous studies. Beginning in 1995, multiple studies have found associations between BV and prevalent HIV infection [4–6, 9, 28, 29, 55–74]. Stronger evidence is provided by numerous prospective cohort studies and meta-analyses, which have consistently shown that both intermediate vaginal microbiota and BV are associated with ~1.5-fold higher risk of HIV acquisition [7, 8, 75]. As detailed above, BV and abnormal microbiota diagnosed by Gram stain are microbiologically heterogeneous. Women with similar vaginal Gram stains often have distinctly different vaginal bacterial communities [76]. Taken together, the data on vaginal dysbiosis as a risk factor for HIV, combined with an evolving understanding of the complexity of the vaginal microbiome, raise a question about the specificity of the relationship between BV and HIV. In particular, vaginal dysbiosis on Gram stain may be a nonspecific marker for individual bacteria or bacterial communities, which are the true drivers of increased HIV susceptibility.

Molecular studies of vaginal microbiota and HIV acquisition

Three studies published or presented during the past year have employed molecular microbiological approaches to expand our understanding of the vaginal microbiota and HIV acquisition [77–79] (Table 1). Last year, Gosmann et al published a prospective cohort study of 236 HIV-uninfected South African adolescent women (18–23 years) [79]. Cervical and vaginal samples were characterized using nucleic acid extraction, amplification with 16S rRNA V4 primer constructs, and sequencing with Illumina MiSeq [80]. The authors defined four vaginal bacterial community groupings (cervicotypes). Cervicotype 1 (CT1) was characterized by high relative abundance of L. crispatus, CT2 had a high relative abundance of L. iners, CT3 was dominated by G. vaginalis, and CT4 was a diverse bacterial community not dominated by L. crispatus, L. iners, or G. vaginalis. In an analysis that excluded women with Chlamydia trachomatis, CT4 was associated with significantly higher risk of HIV acquisition compared to CT1. Two subtypes of Prevotella bivia, P. melaninogenica, Veillonella montpellierensis, Mycoplasma, and Sneathia sanguinegens were significantly more abundant in 31 women who acquired HIV compared to 205 women who remained HIV-uninfected. Noniners Lactobacillus species were associated with protection against HIV.

Table 1.

Studies using molecular techniques to examine the vaginal microbiome and HIV acquisition

Author (Year)	Country/Region	Study Population	N (# HIV events)	Method	Main Findings	Measure of Association	Limitations
Gosmann et al (2017) [92]	South Africa	Women in Females Rising Through Education, Support, and Healthy (FRESH) Study	236 (31)	16S rRNA V4 gene sequencing	Vaginal bacterial communities dominated by Gardnerella vaginalis (CT3) and by taxa other than Lactobacillus, L. iners, and G. vaginalis (CT4) associated with HIV acquisition1. Prevotella bivia 1, Prevotella melaninogenica, Veillonella montpellierensis, Mycoplasma, Prevotella bivia 2, and Sneathia sanguinegens associated with HIV.	CT3: HR 4.22 (95% CI 1.06–16.88), p=0.042 CT4: HR 4.03 (95% CI 1.14–14.27) p=0.031	HIV incidence analyses stratify by the presence of Chlamydia, but do not adjust for other potential confounding factors including for sexual risk behaviors.
Passmore and Williams (2016) [90]	South Africa	Women in CAPRISA 004 Trial	119 (49)	V1-V3 rDNA sequencing	Vaginal communities defined by Prevotella bivia (CST4) associated with inflammation and HIV acquisition1. P. bivia associated with HIV acquisition	aOR 12.7, 95% CI 2.1–77.8, p=0.006	Relatively small sample size Data described in abstract showed only unadjusted ratios.
McClellandet al. (2018) [78]	Eastern and Southern Africa	Female sex workers, pregnant/ postpartum, and HIV-negative women in discordant couples	349 (87)	Broad range PCR with deep sequencing and quantitative PCR	HIV seroconversion with more vaginal microbial diversity Eggerthella species type 1, Gemella asaccharolytica, Leptotrichia/Sneathia, Megasphaera, and Mycoplasma hominis significantly associated with HIV acquisition	Shannon Diversity Index median 0.9, (IQR 0.4–2.3) in HIV sero-converters compared to controls with median 0.7, (IQR 0.1–1.4), p=0.03	Did not explore mechanistic explanations for the associations

Abbreviations: PCR: polymerase chain reaction; CT: cervicotypes; CST: community state types; aOR: adjusted odds ratio; IQR: interquartile range; HR: hazard ratio; CI: confidence interval

¹While CTs and CSTs both describe vaginal communities by relative abundance, the specific processes for defining these communities are different.

The South African Study also explored potential mechanisms through which vaginal microbiota might impact HIV susceptibility. Women with CT4 vaginal communities had 17-fold higher number of activated CD4+ HIV target cells on cervical cytobrushes compared to women with CT1 communities. Strengths of this study included a prospective cohort design, a subset analysis excluding women with chlamydia, and parallel investigation of plausible mechanisms. One limitation of this study is that analyses examining the association between vaginal community types and HIV acquisition did not control for sexual risk behavior. Although the authors did not find statistically significant differences in condom use, frequency or type of sexual acts, or number of sexual partners across cervicotypes, this does not exclude possible confounding by these variables. Second, there were modest numbers of incident HIV infections. Third, this relatively homogenous population of adolescent South African women may not be generalizable to other regions or age groups. Despite these limitations, this study provides the first published evidence that certain vaginal communities and individual bacterial species may shape vaginal mucosal HIV susceptibility.

The next study, by Passmore and Williams, was presented at the 2016 International AIDS Conference in Durban, South Africa [77]. This analysis included 119 South African women from the CAPRISA 004 trial, a Phase IIb clinical trial assessing the effectiveness and safety of 1% tenofovir vaginal gel for preventing HIV infection. Using 16S rRNA V1-V3 variable region primers, then sequencing amplified DNA, 1368 species were identified from cervicovaginal samples. A cluster analysis was used to identify eight broad community state types (CSTs), which differed from those described by Gosmann et al. Community state types 1–3 were all dominated by Lactobacillus species. In contrast, CSTs 4–8 were dominated by species other than Lactobacillus, with P. bivia defining CST4, and distinguishing CST4 it from other CSTs. Community state type 4 was associated with an inflammatory vaginal cytokine profile and with HIV acquisition. Women with high relative abundance of P. bivia in vaginal samples were 19 times more likely to have a pro-inflammatory vaginal cytokine profile (adjusted odds ratio [aOR] 19.2, 95% CI 4.0–92.4, p<0.001), and nearly 13 times more likely to acquire HIV (aOR 12.7, 95% CI 2.1–77.8, p=0.006), compared to other women.

To explore this association, P. bivia’s metagenome was characterized. There was an enrichment of lipopolysaccharide (LPS) biosynthesis, indicating production of this immunostimulatory molecule, in women with CST4. The increased inflammation associated with LPS is a possible mechanism for increased HIV risk. To date, this study has only been presented in a conference symposium, so it is not possible to fully characterize strengths and limitations. Nonetheless, one important strength was the pairing of epidemiological data showing an association with laboratory analyses aimed at uncovering the mechanism of increased HIV susceptibility. There were also some limitations noted. First, it appears that there were only 10 women in the Prevotella-dominant CST4, resulting in wide confidence intervals, and potential for selection bias. Second, it is not clear whether any adjustment was undertaken for sexual risk behaviors. Third, the study population was restricted to younger South African women. It is unclear if these results can be generalized to other populations. Despite the limitations, these data provide potentially important evidence of a strong association between inflammation and P. bivia, possibly mediated by LPS.

At the same conference, McClelland et al presented a nested case-control study of diverse populations of women from six countries in Eastern and Southern Africa, including female sex workers, pregnant and postpartum women, and HIV-seronegative women in discordant couples [78]. There were 87 vaginal microbiota samples from case women who seroconverted to HIV and 262 samples from HIV-seronegative controls. Deep sequencing of broad-range 16S rRNA gene PCR products and highly sensitive taxon-directed qPCR assays were used to characterize the vaginal microbiota. Women who acquired HIV had greater vaginal bacterial community diversity compared to women who remained HIV-seronegative (mean Shannon Diversity Index 1.3 versus 0.9, p=0.02). Using qPCR, five bacterial taxa showed significant associations with HIV acquisition; Eggerthella species type 1, Gemella asaccharolytica, Leptotrichia/Sneathia, Megasphaera, and Mycoplasma hominis. These associations remained significant after adjustment for age, pregnancy, contraceptive use, sex partner number, sex frequency, and recent unprotected intercourse. A strength of this study was inclusion of a diverse population. In addition, the relatively large sample size facilitated adjustment for multiple potential confounders. Data from both deep sequencing data and qPCR assays were included, facilitating examination of concentration-dependent associations between bacteria and HIV acquisition. This study also had limitations. The epidemiological analyses were not paired with data exploring potential mechanisms linking vaginal bacteria to HIV susceptibility. In addition, this study explored associations with multiple bacterial taxa, and it is possible that some associations were observed by chance. Nonetheless, these data highlight potential biological gradients by showing concentration-dependent associations between several types of bacteria and women’s risk of HIV acquisition.

Mechanisms through which vaginal microbiota may influence HIV susceptibility

Numerous studies have focused on mechanisms through which the vaginal microbiota may influence HIV susceptibility. An exhaustive review of this literature is beyond this article’s scope. Nonetheless, understanding the possible mechanistic pathways is important for establishing biological plausibility. Readers are directed to excellent recent reviews by Mirmonsef (2012) [81], Petrova (2013) [82], Mirmonsef and Spear (2014) [83], Cone (2014) [84], and Murphy and Mitchell (2016) [85], for detailed reviews of mechanisms through which the vaginal microbiome may influence HIV susceptibility. A brief summary of key mechanisms is presented here.

Vaginal microbiota associated with BV can recruit mucosal immune cells. In Gosmann et al, women with diverse non-Lactobacillus dominant communities had 17-fold more activated CD4+ HIV target cells compared to women with Lactobacillus crispatus-dominant vaginal communities [79]. The same laboratory used transcriptional profiling to illustrate that epithelial cells and antigen presenting cells (APCs) sense high diversity vaginal communities associated with BV [80]. These APCs use Toll-like receptor-4 signaling to respond to LPS, which activates nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), leading to inflammation and recruitment of lymphocytes. Another recent study points to an alternative mechanism. In this study of US women, numbers of cervical gamma delta 1 (GD1) cells that were protective against HIV were higher in women with normal versus abnormal microbiota by Gram stain [86]. In contrast, vaginal GD2 cells, acting as targets for HIV entry into cells, were associated with abnormal vaginal microbiota.

Humoral immune mediators, including pro-inflammatory chemokines and cytokines, have been studied as a mechanism to explain the association between vaginal dysbiosis and HIV acquisition. Abnormal vaginal microbiota (Nugent score 4–10) has consistently been associated with higher levels of interleukin-1β (IL-1β), a pro-inflammatory cytokine associated with toll-like receptor (TLR) signaling and tissue damage [87–100]. Utilizing data from the CAPRISA 004 trial, cervicovaginal samples from 58 women prior to HIV seroconversion were matched to 58 women who remained HIV-negative [101]. This study showed higher concentrations of interferon gamma inducible protein (IP-10), MIP-1α, MIP-1β, and interleukin-8 (IL-8) in women who seroconverted to HIV compared to HIV-uninfected women. Additionally, MIP-1α and MIP-1β were associated with more diverse vaginal communities. These two chemokines, as well as IP-10, are chemotactic for T cells, monocytes, macrophages, and dendritic cells, all of which are potential HIV target cells [102, 103].

Since the 1990s, it has been evident that BV is associated with the presence of an HIV-inducing factor (HIF) in vaginal secretions [104]. This factor leads to increased HIV-1 replication in T cells and monocytes by activating AP-1 and NF-κB [105, 106]. Mycoplasma hominis, a vaginal bacteria species frequently linked with BV, has been significantly associated with HIF [107].

Lactobacillus-dominant vaginal microbiotas have generally been considered to reflect vaginal health, and have been associated with decreased risk of HIV acquisition [6, 48, 79]. Lactic acid, and associated low pH produced by glycogen metabolism by Lactobacillus species, can inactivate HIV [84, 108, 109]. A recent study also documented significant increases in IL-1RA, an anti-inflammatory cytokine, when human vaginal and cervical epithelial cell lines were treated with lactic acid [110]. This highlights a novel anti-inflammatory mechanism by which lactic acid may impact HIV susceptibility. In addition, some lactobacilli, including L. gasseri, may exert direct anti-HIV effects through bacteriocins, antimicrobial compounds that kill other microorganisms [83, 111].

Disruption of physical barriers, including cervicovaginal mucus and epithelium, may increase women’s HIV susceptibility. Cervicovaginal mucus acts as a physical barrier to HIV [112, 113]. In addition, the virus may diffuse more rapidly in cervicovaginal mucus with high concentrations of L. iners or G. vaginalis [114]. In contrast, there may be more virus trapping with L. crispatus-dominated microbiota. In a recent study by Borgdorff et al, vaginal dysbiosis was associated with cytoskeleton alterations, increased proteolytic activity, and cell death, likely representing epithelial damage [113].

Do high-risk vaginal bacteria increase women’s susceptibility to HIV infection?

In considering the question of whether vaginal microbiota influences women’s susceptibility to HIV infection, it is useful to refer to a set of criteria first proposed by Sir Austin Bradford Hill in 1965 [115], and used extensively in Epidemiology since that time. This list of conditions, often referred to as the Hill criteria, includes strength, consistency, specificity, temporality, biological gradient, plausibility, coherence, experiment, and analogy (Table 2). The Hill criteria are not a checklist to be fulfilled in their entirety to provide proof of causation, but do offer a helpful framework for examining causal inference.

Table 2.

Hill Criteria for Causation [115].

Criteria and explanation as described in Bradford Hill’s original 1965 paper (153)	Evidence from studies using classical methods for BV diagnosis	Evidence from newer studies using molecular methods to characterize vaginal microbiome	Data satisfy criterion1
Strength: “First upon my list I would put the strength of association […] In thus putting emphasis upon the strength of association we must, nevertheless, look at the obverse of the coin. We must not be too ready to dismiss a cause-and-effect hypothesis merely on the grounds that the observed association appears to be slight.”	Multiple prospective studies and meta-analyses generally agree that both BV and intermediate microbiota on vaginal Gram stain are associated with approximately 1.5-fold higher risk of HIV acquisition compared to normal microbiota.	Diverse non-Lactobacillus dominant vaginal bacterial communities, relative abundance of some bacteria, and quantities of some bacteria have been associated with increased risk of HIV acquisition. Risk estimates vary across the three available studies.	Partially
Consistency: “Has it been repeatedly observed by different persons, in different places, circumstances and times?”	Of 15 prospective studies and meta-analyses, all but 1 found a statistically significant association between BV and HIV acquisition	Across all three studies, more diverse bacterial communities were associated with HIV acquisition. Relationships between individual bacterial taxa and HIV acquisition are more consistent for some species (e.g. Sneathia and Mycoplasma) than for others (e.g. Prevotella, non-iners Lactobacillus species).	Mostly
Specificity: “One reason, needless to say, is the specificity of the association […] If as here, the association is limited to specific workers and to particular sites and types of disease and there is not association between the work and other modes of dying, then clearly that is a strong argument in favour of causation.”	Bacterial vaginosis is not a necessary exposure to acquire HIV.	There is limited evidence, but it seems highly unlikely that a particular vaginal community or the presence or high concentrations of a specific type of bacteria is required for HIV acquisition.	Not at all
Temporality: “My fourth characteristics is the temporal relationship of the association—which is the cart and which the horse?”	Numerous prospective studies have demonstrated an association between BV and subsequent acquisition of HIV.	The three available molecular studies all captured data on the vaginal microbiome prior to or very shortly after, HIV infection.	Fully
Biological gradient: “…if the association is one which can reveal a biological gradient, or dose-response curve, then we should look most carefully for such evidence.”	Prospective studies provide inconsistent evidence of a biological gradient for increased HIV risk with increasing Nugent score or numbers of Amsel criteria fulfilled.	There is some evidence from recent studies of increasing HIV risk with increasing vaginal bacterial community diversity, and with higher relative abundance and absolute concentrations of some types of bacteria	Partially
Plausibility: “It will be helpful if the causation we suspect is biologically plausible. But this is a feature I am convinced we cannot demand.”	Multiple biologically plausible mechanisms could explain the relationship between BV and increased susceptibility to HIV.	Studies at the molecular level suggest that more diverse vaginal bacterial communities, and the presence and concentrations of key bacteria, may increase HIV susceptibility.	Mostly
Coherence: “[…] the cause-and-effect interpretation of our data should not seriously conflict with the generally known facts of the natural history and biology of the disease.”	Laboratory studies have mostly shown plausible mechanisms, as noted in the previous row. However, there is not a good animal model of the Lactobacillus-dominated healthy human microbiota associated with lower risk. Ex vivo infection of biopsy specimens from women with different vaginal conditions could be helpful in providing additional data showing coherence between laboratory data and epidemiological studies.		Minimally
Experiment: “Occasionally it is possible to appeal to experimental, or semi-experimental, evidence. For example, because of an observed association some preventive action is taken. Does it in fact prevent?”	Clinical trials of BV interventions have produced moderate reductions in BV prevalence over periods up to one year. However, no effective interventions for reducing BV have been evaluated in HIV prevention trials.	One study has demonstrated that periodic presumptive treatment with metronidazole plus miconazole reduces detection and concentrations of several BV-related bacteria. However, no effective interventions targeting the vaginal microbiome have been evaluated in HIV prevention trials.	Not at all
Analogy: “In some circumstances it would be fair to judge by analogy.”	Prospective studies and clinical trials provide some evidence for an analogous relationship between BV and STIs other than HIV. However, these are also not proven causal relationships, and could be mediated through separate mechanisms.	There is some evidence that the presence of key bacteria may increase susceptibility to STIs other than HIV. However, these are also not proven causal relationships, and could be mediated through separate mechanisms.	Minimally

Abbreviations: BV=bacterial vaginosis; CST=community state types; HIV=human immunodeficiency virus; STIs=sexually transmitted infections

¹Overall degree to which the available evidence satisfies each of the 9 Hill criteria on a scale that includes not at all, minimally, partially, mostly, fully.

Strength of association

Multiple prospective studies and meta-analyses estimate an ~1.5-fold higher risk of HIV acquisition with BV compared to normal vaginal microbiota. Diverse non-Lactobacillus dominant bacterial communities, higher relative abundance of some bacteria, and higher quantities of some bacteria, have been associated with increased risk of HIV acquisition. Effect estimates for significant associations range from an aOR of 2.59 (95% CI 1.26–5.34) for the highest concentrations of Leptotrichia/Sneathia [78], to an aOR of 12.7 (95% CI 2.1–77.8) with P. bivia [77]. Most estimates fall toward the lower end of this range. Additional data will be essential to clarify the strength of these associations.

Consistency

Of the 15 prospective studies and meta-analyses [4–8, 28, 29, 56, 57, 73–75, 116–118], all but one have shown a statistically significant association between BV and HIV acquisition [118]. The three recent studies using newer molecular techniques have also documented an association between more diverse bacterial communities and HIV acquisition. However, relationships between individual bacterial taxa and HIV acquisition are more consistent for some bacteria (e.g. Sneathia and Mycoplasma hominis) than for others (e.g. Prevotella and non-iners Lactobacillus species). These differences may be related to the biology of the vaginal mucosa in different populations, methodological differences, or could have resulted by chance. In general, there is high consistency in studies showing an association between BV, diverse vaginal microbiota, and HIV acquisition. Further work is needed to clarify the relationships between individual bacterial taxa and HIV risk in different populations.

Specificity

The relationship between vaginal microbiota and HIV is not specific, as BV is not required for HIV acquisition. While there are few molecular studies to date, it seems unlikely that a particular bacterial community or species would be a necessary precursor to HIV infection. Importantly, specificity is not essential for determining the existence of a causal relationship.

Temporality

Numerous prospective studies have demonstrated an association between BV and subsequent acquisition of HIV. In addition, all three available molecular studies collected data on the vaginal microbiome prior to or very shortly after HIV infection.

Biological gradient

Prospective studies provide inconsistent evidence for a biological gradient for increased HIV risk with increasing Nugent score [8, 28], or numbers of Amsel criteria fulfilled [4]. There is some evidence from recent molecular studies, that increasing vaginal bacterial community diversity, relative abundance, and absolute concentrations of some bacteria may be associated with increased HIV risk.

Plausibility

Numerous studies demonstrate possible mechanisms through which the vaginal microbiome could contribute to HIV susceptibility. Additional research that more directly links vaginal bacteria to mucosal markers of HIV susceptibility could further strengthen this link.

Coherence

Laboratory studies have primarily addressed mechanisms to explain the association between vaginal microbiota and increased HIV susceptibility. These types of data would seem to pertain more directly to biological plausibility than to coherence between laboratory studies and epidemiological evidence. Future studies should aim to more directly explore the hypothesis that vaginal dysbiosis increases susceptibility. Unfortunately, there is not a suitable animal model of the Lactobacillus-dominant healthy human vaginal microbiota. As such, ex vivo infection of biopsy specimens from women with different vaginal conditions might provide the strongest evidence for coherence.

Experiment

There is some evidence that periodic presumptive treatment (PPT) and suppression approaches can reduce BV [119–122], leading to more optimal vaginal microbiota increased frequency of hydrogen-peroxide producing Lactobacillus species colonization [123], and reduced quantities of BVAB-1, BVAB-2, Atopobium vaginae, Leptotrichia/Sneathia, and Megasphaera [124]. However, no effective interventions for reducing BV or modifying the vaginal microbiota have been evaluated in HIV prevention trials. Such trials, if undertaken in the era of effective HIV prevention interventions including pre-exposure prophylaxis (PrEP) and treatment of HIV-positive partners, would likely require sample sizes in the tens of thousands.

Analogy

The association between the vaginal microbiome and women’s susceptibility to HIV infection represents a unique biological relationship for which it is difficult to identify a closely analogous system. The relationship between BV and other STIs has some parallels. However, the mechanisms of susceptibility to other STIs may not reflect the most important mechanisms mediating HIV susceptibility.

In summary, current data fulfill some of the Hill criteria for assessing whether the vaginal microbiome is causally related to HIV susceptibility. Additional epidemiological studies with prespecified hypotheses will be valuable in establishing which relationships between individual bacterial taxa and communities are consistent across multiple studies and populations. Because many of the current microbiome studies include multiple comparisons, using a global p-value for the entire vaginal microbiota, and delving deeper only if the overall p-value is significant, or using a false discovery rate, may help to reduce the number of associations identified by chance.

Clinical trials of vaginal health interventions would provide the gold standard of evidence, and could be considered if the mechanistic data are sufficiently strong and the proposed interventions can be shown to impact the putative mechanisms driving HIV susceptibility.

Impact of the vaginal microbiome on microbicide and oral pre-exposure prophylaxis efficacy

In the past 10 years, there has been significant investment in both topical and oral PrEP to reduce women’s HIV risk. Of the five major trials, three have shown a benefit [125–127], with the other two citing low adherence as a reason for null results [128, 129]. Of the microbicide trials, only the CAPRISA 004 trial demonstrated significant efficacy of intravaginal tenofovir PrEP [125].

Three recent studies have examined how the vaginal microbiome may impact topical and oral tenofovir-based PrEP. The first of these included data from the CAPRISA 004 trial [125]. Women were randomized to vaginal tenofovir 1% gel versus placebo. Cervicovaginal lavages were analyzed from women prior to acquiring HIV compared to samples from randomly selected HIV-seronegative women [130]. Using a metaproteomic analysis, 188 species were identified. Vaginal microbiomes were classified as Lactobacillus-dominant or non-Lactobacillus dominant. There were no differences in demographics, sexual behaviors, clinical characteristics, or adherence between women with these two vaginal microbiome types. Women with Lactobacillus-dominant vaginal microbiota showed significantly reduced risk of HIV acquisition with vaginal tenofovir gel compared to placebo (HR 0.39, 95% CI 0.20–0.83). In contrast, women with a non-Lactobacillus dominant microbiota treated with tenofovir gel had no difference in HIV acquisition compared to placebo (HR 0.82, 95% CI 0.40–1.65).

To explore the mechanism underlying these results, tenofovir was inoculated in media with G. vaginalis, L. iners, L. crispatus, and an abiotic control [130]. Mass spectrometry illustrated that in vitro, G. vaginalis was associated with a 67.4% depletion of tenofovir levels. This dramatic effect was not present when media was inoculated with L. iners, L. crispatus, or the abiotic control. Thus, metabolism of tenofovir by G. vaginalis could explain the findings of decreased efficacy of vaginal tenofovir in women with non-Lactobacillus dominant microbiota. An abstract presented at the Conference on Retroviruses and Opportunistic Infections (CROI) in early 2017 by Hillier et al further illustrates this point [131]. Lower cervical tissue tenofovir levels were observed in women receiving tenofovir 1% gel when they had BV diagnosed by Nugent score, higher concentrations of G. vaginalis, and higher concentrations of A. vaginae.

In contrast to topical tenofovir, the efficacy of oral PrEP with tenofovir disproxil fumarate does not appear to be affected by the vaginal microbiome [132]. Heffron et al’s study presented data from a secondary analysis from the Partners PrEP study [127]. Efficacy of daily oral tenofovir was the same for women with normal microbiota, intermediate microbiota, and BV by Gram stain. Oral PrEP efficacy also did not differ by presence of Gardnerella/Bacteroides morphotypes on Gram stain.

In summary, recent studies provide strong evidence that the vaginal microbiome impacts the efficacy of topical PrEP, an effect not present with oral PrEP. These results are likely to dramatically change the way vaginal microbicides are developed and tested.

Current approaches for modulating the vaginal microbiome

Current regimens for symptomatic BV provide modest cure rates with frequent recurrences [26, 27, 133, 134]. New approaches to control BV have included alternative drug regimens [135], probiotics [136], biofilm disruptors [137–143], risk factor modification (e.g. cessation of intravaginal practices) [144, 145], and suppressive regimens administered as PPT and periodic directed treatment [119, 120, 123, 146]. Many of these regimens have reduced BV recurrences. Less is known about such regimens’ impact on individual bacterial taxa. One study found that oral PPT with 2 grams of metronidazole each month resulted in more frequent Lactobacillus colonization [123], and another found that monthly intravaginal metronidazole PPT reduced concentrations of BVAB-1, BVAB-2, Atopobium vaginae, Leptotrichia/Sneathia, and Megasphaera measured using quantitative PCR assays [124]. However, producing sustained changes in the vaginal microbiome remains a challenge only partially addressed by current regimens.

Conclusion

Prospective studies have consistently demonstrated an association between BV and increased risk of HIV acquisition. Recently, data from studies using molecular methods have led to the hypothesis that high-risk vaginal microbial communities and the presence and concentrations of key bacterial taxa may be more predictive of women’s HIV risk than a BV diagnosis by traditional methods. One limitation of the current literature is a lack of attention to the potential role of vaginal fungi and viruses (other than herpes and human papilloma viruses) in mediating HIV susceptibility in women. In addition, recently presented and published data examining associations between the vaginal microbiome and HIV acquisition have come from observational studies, and may be most useful for generating new hypotheses about how the vaginal microbiome influences HIV susceptibility.

Given the high prevalence of vaginal dysbiosis, particularly in African women, the question of whether vaginal health interventions could reduce women’s susceptibility to HIV infection is of great importance. Current research priorities should include identifying the vaginal microbiota associated with increased risk, substantiating mechanisms linking vaginal bacteria to mucosal HIV susceptibility, and evaluating the efficacy of interventions for interrupting these mechanisms. If such data are sufficiently convincing, clinical trials of vaginal health interventions should be considered. While clinical trials of new HIV prevention interventions will be challenging in an era where other effective HIV prevention strategies are available, such studies would provide the strongest evidence to prove or disprove the presence of a causal association, and could provide an important and novel approach to HIV prevention in women.