The vaginal microbiome of sub-Saharan African women: revealing important gaps in the era of next-generation sequencing

Search Strategy

To select eligible and relevant literature for this review, we conducted a peer-reviewed article search strategy using important key words. Searches included articles and grey literature including reviews and original research published in PubMed, PubMed central, Google Scholar, Scopus, Web of Science, Evidence-Based Medicine, Biosis preview, Biological Abstract and African Journal Online database.

Identification of Eligible Studies

From the database search, titles, abstracts and full-text versions of articles were identified and screened for potential eligibility. After title, abstract and full-text reviews, irrelevant and non-eligible articles were screened out, leaving only potentially relevant ones. Eligible articles were studies written in English language. Multiple keywords were used for the literature search both alone as well as in combination. Some of the important keywords used for literature search were vaginal microbiome, vaginal microbiota studies, sequencing approach, amplicon marker gene sequencing, next-generation sequencing platforms, vaginal microbiota of African women, postpartum vaginal profile, vaginal microbiota during pregnancy in African cohorts. Original research and critical reviews were both included and studies irrelevant to the scope of this review were excluded described in Fig. 1. All three investigators independently reviewed titles/abstracts and full text for eligibility. The reference lists of eligible articles were also screened to detect relevant articles that were not identified by the initial search strategy.

Evaluation of eligible studies

All investigators independently extracted data from the selected search database and downloaded article. Any discrepancies in data extraction and risk of bias assessment were resolved by consensus. All authors reviewed article, titles and abstracts independently and retrieved full articles that potentially met the inclusion criteria. Having identified the studies that met the inclusion criteria, full text versions of these articles were read and saved in personal devices.

Selection of eligible studies

Of 550 unique titles/abstracts identified from the database search, 29 were excluded after title review and 305 after abstract review, leaving 216 eligible articles for full text review (Fig. 1). Of these, 36 articles were discarded as duplicates as they were found more than once in the selected database search engine. The remaining 180 titles and abstracts of articles were reviewed again and another 33 articles were excluded, based on their irrelevance or inability to meet specified criteria. The remaining 147 full-text studies were retrieved and read in full (Fig. 1).

The vaginal ecosystem

The vaginal epithelium comprises of three cell layers, superficial, intermediate, and basal. The epithelial mucosa of the lower genital tract is extensively populated by commensal microorganisms, while the tissues of the upper genital tract are not colonized by commensals thus are less prone to infection (Rampersaud, Randis & Ratner, 2012). The vaginal microbiome is distinctive for its relatively simple biodiversity, low species richness and numerous Lactobacillus species Human Microbiome Project Consortium (2012). Lactobacillus stand out as key players in modulating reproductive health in post pubertal women by exerting a protective effect on the vagina. The mechanisms by which Lactobacilli modulate reproductive health of women is not known with certitude but may be by it acting as a competitor with other pathogenic organisms for nutrients, epithelial cell receptors and space (Boris et al., 1998). Other putative mechanisms are the release of metabolites and secretion of bacteriocins to maintain a low hostile vaginal pH (Martin & Suarez, 2010) and the production of lactic acid which protects the vagina from colonization by other species (Witkin et al., 2013).

A vaginal microenvironment less dominated by Lactobacilli predisposes to adverse clinical conditions like BV (Allsworth & Peipert, 2007; Srinivasan & Fredricks, 2008) or non-specific vaginitis (Amsel et al., 1983). Due to the multifaceted function of the vagina, coupled with its anatomical location, it may be influenced by hormones, menstruation, douching practices, contraceptives, sexual intercourse and the gastrointestinal microflora from the rectum (Reid, 2018). Several studies have reported a Lactobaccillus depleted vaginal profile in African-American and Hispanic women (Shendure, Porreca & Reppas, 2005; Ravel et al., 2011; Zhou et al., 2010b). Jespers et al. (2014) noted similar findings in a cohort of African women. Interestingly, others have also reported a high prevalence of BV in sub-Saharan African women (Gautam et al., 2015; Torrone et al., 2018). With the advent of next generation sequencing, explicit identification of vaginal microbiota has been made possible. This move has also provided a platform for scientists to make comparisons across populations and construct novel research questions in vaginal microbiology and ecology.

Next generation sequencing

NGS describes a method of sequencing where millions of oligonucleotides sequencing fragments are executed in parallel, giving rise to large number of sequencing outputs (Mendz, Kaakoush & Quinlivan, 2016). Until the HMP was established, therapeutic interventions and treatment of vaginal disorders has been unsuccessful because the identification of the complex vaginal microbial community depended on the tripods of clinical diagnosis, microscopy and basic culture technique (White et al., 2011). DNA Sequencing was first elaborated by Sanger in a method known as Sanger sequencing or the chain-terminator methods (Sanger, Nicklen & Coulson, 1977). It was developed in 1977 and remains the “gold standard” in molecular diagnostics. It operates by utilizing DNA polymerase to generate a complementary copy to a single stranded DNA template which bind to a given primer. Due to primer binding, the preceding bases of the sequences produced are usually of poor quality (Sanger, Nicklen & Coulson, 1977; Adams, 2008). In addition, it is time consuming and expensive. With these limitations, Sanger sequencing has been replaced by other powerful next-generation sequencing methods which has improved the identification of the myriads of microbes even in larger scale (NIH et al., 2009; Human Microbiome Project Consortium, 2012). A major application of NGS is in phylogenetic sequencing analysis.

Application of next generation sequencing

Protocol for 16S rRNA gene sequencing

16S gene sequencing has shown its efficacy in both deciphering bacterial species in environmental specimen and establishing phylogenetic relationship between them (Shah et al., 2011; Eren, Ferris & Taylor, 2011). This analysis has a robust but simplified protocol given that it requires only polymerase chain reaction (PCR) and sequencing. First, an amplicon of the 16S gene is obtained through PCR. Amplicons are then sequenced by targeting the hypervariable regions of choice. The sequence obtained can be matched with a reference sequence from an existing DNA database. These signature nucleotides (reference sequences of 16S rRNA gene) allows for taxonomical classification and identification by basis of similarities to already known sequences in preceding databases (Chanama, 1999; Barghoutti, 2011; Mizrahi-Man, Davenport & Gilad, 2013). Furthermore, several bioinformatic pipeline can be used to analyze the resulting sequences including QIIME 2 (Bolyen et al., 2019), MOTHUR, USEARCH-UPARSE (for OTU-level), DADA2, USEARCH-UNOISE3(for ASV-level) (Prodan et al., 2020). Existing NGS platforms for 16S rRNA sequencing are described in Table 1.

Vaginal microbiota in non-pregnant African women

The vaginal microbial communities have been studied in multiple levels, from morphological descriptions to understanding the genetic signature of microbes and how the mixtures of microbes could promote or disrupt reproductive outcome. By microscopy, the vaginal microbiota of Black women are reported to correlate with high Nugent Scores and a low proportion of Lactobacilli compared to their European counterparts (Nugent, Krohn & Hillier, 1991; Royce et al., 1999; Ness et al., 2003; Jespers et al., 2014). These observations were further buttressed by terminal restriction, fragment polymorphism and shallow profiling of the 16S rRNA ribosomal gene (Zhou et al., 2007; Zhou et al., 2010a; Zhou et al., 2011). By pyrosequencing, Zhou et al. observed a higher prevalence of Lactobacillus specie in Black women (33%) compared to the 7% observed in Caucasians. Furthermore, only one or two species of Lactobacillus were found in the few Black participant with a Lactobacillus profile (Zhou et al., 2010b). Similarly, Ravel et al. (2011) characterized the vaginal microbiota of 396 women by pyrosequencing of the V1 and V2 region of the 16S rRNA gene and identified a high prevalence of BVAB in African-American women (39%) compared to the lower prevalence in Asians (18%) and Caucasians (9%). The absence of L. jensenii vagitype and minute proportion of L. crispatus in African Americans was another important observation noted in their study (Ravel et al., 2011). In keeping with this, by sequencing the V1-V3 region of the 16S rRNA gene, Fettweis et al. also described the vaginal profile of Black women to be depleted of Lactobacillus and rich in BVAB, including Prevotella and Sneathia (Fettweis et al., 2014). It should be noticed that these studies highlighted are reports on African women living outside Africa. Results obtained from characterizing the vaginal microbiome of African women living in Africa appear to deviate from what has been observed among Africans in the diaspora. This raises important questions about the influence of geography on the vaginal microbiome. With NGS technology, a few studies have provided insight into the vaginal microbiome of women in Africa. In an 8-week longitudinal cohort study, Jespers et al. (2017) studied the vaginal microbiota of South African, Rwandan and Kenyan women and these were reported to be relatively stable and dominated by L. iners (75%) and L. crispatus (35%). Two other studies in the South African population also reported an abundance of L. crispatus and L. iners, including an heterogenous mix of CST IV microbes in the vaginal microbiota of these women in Africa (Anahtar et al., 2015; Bayigga et al., 2019).

Similarly, Lennard et al. (2017), observed a vaginal microbiota dominated by Lactobacillus species and some proportions of BVAB. No remarkable differences were found between the vaginal microbiome of Nigerian and Swedish women. Anukam and colleagues reported the presence of L. gasseri, L. crispatus and high proportions of L. iners in Nigerian women (Anukam et al., 2006), which is similar to the vaginal microbiome profile that had earlier been reported in Swedish women (Vasquez et al., 2002). Bacterial vaginosis-associated vaginal profile has also been reported in African women (Torrone et al., 2018; Jespers et al., 2014). By Illumina sequencing, the vaginal microbial profile of Tanzania women was characterized and found to have a significant proportion of Prevetolla bivia an observation made in only a small proportion of Caucasian and African-American women in North America (Hummelen et al., 2010).

Similarly, Lennard et al. (2017) reported a vaginal microbiota dominated by Gardnerella, Prevotella and Lactobacillus species. Furthermore, in a longitudinal study, Gossman and colleagues sequenced the V4 region of the 16S rRNA gene and reported a diverse vaginal microbiota dominated by G. vaginalis, Prevotella, Megasphaera, Sneathia, and Shuttleworthia in 58% of the study cohort. Only few subjects had a Lactobacillus profile dominated by L. iners and L. crispatus (Gosmann et al., 2017). The higher prevalence of BV in Black women compared to White may be explained by differences in host genetics (Ness et al., 2003; Gajer et al., 2012; Hickey et al., 2013). Besides ethnic influence and geographical consideration the vaginal microbiome of African women may also vary due to diet (Faucher et al, 2019; Tuddenham et al., 2019), innate/adaptive immunity (Jespers et al., 2017; Torcia, 2019), hormonal flunctuation (Gajer et al., 2012; Van de Wijgert et al., 2013) and other confounding factors (Koumans et al., 2007; Peipert et al., 2008). Given the inconsistency in reports from various studies on the vaginal microbiome of African women, future studies are definitely necessary.

Vaginal microbiome of African women during pregnancy

Pregnancy represents a unique phase, characterized by a suspension of the menstrual cycle vaginal microbiome (Genc & Onderdonk, 2011). During pregnancy, the vaginal microbiome is more enriched with Lactobacillus than in the non-pregnant state (Romero et al., 2014; Freitas et al., 2017). Several studies have described the vaginal microbiota in pregnant women. These studies have also noted significant differences in the vaginal profile of Black and White women. African-American ethnicity increases the likelihood for having an absence of protective Lactobacilli which predisposes to preterm birth and other pregnancy complications (Beigi et al., 2005; Larsson et al., 2007; Klatt et al., 2010). Since no prior study has described in details the vaginal microbiome profile of African women during pregnancy in a longitudinal fashion, researchers continue to rely on results extrapolated from African women in the diaspora. Although ethnicity may have a significant influence on the vaginal microbiome, geographical variations may also be contributory. The study of Hyman et al. (2014) observed a low proportion of Lactobacillus in Black women who encountered preterm birth (Hyman et al., 2014). Fettweis and colleagues made a similar observation (Fettweis et al., 2019). These findings were further reinforced by the observations made in a longitudinal study of a cohort comprising of 23 White, 5 Black and 13 Asian healthy women. A large proportion of L. jensenni and L. gasseri were reported in White and Asian women but no traces of L. gasseri and L. jensenni were found in the vaginal samples of the Black women in this cohort (MacIntyre et al., 2015). Conversely, a study conducted in Burkina Faso which features HIV- infected pregnant women at 36–38 weeks’ gestation reported a large number of women having a Lactobacillus-dominant profile comprising of three distinct clusters. The first cluster comprised of L. iners (77%), L. crispatus (11%), L. fornicalis (3.9%), L. gasseri (3.2%) and L.vaginalis (0.5%). The second cluster comprised of coagulase-negative Staphylococcus while the third group of bacteriomes were a mixture of microbes of the CST IV type, dominated by Gardnerella species (Frank et al., 2012). Obviously, a Lactobaccillus depleted and BV-dominated vaginal profile correlates significantly with STI and HIV (Bayigga et al., 2019), yet the study of Frank et al. reported some Lactobacillus vagitypes in the vaginal profile of the HIV-positive pregnant women. To bridge the gap in these discrepancies, a geographically tailored approach to vaginal microbiome science is required.

Vaginal microbiome of African women in the postpartum period

Following the changes that occur in a woman’s physiology during the postpartum, the vaginal microbiome profile is dramatically altered. During pregnancy, estrogen in maternal circulation rises (Roy & Mackay, 1962; Siiteri & MacDonald, 1966), however during the postpartum elevated level of estrogen falls dramatically due to expulsion of the placenta (Nott et al., 1976; O’Hara et al., 1991). This suggests why any estrogen driven Lactobacillus during pregnancy are significantly depleted postpartum (MacIntyre et al., 2015). Only few studies have successfully described the composition of the vaginal profile in postnatal women of African descent using 16S rRNA gene sequencing. These studies have observed a predominance of BVABs, Prevotella, Anaerococcus, Streptococcus, Atopobium and Peptoniphilus than Lactobaccillus in numerous postnatal women (Poretsky et al., 2014; MacIntyre et al., 2015; DiGiulio et al., 2015; Doyle et al., 2018). Notable is a study which described the vaginal microbial profile of rural Malawian women postpartum as being dominated by Gardnerella vaginalis (75.7%), with minute proportions of L. crispatus and L. iners in 30.4% of the study population (Doyle et al., 2018). The report Doyle’s group presented is similar to other observations on the postpartum microbiome in several other populations (Poretsky et al., 2014; MacIntyre et al., 2015; DiGiulio et al., 2015; Doyle et al., 2018). These observations are interesting, given that these studies featured participants from different ethnicities with variations in sample size, sample collection methods and laboratory methods. It therefore appears that the postpartum microbiome may neither be influenced by ethnicity nor geography. Till date, only the study of Doyle et al. (2018) has described the vaginal microbiota in an African population (rural Malawian women) postpartum employing 16S rRNA sequencing (Table 2). This highlights the need for further studies on the vaginal microflora during the postpartum. Another transition requiring further study is the period of restoration from the postpartum vaginal profile to the interpregnancy (normal) profile. While MacIntyre et al. (2015) focused on a mixed ethnic cohort at 6 weeks postpartum, Doyle’s group focused on a postpartum cohort one week after delivery and followed the cohort up for up to one year, yet reported no trace of Lactobacillus restoration (Doyle et al., 2018). Another group observed a cohort of postnatal women for one year, yet no profound vaginal Lactobacillus was observed (DiGiulio et al., 2015). A large longitudinal study is therefore recommended to establish the composition of the postpartum vaginal microbiome accurately and to provide more insight into how a Lactobacillus profile is restored after lochia regression. The vaginal microbiome composition of sub-Saharan African women is described in Table 2.

Alternative platform for bacterial identification

Over a decade after the recommendations by the HMP, we have witnessed a growing body of literatures deployed the 16S rDNA sequencing for bacteria identification. Although it’s been a reliable and convenient method of bacterial species identification, it has some shortfalls. It is difficult for bacteria that share similar gene sequence to be differentiated at specie level. When sequences are aligned wrongly, bacteria species are matched incorrectly. Other pitfalls with this technique are hitches with purity of bacteria isolates and sequencing artefacts which introduce errors into a DNA database which mostly likely is interpreted as an existing or reference database for new studies thus hampering accurate bacterial identification (Tshikhudo et al., 2013). Alternative cutting-edge technologies are recommended to facilitate bacteria identification even further (Table 3).