Characterization of the vaginal microbiota among sexual risk behavior groups of women with bacterial vaginosis

Abstract

Background: The pathogenesis of bacterial vaginosis (BV) remains elusive. BV may be more common among women who have sex with women (WSW). The objective of this study was to use 454 pyrosequencing to investigate the vaginal microbiome of WSW, women who have sex with women and men (WSWM), and women who have sex with men (WSM) with BV to determine if there are differences in organism composition between groups that may inform new hypotheses regarding the pathogenesis of BV.

Methods: Vaginal swab specimens from eligible women with BV at the Mississippi State Department of Health STD Clinic were used. After DNA extraction, 454 pyrosequencing of PCR-amplified 16S rRNA gene sequences was performed. Sequence data was classified using the Ribosomal Database Program classifer. Complete linkage clustering analysis was performed to compare bacterial community composition among samples. Differences in operational taxonomic units with an abundance of ≥ 2% between risk behavior groups were determined. Alpha and beta diversity were measured using Shannon's Index implemented in QIIME and Unifrac analysis, respectively.

Results: 33 WSW, 35 WSWM, and 44 WSM were included. The vaginal bacterial communities of all women clustered into four taxonomic groups with the dominant taxonomic group in each being Lactobacillus, Lachnospiraceae, Prevotella, and Sneathia. Regarding differences in organism composition between risk behavior groups, the abundance of Atopobium (relative ratio (RR)=0.24; 95%CI 0.11-0.54) and Parvimonas (RR=0.33; 95%CI 0.11-0.93) were significantly lower in WSW than WSM, the abundance of Prevotella was significantly higher in WSW than WSWM (RR=1.77; 95%CI 1.10-2.86), and the abundance of Atopobium (RR=0.41; 95%CI 0.18-0.88) was significantly lower in WSWM than WSM. Overall, WSM had the highest diversity of bacterial taxa.

Conclusion: The microbiology of BV among women in different risk behavior groups is heterogeneous. WSM in this study had the highest diversity of bacterial taxa. Additional studies are needed to better understand these differences.

Figure 1. Bar graph of the most abundant OTUs within the total sample and within each sexual risk behavior group.

The bacteria group represents sequences which cannot be classified beyond the kingdom (bacteria) level.

Figure 2. Heatmap with colors showing the abundance of microbial taxa (in proportion) found in the vaginal bacterial communities of 112 women with BV (color key is indicated to the left of the heatmap diagram).

(A) Complete linkage clustering of samples based on the species composition and abundance of vaginal bacterial communities that define clusters I-IV. (B) Nugent score, pH measurement, and sexual risk behavior group for each of the 112 vaginal bacterial community samples (color keys for Nugent scores, pH measurements, and sexual risk behavior groups are to the right and the left of the heatmap diagram). (C) Complete linkage clustering of taxa based on Spearman’s correlation coefficient profiles. (D) Shannon diversity indices calculated from the 112 vaginal bacterial communities.

Figure 3. Pie chart representation of the vaginal bacterial communities comprising clusters I-IV within each sexual risk behavior group of women.

The number of women from each sexual risk behavior group is indicated in parentheses. Note that 3 samples from the WSWM group contained taxonomic groups that were considered as outliers (i.e., had a major proportion of unidentified bacteria, Enterobacteriaceae, or Serratia) and were not included in this analysis.

Figure 4. Forest plot depicting the differences in abundance estimates (blue dot) and 95% confidence intervals (CIs) (solid red line) among sexual risk behavior groups of each of the taxonomic groups with an abundance of ≥2% across the entire sequence library.

A relative ratio of 1 (dotted vertical line) signifies that both comparison groups have the same mean abundance. If the solid red line for sexual risk behavior group comparisons does not overlap with a relative ratio of 1, a statistically significant difference between the two comparison groups is supported.

Figure 5. Box plot of Shannon’s diversity for each sexual risk behavior group.

The red plus signs depict the average value of Shannon’s diversity for each sexual risk behavior group; blue dots represent outliers. The WSM group had the highest diversity compared to the other sexual risk behavior groups.

Figure 6. Unweighted beta-diversity Unifrac analysis.

This figure depicts a distinct cluster within the WSM group (indicated in green) that is not present within the WSW or WSWM groups (indicated in blue and red, respectively).