Use of an Ethinyl Estradiol/Etonogestrel Vaginal Ring Alters Vaginal Microbial Communities in Women With HIV

Abstract

Background: Human immunodeficiency virus 1 (HIV-1) antiretroviral therapy (ART) alters hormonal contraceptive levels delivered via intravaginal ring (IVR) in a regimen-specific manner. We explored the role of the IVR on vaginal microbial communities, vaginal short chain fatty acids (SCFAs), vaginal HIV shedding, and the effect of vaginal microbes on hormone concentrations in cisgender women with HIV (WWH).

Methods: Vaginal microbes were assessed by 16S RNA sequencing of weekly vaginal swabs, vaginal SCFA by mass spectrometry, HIV-1 shedding by nucleic acid amplification on vaginal aspirates, and bacterial vaginosis by Nugent scoring from 74 participants receiving an etonogestrel/ethinyl estradiol (ENG/EE) intravaginal ring while on no ART (n = 25), efavirenz-based ART (n = 25), or atazanavir-based ART (n = 24).

Results: At baseline, microbial communities of the 64 substudy eligible participants robustly classified as Lactobacillus crispatus-dominant (n = 8), Lactobacillus gasseri-dominant (n = 2), Lactobacillus iners-dominant (n = 17), or mixed anaerobic communities (n = 37). During IVR therapy, there was an increased probability of Lactobacillus-dominant community state types (CSTs) (odds ratio = 1.61, P = .04). Vaginal CSTs were associated with Nugent scores. Bacterial vaginosis-associated bacteria were associated with significantly higher and L. iners with lower Nugent scores (all P adjusted <.1). Lactic acid levels were correlated with the relative abundance of Lactobacillus species (r2 = 0.574; P < .001). Vaginal shedding of HIV-1 was less common in women with L. crispatus-dominant microbiomes (P = .04). Mixed anaerobic vaginal communities modulated EE concentrations in a regimen-specific manner.

Conclusions: Combined ENG/EE IVR therapy was associated with an increase in Lactobacillus-dominant vaginal microbial communities in WWH and may benefit those with bacterial vaginosis. EE levels were altered by the vaginal microbiota.

Keywords: HIV transmission; contraceptive therapy; intravaginal ring; vaginal microbiome.

Figure 1.

Schematic of the study design. The ethinyl estradiol (EE)/etonogestrel (ENG) intravaginal ring was placed at week 0 (Ring Insertion) and removed at week 3 (Ring Removal). 16S rRNA sequencing was performed at study entry (week 0), weekly during the 3 weeks of ring placement (week 1–3), and 1 week following ring removal (week 4). Vaginal short chain fatty acids (SCFAs) were measured at week 0 or 1 depending on sample availability and weeks 2 and 4. Vaginal HIV RNA/DNA for vaginal shedding of HIV-1 was measured at entry and week 3.

Figure 2.

Heatmap of species-level relative abundances sorted by treatment group and vaginal CST. Rows 4–6 show measured levels of EE, ENG, and progesterone. Vaginal CST I–IV were assigned at each visit with CST I, L. crispatus; CST II, L. gasseri; CST III, L. iners; and CST IV, mixed anaerobic community. Nugent scores are presented with values of 0–10 or by further classification (Nugent2) of BV negative (0–3), intermediate BV (4–6), or BV positive (7–10). Gray denotes no slide available. Abbreviations: ART, antiretroviral therapy; BV, bacterial vaginosis; CST, community state type; EE, ethinyl estradiol; ENG, etonogestrel.

Figure 3.

Vaginal CST, EE/ENG levels, and ART-vaginal microbe interactions. A, Vaginal CST by study visit. B, Vaginal CST by treatment group and study visit. C, Regression estimates and 95% CI for EE and ENG drug levels as a function of treatment group, CST, and study visit. Similar to the main study, EE (P = .03) and ENG (P < .001) levels were significantly lower in the EFV group and ENG (P < .001) levels were significantly higher in the ATV group. EE levels were increased by CST IV (CST IV vs CST I; P = .02). ATV blunted the CST IV-associated increase in EE (ATV CST IV, P = .02). D, Boxplot of contraceptive levels by treatment group and CST where the line defines the median, the box the interquartile range, and the whiskers the minimum and maximum values. Abbreviations: ART, antiretroviral therapy; ATVr, atazanavir-ritonavir; CI, confidence interval; CST, community state type; EE, ethinyl estradiol; EFV, efavirenz; ENG, etonogestrel; NS, not significant.

Figure 4.

Association of microbes with Nugent scores and HIV vaginal shedding. A, Linear regression of species by Nugent score. Positive estimates denote species that are more abundant in cisgender women with higher Nugent scores. Negative estimates denote species that are more abundant in women with lower Nugent scores. B, Zero-inflated negative binomial regression of species by vaginal shedding of HIV. Positive estimates denote species that are more abundant in women with viral shedding. Negative estimates denote species that are more abundant in women without viral shedding. Abbreviations: HIV, human immunodeficiency virus; NS, not significant.