Vaginal Microbiota and Mucosal Pharmacokinetics of Tenofovir in Healthy Women Using a 90-Day Tenofovir/Levonorgestrel Vaginal Ring

Abstract

Background: A relationship between the vaginal microbiota and tenofovir (TFV) concentrations and activity after topical administration has been previously reported.

Objective: CONRAD A15-138 was a randomized, placebo-controlled Phase I study aimed at characterizing the safety, pharmacokinetics (PK), and pharmacodynamics (PD) of TFV and levonorgestrel (LNG) administered through a vaginal ring (IVR) for 90 days. Herein, we describe changes from baseline in the vaginal microbiota with IVR use and the impact of the vaginal microbiota on mucosal TFV PK.

Methods: The study screened 68 participants and randomized 47 (37 TFV/LNG, 10 placebo), assessing the vaginal microbiota by sequencing the V3-V4 regions of 16S rRNA genes prior to IVR insertion and monthly for 3 months. Concentrations of TFV in vaginal fluid (VF), and TFV and TFV-diphosphate (TFV-DP) in vaginal tissue, and modeled PD against HIV-1 in vitro were measured before and after treatment.

Results: There were no clinically significant changes in relative abundance of vaginal bacterial phylotypes from pre-insertion baseline at any month among active and placebo IVR users. There were no significant changes in community state type (CST) with IVR use. Participants with diverse, anaerobic CST IVA/B microbiota had higher in vivo release of TFV from the IVR compared to women with Lactobacillus-dominated (LbD) microbiota, who had expected in vivo TFV release rates. Median VF TFV concentrations were significantly higher among women with CST IVA/B microbiota in months 1 (3,135 ng/mg VF) and 2 (3,800 ng/mg). Women with LbD microbiota had significantly higher median VF TFV concentration (1,423 ng/mg) and median TFV (103 ng/mg) and TFV-DP (5,877 fmol/mg) tissue concentrations versus women with CST IVA/B microbiota at month 3. All women demonstrated a significant increase from pre-insertion baseline of in vitro HIV-1 inhibition by VF (p values <0.05). PD differences in tissue according to CST, however, were not statistically significant.

Conclusion: TFV/LNG IVR use did not change the vaginal microbiota nor increase the incidence of CST IVA/B. Vaginal microbiota, and in particular CST IVA/B, possibly through increased vaginal pH, impacted in vivo TFV release and cervicovaginal (CV) PK, but both PK and PD data suggest CV protection against HIV-1.

Clinical trial registration: ClinicalTrials.gov (#NCT03279120).

Keywords: PrEP; multipurpose prevention technology; pre-exposure (PrEP) prophylaxis; tenofovir diphosphate; vaginal microbiome.

Figure 1

(A–C) Correlation of tenofovir (TFV) vaginal fluid concentration with the relative abundance of Gardnerella vaginalis at months 1, 2, and 3. Green dots represent participants who had Lactobacillus-dominated (LbD) vaginal microbiota at the time of sampling and their individual TFV vaginal fluid concentration and relative abundance of Gardnerella vaginalis. Blue dots represent participants who had diverse anaerobe-dominated (CST IVA/B) vaginal microbiota at the time of sampling and their individual TFV vaginal fluid concentration and relative abundance of Gardnerella vaginalis.

Figure 2

(A–C) Correlation of tenofovir (TFV) vaginal fluid concentration with the relative abundance of Prevotella species at months 1, 2, and 3. Green dots represent participants who had Lactobacillus-dominated (LbD) vaginal microbiota at the time of sampling and their individual TFV vaginal fluid concentration and relative abundance of Prevotella species. Blue dots represent participants who had diverse anaerobe dominated (CST IVA/B) vaginal microbiota at the time of sampling and their individual TFV vaginal fluid concentration and relative abundance of Prevotella species.

Figure 3

(A–D) Correlation of TFV (A) and TFV-DP (C) tissue concentrations with relative abundance of G. vaginalis species and fit of the linear model with CST IV samples at the end of treatment with TFV tissue concentrations (B) and TFV-DP tissue concentrations (D). Green dots represent participants who had Lactobacillus-dominated (LbD) vaginal microbiota at the time of sampling and their individual TFV vaginal fluid concentration and relative abundance of Gardnerella vaginalis. Blue dots represent participants who had diverse anaerobe-dominated (CST IVA/B) vaginal microbiota at the time of sampling and their individual TFV vaginal fluid concentration and relative abundance of Gardnerella vaginalis. Black dots in the linear fit model represent participants who had diverse anaerobe-dominated (CST IVA/B) vaginal microbiota at the time of sampling and their individual TFV vaginal fluid concentration and relative abundance of Gardnerella vaginalis.

Figure 4

Percent inhibition of HIV in vitro by cervicovaginal fluid lavage (CVL) supernatant obtained from study participants using the TFV/LNG IVR at baseline (BL) pre-insertion, month 1 (M1), and end of treatment month 3 (M3). Red dots represent participant data from individuals with diverse anaerobic CST IVA/B microbiota at sampling. Blue dots represent participant data from individuals with Lactobacillus-dominated (LbD) microbiota at sampling. p values < 0.01 paired change from BL in percent HIV inhibition at months 1 and months 3 for both LbD and CST IVA/B groups.

Figure 5

Cumulative (CUMU) p24 antigen production from cervicovaginal (CV) tissue biopsies obtained from all participants at baseline and from participants using the placebo IVR (BL). CUMU p24 antigen production from CV tissue biopsies obtained from participants using the TFV/LNG IVR at end of treatment (EOT). Red dots represent participant data from individuals with diverse anaerobic CST IVA/B microbiota at sampling. Blue dots represent participant data from individuals with Lactobacillus-dominated (LbD) microbiota at sampling. p values > 0.05 for change in p24 antigen production from tissues at EOT compared to BL.