Longitudinal Assessment of the Vaginal Microbiome with Associated Clinical Predictors and HIV-1 Shedding Across the Menstrual Cycle in HIV-Infected Women on Antiretroviral Therapy

Background:

Alterations of female genital tract (FGT) microbiota have been associated with bacterial vaginosis (BV), sexually transmitted infections (STI), hormonal factors, and FGT HIV shedding. However, these associations are not well characterized in HIV-infected women on antiretroviral therapy (ART); thus we characterized the FGT microbiome and its association with clinical factors and HIV shedding in this population.

Methods:

HIV-infected women on ART with regular menstrual cycles, suppressed plasma HIV-1 RNA and without clinical BV/STIs were prospectively enrolled. Cervicovaginal lavage (CVL) and plasma were collected at up to 6 twice-weekly visits during one menstrual cycle for HIV RNA and DNA measurement. Plasma estradiol/progesterone (E2/P4) were quantified and used to characterize menstrual cycle phase. CVL underwent Illumina MiSeq 16S rDNA sequencing. Dirichlet Multinomial Mixtures was used to cluster each participant-visit into a microbiome community type (mCT) based on similar abundances of bacterial taxa. Repeated measures mixed logistic models described the associations between mCTs and clinical factors.

Results:

Among 109 visits by 20 participants with median age 38 (range, 24-48) years, 3 FGT mCTs were identified: low- (mean Shannon Index (SI) 0.14, standard deviation (SD) 0.19), N=40), intermediate (int.)- (SI 1.27, SD 0.49, N=28) and high-diversity mCTs (SI 2.07, SD 0.34, N=41), each with varied distributions of bacterial taxa (Fig. 1). mCT was stable among most women; 9 changes in mCT occurred during follow-up among 6 (30%) participants. Body mass index (BMI) was positively associated with the int.- vs. the high-diversity mCT (odds ratio per unit BMI (kg/m²) 1.46, 95% confidence interval 1.07-2.00, P=0.019). mCT was not associated with age, recent sexual activity, serum E2/P4 levels, or menstrual cycle phase (all p>0.05). CVL HIV RNA and DNA shedding were rare (detected in 18/108 (16.7%) and 36/108 (33.3%) visits among 8 and 14 participants, respectively) and did not significantly vary by mCT (Fig. 2, both p>0.05).

Conclusion:

Virologically suppressed HIV-infected women had varied yet stable mCTs across the menstrual cycle which were associated with BMI but not associated with cervicovaginal HIV RNA or DNA shedding.