Impact of Vaginal Estradiol on the Genitourinary Syndrome of Menopause, Vaginal Microbiome and Mucosal Immune Mediators in Women With HIV

Abstract

Background

Women with human immunodeficiency virus (HIV, WWH) experience early onset menopause and symptoms may impact antiretroviral therapy (ART) adherence. Vaginal estradiol is safe and effective for treatment of the genitourinary syndrome of menopause (GSM) but has not been studied in WWH. The study objective was to test whether vaginal estradiol would improve GSM symptoms in menopausal WWH.

Methods

Menopausal WWH were randomized 1:1 to 12 weeks of open-label vaginal estradiol (n = 25) versus no treatment (n = 26). Participants had at least 1 GSM vulvovaginal symptom in the month prior (dryness, itching, irritation, soreness, or dyspareunia) and clinical atrophy. The primary outcome was difference in mean symptom severity (scale 0–3) defined by the composite vaginal symptom index (VSI) between baseline and week 12. Changes in severity of participants’ most bothersome symptom (MBS), vaginal maturation index (VMI), vaginal microbiome, and mucosal immune mediators were assessed. Results were compared by paired t test or Wilcoxon signed rank tests; a mixed effect model examined differences in VSI improvement between groups.

Results

Mean age was 59 years, 78% self-reported as Black, and all were on ART. Both groups had significant reductions in VSI, however the magnitude was significantly greater in the estradiol group (P < .0001). Estradiol and younger age were associated with VSI improvement. VMI and MBS improved significantly only in the estradiol group. There were no significant changes in the microbiome or mucosal mediators.

Conclusions

The favorable response to estradiol suggests that treatment perhaps initiated earlier may be beneficial for GSM in WWH and should be studied in larger clinical trials.

Graphical Abstract

Graphical Abstract.

https://tidbitapp.io/institutional-portal/clinical-infectious-diseases/tidbits/impact-of-vaginal-estradiol-on-the-genitourinary-syndrome-of-menopause-vaginal-microbiome-and-mucosal-immune-mediators-in-women-living-with-hiv-055d0327-ec0f-4745-9618-705b91948cfa/update

In a pilot clinical study of vaginal estradiol versus no treatment for the genitourinary syndrome of menopause in women with human immunodeficiency virus (HIV-1), estradiol treatment and younger age associated with symptom improvement; microbiome and mucosal mediators did not change.

Postmenopausal urinary and vulvovaginal symptoms of dryness, itching, irritation, soreness, and dyspareunia have been defined as the genitourinary syndrome of menopause (GSM) [1, 2]. GSM is estimated to occur in 50% of women, but 25% or less receive treatment [3, 4]. Studies in women without human immunodeficiency virus (HIV) demonstrated safety and efficacy of topical vaginal estradiol [3, 5–8], and both topical and oral estradiol are associated with beneficial changes in the vaginal microbiome, including increases in protective lactic acid-producing lactobacilli and decreases in diverse anaerobes [9, 10]. It is unclear if the vaginal microbiome contributes to GSM or treatment response [11]. However, interventions that promote a lactobacillus-dominant microbiome may reduce the risk for urinary tract infections, bacterial vaginosis (BV), HIV, and other sexually transmitted infections (STI) [12–14].

Women with HIV (WWH) often enter menopause earlier than women without HIV and report increased severity of symptoms [15–18]. This has been linked to reduced quality of life and decreased adherence to antiretroviral therapy (ART), which may lead to reduced HIV control and increased viral transmission [17, 19–22]. Compared to menopausal women without HIV, WWH have more vaginal dysbiosis and less protective lactobacilli [23], changes linked to vaginal HIV shedding [24, 25]. Therefore, interventions to reduce GSM and improve the microbiome in WWH could prove highly beneficial. The effects of vaginal estradiol in menopausal WWH are under researched. We conducted an open-label pilot clinical study to test the hypothesis that treatment of menopausal WWH with 12 weeks of vaginal estradiol would improve vaginal symptoms and promote favorable changes in the vaginal microbiome.

MATERIALS AND METHODS

Study Design

This study was approved by the Albert Einstein College of Medicine IRB and registered in clinicaltrials.gov (NCT04079218). Participants were recruited from the Bronx and Brooklyn Multicenter AIDS Cohort/Women's Interagency HIV Combined Cohort Study and Montefiore Medical Center (MMC) clinics. Participants provided written informed consent. Menopausal women with HIV-1 (defined as having amenorrhea for at least 12 months) ages 45–70 were eligible if they had at least 1 of 5 vulvovaginal (herein referred to as vaginal) symptoms of moderate-severe intensity at least once per week in each of the 4 weeks prior to screening (dryness, itching, soreness, irritation) or dyspareunia at least once in the prior month (if sexually active). All participants had atrophy on exam. Exclusion criteria included contraindications to estradiol (eg breast cancer, estrogen dependent neoplasia, myocardial infarction or stroke, thromboembolic disease) and vaginal infections. See Supplementary Table 1 for all exclusions. Participants agreed not to use other vaginal products including lubricants during the study. Enrollment occurred one month after screening. Participants were randomized 1:1 to vaginal estradiol or no treatment. Vaginal estradiol tablets (Vagifem 10 μg) with applicators were provided by Novo Nordisk (Plainsboro, New Jersey, USA). Estradiol was applied daily for 2 weeks and then twice weekly for an additional 10 weeks. Participants returned 6 and 12 weeks later for assessment of safety, vaginal symptoms, adherence (self-report and used applicator collection), pelvic exams, vaginal swab and cervicovaginal lavage (CVL) collection. During the informed consent process, participants were educated on the estrogen decline during menopause, its association with GSM, available treatment options, and given a letter with treatment information to share with their provider after study completion.

Vaginal Symptoms

The severity of each of five vaginal symptoms was assessed at baseline and week 12 using a scale of 0–3 corresponding to none, mild, moderate, or severe. A composite vaginal symptom index (VSI) was calculated as the mean severity of the five symptoms at each visit [7, 10, 26]. Participants also identified a most bothersome vaginal symptom (MBS) at baseline and similarly rated the severity of that symptom [7]. Change in severity of VSI and MBS were compared between baseline and week 12. Twenty-seven participants were sexually inactive at enrollment. Thus, we performed sensitivity analyses with 2 modified VSI scores; the first omitted data for dyspareunia only for those sexually inactive participants and the second omitted dyspareunia for all participants. We defined a change in VSI or MBS of 1 or greater as clinically meaningful based on prior studies in women without HIV [7, 10, 26]. Adverse events (AE) were graded per the Division of AIDS Female Genital Grading Table for Use in Microbicide Studies and reviewed by a Protocol Safety Review Team composed of 2 co-investigators and an external gynecologist.

Clinical Assessment and Sampling

Clinical signs of GSM including pallor, epithelial integrity, epithelial thickness, and presence of secretions were assessed by visual inspection and the severity of each scored on a scale of 0–3 [27]. A compositive vaginal atrophy score was calculated as the mean severity of the four measures. Samples were collected in the following order: (i) vaginal swab for pH, Nugent score, wet prep; (ii) swabs for metagenomic sequencing; (iii) cell sampling of the upper third of the vaginal wall for vaginal maturation index (VMI) [28]; (iv) CVL with 10 mL of normal saline for quantification of immune mediators [23]. Blood and vaginal biopsies were obtained at enrollment and week 12 for measurement of CD4 count, human immunodeficiency virus type 1 (HIV-1) plasma viral load and future studies respectively. Screening for BV, trichomonas and candidiasis was conducted at all visits by assessment of Amsel criteria and microscopic examination of swabs (Supplementary Table 1). Testing for gonorrhea and chlamydia was not performed as rates are relatively low in women aging with HIV [29, 30].

The VMI was determined from Papanicolaou stained slides of vaginal cells at baseline and week 12 [28]. Slides were read by MMC cytopathologists blinded to randomization status. Slides for 5 samples (2 no treatment, 3 estradiol) were poorly preserved or had scant cellularity and were excluded from analyses.

Metagenomic Sequencing and Analysis

The vaginal microbiome was characterized by metagenomic sequencing and taxonomy assigned using MetaPhlAn 4. Details are provided in Supplemental Methods. Alpha diversity was quantified with Shannon diversity index (SDI). Bray–Curtis dissimilarity was calculated using diversity.R and principal coordinate analysis (PCoA) was performed with skbio.stats.ordination.pcoa [31]. A hierarchically clustered heatmap was created with hclust2 1.0.0, using parameters ftop 50, f_dist_f correlation, and fperc 99. Vaginal community state types (CST) were defined based on ≥50% relative abundance of L. crispatus (CST I), L. gasseri (CST II), L. iners (CST III), diverse anaerobes (CST IV), and L. jensenii (CST V). If the proportion of one or more of CST I, II, and V was ≥ 50%, the sample was categorized as protective lactobacillus (LB) group [32]. The relative abundance of individual species was compared between baseline and week 12 in both groups using Wilcoxon signed rank tests with and without a centered log ratio transformation.

Measurement of Immune Mediators in CVL

The concentrations of mucosal immune mediators were measured by Luminex and ELISA and chosen for their association with inflammation, HIV, the vaginal microbiome, and reproductive aging[14, 23] (Details in Supplemental Methods).

Statistical Analysis

Demographic and clinical variables were analyzed between groups using χ² or Fisher exact tests (categorical variables) or Mann–Whitney U and unpaired t tests (continuous variables). Differences in VSI (primary outcome), MBS, VMI, clinical atrophy scores, pH, Nugent score, and immune mediator concentrations were analyzed between baseline and week 12 using paired t tests or Wilcoxon signed rank tests (GraphPad Prism Software, version 10, La Jolla, California, USA). A mixed effects model including age (years), time in menopause (years), randomization status (no treatment vs estradiol), visit (baseline vs week 12) and the interaction between randomization to estradiol and visit was used to assess improvement in VSI (scale 0–3) between groups. Age (range 21 years) and time in menopause (range 44 years) were included in the model as these may impact response to estradiol treatment. We also hypothesized that changes in the microbiome and epithelium might contribute to the VSI and performed an additional model including Nugent score, alpha diversity, and VMI.

RESULTS

In total, 79 participants were screened; 14 were ineligible due to vaginal infection (n = 6) or medical conditions (n = 8). Sixty participants returned for enrollment and randomization to vaginal estradiol (n = 33) or no estradiol treatment (control) (n = 27). Twenty-five participants randomized to estradiol, and 26 as controls completed the study (Figure 1). There were no significant differences between groups in demographic and clinical characteristics including menopausal duration (Table 1). The mean age was 59 years, 78% identified as Black; the majority with well-controlled HIV on ART. Forty-seven percent were sexually active (24/51), 92% had male partners, and all reported some form of vaginal penetration.

Figure 1.

Recruitment, enrollment, randomization, and follow-up. Diagram showing participants who were recruited, enrolled, and randomized to vaginal estradiol or no treatment. * Immunosuppressant, prior endometrial cancer, antibiotic use, inconsistent reporting of vaginal symptoms, hospitalized for depression, vaginal lesion, and vaginal condition needing gynecological follow-up, HIV care needed prioritization. Abbreviations: HIV, human immunodeficiency virus; WDC, withdrew consent. Created in BioRender. Mahant, A. (2025) https://BioRender.com/1mxdfp4

Table 1.

Demographic and Clinical Characteristics of the Study Population

Characteristics	No Treatment (n = 26)	Estradiol	P Value
		(n = 25)
Age, y	60 (4.9)	59 (4.7)	.87
Race
Black	20 (77%)	20 (80%)	.94
White	2 (8%)	2 (8%)
Other	4 (15%)	3 (12%)
Hispanic ethnicity	9 (35%)	5 (20%)	.35
Education
Completed high	18 (69%)	18 (72%)	.93
school or greater
Smoking
Never	10 (38%)	5 (20%)	.32
Past	10 (38%)	11 (44%)
Current	6 (23%)	9 (36%)
Alcohol use
Never	9 (35%)	8 (32%)	.92
Past	7 (27%)	8 (32%)
Current	10 (38%)	9 (36%)
Drug use
Never	4 (15%)	7 (28%)	.3
Past	18 (69%)	12 (48%)
Current	4 (15%)	6 (24%)
Years in menopause	8.5 (5, 20)	10 (7, 18)	.6
Sexually active in prior 3 m	11 (42%)	13 (52%)	.58
Sexual partners
Men	8 (73%)	11 (85%)	.72
Women	1 (9%)	1 (8%)
Men and women	2 (18%)	1 (8%)
History of hysterectomya
	1 (4%)	0 (0%)	>.99
History of hysterectomy with BSO
	1 (4%)	3 (12%)	.28
History of unilateral SOb
	0 (0%)	1 (4%)	.49
BMI	30 (27, 35)	28 (25, 32)	.46
Self-reported estradiol adherence		23 (92%)
Used estradiol applicators returned		21 (85%)
ART adherence	24 (92%)	21 (84%)	.42
≥95%
Suppressed HIV-1 plasma viral load	21 (81%)	20 (80%)	>.99
(≤ 40 copies/mL, %)
CD4 count (cells/mm3)	891	731	.3
	(548, 1098)	(492, 958)

Abbreviations: ART, antiretroviral therapy; BMI, body mass index; BSO, bilateral salpingoophorectomy; HIV-1, human immunodeficiency virus type 1; SO, salpingoophorectomy.

Categorical variables reported as number, n (%) and continuous variables reported as mean ± standard deviation (SD) or median with interquartile range (IQR). Pairwise comparisons were assessed with P values calculated with χ² or Fisher exact tests for categorical values or unpaired t tests or Mann–Whitney tests for continuous variables.

^aParticipant aged 65, ^bParticipant aged 57.

Greater Improvement in Vaginal Symptoms in Those Treated With Vaginal Estradiol

The median VSI was similar at enrollment and decreased significantly over 12 weeks in both the estradiol (1.0 to 0, P < .0001) and no treatment groups (1.1 to 0.4, P = .004), although the magnitude of change was greater and median VSI was significantly lower at week 12 in the estradiol group compared to no treatment, P < .0001 (Figure 2A, Supplementary Table 2). Because sexually inactive participants could not report dyspareunia, we performed sensitivity analyses either including dyspareunia (sexually active participants only) or omitting dyspareunia (all participants); the findings were not altered (Supplementary Table 2). At baseline, dryness was the predominant MBS reported (n = 25/51) followed by itching (n = 18/51), and dyspareunia (n = 6/24 sexually active) (Table 2). Participants treated with estradiol who reported MBS of dryness or itching had a significant reduction in median reported severity from 2 to 0 between baseline and 12 weeks (P = .002 and P = .001 respectively), whereas participants randomized to no treatment reported no significant changes in severity of any MBS. Although the median severity of dyspareunia decreased in both groups, we were not able to perform statistical analysis because of limited occurrences. One participant reported dyspareunia as their MBS at baseline but also reported no sex partners in the prior 3 months. Excluding this participant did not change the results. The VSI median severity was lower than the individual MBS scores, which is expected as VSI is a composite score, but the severity of both VSI and MBS decreased significantly in the estradiol group as did the severity scores for individual vaginal symptoms (Supplementary Table 3).

Figure 2.

Vaginal symptoms and clinical measures of atrophy and mucosal estrogenization improve in participants treated with vaginal estradiol. Changes in the vaginal symptom index (VSI) (A), clinical atrophy score (B), and vaginal maturation index (VMI) (C) between baseline (BL) and the 12-wk endpoint in participants randomized to no treatment versus vaginal estradiol. Scales of 0–3 for VSI and clinical atrophy correspond to none, mild, moderate, or severe symptoms or clinical atrophy findings. Pairwise comparisons were performed with P values calculated using paired t tests or Wilcoxon signed rank test. *P < .05, **P < .01, ***P < .001.

Table 2.

Change in Most Bothersome Vulvovaginal Symptom Severity Over 12 Weeks

	No Treatment BL (n = 26)	No Treatment Week 12 (n = 26)	P Value		Estradiol BL (n = 25)	Estradiol Wk 12 (n = 25)	P Value	Difference at 12 Weeks No Treatment Versus Estradiol
Vaginal dryness (n = 14/26)	2 (2, 3)	1 (0.75, 3)	.08	Vaginal dryness (n = 11/25)	2 (1, 2)	0 (0, 1)	.002	0.003
Vaginal itching (n = 7/26)	2 (1, 2)	1 (1, 2)	.38	Vaginal itching (n = 11/25)	2 (2, 3)	0 (0, 0)	.001	.01
Dyspareunia (n = 4/11)a	2 (1.3, 2.8)	0 (0, 2.3)		Dyspareunia (n = 2/13)	2.5 (2, 3)	0 (0, 0)	…	…
Vaginal soreness	NA	…		Vaginal soreness (n = 1/25)	2 (2, 2)	0 (0, 0)	…	…
Vaginal irritation (n = 1/26)	1 (1, 1)	1 (1, 1)		Vaginal irritation	NA	…	…	…

Median (IQR) severity of participants’ most bothersome symptom identified at baseline and at 12 weeks. Severity scores reported on scale from 0 (none) to 3 (severe) symptoms. P values calculated with Wilcoxon test for difference in median severity of symptoms between baseline and 12 weeks within groups and with Mann Whitney and unpaired t-tests between the no treatment and estradiol treatment groups at the 12-week end point.

^aOne participant reported dyspareunia as their MBS at baseline but had no partners in the prior 3 months. Removal of this participant's data did not change the results.

There were similar numbers of genitourinary AEs reported between the groups; mostly vaginal discharge and one episode of vaginal candidiasis, BV and trichomonas in each group. There were 4 serious AEs in the estradiol arm; none were drug related (Supplementary Table 4).

VMI Improved in the Estradiol Treatment Group

There was a nonsignificant trend toward improvement in the clinical vaginal atrophy score in the estradiol group (P = .13) but no change in the no treatment group between baseline and 12-week (Figure 2B). The VMI, which was similar at enrollment in both groups, increased only in the estradiol group (P = .048) (Figure 2C).

Vaginal Estradiol Treatment Was not Associated With Significant Changes in Microbial Diversity or Mucosal Immune Mediators

To explore molecular changes that might be associated with clinical improvement, we compared changes in the microbiome and mucosal immune mediators (Table 3). There were no significant differences at baseline or over time in any of the CVL mediators measured including interleukin (IL)-1α, IL-8, IP-10, MCP-1, SLPI, immunoglobulin A (IgA) or immunoglobulin G (IgG). There were also no significant differences in baseline vaginal pH, Nugent score, or SDI and over time just a trend toward a reduction in SDI in the estradiol arm at week 12 (P = .053) (Table 3 and Supplementary Figure 1). There were no differences in baseline CST across groups. The majority of participants were CST IV (57%) with few transitions during the study. Five shifted after estradiol treatment, but only 3 were beneficial changes from CST IV to protective lactobacillus group (n = 2) or CST III (n = 1). Seven participants in the no treatment group shifted CSTs; 3 of these were suboptimal changes from protective lactobacillus to CST IV (n = 2) or CST III (n = 1) (Figure 3A, Supplemental Figures 2 and 3). Beta diversity did not differ by randomization status or visit (Supplementary Figure 4) and a heat map generated by hierarchical clustering according to Bray-Curtis dissimilarity also failed to identify differences in vaginal microbiome by randomization or visit (Figure 3B). There were no differences in relative abundance of individual taxa comparing baseline to week 12 after correction using raw values or after centered log ratio transformation (Supplementary Tables 5 and 6).

Table 3.

Longitudinal Levels of Cervicovaginal Mucosal Immune (log10) Mediators and Measures of the Vaginal Microbiome

Level	No Treatment Baseline	No Treatment Week 12	P Value	Estradiol Baseline	Estradiol Week 12
Soluble mucosal mediators	n = 26	n = 26		n = 25	n = 25	P Value
Log10 IL-1α (pg/mL)	1.9 (0.67)	1.9 (0.53)	.59	1.9 (0.7)	2.0 (0.66)	.18
Log10 IL-8 (pg/mL)	2.3 (1.1)	2.4 (0.98)	.27	2.6 (0.68)	2.5 (0.64)	.64
Log10 IP-10 (pg/mL)	2.0 (0.85)	2.0 (0.91)	.92	2.2 (0.74)	2.2 (0.55)	.75
Log10 MCP-1 (pg/mL)	1.1 (0.97)	1.2 (0.77)	.75	1.3 (0.75)	1.2 (0.80)	.45
Log10 SLPI (pg/mL)	4.5 (0.72)	4.4 (0.73)	.37	4.4 (0.58)	4.6 (0.51)	.12
Log10 IgA (μg/mL)	−0.03 (1.0)	0.09 (0.97)	.42	0.57 (0.94)	0.31 (1.1)	.09
Log10 IgG (μg/mL)	0.85 (0.55)	0.94 (0.59)	.32	1.1 (0.72)	0.98 (0.66)	.44
CST microbiome	n = 24	n = 24		n = 20	n = 20	…
Vaginal pH	4.6 (4.5, 5)	4.5 (4.3, 5)	.91	4.7 (4.4, 5.2)	4.7 (4.4, 5)	.27
Nugent score	4.8 (3.3, 7.5)	4 (2.6, 7.5)	.61	4.5 (3.3, 6)	4.5 (2.8, 6)	.35
Shannon diversity index	0.80 (0.66)	0.79 (0.62)	.96	1.00 (0.80)	0.75 (0.59)	.053
CST IV	13 (54%)	14 (58%)	>.99	12 (60%)	11 (55%)	>.99
CST III	5 (21%)	6 (25%)	>.99	4 (20%)	4 (20%)	>.99
Protective lactobacilli (LB) (CST I, II, V)	6 (25%)	4 (17%)	.72	4 (20%)	5 (25%)	>.99
Change in CST between baseline and wk 12	…	7 (29%)		…	5 (25%)	>.99

Abbreviations: CST, community state type; Ig, immunoglobulin; IL, interleukin; IP, interferon gamma-induced protein; LB, lactobacilli; MCP, macrophage chemotactic protein; SLPI, secretory leukocyte protease inhibitor.

Categorical variables reported as number, n (%) and continuous variables reported as mean ± standard deviation (SD) or median with interquartile range (IQR). Pairwise comparisons were assessed with P values calculated with χ² or Fisher exact tests for categorical values or paired t tests or Wilcoxon signed rank for continuous variables.

Figure 3.

The vaginal microbiome does not differ by visit or treatment status. A, Relative abundance of the vaginal microbiome (taxa >2% total included) in participants randomized to no treatment (left) versus estradiol (right). X axis lists participant samples longitudinally with baseline (BL) followed by wk 12. B, Hierarchical clustering by Bray–Curtis dissimilarity by treatment status (randomization to no treatment vs estradiol) and by visit (baseline vs wk 12).

Vaginal Estradiol Treatment Mediates Improvement in VSI

A mixed effects model demonstrated that randomization to estradiol was the most significant variable associated with improvement in VSI (coefficient −0.62, P = .001). Visit (week 12 vs baseline) and younger age were also associated with reduction in VSI (Table 4). The results did not change when VMI, Nugent score, or SDI were added to the model.

Table 4.

Mixed Effects Model for Predictors of Improvement in VSI Over 12 Weeks (n = 51)

Variables	Model
	Coefficient	P Value
Age, years	−0.03	.02
Time in menopause, y	0.01	.12
Baseline randomization	0.072	.63
Visit	−0.39	.003
Randomization estradiol:visit	−0.62	.001

DISCUSSION

To our knowledge, this is the first study of GSM treatment in WWH, and results demonstrate that vaginal estradiol led to a significantly greater improvement in the composite VSI and MBS compared to the no treatment group. Although both groups showed some improvement in VSI over the 12-week study period, the mixed effect model confirmed that estradiol treatment was the most significant factor associated with improvement. We also found a modest association with younger age and improvement in VSI, which has not been reported in women without HIV, but if confirmed by larger studies could inform clinical practice. As estrogen levels begin to decline during perimenopause, GSM symptoms can occur even before menses ceases. It is further known that GSM worsens with time emphasizing the importance of timely treatment [1, 4]. Future larger clinical trials including both younger and older menopausal participants in various stages of the menopausal transition may help confirm our findings and provide additional data on ideal timing for treating GSM in WWH.

Severity of the predominant MBS of dryness and itching (also common in women without HIV) decreased significantly only in the estradiol treatment group. The changes (Δ) over time in median VSI (Δ1) and median MBS severity (Δ2) in the estradiol treatment arm are similar to what has been reported in women without HIV [5, 7, 8, 26, 33]. Fewer participants reported dyspareunia as their MBS compared to studies in women without HIV, but over half of our study participants were sexually inactive. Among those that were sexually active, 25% (6/24) reported an MBS of dyspareunia.

In addition to HIV status, there were several other characteristics of our study population that may have contributed to differences in the study outcomes. Our study included 78% Black women, whereas the vast majority of published studies included mostly White women. The mean VMI was higher at enrollment (54 ± 16) in our cohort compared to published data (46.6 ± 0.8), but Black race and obesity, characteristics of our population, are associated with a higher VMI [28]. The vaginal microbiome of our study population was characterized by a predominance of anaerobes and low abundance of lactobacilli, consistent with an aging phenotype [9, 23, 34, 35, 36]. However, unlike some studies in menopausal women without HIV, estradiol treatment did not significantly change the microbiome, despite a trend toward a decrease in SDI [9, 10, 37]. We speculate that this may reflect HIV status and the predominance of Black women in our cohort as both are associated with dysbiosis independent of menopause [23, 32, 38, 39, 40]. The 12-week endpoint was based on prior studies in women without HIV, but we hypothesize that earlier and/or longer treatment may be needed to improve the microbiome and overcome the additional effects of HIV and long duration of menopause (8–10 year) in our population. Similar to GSM studies in women without HIV, we did not detect significant changes in the mucosal immune mediators evaluated [41–43], although it is possible that a more comprehensive proteomic study may identify molecules in CVL associated with clinical improvement or failure to respond.

Mechanisms underlying the GSM are not fully defined and are likely complex, including the impact of decreased estrogen on the vaginal microbiome, epithelial barrier function and inflammation. Several studies in women without HIV demonstrated an increased odds of having a CST IV (anaerobic dominant) microbiome and decreased odds of having protective lactobacillus-dominant microbiomes with vaginal atrophy and dryness [11, 34–36]. In a study comparing vaginal estradiol with moisturizer in women without HIV, similar improvements in GSM symptoms were observed in both groups but beneficial changes in the microbiome occurred in only estradiol treated participants, suggesting a role for additional factors[7, 9]. This notion is supported by our study where we found significant subjective improvement (VSI and MBS) but no significant change in the vaginal microbiome. The increase in VMI observed with estradiol treatment suggests that restoration of epithelial integrity may contribute to clinical improvement, consistent with studies showing that vaginal atrophy is associated with decreased expression of genes associated with epithelial integrity [34].

The reasons for improvement in women in the no treatment arm are unclear. The only “intervention” these women received was vaginal sampling and CVL. Though participants were asked to refrain from using vaginal lubricants and moisturizers, shown to improve symptoms in women without HIV[2, 7], surreptitious use cannot be excluded. Education about GSM received during informed consent may have impacted participants’ assessment of vaginal symptoms.

The lack of a placebo limits the strength of our findings as does the reliance on self-report and return of used applicators as adherence measures, rather than tissue estradiol levels. We did not formally assess acceptability or tolerability of estradiol treatment but did monitor AEs. Vaginal discharge was the most common genitourinary AE (3%–3.7%) and no participants randomized to estradiol stopped treatment early due to AE. Future behavioral and qualitative studies investigating acceptability and tolerability of vaginal estradiol in WWH should be conducted.

US guidelines do not explicitly recommend vaginal estradiol for WWH but state that recommendations for menopausal care for WWH do not differ from women without HIV [44]. The European AIDS Clinical Society and British Menopause Society recommend screening for menopausal symptoms in WWH at age 40 and 45, respectively, and both recommend vaginal estradiol be offered for GSM if no contraindications [21, 45, 46 ]. Despite this, providers may not be confident treating menopausal symptoms without more data, clear guidance and training [20, 21, 47–49]. The recent decision by the Food and Drug Administration (FDA) to remove the black box warning on hormone replacement products will reduce a major barrier to GSM treatment [50, 51]. In conclusion, we found significant improvements in participants’ subjective measures of composite VSI and MBS of dryness and itching and few AEs. This study, combined with experience in women without HIV, supports vaginal estradiol treatment for WWH and GSM.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions . K. M., M. J. K., and B. C. H. conceptualized the study. K. M., M. J. K., B. C. H., M. G., A. M. M., J. C., L. K., J. M., J. A., H. E., S. G., A. C., and A. C. contributed to the methodology. K. M., M. G., J. C., T. W., and J. M. contributed to the analysis. K. M. wrote the main manuscript text. K. M., J. C., A. M. M. and T. W. prepared the figures. All authors reviewed the manuscript.

Acknowledgments . The authors gratefully acknowledge the contributions of the study participants. The authors also thank Dr Susan Cu-Uvin and Dr Nir Barzilai, the staff of the Bronx and Brooklyn MWCCS sites and the staff of the Einstein Clinical Research Center. They also thank Novo Nordisk (Plainsboro, NJ) for providing vaginal estradiol (Vagifem) for the study.

Financial support. This study was supported by grants from the National Institute on Aging (NIA), Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), the National Institute of Allergy and Infectious Diseases (NIAID), and the National Center for Advancing Translational Sciences at the National Institutes of Health (NIH) (grant numbers K23 AG062400, R01 HD098977, R01 AI134367, and UM1 TR004400). This work was also supported by the Einstein-Rockefeller-CUNY Center for AIDS Research, Albert Einstein College of Medicine (P30 AI124414) which is supported by NIH co-funding and participating Institutes and Centers: NIAID, National Cancer Institute (NCI), NICHD, National Heart, Lung, and Blood Institute (NHLBI), National Institute on Drug Abuse (NIDA), National Institute of Mental Health (NIMH), National Institute on Aging (NIA), FIC, and Office of AIDS Research (OAR). Additional funding was provided by an Emerging Scholars in Aging Award from the Albert Einstein Jack and Pearl Resnick Gerontology Center, Albert Einstein College of Medicine and an Irma L. and Abram S. Croll Charitable Trust Research Grant to K. M.

Some of the data in this manuscript were collected by the MACS/WIHS Combined Cohort Study (MWCCS) Bronx Clinical Research Site (CRS) (Kathryn Anastos, David Hanna, and Anjali Sharma), U01-HL146204 and Brooklyn CRS (Deborah Gustafson and Tracey Wilson), U01-HL146202. The contents of this publication are solely the responsibility of the authors and do not represent the official views of the NIH. The MWCCS is funded primarily by the NHLBI, with additional co-funding from the NICHD, NIA, National Institute of Dental and Craniofacial Research (NIDCR), NIAID, National Institute of Neurological Disorders and Stroke (NINDS),NIMH, NIDA, National Institute of Nursing Research (NINR), NCI, National Institute on Alcohol Abuse and Alcoholism (NIAAA), National Institute on Deafness and Other Communication Disorders (NIDCD), National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institute on Minority Health and Health Disparities (NIMHD), and in coordination and alignment with the research priorities of the NIHOAR. MWCCS data collection is also supported by UL1-TR000004 (UCSF CTSA), UL1-TR003098 (JHU ICTR), UL1-TR001881 (UCLA CTSI), P30-AI-050409 (Atlanta CFAR), P30-AI-073961 (Miami CFAR), P30-AI-050410 (UNC CFAR), P30-AI-027767 (UAB CFAR), P30-AI-124414 (ERC-CFAR), P30-MH-116867 (Miami CHARM), UL1-TR001409 (DC CTSA), KL2-TR001432 (DC CTSA), and TL1-TR001431 (DC CTSA).