Microbiota, Gender-Affirming Hormone Therapy, and Inflammatory Biomarkers in Transgender Women with HIV: Potential Implications for Cardiovascular Disease

Abstract

Purpose:

The intersecting disparities of human immunodeficiency virus (HIV) and cardiovascular disease (CVD) among transgender women have raised questions about the role of the gut microbiota and gender-affirming hormone therapy (GAHT) in the pathogenesis of CVD in the context of HIV. The purpose of this study was to provide an early exploration of the associations between these possible mechanisms driving inflammatory CVD risk markers among transgender women with HIV.

Methods:

We conducted a preliminary study with 21 transgender women with HIV exploring the relationship between GAHT use (self-report), gut/rectal microbiota composition (rectal swabs), and inflammatory markers linked to CVD (plasma). Microbiota measures included alpha (richness, evenness, and Shannon diversity) and beta (Bray–Curtis, un/weighted UniFrac) diversity metrics. Inflammatory biomarkers included intestinal fatty-acid binding protein, monocyte chemoattractant protein-1, soluble CD163, intercellular adhesion molecule 1, tumor necrosis factor alpha (TNFa), soluble TNF receptor I (sTNF-I), sTNF-II, interleukin (IL)-6, IL-8, IL-1b, IL-1a, soluble CD14, d-dimer (domain dimer), vascular cell adhesion molecule 1, and high-sensitivity C-reactive protein. Wilcoxon rank sum test, log-level regression, Spearman’s rho, permutational multivariate analysis of variance, and differential abundance testing assessed relationships between constructs.

Results:

Key inflammatory markers linked to CVD were associated with GAHT use—an increased sTNF-I and sTNF-II levels and decreased IL-1a levels. Microbiota composition was not related to GAHT use but was variably associated with inflammatory biomarkers related to CVD risk.

Conclusions:

Although preliminary, these findings suggest a potential association between inflammation linked to CVD risk and microbiota composition and GAHT. The results contribute to the characterization of interconnecting factors that may inform understanding and interventions to enhance overall health and well-being in transgender women with HIV. Further research is essential to elucidate the mechanisms underlying these associations, ultimately striving for health equity.

Introduction

Transgender women are at elevated risk for acquiring human immunodeficiency virus (HIV), with 49 times higher odds of HIV-1 acquisition compared with cisgender people. ¹ Transgender women with HIV are less likely to achieve viral suppression and are at an elevated risk of cardiovascular disease (CVD) compared with cisgender women.^2–5 A study with transgender women reported increased risk of myocardial infarction and stroke compared to cisgender women and increased risk of venous thromboembolism compared to both cisgender men and cisgender women. ⁶ Although HIV is an established driver of increased CVD risk, there is evidence to suggest that transgender women with HIV may have even greater CVD risk compared with cisgender people. ⁷

Despite the dual and interacting disparities in HIV and CVD risk, there is limited knowledge of the potential mechanisms of CVD pathogenesis in transgender women with HIV. Although there are established behavioral CVD risk factors that transgender women with HIV face due to marginalization (tobacco use, substance use, depression, lower rates of physical activity), the biological pathways of CVD within the context of HIV unique to transgender women are understudied.^8–10 HIV infection has been associated with alterations in the intestinal microbiota, including structural impairment, disruption of intestinal immune homeostasis, increased inflammation, and disturbances in the diversity and species of gut microorganisms.^11–15 Indeed, multiple HIV-related clinical, virologic, and immunological markers of disease progression and persistence have been linked to gut microbial dysbiosis. ¹⁶

In addition, disruptions in the microbial population that are caused by HIV infection have been linked to systemic inflammation and cardiometabolic complications such as hypertension, dyslipidemia, chronic kidney disease, obesity, and type 2 diabetes, all of which can lead to CVD morbidity and mortality.^17–24 Numerous studies in humans and animals show increased inflammatory markers and oxidative stress lead to the development of CVD.^18,25 Although the sources of inflammation are multifactorial, recent data indicate that microbial dysbiosis may represent one of the primary sources of inflammation and oxidative stress that are precursors of CVD.^18,19,26

Gender-affirming hormone therapy (GAHT) may also influence disruptions in the gut microbiota. More than 70% of transgender women use GAHT, including estrogen, progestin, and/or anti-androgens (i.e., testosterone antagonist).^27–29 Although there is a paucity of studies on GAHT and the gut microbiota, levels of endogenous sex steroid hormone levels (testosterone and estradiol) have been shown to be correlated with diversity and gut microbial composition.^30,31 Exogenous hormones may also influence and disrupt the microbiota; animal models have shown that undergoing a gonadectomy (removal of testes or ovaries) and being treated with hormone replacement therapy had an impact on the composition of gut microbiota in mice. ³² Thus, it is plausible that GAHT influences the microbiota of transgender women with HIV.

GAHT may also contribute to varying CVD risk among transgender women. The use of oral estrogen has been linked to threefold higher rates of cardiovascular mortality ²⁹ and 20-fold higher rates of venous thrombosis and/or pulmonary embolism compared with the general population. ²⁸ A study also showed a higher incidence of myocardial infarction and transient ischemic attack among transgender women using ethinyl estradiol. ³³ However, transdermal administration of estrogen has been shown to have little to no impact on CVD risk factors and outcomes in these studies.^28,29,33 Overall, among transgender women with HIV, there is a lack of conclusive data on direct effects of GAHT on CVD risk.^28,34

Taken together, evidence indicates that HIV is associated with gut microbial dysbiosis with subsequent HIV pathogenesis and potential CVD risk. Furthermore, there is potential for GAHT to influence both disruptions in the gut microbiota and CVD risk, especially among persons with HIV (Fig. 1). Despite the high prevalence of HIV, CVD morbidity and mortality, and utilization of GAHT among transgender women, there is a paucity of explorations into the connections between them. Therefore, there is a need to better delineate the associations between the microbiota, CVD risk and its associated mechanisms (e.g., inflammatory processes), and GAHT among transgender women with HIV. Thus, the aims of this preliminary study were to investigate, among transgender women with HIV, (1) gut/rectal microbiota composition, (2) the association between microbiota composition and inflammatory markers linked to CVD risk, (3) differences in microbiota composition by GAHT use, and (4) differences in inflammatory markers by GAHT use.

FIG. 1.

Conceptual model. Dashed lines indicate associations preliminarily supported by current findings. CVD, cardiovascular disease; GAHT, gender-affirming hormone therapy.

Materials and Methods

Participants and procedures

This study is an exploratory secondary analysis of existing datasets. The study leveraged previously collected data from the University of Miami Center for AIDS Research’s (CFAR) transgender health registry and from a University of Miami CFAR pilot study evaluating subclinical CVD in transgender women with HIV. ³⁵ This transgender health registry established a repository of biomarkers and psychosocial survey data collected from transgender women for future analyses. All participants provided informed consent. Biological samples were collected from N = 21 transgender women with HIV on antiretroviral therapy (ART), receiving GAHT with estrogen with or without anti-androgen therapy (n = 15 with GAHT use and n = 6 without GAHT use), and with no previous diagnosis of CVD. History of orchiectomy was an exclusion criterion in an attempt to have a more homogeneous group in view of the small sample size. Specifically, stored samples from plasma and rectal swabs from 2018 to 2020 were analyzed along with cross-sectional psychosocial survey data collected as part of the registry. On the same day, a blood draw was completed by staff in EDTA tubes, and rectal swabs were self-collected via FLOQSwabs 519C without reagents and immediately stored at −80°F. Participants were recruited from the University of Miami infectious disease clinic. Procedures were approved by the University of Miami Institutional Review Board.

Microbiota 16S rRNA gene sequencing

To assess the gut microbiota, 16S ribosomal ribonucleic acid (rRNA) gene total DNA was extracted and purified from rectal swabs using the PowerSoil Pro kit (Qiagen; Hilden, Germany). Extracted DNA was amplified using a dual-indexing quantitative polymerase chain reaction method for 16S rRNA gene amplicon sequencing utilizing the 341F-806R primers to target the V3–V4 region of the 16S small subunit rRNA. Amplified samples were then normalized, pooled, and sequenced on an Illumina MiSeq platform with 15% phiX. ³⁶ A water control was extracted simultaneously with study specimens to monitor for contamination during the sequencing process. To remove the adaptor and primers from the resulting sequencing files, the Cutadapt tool was utilized. The DADA2 R package pipeline was then used to filter, trim, check the quality of sequences, remove chimeras, and produce an amplicon sequence variants table. ³⁷ The GreenGenes database was used for taxonomic identification. A 5% prevalence threshold was set to remove rare/low-abundance bacteria across all samples.

Inflammatory marker assays

From stored plasma samples, a series of inflammatory biomarker assays was completed. Assays were performed by the University of Miami CFAR Laboratory Sciences Core and tested once. Plasma levels of cytokines were measured using customized magnetic bead panels (R&D Systems, Minneapolis, MN) following the manufacturer’s instructions. Interleukin-6 (IL-6) was measured using a high-sensitivity kit, and the remaining analytes were measured using discovery kits. Acquisition was via a FlexMap 3D instrument (Luminex Corporation, Austin, TX). The median fluorescent intensity data were analyzed with MILLIPLEX^™ Analyst Software V.3.5 (EMD Millipore; Burlington, MA). Cytokine concentrations were determined based on standard curves and expressed in pg/mL.

Measures

Demographics and HIV characteristics

Participants reported on age, race, number of years with HIV, if they had a detectable HIV viral load (yes/no), and CD4 count (above 200; between 200 and 500; above 500).

GAHT use

Participants reported if they were currently using estrogen-based GAHT (yes/no) and, if yes, the type of GAHT use (Premarin, estrogen, estrogen patch, spironolactone, finasteride, Aldactone, or other).

Microbiota composition

Prevalence and overall relative abundance of bacteria were measured per standard practice. ³⁸ Bacterial community composition was assessed through alpha and beta diversity metrics. Alpha diversity is the microbial diversity of bacteria within a sample. Richness refers to the total number of observed bacteria within a sample, and evenness refers to the abundance of bacteria in a sample. Evenness was examined via the normalized Pielou’s evenness metric. The Shannon Diversity Index was also used, which weighs both the richness and evenness of the sample. Beta diversity is the change in diversity of bacterial species from one environment to another. The Bray–Curtis index measures the compositional similarity between samples, taking into account the relative abundances of the different bacteria. Unweighted UniFrac calculates distances based on bacterial abundance, and weighted UniFrac does the same but weights the branch length with abundance difference.

Inflammatory markers associated with CVD risk

Assays included intestinal fatty-acid binding protein (I-FABP), monocyte chemoattractant protein-1 (MCP-1), soluble CD163 (sCD163), intercellular adhesion molecule 1 (ICAM-1), tumor necrosis factor alpha (TNF-a), soluble tumor necrosis factor receptor I (sTNF-I), soluble tumor necrosis factor receptor II (sTNF-II), IL-6, interleukin 8 (IL-8), interleukin 1 beta (IL-1b), interleukin 1 alpha (IL-1a), soluble CD14 (sCD14), domain dimer (d-dimer), vascular cell adhesion molecule 1 (VCAM-1), and high-sensitivity C-reactive protein (hs-CRP). ³⁹ Markers were chosen based on previously published findings from the national Multicenter AIDS Cohort Study/Women’s Interagency HIV Study Combined Cohort Study identifying these specific biomarkers as connected to CVD risk and also established clinical biomarkers related to vascular dysfunction.^39,40 Samples where a biomarker was below the detection limit were adjusted to reflect the minimum detectable concentration for each respective biomarker (see Supplementary Table S1 for sensitivity of assays).

Data analysis

Differences in microbiota composition by GAHT use

To test if alpha diversity metrics (richness, evenness, and Shannon diversity) were different between GAHT groups (use vs. no use), the Wilcoxon rank sum test was utilized. Dissimilarity of beta diversity (Bray–Curtis index, unweighted UniFrac, and weighted UniFrac distance calculation) between GAHT groups was evaluated using the Adonis permutational multivariate analysis of variance (PERMANOVA) test. Furthermore, differential abundance testing was used to investigate bacterial abundance differences between GAHT groups. The DeSeq2 package was utilized; this package accounts for sequencing depth and multiple comparisons by controlling the false discovery rate using the Benjamini–Hochberg method. ⁴¹ Analysis was completed on the phylum and genus level. For the differential abundance testing, age, race, and years with HIV were explored as covariates. The results did not change; thus, unadjusted models were retained due to the small sample size.

Differences in inflammatory markers related to CVD by GAHT use

Linear regressions with natural log-transformed cytokine data (log-level regressions) were fit to test the association between GAHT use (predictor) and inflammatory biomarkers (outcomes). Age and years with HIV were explored as covariates. The results did not change; thus, unadjusted models were retained due to the small sample size.

Relationship between microbiota composition and inflammatory markers related to CVD

Inflammatory biomarkers were log transformed, and log-level regressions were conducted to determine the association between microbial communities (alpha [richness, evenness, Shannon diversity]) and inflammatory biomarkers. Spearman’s correlation was used to test the association between inflammatory biomarkers and the relative abundance of the identified topmost abundant bacteria genera.

Alpha was set to 0.05 and adjusted using the Benjamini–Hochberg procedure as necessary. ⁴¹ Analysis and visualization were performed in R. Gene sequence data are available through the National Center for Biotechnology Information Sequence Read Archive (PRJNA1073553).^36,37,42

Results

Participant characteristics

A total of 21 samples were analyzed from transgender women with HIV who were on ART. Participants had a mean age of 48.6 years (minimum = 23, maximum = 69; interquartile range [IQR] = 23). Approximately half of the sample was Black (49%), and the other half was White (51%). On average, individuals were living with HIV for 18.3 years, ranging from <1 to 29 years (median = 19; IQR = 19). The majority of the sample had an undetectable HIV viral load (n = 17, 81%), one participant reported a detectable viral load (n = 1, 5%), and 14% had missing viral load data (n = 3). Nearly half had a CD4 count of above 500 (n = 10, 48%), 24% between 200 and 500 (n = 5), and 29% (n = 6) had missing data. Missing data for the current study are a product of this being an analysis on stored samples and not specifically designed for this particular study; thus, data were limited to what was collected at the time of specimen storage. A majority reported current estrogen-based GAHT use (n = 15, 71%) with a little under a third reporting no active GAHT use (n = 6, 29%). The types of GAHT reported included estrogen (n = 16, 76%), spironolactone (n = 6, 29%), and finasteride (n = 1, 5%).

Microbiota composition by GAHT use (GAHT → microbiota)

Figures 2 and 3 display the overall prevalence and relative abundance of bacteria in the sample. Regardless of GAHT status, at the genus level, the microbiota was dominated by Finegoldia, Prevotella, Corynebacterium, and Anaerococcus, and Firmicutes were the dominating phyla. There were no significant differences in alpha diversity metrics by GAHT use via Wilcoxon rank sum tests. Specifically, there was no difference in richness (GAHT use median = 210, no use median = 166, p = 0.293), evenness (GAHT use median = 0.78, no use median = 0.77, p = 0.791), or Shannon diversity (GAHT use median = 4.15, no use median = 3.85, p = 0.424).

FIG. 2.

Overall prevalence of bacteria in the sample (microbiota composition).

FIG. 3.

Overall relative abundance of bacteria in the sample (microbiota composition).

For beta diversity, the Adonis (PERMANOVA) test yielded that the bacterial composition did not vary between the GAHT groups (Bray–Curtis index F(1,19) = 0.001, p = 0.976; unweighted UniFrac, F(1,19) = 0.384, p = 0.521; weighted UniFrac, F(1,19) = 0.325, p = 0.608). Differential abundance testing revealed no bacteria were significantly differentially abundant between individuals who used GAHT and those who did not use GAHT, on both the phylum and genus level. Figures 4 and 5 display the topmost phylum and genus by GAHT use.

FIG. 4.

Topmost phylum by gender-affirming hormone therapy use.

FIG. 5.

Topmost genus by gender-affirming hormone therapy use.

Inflammatory markers and GAHT use (GAHT → inflammation)

Log-level regressions assessed if GAHT use was associated with inflammatory biomarkers related to CVD. Out of the 15 biomarkers, three were significantly related to GAHT use. Specifically, there were predicted expected increases in sTNF-I (+38.8%, p = 0.028) and sTNF-II (+31.8%, p = 0.008), and a predicted expected decrease in IL-1a (−63.8%, p = 0.008) among participants that had active GAHT use as compared with participants that did not use GAHT as noted in Table 1.

Table 1.

Log-Level Regression Models Examining the Association Between Gender-Affirming Hormone Therapy Use and Inflammatory Biomarkers Related to Cardiovascular Disease (N = 21)

Predictor: GAHT use (binary active use vs. no use)
Model #	Outcomes (log)	R 2	b	SE	95% CI	p	Expected % change in outcome for GAHT use vs. no GAHT
1 a	sTNF-I	0.229	0.328	0.138	(0.039 to 0.616)	0.028	+38.8%
2 a	sTNF-II	0.317	0.276	0.093	(0.081 to 0.470)	0.008	+31.8%
3 a	IL-1a	0.316	−1.016	0.343	(−1.733 to −0.299)	0.008	−63.8%
4	IL-1b	0.018	−0.181	0.303	(−0.815 to 0.454)	0.558	n/a
5	MCP-1	0.000	−0.012	0.225	(−0.484 to 0.459)	0.956	n/a
6	IL-8	0.001	0.036	0.304	(−0.600 to 0.673)	0.906	n/a
7	IL-6	0.000	0.012	0.240	(−0.490 to 0.515)	0.960	n/a
8	TNFa	0.057	0.179	0.166	(−0.169 to 0.527)	0.295	n/a
9	sCD163	0.021	−0.123	0.193	(−0.528 to 0.282)	0.533	n/a
10	d-dimer	0.013	0.187	0.376	(−0.600 to 0.973)	0.625	n/a
11	VCAM-1	0.000	−0.017	0.231	(−0.502 to 0.467)	0.942	n/a
12	ICAM-1	0.001	−0.089	0.693	(−1.539 to 1.362)	0.900	n/a
13	sCD14	0.003	0.027	0.113	(−0.210 to 0.264)	0.815	n/a
14	hs-CRP	0.013	−0.289	0.573	(−1.487 to 0.910)	0.620	n/a
15	I-FABP	0.136	0.502	0.290	(−0.106 to 1.109)	0.100	n/a

Outcomes vertically on the left side represent each model run (i.e., each row = a regression model); all rectal cytokine biomarkers were log transformed.

^aThe overall model was significant; % change calculated via (exp(b)−1) × 100.

CI, confidence interval; d-dimer, domain dimer; GAHT, gender-affirming hormone therapy; hs-CRP, high-sensitivity C-reactive protein; ICAM-1, intercellular adhesion molecule 1; I-FABP, intestinal fatty-acid binding protein; IL, interleukin; MCP-1, monocyte chemoattractant protein-1; n/a, not applicable; sCD14, soluble CD14; sCD163, soluble CD163; SE, standard error; sTNF-I, soluble tumor necrosis factor receptor I; sTNF-II, soluble tumor necrosis factor receptor II; TNFa, tumor necrosis factor alpha; VCAM-1, vascular cell adhesion molecule 1.

Microbiota composition and inflammatory markers (microbiota → inflammation)

Log-level regressions assessed if microbial communities (alpha diversity [richness, evenness, Shannon diversity]) were associated with inflammatory biomarkers related to CVD. Out of the 15 biomarkers, only one was significantly associated with the three alpha diversity metrics. Specifically, for every one-unit increase in richness, there was a predicted expected 59.8% decrease in IL-1a (p = 0.021); for every one-unit increase in evenness, there was a predicted expected 94.7% decrease in IL-1a (p = 0.049); and for every one-unit increase in Shannon diversity, there was a predicted expected 39.9% decrease in IL-1a (p = 0.021). Results are presented in Table 2.

Table 2.

Log-Level Regression Models Examining the Association Between Alpha Diversity and Interleukin 1 Alpha (N = 21)

		R 2	b	SE	95% CI	p	Expected % change in IL-1a for every one-unit increase in diversity predictor (%)
Outcome: IL-1a (Log)
Model 1	Richness	0.249	−0.006	0.002	(−0.011 to −0.001)	0.021	−59.8
Model 2	Evenness	0.189	−2.931	1.391	(−5.841 to −0.020)	0.049	−94.7
Model 3	Shannon diversity	0.251	−0.509	0.202	(−0.932 to −0.087)	0.021	−39.9

Out of 15 inflammatory biomarkers, only one (IL-1a) was associated with microbiota diversity predictors; thus, only results for IL-1a are shown. Predictors were run in separate models; all three models had overall model significance. % change calculated via (exp(b)−1) × 100. Full data for all models are available upon request.

IL-1a, interleukin 1 alpha.

Spearman’s correlations were used to test the association between the relative abundance of the identified topmost abundant bacteria genera and inflammatory biomarkers related to CVD. Bacteroides was moderately positively correlated with sTNF-I (rho = 0.58, p = 0.006), sTNF-II (rho = 0.55, p = 0.010), TFN-α (rho = 0.45, p = 0.041), and sCD14 (rho = 0.57, p = 0.007). Faecalibacterium (rho = −0.52, p = 0.016) and Blautia (rho = −0.44, p = 0.048) were both moderately negatively correlated with IL-1a. Streptococcus was moderately negatively correlated with TFN-α (rho = −0.48, p = 0.026), and IL-8 was moderately negatively correlated with Eubacterium (rho = −0.44, p = 0.048). Results are presented in Table 3.

Table 3.

Spearman’s Rho Associations Between Relative Abundance of the Identified Topmost Abundant Bacteria Genera and Inflammatory Biomarkers (N = 21)

	I-FABP	sTNF-I	sTNF-II	MCP-1	IL-8	IL-1a	TNFa	IL-1b	IL-6	sCD163	d-dimer	VCAM-1	ICAM-1	sCD14	hs-CRP
Collinsella	0.28	0.28	−0.01	0.25	−0.07	−0.32	0.19	−0.11	0.35	0.32	0.34	0.25	−0.12	0.05	0.15
Prevotella	0.06	0.29	0.24	0.22	0.08	−0.34	0.26	−0.10	0.33	0.14	0.13	−0.06	−0.09	−0.12	0.04
Finegoldia	0.05	−0.27	−0.15	−0.14	−0.01	0.41	−0.15	−0.08	−0.04	−0.15	−0.22	−0.22	0.19	0.19	−0.33
Streptococcus	0.00	−0.24	−0.28	−0.28	0.05	−0.22	−0.48*	0.00	−0.24	−0.27	−0.13	0.17	−0.14	−0.34	0.03
Faecalibacterium	−0.10	0.10	−0.13	−0.26	−0.26	−0.52*	0.06	−0.04	0.22	0.22	0.31	−0.02	−0.09	−0.36	0.12
Peptoniphilus	0.06	−0.14	0.22	−0.02	−0.07	0.32	−0.10	0.04	−0.18	−0.28	−0.22	−0.17	0.28	−0.03	−0.06
WAL_1855D	−0.13	−0.29	−0.04	0.16	0.05	0.40	0.12	0.09	−0.03	0.12	0.00	−0.10	0.16	−0.07	0.12
Dialister	−0.35	−0.01	0.22	0.10	0.33	−0.17	0.35	0.14	0.02	0.39	−0.23	−0.33	0.18	−0.20	0.23
Bacteroides	−0.31	0.58**	0.55*	0.30	0.32	0.21	0.45*	0.31	−0.02	0.38	0.23	0.06	0.24	0.57**	0.28
Corynebacterium	0.27	−0.10	−0.25	−0.15	−0.31	0.26	−0.09	−0.21	−0.31	−0.13	−0.16	0.14	0.04	−0.01	−0.33
Anaerococcus	0.07	0.09	0.18	0.23	0.26	0.43	−0.24	0.15	−0.12	−0.20	−0.18	0.13	0.33	0.25	−0.02
Catenibacterium	0.03	0.20	0.11	0.21	0.08	−0.42	0.18	−0.03	0.26	0.18	0.17	0.17	−0.11	−0.02	0.22
Blautia	−0.03	0.12	−0.12	−0.17	−0.19	−0.44*	0.06	−0.09	0.18	0.30	0.33	0.07	−0.04	−0.25	0.15
[Eubacterium]	−0.03	0.11	−0.24	−0.27	−0.44*	−0.25	−0.01	−0.11	0.17	0.30	0.30	−0.09	−0.29	−0.28	0.05
Campylobacter	0.10	0.16	0.27	0.15	0.36	0.04	0.21	−0.12	−0.11	0.14	−0.21	0.00	0.09	0.18	−0.16
Actinomyces	−0.18	−0.31	0.08	0.13	0.30	0.31	−0.09	−0.04	−0.25	0.02	−0.08	0.13	0.16	0.07	−0.04
Porphyromonas	−0.26	−0.35	−0.04	0.13	0.40	0.03	−0.30	0.30	−0.09	−0.18	−0.34	−0.04	0.10	−0.17	0.16

*p < 0.05.

**p < 0.01.

Discussion

Among a sample of transgender women with HIV, the goal of this study was to characterize the associations between (1) the microbiota, (2) select inflammatory markers that have implications for CVD, and (3) GAHT. Although transgender women experience a dual disparity in both HIV and CVD, the underlying mechanisms behind this disparity remain poorly understood. The current investigation was prompted by evidence suggesting that HIV is associated with gut microbial dysbiosis with potential for subsequent CVD risk and the potential for GAHT to influence both disruptions in the gut microbiota and inflammatory markers linked to CVD risk.

This is among the first studies to assess and characterize these relationships among transgender women living with HIV, aiming to enhance our understanding of the intersecting nature of HIV and CVD disparities. Furthermore, studies on the microbiota of transgender women have focused on specimens collected from the vagina of transgender women who have had vaginoplasty. ⁴³ To our knowledge, this is among one of the initial studies to examine the microbiota via rectal swabs among transgender women, a practical site to examine the composition of the gut microbiota given that stool is a highly representative sample type for the colon (see Lacunza et al. 2023). ⁴⁴ Overall, our findings showed varied associations between microbial dysbiosis, inflammatory markers (that have implications for CVD risk), and GAHT use.

Our results showed no significant difference in alpha or beta diversity of the microbiota by GAHT use. Studies of cisgender, pre-menopausal women have found a lack of association between estrogen levels and microbiota diversity. ⁴⁵ However, cisgender males and cisgender females with hyperandrogenism (e.g., polycystic ovary syndrome) show an inverse relationship with level of estradiol and alpha diversity. ⁴⁶ Given mixed extant findings and that the current study had a small sample size and did not assess estradiol levels (only GAHT use vs. not), the findings should be interpreted with caution.

Key inflammatory markers that have implications for CVD were significantly associated with GAHT use. Specifically, individuals with active GAHT use had higher levels of sTNF-I and sTNF-II and lower levels of IL-1a compared with those without GAHT use. Previous studies in other inflammatory conditions have demonstrated a positive relationship of estrogen hormones and inflammatory markers such as TNF.^47,48 Increased sTNF-I and sTNF-II have been implicated in CVD risk. Specifically, increased concentrations of these biomarkers have been directly associated with the presence of coronary artery disease and angina and increased risks of cardiovascular events and mortality.^49–51 Notably, increased sTNF-I has been associated with cardiovascular events even among those with a coronary artery calcium score of zero, indicating increased concentrations put even those at traditionally “low” risk at risk for cardiovascular events.

The decreased IL-1a, which is a proinflammatory cytokine, found in those using GAHT, indicates that GAHT may also have anti-inflammatory properties. Aligned with a recent study, after 12 months of transdermal GAHT, systemic and endothelial inflammatory marker concentrations decreased among transgender women. ⁵² More generally, sex hormones have been associated with a decrease in production of IL-1 by human peripheral monocytes. ⁵³ The IL-1 family of cytokines, which can be segregated into many subfamilies, have emerged as orchestrators of various inflammatory diseases and as key mediators at the interface between the microbiota and human health and disease.

Microbiota composition was variably related to inflammatory biomarkers such that both positive and negative associations existed. Recent advances in understanding how the microbiota can influence both the physiology and the pathogenesis of disease in humans have highlighted the importance of gaining a deeper insight into the complexities of the host–microbial dialog. In this study, the association between relative abundance of the identified topmost abundant bacteria genera and inflammatory biomarkers related to CVD demonstrated that among our population, Bacteroides was positively correlated with some inflammatory markers (sTNF-I, sTNF-II, TFN-a, and sCD14). Previous studies in other populations have reported that Bacteroides can be proinflammatory or anti-inflammatory depending on their location and environment.^54,55

In addition, we found that Faecalibacterium and Blautia were both moderately negatively correlated with IL-1a as previously described in the literature.^56,57 In addition, two species of Firmicutes were negatively associated with biomarkers. Specifically, Streptococcus was negatively correlated with TFN-a, and Eubacterium was negatively correlated with IL-8, which have also been found by studies in different populations.^58,59 Taken together, these findings support the role that microbiota composition has in contributing to inflammatory processes.

Indeed, the gut microbiota plays a significant role in the connection between inflammatory markers and CVD. ⁶⁰ It produces several bioactive metabolites, including trimethylamine N-oxide (TMAO) and short-chain fatty acids (SCFAs), which influence systemic inflammation and contribute to CVD risk.^61,62 Elevated TMAO levels are associated with an increased risk of CVD and atherosclerosis, partly by promoting inflammation and oxidative stress. Yet, SCFAs, such as butyrate, acetate, and propionate, have been shown to exhibit anti-inflammatory and cardioprotective properties by modulating the release of proinflammatory cytokines, such as IL-1a, IL-6, and TNF-a, and enhancing the production of anti-inflammatory cytokines, thereby reducing the risk of CVD.^63,64 Therefore, further understanding of the gut microbiota and its complex relation to biomarkers of inflammation in the context of HIV is necessary to identify targets to decrease CVD risk in transgender women with HIV.

As discussed, the current study also found that both GAHT use and microbiota diversity were associated with lower IL-1a levels. Future research should attend to the role of anti-inflammatory processes (e.g., via IL-1a) within the interconnected network of HIV, CVD, and microbiota, specifically among transgender women.

Limitations

Despite this being one of the first studies to preliminarily explore the role of the microbiota in CVD risk among transgender women with HIV, limitations should be noted. This study leveraged data collected, thus the sample size was small, and the group sizes were disproportional, with only a few transgender women not on GAHT. Due to the limitation of previously collected data, we did not have access to ART adherence/regimens and had very limited data on viral load and CD4. We also did not have a cohort of transgender women without HIV for comparison. Also, although we have some data on anti-androgen therapy, we were not able to specifically examine its relationship with CVD risk, which is an important emerging pathway to explore. Future research needs to collect detailed information on these potentially influential variables and comparators in larger samples adequate for analyses.

HIV information (viral load/CD4 count) was self-reported, introducing recall bias, which resulted in limited collection of these data. In addition, estradiol levels were not objectively collected for GAHT use, introducing self-report bias and the inability to examine how the variability in hormone levels may affect outcomes. Relatedly, despite working from stored samples, given the limited scope of work/funding, we were not able to test for hormone levels (i.e., luteinizing hormone, follicle-stimulating hormone, and estradiol); the focus was contained to the microbiota analysis. Furthermore, we investigated inflammatory markers previously linked to CVD and not direct measures of CVD risk. The inflammatory markers utilized are also associated with other health outcomes (e.g., diabetes), so the results may speak to the larger picture of comorbid health disparities for transgender women living with HIV. ⁶⁵

Taken together, these results should be considered descriptive to support future work to better characterize relationships among HIV, the microbiota, GAHT, and CVD risk. To note, a more holistic integrated model that examines the interaction between the psychobehavioral and biological risk factors of CVD for transgender women with HIV is needed to fully explore the complexity of this dual disparity. While acknowledging the study’s limitations, these initial findings pave the way to a deeper understanding of the intricate associations among these factors. Future research endeavors in this area should address these limitations by employing larger and more diverse cohorts, adopting longitudinal designs, utilizing objective and more nuanced measures, and examining specific GAHT regimens. Longitudinal studies would provide more informative insight into establishing causal relationships and understanding changes over time.

Conclusions

In this study, we characterized the gut microbiota of transgender women with HIV and examined the potential influence of GAHT use. The findings from this study suggest a potential association between biomarkers of inflammation linked to CVD risk and microbiota composition and GAHT. Although this was a small study, the findings contribute to the characterization of interconnecting factors that may help identify potential mechanisms of CVD risk among transgender women with HIV. This is an area where further research is needed for developing targeted interventions to reduce the burden of CVD in transgender women with HIV and improve their overall health and well-being.

Authors’ Contributions

Conceptualization: C.M. and T.R.G. Investigation: C.M. and M.L.A. Data curation: C.A.B., N.K., Su.P., M.R., and Sa.P. Formal analysis: C.A.B. and T.R.G. Visualization: C.A.B. Funding acquisition: C.M. and T.R.G. Methodology: C.M., T.R.G., N.K., M.L.A., and B.E.H. Project administration: N.F.N., V.R., V.L., and T.B. Resources: N.K., Su.P., M.R., and Sa.P. Supervision: M.L.A., B.E.H., N.K., and D.J. Writing—original draft: T.R.G., C.A.B., C.M., and V.R. Writing—critical review and editing: M.L.A., L.A.F., B.E.H., D.J., N.F.N., V.L., and T.B.

Abbreviations

ART

antiretroviral therapy

CVD

cardiovascular disease

d-dimer

domain dimer

GAHT

gender-affirming hormone therapy

HIV

human immunodeficiency virus

hs-CRP

high-sensitivity C-reactive protein

ICAM-1

intercellular adhesion molecule 1

I-FABP

intestinal fatty-acid binding protein

IL-1a

interleukin 1 alpha

IL-1b

interleukin 1 beta

IL-6

interleukin 6

IL-8

interleukin 8

MCP-1

monocyte chemoattractant protein-1

rRNA

ribosomal ribonucleic acid

sCD14

soluble CD14

sCD163

soluble CD163

sTNF-I

soluble tumor necrosis factor receptor I

sTNF-II

soluble tumor necrosis factor receptor II

TNF-a

tumor necrosis factor alpha

VCAM-1

vascular cell adhesion molecule 1

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Cite this article as: Glynn TR, Broedlow CA, Rodriguez V, Nogueira NF, Londono V, Brophy T, Pallikkuth S, Roach M, Pahwa S, Fein LA, Hurwitz BE, Jones D, Alcaide ML, Klatt N, Martinez C (2026) Microbiota, gender-affirming hormone therapy, and inflammatory biomarkers in transgender women with HIV: potential implications for cardiovascular disease, Transgender Health 11:1, 114–125, DOI: 10.1089/trgh.2024.0042.