Transgender Women with HIV Demonstrate Unique Non-Alcoholic Fatty Liver Disease Profiles

Abstract

Purpose:

Non-alcoholic fatty liver disease (NAFLD) prevalence and severity may be higher in people with human immunodeficiency virus (HIV) than the general population, and vary with sex and age. We explored NAFLD characteristics by gender.

Methods:

Adult transgender women (TW), cisgender women (CW), and cisgender men (CM) with HIV on antiretroviral therapy and without other known causes of liver disease underwent screening for NAFLD (2017–2020). Circulating factors associated with NAFLD were measured. Hepatic steatosis and fibrosis were assessed using transient elastography by controlled attenuation parameter (CAP) and liver stiffness measurement (LSM), respectively. Analysis of variance/Wilcoxon testing compared normally/non-normally distributed variables, respectively. Logistic regression evaluated factors associated with CAP and LSM.

Results:

Participants (n=194) had median age 48 years and body mass index 28.3 kg/m²; 42% were CM, 37% TW, and 21% CW; 95% were non-white; and 16% had diabetes, 40% dyslipidemia, and 49% hypertension. NAFLD prevalence was 59% using CAP ≥248 dB/m (≥S1 steatosis), 48% using CAP ≥260 dB/m (≥S2 steatosis), and 32% using CAP ≥285 dB/m (≥S3 steatosis). Compared to CM and CW, TW had the highest median CAP scores, were more likely to have ≥S2 steatosis, and had the highest insulin resistance, interleukin-6, and fetuin-A values. TW off versus on gender-affirming hormone therapy (GAHT) had slightly higher median CAP scores.

Conclusion:

TW on GAHT had less hepatic steatosis than TW not on GAHT, although overall NAFLD severity was greater than expected for TW compared to CM and CW. The effects of estrogen supplementation and androgen deprivation on liver health in TW require further study.

Introduction

Non-alcoholic fatty liver disease (NAFLD) affects 25% of adults in the general population and includes a disease spectrum spanning steatosis, non-alcoholic steatohepatitis (NASH), and end-stage liver disease.^1,2 NAFLD is defined as hepatic steatosis (>5% excess lipid accumulation in hepatocytes), either by imaging or histology, following exclusion of other causes of steatotic liver disease (significant alcohol consumption, long-term use of a steatogenic medication, hepatitis C virus [HCV], etc.).³ NAFLD rates increase with age and are associated with the risk of progression to advanced liver disease, the metabolic syndrome, and cardiovascular disease (CVD).^4,5

NAFLD prevalence and severity are higher in men than women of reproductive age.⁶ However, after menopause, NAFLD occurs at a higher frequency in women, suggesting a potential protective effect of estrogen.^7,8 In addition, NAFLD risk is increased among women of reproductive age with altered sex hormone levels (e.g., polycystic ovary syndrome, Turner syndrome, oophorectomy), whereas it is reduced among post-menopausal women on hormone replacement therapy.⁹ Nonetheless, once NAFLD is established, women have a higher risk of advanced fibrosis than men, especially after age 50.¹⁰ Higher levels of testosterone, associated with increased NAFLD risk in young women, may play a role in the progression to NASH in this population.^11,12

Among people living with human immunodeficiency virus (HIV) (PWH), NAFLD prevalence in observational studies is frequently >30%, and may have unique features.¹³ HIV is characterized by persistent inflammation and immune activation, and the combination of HIV- and NAFLD-associated inflammation may accelerate hepatic injury, fibrosis, and organ dysfunction, as evidenced by higher rates of progression from NAFLD to NASH in PWH (63% vs. 37% in people without HIV).¹⁴

To date, NAFLD has not been broadly assessed in diverse groups of PWH, including transgender persons. In this hypothesis-generating, pilot study, we explored NAFLD prevalence and characteristics among transgender women (TW), cisgender women (CW), and cisgender men (CM) with HIV undergoing screening for hepatic steatosis, with the aim of better understanding features that may be unique to each population.

Methods

Study design

This study is a secondary analysis nested within an ongoing, cross-sectional, observational, single-center, pilot study.

Study population

Participants were approached (2017–2020) at random at the Thomas Street Health Center, a freestanding HIV clinic in Houston, Texas. Inclusion criteria were documented HIV and age ≥18 years. Exclusion criteria were active, untreated opportunistic or acquired immune deficiency syndrome-defining illness, other acute infection, untreated autoimmune disorders, prednisone use ≥5 mg/day, and current pregnancy. Participants with other known causes for liver disease, such as active hepatitis B virus (HBV), HCV, current heavy alcohol use (>3 drinks per day), and diagnosis of autoimmune or genetic disorders, were excluded from this analysis.

Study procedures

The study was approved by the institutional review board of UTHealth Houston. Written informed consent was provided by all participants before initiation of study procedures. Demographics, medical history, medications, substance use history, and values for the most recent HIV-1 ribonucleic acid (RNA), cluster of differentiation (CD) 4⁺ T-cell count, aspartate aminotransferase, alanine aminotransferase (ALT), fasting glucose, total cholesterol, high-density lipoprotein, low-density lipoprotein (LDL) cholesterol, and triglyceride (TG) levels were obtained from medical records. The homeostatic model assessment of insulin resistance (HOMA-IR) value was calculated as ([insulin×glucose]/22.5). All study evaluations were performed in the fasting (>8 h) state.

Circulating factors (i.e., adipokines, cytokines, hormones, henceforth referred to in aggregate as biomarkers) chosen for their known association with NAFLD physiology in the general population were adiponectin, soluble CD (sCD) 14, E-selectin, fatty acid-binding protein (FABP)-4, fetuin-A, fibroblast growth factor (FGF) 21, FABP-2, resistin, insulin, chitinase 3-like (CHI3L) 1, soluble interleukin (IL)-1 receptor-like-1 (ST2), proprotein convertase subtilisin/kexin type 9 (PCSK9), IL-6, transforming growth factor-beta (TGFβ)-1, chemokine ligand (CXCL) 4, and plasminogen activator inhibitor (PAI)-1. Plasma and serum were stored at -70°C until batched biomarker measurement could occur centrally by enzyme-linked immunosorbent assay or Luminex at the end of study. All kits were purchased from R and D Systems (Minneapolis, MN); assays were performed per manufacturer's instructions. As coefficients of variation were within expected ranges, no assay was repeated.

Hepatic fat content (controlled attenuation parameter [CAP]) and liver stiffness measurement (LSM) were obtained by a single operator using transient elastography FibroScan^® Compact 530 model (Echosens, Paris, France). Based on abdominal circumference, M probe was used in most cases and XL probe was used in 23% of participants. The following criteria were applied to define the result of transient elastography as reliable: at least 10 validated measurements and an interquartile range (IQR) less than 30% of the median LSM.¹⁵ The following cutoffs for hepatic steatosis were used: CAP ≥248 dB/m (≥S1 steatosis), CAP ≥260 dB/m (≥S2 steatosis), and CAP ≥285 dB/m (≥S3 steatosis).^16–19 The validated LSM cutoff value utilized for significant fibrosis (≥F2) was 8.6kPa.¹⁷

Analysis

SAS version 9.4 (SAS Institute, Cary, NC) was used for analysis, with a two-sided α=0.05 used to determine statistical significance. Descriptive statistics (mean, standard deviation, median, IQR, and frequency) were generated for cross-sectional characterization of the study population and key outcome variables. Normality assumption was checked for continuous variables. Analysis of variance (ANOVA) was used for normally distributed continuous variables, and Wilcoxon test if non-normal. For categorical variables, p-value was based on likelihood ratio test. This was a hypothesis-generating pilot study, and all analyses were exploratory and reported without adjustment for multiple testing. Logistic regression assessed factors associated with hepatic steatosis and/or increased liver stiffness. The multivariate model was obtained through stepwise selection with inclusion/exclusion p-value thresholds of 0.30/0.05, respectively. Further analyses were conducted restricting TW to TW on gender-affirming hormonal therapies (GAHTs).

Results

Study population

A total of 194 participants enrolled. Thirty participants with other known causes for liver disease, including active HBV, HCV, and current alcohol heavy use, were excluded. Overall, the population self-identified as 42% (n=69) CM, 37% (n=60) TW, and 21% (n=35) CW; 51% were non-Hispanic Black and 45% Hispanic (Table 1).

Table 1.

Clinical and Demographic Characteristics

	n	TW	n	CM	n	CW
Study population
Age†,‡,§	60	40.5 (34, 48.5)	69	49 (42, 58)	35	54 (49, 63)
BMI§	59	28.4 (23.9, 34.5)	69	26.8 (24.4, 29.8)	35	30.1 (26, 39.9)
Race§	60		69		35
White		3 (5.0%)		5 (7.3%)		0 (0%)
Black		32 (53.3%)		25 (36.2%)		26 (74.3%)
Hispanic		25 (41.7%)		39 (56.5%)		9 (25.7%)
Current smoker	57	21 (36.8%)	67	18 (26.9%)	34	13 (38.2%)
Current drug use†,‡	51	11 (21.6%)	67	5 (7.5%)	35	2 (5.7%)
CVD history	60	3 (5.0%)	68	5 (7.4%)	34	0 (0%)
Hypertension	60	26 (43.3%)	68	33 (48.5%)	34	22 (64.7%)
Dyslipidemia	60	24 (40.0%)	68	27 (39.7%)	34	15 (44.1%)
Diabetes	60	7 (11.7%)	68	13 (19.1%)	34	6 (17.7%)
HBV history	60	3 (5.0%)	68	6 (8.8%)	34	3 (8.8%)
HCV history	60	3 (5.0%)	67	10 (14.9%)	34	4 (11.8%)
Cirrhosis‡	60	0 (0%)	68	4 (5.9%)	34	3 (8.8%)
Years living with HIV‡	59	13 (8, 20)	67	14 (9, 22)	33	20 (13, 23)
Years on ART‡	55	9 (7, 14)	60	10 (8, 14.5)	28	13 (10.5, 16.5)
CD4 T-cell count (cells/μL)	59	561 (420, 861)	65	593 (434, 764)	35	667 (424, 925)
HIV-1 RNA <50 copies/mL†,§	57	45 (79.0%)	66	61 (92.4%)	35	26 (74.3%)
ART regimen
PI	59	11 (18.6%)	67	10 (14.9%)	34	8 (23.5%)
NNRTI	59	14 (23.7%)	67	15 (22.4%)	34	19 (55.9%)
INSTI	59	37 (62.7%)	67	43 (64.2%)	34	8 (23.5%)
Past use of AZT/D4T‡	59	4 (6.8%)	66	8 (12.1%)	34	9 (26.5%)
Metabolic panel
Fasting glucose (mg/dL)	60	92 (82, 105)	67	93 (81, 107)	35	93 (85, 112)
Total cholesterol (mg/dL)	54	179 (145, 210)	65	169 (143, 190)	35	174 (151, 208)
HDL cholesterol (mg/dL)†,‡,§	54	50 (37, 60)	65	42 (36, 49)	35	59 (46, 69)
LDL cholesterol (mg/dL)	54	97 (75, 115)	65	94 (71, 117)	34	96 (60, 125)
Triglycerides (mg/dL)§	54	121 (91, 194)	65	131 (87, 170)	35	103 (83, 113)
AST (IU/L)†	60	19 (16, 29)	66	25 (18, 32)	35	20 (16, 27)
ALT (IU/L)†,§	60	23 (18, 37)	66	33 (26, 47)	35	24 (18, 31)
ALT >40 (IU/L)§	60	12 (20.0%)	66	21 (31.8%)	35	4 (11.4%)
HOMA-IR	60	2.79 (1.52, 3.93)	65	2.26 (1.39, 4.45)	34	2.27 (1.24, 3.45)

Frequency or median (IQR) presented.

^† p<0.05 TW versus CM.

^‡ p<0.05 TW versus CW.

^§ p<0.05 CW versus CM.

ALT, alanine aminotransferase; ART, antiretroviral therapy; AST, aspartate aminotransferase; AZT, zidovudine; BMI, body mass index; CD, cluster of differentiation; CM, cisgender men; CVD, cardiovascular disease; CW, cisgender women; d4T, stavudine; HBV, hepatitis B virus; HCV, hepatitis C virus; HDL, high-density lipoprotein; HIV, human immunodeficiency virus; HOMA-IR, Homeostatic Model Assessment for Insulin Resistance; INSTI, integrase strand transfer inhibitor; IQR, interquartile range; LDL, low-density lipoprotein; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; RNA, ribonucleic acid; TW, transgender women.

Among TW, 30 (50%) self-reported current use of estradiol in any form, and 19 (31%) current spironolactone use. However, dosage, time on GAHT, and hormone concentrations were not recorded. No CM reported use of testosterone in any form. TW were younger; had similar median body mass index (BMI) and history of CVD, hypertension, dyslipidemia, diabetes mellitus (DM), or HBV/HCV; less likely to have clinically diagnosed cirrhosis; and more likely to currently use illicit drugs. Group characteristics remained similar when restricting analysis to TW on GAHT. Of note, the median (IQR) age of CW was 54 (49, 63) years, above the median age of menopause in the United States (51 years).²⁰

TW had shorter median time living with HIV, had shorter median time on antiretroviral therapy (ART), and were less likely to have history of zidovudine or stavudine use compared to CW, while showing no difference with CM. The three populations had similar median CD4⁺ T cell count and current ART, although CM were more likely to have an undetectable HIV-1 RNA compared to TW and CW. Metabolic profiles were similar among the three groups, except that CW were less likely to have ALT >40 IU/L.

Hepatic steatosis and stiffness

A total of 97 (59%) participants had CAP ≥248 dB/m, 78 (48%) participants had CAP ≥260 dB/m, and 52 (32%) participants had CAP ≥285 dB/m. Median CAP score for TW, CW, and CM were 265 (IQR 225, 303), 259 (IQR 236, 300), and 250 (IQR 228, 295) dB/m, respectively. TW were more likely to have median CAP score >260 dB/m (TW 54%, CW 49%, CM 44%) and also >285 dB/m (TW 39%, CW 29%, CM 29%), regardless of GAHT use, although no statistically significant difference was found. Notably, TW who were off GAHT (n=28) had higher median CAP score than TW on GAHT (n=29) (277 [IQR 223, 297] vs. 265 [IQR 225, 305] dB/m).

LSMs were generally not elevated and similar among the three groups: TW=4.3 (IQR 3.9, 5.3) kPa, CW=4.3 (IQR 3.6, 6.1) kPa, and CM=4.6 (IQR 4, 6.1) kPa. Frequency of clinically significant fibrosis, defined as LSM ≥8.6 kPa, was similar for all groups (8.5% of total population).

Controlling for other covariates, age, gender, and race was not associated with significant differences in CAP score or LSM. However, being a CW was associated with 8 dB/m lower median CAP score compared to CM. When restricting to TW on GAHT, being a TW was associated with 7.5 dB/m higher median CAP score compared to CM. In multiple quantile regression adjusted for age, gender, and race, higher CAP score was associated with higher BMI (p<0.0001), E-selectin (p<0.05), and TG levels (p<0.05), and lower resistin levels (p<0.01). These associations persisted when restricting analysis to TW on GAHT.

In a model exploring gender-race interactions and restricted to TW on GAHT, being a Black/African-American CW was associated with 4 kPA lower LSM compared to Black/African-American CM. In multiple quantile regression adjusted for age, gender, and race, higher LSM was associated with higher adiponectin (p<0.05), HOMA-IR (p<0.0001), and BMI (p<0.0001), and lower insulin (p<0.05) and platelet levels (p<0.0001). These associations persisted when restricting analysis to TW on GAHT.

Biomarker concentrations by gender

Biomarker concentrations are presented in Table 2. Briefly, TW had the highest levels of IL-6, fetuin-A, PAI-1, insulin, and resistin, and the lowest levels of FGF21, FABP-2, and CHI3L1 compared to CM and CW. Notably, TW had levels of adiponectin, sCD14, E-selectin, FABP-4, ST2, PCSK9, CXCL4/PF4, and TGFβ1 in-between CW and CM. In addition, when restricting analysis to TW on GAHT, TW had higher levels of fetuin-A and ST2 compared to CM, and higher levels of PAI-1/SERPINE-1 compared to CW (all p<0.05). No striking correlation was noted between CAP score or LSM and biomarker concentrations.

Table 2.

Biomarkers of Systemic Inflammation

Biomarker	TW n=60	CM n=67	CW n=34
Adiponectin (ng/mL)§	3847 (1800, 6809)	2941 (1225, 5627)	4894 (2625, 8497)
sCD14 (μg/mL)§	1.68 (1.42, 2.13)	1.68 (1.36, 1.84)	1.91 (1.6, 2.09)
E-Selectin (ng/mL)	42 (35, 53)	47 (34, 56)	41 (32, 55)
FABP4 (pg/mL)‡,§	18736 (10640, 24124)	14529 (8924, 20898)	30586 (19039, 39295)
Fetuin-A (μg/mL)	1068 (833, 1291)	905 (695, 1209)	986 (788, 1166)
FGF21 (pg/mL)	163 (97, 298)	166 (76, 249)	182 (91, 224)
FABP2 (ng/mL)†	0.87 (0.46, 1.61)	1.31 (0.91, 2.04)	1.15 (0.78, 2.24)
Resistin (ng/mL)	10.04 (6.48, 14.03)	8.16 (5.91, 11.44)	8.84 (5.33, 14.15)
CHI3L1 (pg/mL)	29973 (19857, 51011)	31288 (13983, 54817)	42024 (21269, 71546)
ST2 (pg/mL)§	9721 (6696, 12458)	10999 (7246, 16259)	7415 (5424, 12909)
PCSK9 (pg/mL)	54197 (25660, 82264)	47218 (28197, 72744)	58064 (31432, 91755)
IL-6 (pg/mL)†	3.57 (0.96, 11.04)	1.71 (0.63, 3.23)	2.46 (1.06, 4.11)
CXCL4/PF4 (ng/mL)	12649 (10260, 17900)	11500 (8795, 16132)	12699 (8781, 18800)
PAI1/SERPINE1 (pg/mL)	143448 (107109, 171614)	138334 (109328, 160578)	125508 (95536, 146878)
TGFβ1 (pg/mL)	81721 (27375, 134537)	87721 (23970, 124012)	62927 (29862, 117933)
Insulin (pmol/L)	70 (36, 125)	56 (36, 95)	60 (33, 79)

Frequency or median (IQR) presented.

^† p<0.05 TW versus CM.

^‡ p<0.05 TW versus CW.

^§ p<0.05 CW versus CM.

CHI3L1, chitinase 3-like 1; CXCL4/PF4, chemokine ligand 4/platelet factor 4; FABP, fatty acid-binding protein; FGF, fibroblast growth factor; IL, interleukin; PAI-1, plasminogen activator inhibitor 1; PCSK9, proprotein convertase subtilisin/kexin type 9; sCD, soluble CD; ST2, soluble IL-1 receptor-like-1; TGFβ-1, transforming growth factor-beta1.

Discussion

We explored gender differences in NAFLD prevalence, severity, and associated factors in a cohort of PWH being screened for hepatic steatosis. Sex differences in NAFLD prevalence, risk factors, progression, and clinical outcomes in persons without HIV are known.⁶ However, there is a paucity of data on hepatic steatosis among TW on and off GAHT, and even less knowledge specific to TW living with HIV. Therefore, this novel, descriptive study is a first step toward understanding the spectrum of liver disease in this unique population.

NAFLD prevalence among PWH using transient elastography as diagnostic tool varies from 35% to 48%, depending on the CAP score cutoff utilized to define hepatic steatosis. However, these studies included mostly or all CM. In our ethnically and gender diverse cohort of overweight PWH on stable ART, NAFLD prevalence was 59% using CAP ≥248 dB/m (≥S1 steatosis), 48% using CAP ≥260 dB/m (≥S2 steatosis), and 32% using CAP ≥285 dB/m (≥S3 steatosis), and frequency of traditional risk factors was high (median BMI 28.3 kg/m², 16% DM, 40% dyslipidemia, and 49% hypertension).

Interestingly, TW in our cohort had the highest median CAP scores, despite being younger than CM and CW, and were more likely to have moderate-to-severe hepatic steatosis regardless of GAHT use status. This is surprising, given the known relationship between increasing age and NAFLD prevalence. Similarly, after adjusting for age, gender, and race, being a CW was associated with 8 dB/m lower, and being TW on GAHT was associated with 7.5 dB/m higher, median CAP score compared to CM. In multiple models, β-estimates for TW and CW had consistently opposite trends (data not shown). Although our sample size was small, these could signify risk of greater NAFLD disease severity for TW.

Gender and GAHT may modulate NAFLD risk for TW. In this study, TW on GAHT had less hepatic steatosis than TW not on GAHT, raising the question of whether estrogen could have a protective role. Greater hepatic fatty acid (FA) synthesis in CM and lesser ability to clear FA from plasma and production and export TG promote hepatic lipid deposition compared to CW.^21–23 Estrogens act directly in the liver through nuclear estrogen receptor (ER)-α in hepatocytes and Kupffer cells, and indirectly by regulating growth hormone anabolic action.⁹ Ovariectomy in mice leads to hepatic steatosis through increased hepatic insulin resistance and de novo lipogenesis (DNL), and decreased FA oxidation.²⁴ Female ER-α-deficient mice also develop hepatic steatosis, and liver-specific ER-α-deficient mice do not have regression of liver fat accumulation with estrogen treatment.^9,24,25 Further prospective studies are needed to evaluate this relationship.

While estrogens play a role in hepatic health, so do androgens. Hepatic androgen receptor (AR)-deficient male mice develop hepatic steatosis when fed a high-fat diet, a finding modulated through decreased hepatic peroxisome proliferator-activated receptor-α expression and subsequent decreased FA oxidation, increased DNL, and hepatic insulin resistance.^26,27 Likewise, orchiectomy in mice results in insulin resistance and hepatic steatosis that reverse with testosterone administration.²⁸ In CM, androgens promote glucose and energy homeostasis by AR and ER signaling, and androgen deprivation contributes to metabolic syndrome, DM, and NAFLD.^29–31

The evidence suggesting that androgen deprivation is associated with NAFLD risk in CM males is highly relevant to TW on GAHT.³⁰ Contrary to this, Nelson et al. reported that of 12 TW (11 on estradiol, 4 on anti-androgen therapy, and another 4 with bilateral orchiectomy), 2 TW with testes, but suppressed serum total testosterone concentrations, had hepatic fat content by magnetic resonance spectroscopy similar to the four TW with orchiectomy; the remaining six TW (all with testes, four with nonsuppressed testosterone) had significantly higher hepatic fat content. The four TW with the highest testosterone levels also had the lowest circulating levels of estrogen. The same TW had the greatest hepatic fat content and the worst insulin sensitivity.³²

While estrogens improve glucose tolerance and insulin sensitivity in CM and CW, they have been shown to worsen insulin sensitivity in TW (Lake et al., unpublished data).³³ In our study, TW were least likely to have DM, but had the highest insulin levels and HOMA-IR values, a finding further exacerbated when restricting analysis to TW on GAHT. As insulin resistance is closely linked with NAFLD, this finding additionally supports the potential for exacerbated metabolic dysregulation in TW.

It is also worth discussing gender differences in select circulating biomarkers in our cohort. PCSK9 promotes LDL receptor degradation, increasing circulating LDL cholesterol, and PWH have elevated PCSK9 levels compared to people without HIV.^34–36 In the general population, PCSK9 levels are significantly higher in CW than CM, and in post-menopausal women compared to pre-menopausal women regardless of estrogen use.³⁷ Likewise, in our study, CW of post-menopausal age had higher levels than TW (on or off GAHT), and CM had the lowest levels. To our knowledge, this is the first description of PCSK9 levels in TW.

ST2 is a proinflammatory member of the Toll-Like Receptor family that has been implicated in hepatic steatosis and CVD.^38,39 In the general population, circulating ST2 levels are higher in CM than in CW. In our study, concentrations in TW were in between those of CM and CW (all p<0.05 when restricting analysis to TW on GAHT, data not shown), suggesting estrogen therapy in HIV+ TW may exacerbate this inflammatory pathway.

In the liver, chronic IL-6 exposure is proinflammatory, promotes hepatic gluconeogenesis and hepatic insulin resistance, and is associated with increased NAFLD risk.^40,41 IL-6 also increases with age.⁴² Men and women with HIV have equally elevated IL-6 levels even when virologically suppressed, although recent in vitro data suggest that estrogen exposure may enhance innate immune activation in PWH, increasing IL-6.^43,44 In our study, IL-6 levels in CM were similar to levels previously described, but TW had significantly higher levels regardless of GAHT use and notwithstanding their younger age.⁴⁵ Lower rates of virologic suppression among TW may partially explain these findings. Frequency of other factors associated with elevated IL-6 like smoking, DM, CVD, and HBV/HCV was similar across genders.

TW also had the highest levels of fetuin-A (TW on GAHT vs. CM, p<0.05, data not shown). Fetuin-A has been associated with metabolic syndrome, insulin resistance, and increased DM risk, and its levels are increased in people with NAFLD, although its pathogenic role is incompletely understood.^46–48 In male mice, Fetuin-A transcription is upregulated by estrogen.⁴⁹

Strengths and limitations

Our study has several strengths. First, few data exist describing NAFLD in TW, and understanding of NAFLD in PWH is emerging.⁵⁰ Second, the demographic diversity of our cohort reflects the communities of PWH in the United States and will significantly add to the available literature by examining gender differences. This analysis was meant to be descriptive and hypothesis generating, and as such is not a limitation. Limitations do include the cross-sectional, observational nature. In addition, due to small subgroup sample sizes, statistical power to readily identify between-group differences was limited, although differences believed to have potential clinical relevance are highlighted. Finally, GAHT use by TW was self-reported and estradiol and testosterone levels were not available for the full cohort to expand or support findings.

Conclusion

In conclusion, an improved understanding of the burden, pathogenesis of, and therapeutic options for NAFLD among PWH is needed, including sex and gender differences. The effects of estrogen supplementation and androgen deprivation on liver health in TW are virtually unstudied, but could profoundly affect the health of this population.

Abbreviations Used

ALT

alanine aminotransferase

ANOVA

Analysis of variance

AR

androgen receptor

ART

antiretroviral therapy

AST

aspartate aminotransferase

AZT

Zidovudine

BMI

body mass index

CAP

controlled attenuation parameter

CD

cluster of differentiation

CHI3L

chitinase 3-like

CM

cisgender men

CVD

cardiovascular disease

CW

cisgender women

CXCL

chemokine ligand

CXCL4/PF4

chemokine ligand 4/platelet factor 4

d4T

Stavudine

DNL

de novo lipogenesis

DM

diabetes mellitus

ER

estrogen receptor

FA

fatty acid

FABP

fatty acid-binding protein

FGF

fibroblast growth factor

GAHT

gender-affirming hormone therapy

HBV

hepatitis B virus

HCV

hepatitis C virus

HDL

high-density lipoprotein;

HIV

human immunodeficiency virus

HOMA-IR

homeostatic model assessment of insulin resistance

IL

interleukin

INSTI

integrase strand transfer inhibitor

IQR

interquartile range

LDL

low-density lipoprotein

LSM

liver stiffness measurement

NAFLD

Non-alcoholic fatty liver disease

NASH

non-alcoholic steatohepatitis

NNRTI

non-nucleoside reverse transcriptase inhibitor

PAI

plasminogen activator inhibitor

PCSK9

proprotein convertase subtilisin/kexin type 9

PI

protease inhibitor

PWH

people living with HIV

RNA

ribonucleic acid

sCD

soluble CD

ST2

soluble IL-1 receptor-like-1

TG

triglyceride

TGFß

transforming growth factor-beta

TW

transgender women

Authors' Contributions

J.E.L.: conceptualization, supervision, project administration, methodology, data interpretation, and writing—reviewing and editing. A.N.H.: data curation and writing—original draft. H.F.: formal analysis. H.M.: methodology and formal analysis. A.S.: investigation. N.S.U.: conceptualization, methodology, and data interpretation. K.E.C.: writing—reviewing and editing.

Disclaimer

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

Author Disclosure Statement

J.E.L. receives research support from Gilead Sciences, and serves as a consultant to Theratechnologies, unrelated to the work. A.N.H. declares no conflict of interest. H.F. declares no conflict of interest. H.M. is the co-founder of SmartCT Studio LLC, which is not involved in this study. A.S. declares no conflict of interest. N.S.U. declares no conflict of interest. K.E.C. serves on the scientific advisory board for Theratechnologies, Novo Nordisk, and BMS and has received grant funding from Boehringer-Ingelheim, BMS, and Novartis.

Funding Information

J.E.L.'s effort on this project was funded by National Institutes of Health grants K23AI110532 and R01DK126042 to J.E.L. All other project costs were supported by nongovernmental sources.

Cite this article as: Lake JE, Hyatt AN, Feng H, Miao H, Somasunderam A, Utay NS, Corey KE (2023) Transgender women with HIV demonstrate unique non-alcoholic fatty liver disease profiles, Transgender Health 9:5, 413–420, DOI: 10.1089/trgh.2022.0182.