Changes in liver steatosis in HIV-positive women are associated with the BMI, but not with biomarkers

Abstract

The prevalence of non-alcoholic fatty liver disease (NAFLD) is higher in HIV-infected patients compared to the general population. While metabolic risk factors such as obesity, insulin resistance and the metabolic syndrome have been identified as key risk factors in all individuals, there is limited information regarding the mechanisms that contribute to the higher prevalence among individuals living with HIV, particularly among women and ethnic minorities. The aim of this study was to determine the association, over two time points, of a panel of biomarkers with liver steatosis in a cohort of HIV-seropositive women and age-matched negative controls and to investigate whether the association differed by HIV status. To this effect, plasma samples obtained from 105 HIV-positive and –negative participants enrolled in the Women’s Interagency HIV study (WIHS) Washington DC site were assayed for biomarkers associated with inflammation, adipose tissue function, fibrinolysis, gut permeability and hepatocyte apoptosis/necrosis. Their association with liver steatosis, measured using Controlled-Attenuation Parameter (CAP) scores determined by transient elastography, were then analyzed. HIV positivity was associated with lower median IL-17A and higher IL-22 and sCD14 values. There were no statistically significant associations between HIV status, biomarkers or covariates with CAP measurement over two time points. However, IL-1β levels were associated with higher CAP scores at the second visit. Across all statistical models, an increase in BMI was associated with an increase in CAP measurements. No statistically significant associations were found between viral load history, CD4 + T-cell count, biomarkers and covariates, including ART use, on CAP measurements. These results confirm that BMI is a key risk factor for liver steatosis independent of HIV status. The potential contributions to NAFLD of differences in IL-1β, Th17-family cytokines and gut permeability between HIV-positive vs. negative individuals require further study.

1. Introduction

Non-alcoholic fatty liver disease (NAFLD), defined as the presence of hepatic steatosis in individuals who consume little or no alcohol [1], has become the most common chronic liver disease in the U.S. and many other parts of the world, with an incidence of over 25% in non– HIV-infected individuals and over 35% in HIV-infected patients [2]. The clinical presentation of NAFLD can range from a non-progressive form, characterized by simple hepatic steatosis, to progressive forms characterized by inflammation such as non-alcoholic steatohepatitis (NASH) and fibrosis [1,3]. In either the progressive or non-progressive form, for the general population the etiology of NAFLD is clearly related to metabolic risk factors such as obesity, insulin resistance and the metabolic syndrome [1,4,5]. However, for individuals living with HIV who have these risk factors, the prevalence of NAFLD appears to be higher than could be expected from those factors alone [2,6]. Hence, individuals living with HIV may have an additional risk that is tied to their HIV status [7]. There is limited information regarding the mechanisms that contribute to the higher prevalence and severity of NAFLD among HIV-infected patients as well as the role of gender, race and ethnicity on these risk factors [7,8]. Unfortunately, most of the available data of NAFLD in HIV-infected populations is disproportionately from studies of male subjects [2,9]. Considering that women and minorities living with HIV have a greater prevalence of obesity and are at greater risk to experience poorer outcomes and increased weight gain from HIV treatment [7,10], there is a clear need for studies of the natural history of NAFLD in these HIV-infected populations.

A growing amount of evidence has suggested that NAFLD has a strong genetic component, involving multiple genetic loci [7,8,11]. Among these are genes that regulate lipid metabolism, oxidative stress and inflammation [8,11]. These genes appear to modulate not only the susceptibility, but the severity and the progression of the disease as well [8,11]. Moreover, these seem to be also modulated by epigenetic factors [11]. While a “two-hit hypothesis” was used to explain the pathogenesis of NAFLD [1], it has now become evident that NAFLD is a multi-factorial disease, where “multiple hits” act together on genetically-predisposed subjects. Such hits may include insulin resistance, hormones secreted from the adipose tissue, oxidative stress, intestinal microbiota, inflammatory stimuli and genetic and epigenetic factors [1,10]. One explanation for the differences in prevalence between HIV-negative and positive individuals might be the impact that HIV infection has on several of these elements. For example, individuals living with HIV, even when taking anti-retroviral therapy (ART), have increased intestinal permeability with subsequent endotoxin leakage from the gut, which in turn promotes the activation of the innate immune system, leading to pro-inflammatory cytokine release [12,13]. HIV infection and ART are also known to alter adipose tissue biology, inducing alterations in the release of pro-inflammatory cytokines and adipokines, leading to basal systemic inflammation and further insulin resistance [14]. Moreover, inflammatory cytokines including the pro-fibrotic TGFβ, are potent inducers of the secretion by adipose tissue of PAI-1, a key component of the fibrinolytic system that has a pro-fibrotic effect [15,16].

The aim of this study was to determine the association, by HIV status, of a panel of biomarkers of inflammation, adipose tissue function, fibrinolysis, gut permeability and hepatocyte apoptosis/necrosis with changes in liver steatosis measures (CAP scores) over time [17] using data from participants in the WHIS Washington DC site.

2. Materials and methods

2.1. Ethics statement

These studies were approved by Institutional Review Boards from Georgetown University (1993-077) and the University of Louisville (12-043). All procedures were in accordance with ethical standards of the IRBs and the Helsinki Declaration of 1975, as revised in 2000. All participants provided written informed consent before enrollment in the study.

2.2. Study population

The Women’s Interagency HIV Study (WIHS) is a large, comprehensive prospective cohort study designed to investigate the progression of HIV disease in women and includes both HIV-positive and -negative women. WIHS methods and baseline cohort characteristics have been described previously [18,19]. Participants from the Washington, DC WIHS site with at least one valid liver steatosis measurement CAP from 2015 to 2018 were included in this study. These included a total of 105 women, of them 78 HIV-positive and 27 HIV–negative. Among the 105 participants, 82 (78.1%) contributed plasma samples and CAP measurements at two different visits. These visits are referred to as “participant-visit”. 23 participants (21.9%) provided only one sample and CAP measurement. Any subjects seropositive for HBV or HCV were excluded from this study.

2.3. HIV status

HIV status (HIV-positive/HIV-negative) was assessed using enzyme-linked immunosorbent assay (ELISA) with Western blot on all participants at their first visit and at every visit for those who were HIV negative. HIV-positive participants included all women who were positive at baseline and those who seroconverted during follow-up.

2.4. Covariates

Age was calculated in years as the difference between participant’s self-reported date of birth and first visit date and reported as a continuous variable in years. Race/ethnicity was reported at baseline and categorized as white, non-Hispanic, black, non-Hispanic, Hispanic, and Other (Asian/Pacific Islander/American Indian/Other). Body Mass Index (BMI) was reported as a continuous variable in kg/m². Alcohol use in a typical week was collected and categorized as None (0 drinks/week), Light/Moderate (1–7 drinks/week), Moderate/Heavy (8–12 drinks/week), and Heavy (13 + drinks/week). Follow-up time was calculated between the first and second visit and reported in months. Among HIV-positive participants, current viral load, CD4 + T-cell count and viral load history were obtained at both visits. HIV RNA plasma viral load values of <20 copies/mL were categorized as undetectable. CD4 + T-cell count (cell/mm³) was dichotomized into “greater than 350 cells/mm³ and “less than or equal to 350 cell/mm³. Viral load history was categorized into low, intermediate and high probability of viral load detection was determined using a group-based trajectory analysis done in previous research [20]. We also obtained information on the use of one or more the anti-retroviral medication classes: 1) Nucleoside Reverse Transcriptase Inhibitors (NRTIs); 2) Non-nucleoside Reverse Transcriptase Inhibitors (NNRTIs); 3) Protease Inhibitors (Pis); 4) Entry Inhibitors (EIs); 5) and Integrase Inhibitors (IIs), with most participants using more than a single ART class. Transient Elastography (FibroScan) was used to determine both liver stiffness and steatosis through CAP measurements. CAP measures the degree of ultrasound attenuation by hepatic fat at the central frequency of the FibroScan probe simultaneously with liver stiffness measurement [17]. The cut point for liver steatosis was 238 dB/m [21].

2.5. Biomarkers

Plasma samples were kept frozen at −70 °C until assay. The following measurements were performed on recently thawed and centrifuged plasma samples:

2.5.1. Biomarkers of inflammation

The concentrations of IL-1β, IL-6, TNFα, IL-10, IL-17A and IL-22 were measured using MSD V-Plex Proinflammatory Panel 1 and Th17 Panel 1 kits (Meso Scale Diagnostics, Rockville, MD). The concentration of total TGFβ1 was measured using a Duo Set ELISA kit following acid activation and neutralization (R&D Systems, Minneapolis, MN). The concentration of C-Reactive Protein (CRP) was measured using a Duo Set ELISA kit from R&D Systems, Minneapolis, MN.

2.5.2. Biomarkers of adipose tissue function

The concentrations of Adiponectin, Leptin, Resistin and total PAI-1 were measured using a HADCYMAG-61K Milliplex multiplex assay kit (Burlington, MA).

2.5.3. Fibrinolytic markers

The concentrations of Urokinase (uPA) and soluble Urokinase Plasminogen activator receptor (uPAR) were measured using Duo Set ELISA kits from R&D Systems, Minneapolis, MN.

2.5.4. Biomarkers of gut permeability and endotoxin exposure

The concentrations of soluble CD14 and Intestinal fatty-acid binding protein-B (I-FABP) were measured using Duo Set ELISA kits from R&D Systems, Minneapolis, MN.

2.5.5. Biomarkers of hepatocyte necrosis/apoptosis

The concentrations of cytokeratin-18 (CK18)-derived peptides M65 (a marker of total cell death, apoptosis and necrosis) and M30 (specific for apoptosis) were measuring using M65 ELISA CK18 and M30 Apoptosense CK18 kits from Diapharma, Cincinnati, OH).

2.6. Statistical analysis

Descriptive statistics were generated for the outcome: CAP measurement, inflammatory biomarkers and other covariates using frequencies and medians and interquartile ranges. With the exception of race/ethnicity and viral load history, descriptive statistics were generated using both time points. Differences in the descriptive statistics by HIV status were tested using X² test (categorical variables) and Wilcoxon test (continuous variables). Median and interquartile ranges by HIV Status and visit were also generated for each of the biomarkers. Differences in values at the first and last visit (within each HIV Status category) were tested using Wilcoxon test. Each biomarker was modeled separately using median regression using the SAS procedure PROC QUANTREG at the first and second visit to assess difference by HIV status, adjusting for age and BMI. Median regression was used because the biomarker values were not normally distributed. Next, generalized linear modeling with repeated measures (SAS procedure PROC GLM) was used to assess the impact of HIV status and each inflammatory biomarker separately on liver steatosis over time, adjusting for covariates and follow-up time. This method was repeated among HIV-positive participants, with the addition of CD4 + T-cell count and viral load history variables as covariates. Additionally, we modeled the association of liver steatosis (CAP score ≥ 238) and biomarkers using a repeated measures logistic model, adjusting for HIV status, age, and BMI.

Statistical significance was set at p<0.0029 to account for multiple comparisons (Bonferroni adjustment). All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, North Carolina, USA).

3. Results

3.1. Subject characteristics, measurements of HIV infection and liver steatosis.

The baseline characteristics of our sample are shown in Table 1. The follow-up time between first and second visits ranged from 12 to 32 months (average; 22.0 ± 5.9 months). The overall median age was 49.0 years (IQR: 44.0–54.0) with HIV-positive participants reporting a median age of 48.0 years (IQR: 44.0–55.0) and HIV-negative participants reporting a median age of 51.0 years (IQR: 43.0–57.0). The majority of participants were black, non-Hispanic (Overall: 83.8% HIV-positive: 84.6%; HIV-negative: 81.5%). The median BMI for all study participants was 29.4 kg/m² with HIV-negative participants reporting higher BMI than HIV-positive participants (33.5 kg/m² [IQR: 30.6–41.6] vs 28.5 kg/m² [IQR: 24.4–32.8]; p < 0.0001). 10.8% of the participants reported having diabetes (HIV-positive: 9.0% vs HIV-negative: 18.0%; p = 0.1092). Alcohol consumption in individuals living with HIV was relatively low, with 89.4% reporting none to only light/moderate alcohol consumption. 74.4% of the HIV- negative individuals also reported none to only light/moderate consumption. Among HIV positive participants, 74.8% had an undetectable viral load, 87.7% had a CD4 + T-cell count greater than 350 cells/mm³ and 88.5% had a history of low viral load detection. The distribution of anti-retroviral medication types was as follows: 85.5% use one or more NRTIs, 38.1%% used one or more NNRTIs, 30.3% used one or more Pis, 1.3% used one or more EIs and 43.9% used one or more IIs. The median liver CAP value was 231.0 dB/m (IQR: 206.0–237.0), with HIV-negative participants having a higher CAP value than HIV-positive participants (259.0 dB/m [IQR: 218.0–305.0] vs 229.0B/m [IQR: 205.0–263.0]; 0.0125) (Table 1). The prevalence of liver steatosis in HIV-positive and -negative participants were 40.0% and 59.0% (p = 0.0328), respectively. The median, interquartile ranges and p-values testing the difference in values (first visit vs last visit) for each inflammatory biomarker are reported by HIV status in Table 2.

Table 1.

Characteristics of Study Participants.

	HIV Positive	HIV negative	Total	p-value
Participants, N (%)	78 (74.3%)	27 (25.7%)	105	–
Participant visits, N (%)	155 (79.9%)	39 (20.1%)	194	–
Age (years)**, median (IQR)	48.0 (44.0–52.0)	51.0 (43.0–57.0)	49.0 (44.0–54.0)	0.1321
Race/Ethnicity*, n (%)
White, non-Hispanic	7 (9.0%)	1 (3.7%)	8 (7.6%)	0.0051
Black, non-Hispanic	66 (84.6%)	22 (81.5%)	88 (83.8%)
Hispanic	4 (5.1%)	2 (7.4%)	6 (5.7%)
Other	1 (1.3%)	2 (7.4%)	3 (2.9%)
BMI**, median (IQR) kg/m2	28.5 (24.4–32.8)	33.5 (30.6–41.6)	29.4 (25.6–33.9)	<0.0001
Diabetes**, n (%)
Yes	14 (9.0%)	7 (18.0%)	21 (10.8%)
No	141 (91.0%)	32 (82.1%)	173 (89.2%)
Alcohol Use**, n(%)
None	72 (45.5%)	9 (23.1%)	81 (41.8%)	0.0064
Light/Moderate	68 (43.9%)	20 (51.3%)	88 (45.4%)
Moderate/Heavy	5 (3.2%)	2 (5.1%)	7 (3.6%)
Heavy	10 (6.5%)	7 (18.0%)	17 (8.8%)
Missing	0 (0.0%)	1 (2.6%)	1 (0.5%)
Viral Load Detection**, n (%)
Yes	39 (25.2%)	–	39 (25.2%)	–
No	116 (74.8%)	–	116 (74.8%)
CD4 + T-cell Count**, n (%)
Greater than 350 cell/ul	136 (87.7%)	–	136 (87.7%)	–
Less than or equal to 350 cell/ul	19 (12.3%)	–	19 (12.3%)
Use of Nucleoside Reverse Transcriptase Inhibitors**, n (%)
Yes	133 (85.8%)	–	133 (85.8%)	–
No	22 (14.2%)	–	22 (14.2%)
Use of Non-nucleoside Reverse Transcriptase Inhibitors**, n (%)
Yes	59 (38.1%)	–	59 (38.1%)	–
No	96 (61.9%)	–	96 (61.9%)
Use of Protease inhibitor**, n (%)
Yes	47 (30.3%)	–	47 (30.3%)	–
No	108 (69.7%)	–	108 (69.7%)
Use of Entry Inhibitors**, n (%)
Yes	2 (1.3%)	–	2 (1.3%)	–
No	153 (98.7%)	–	153 (98.7%)
Use of Integrase Inhibitors**, n (%)
Yes	68 (43.9%)	–	68 (43.9%)	–
No	87 (56.1%)	–	87 (56.1%)
Viral Load Trajectory History*, n (%)
High Viremia	0 (0.0%)	–	0 (0.0%)	–
Intermediate Viremia	9 (11.5%)	–	9 (11.5%)
Low viremia	69 (88.5%)	–	69 (88.5%)
CAP Measurement**, median (IQR) dB/m	229.0 (205.0–263.0)	259.0 (218.0–305.0)	231.0 (206.0–237.0)	0.0125
Prevalence of Liver Steatosis (≥238 dB/m) **, n(%)	62 (40.0%)	23 (59.0%)	85 (43.8%)	0.0328

IQR - Interquartile Range;

^*By participant;

^**At both time points;

p-values were not generated for HIV-positive specific variables

Table 2.

Median and Interquartile Range of Biomarkers and CAP Measurements.

	HIV-Positive			HIV-Negative
	First Visit	Second Visit	p-value	First Visit	Second Visit	p-value
Inflammatory markers
IL-1β (pg/mL)	0.2 (0.1–0.3)	0.2 (0.1–0.3)	0.3150	0.3 (0.1–0.5)	0.3 (0.1–0.5)	0.5093
IL-6 (pg/mL)	0.6 (0.5–0.9)	0.7 (0.4–1.1)	0.9002	0.9 (0.5–1.5)	0.8 (0.5–1.1)	0.7827
TNFα (pg/mL)	1 (0.7–1.4)	0.9 (0.7–1.3)	0.5537	1.2 (0.9–1.6)	1.4 (0.9–1.7)	0.7534
IL-10 (pg/mL)	0.1 (0.1–0.2)	0.1 (0.1–0.1)	0.1096	0.2 (0.1–0.3)	0.2 (0.1–0.2)	0.7261
IL-17A (pg/mL)	0.2 (0.1–0.4)	0.2 (0.1–0.3)	0.3251	0.4 (0.3–0.8)	0.5 (0.3–0.7)	0.8431
IL-22 (pg/mL)	0.4 (0.3–0.7)	0.4 (0.2–0.6)	0.1394	0.1 (0–0.8)	0.2 (0–0.6)	0.7317
TGFβ1 (pg/mL)	7.2 (3.4–10.7)	8.1 (3.8–11.6)	0.5597	10.9 (8.2–14.5)	13.3 (11.5–19.7)	0.0552
CRP (μg/mL)	1.4 (0.4–2.8)	1.5 (0.6–4.1)	0.4285	2 (0.9–3.6)	1.6 (1.1–2.4)	0.5300
Adipose tissue markers
Leptin (ng/mL)	2.5 (0.7–4.1)	2.4 (0.6–3.9)	0.9841	3.3 (2.2–3.9)	4.0 (2.1–5.1)	0.5228
Adiponectin (ng/mL)	2.1 (1.5–3)	2 (1.4–2.9)	0.9032	1.8 (1.3–2.6)	2.1 (1.6–2.5)	0.4113
Resistin (ng/mL)	10.9 (8.3–13.8)	10.7 (7.6–13.9)	0.9324	11.3 (6.9–14.7)	11.5 (9.9–14.9)	0.6701
PAI-1 (ng/mL)	5.4 (3.4–7.5)	5.8 (4.1–7.7)	0.2502	7.8 (5.2–12.9)	11.3 (6.5–18.2)	0.2856
Fibrinolytic System markers
uPA (pg/mL)	244 (86–389)	252 (104–398)	0.8136	212 (142–362)	243 (68–503)	1.0000
uPAR (ng/mL)	1.4 (1.0–1.8)	1.3 (1.1–1.7)	0.7395	1.6 (1.0–2.1)	1.3 (0.9–2.3)	0.4745
Gut permeability markers
I-FABP (pg/mL)	879 (589–1496)	912 (538–1327)	0.5089	588 (359–773)	516 (325–786)	0.6924
sCD14 (ng/mL)	1.1 (0.8–1.5)	1.1 (0.7–1.4)	0.3363	0.6 (0.6–0.9)	0.6 (0.6–0.7)	0.9273
Hepatocyte necrosis – apoptosis markers
M65 (U/L)	60 (30.8–102.1)	65.8 (29.5–123.4)	0.5626	41.4 (22.9–134.8)	81.4 (30–142.5)	0.4946
M30 (U/L)	16.4 (8.9–29.7)	15.6 (8.5–29.4)	0.7230	13 (3–38)	12.5 (8–44.8)	0.9393
M30/M65 ratio	0.3 (0.2–0.5)	0.3 (0.1–0.5)	0.4392	0.2 (0.1–0.5)	0.2 (0.2–0.4)	0.8966
CAP Measurement (dB/m)	230 (195–269)	229 (209–258)	0.8805	269 (218–305)	239 (204–317)	0.7263

Of a total of 105 participants, 82 contributed samples and data for two visits, 23 for a single one. “Participant-visits” were used to analyze the results as defined under “Materials and Methods”.

3.2. Association of HIV status with biomarkers at first and second visit

At the first visit, being HIV positive was associated with a lower median IL-17A value (β: −0.2593 95% CI −0.409 to −0.110; p = 0.0008), and higher median IL-22 (β: 0.324; 95% CI 0.118 to 0.530; p = 0.0024) and median sCD14 values (β: 0.479; 95% CI 0.316 to 0.641; p < 0.0001). There were no effects of age and BMI on the median biomarker values at the first visit. At the second visit, being HIV positive was again associated with a higher median sCD14 value (β: 0.412; 95% CI 0.173 to 0.652; p = 0.0009) and a lower median IL-17A value (β: −0.333; 95% CI −0.485 to −0.180; p < 0.0001). There were no effects of age and BMI on the median biomarker values at the second visit.

3.3. Association of CAP measures with HIV status and biomarkers over time

Beta estimates for the association of CAP measure and biomarkers over time are presented in Table 3. There were no statistically significant association of HIV status, the biomarkers and covariates with CAP measurement over time. However, the biomarker IL-1β was positively associated with a higher CAP measurement (β: 18.79; 95% CI 4.81 to 32.78; p = 0.0091) in the second visit among all participants, regardless of HIV status. Across all statistical models, the BMI was positively associated with higher CAP measurement at the first (Median β value: 5.77; IQR β:5.52 to 5.86) and second (Median β value: 5.00; IQR β: 4.9 to 5.25) visits. There was no statistically significant effect of follow-up time on CAP measurement.

Table 3.

Beta Estimates (95% CI) of Biomarkers at the First and Second Visit.

	Overall*		HIV-positive only**
	CAP Measurement at First Visit	CAP Measurement at Second Visit	CAP Measurement at First Visit	CAP Measurement at Second Visit
Inflammatory markers
IL-1β (pg/mL)	−19.69 (−42.2–2.81)	18.79 (4.81–32.78)	−2.57 (−36.78–31.64)	15.15 (−3.62–33.92)
IL-6 (pg/mL)	−0.94 (−17.21–15.33)	7.29 (−9.77–24.36)	−14.07 (−31.24–3.1)	8.78 (−9.73–27.29)
TNFα (pg/mL)	−13.47 (−28.3–1.36)	7.85 (−9.77–25.48)	−10.63 (−26.46–5.2)	5.93 (−13.82–25.68)
IL-10 (pg/mL)	44.81 (−40.63–130.24)	−41.72 (−168.88–85.43)	47.18 (−54.19–148.56)	−70.21 (−206.17–65.75)
IL-17A (pg/mL)	−11.86 (−35.41–11.69)	5.4 (−20.7–31.5)	−7.24 (−30.85–16.37)	7.8 (−22.53–38.14)
IL-22 (pg/mL)	−4.62 (−24.94–15.69)	−5.72 (−30.64–19.19)	−21.21 (−42.37–0.05)	−5.89 (−31.78–19.99)
TGFβ1 (pg/mL)	1.11 (−2.23–4.45)	−0.62 (−3.62–2.38)	1.54 (−1.91–4.99)	−1.52 (−4.79–1.75)
CRP (μg/mL)	−3.7 (−7.8–0.4)	0.62 (−3.15–4.4)	−3.44 (−7.53–0.64)	1.09 (−3.27–5.45)
Adipose tissue markers
Leptin (ng/mL)	0 (−0.01–0.01)	0 (−0.01–0)	0 (−0.01–0.01)	0 (−0.01–0.01)
Adiponectin (ng/mL)	−11.46 (−26.78–3.85)	5.3 (−8.45–19.06)	−9.11 (−24.27–6.05)	3.88 (−9.71–17.47)
Resistin (ng/mL)	0 (0–0)	0 (0–0)	0 (0–0.01)	0 (−0.01–0)
PAI-1 (ng/mL)	0 (0–0)	0.01 (−0.04–0.05)	0 (0–0.01)	0 (0–0.01)
Fibrinolytic System markers
uPA (pg/mL)	−0.01 (−0.07–0.04)	−0.02 (−0.06–0.03)	−0.02 (−0.07–0.04)	−0.02 (−0.07–0.03)
uPAR (ng/mL)	0 (−0.02–0.01)	−0.01 (−0.03–0.01)	0 (−0.02–0.02)	−0.01 (−0.03–0.01)
Gut permeability markers
I-FABP (pg/mL)	0 (−0.01–0.02)	0 (−0.01–0.01)	0 (−0.02–0.02)	0 (−0.01–0.02)
sCD14 (ng/mL)	−5.39 (−28.83–18.05)	−28.33 (−50.64–6.01)	−5.1 (−28.22–18.02)	−21.61 (−44.6–1.38)
Hepatocyte necrosis – apoptosis markers
M65 (U/L)	−0.02 (−0.11–0.06)	−0.05 (−0.18–0.07)	−0.02 (−0.1–0.06)	−0.05 (−0.17–0.08)
M30 (U/L)	−0.03 (−0.23–0.18)	0.01 (−0.27–0.29)	−0.03 (−0.22–0.17)	−0.03 (−0.33–0.27)
M30/M65 ratio	38.64 (−6.9–84.17)	−3.27 (−45.38–38.84)	41.42 (−7.69–90.54)	−13.33 (−58.63–31.96)

^*Multivariate results: adjusted for HIV status, age, BMI and follow-up;

^**Multivariate results: adjusted for HIV status, age, BMI, viral load history and CD4 count and follow-up

3.4. Association of viral and immune factors with liver steatosis among HIV-positive participants

Beta estimates for the association of CAP measure and biomarkers among HIV-positive participants over time are presented in Table 3. There were no statistically significant associations of viral load history, CD4 + T-cell count, biomarkers and covariates on CAP measurement over time in the HIV-positive only analyses. There was no statistically significant effect of follow-up time on CAP measurement in the HIV-positive only analyses. When the potential effect of ART on liver steatosis was investigated, addition of ART classes to our model did not change the association between BMI and CAP and the Biomarkers and CAP. Among the different classes of ART, II was the only class that was associated with CAP measurements, but only at time point 1.

3.5. Risk factors of steatosis

In order to explore risk factors of steatosis in women, we modeled the association of liver steatosis (CAP score ≥ 238) and biomarkers using a repeated measures logistic model, adjusting for HIV status, age, and BMI. In the adjusted model, only BMI was associated with liver steatosis. For every unit increase in BMI was associated with 24% increase in odds of liver steatosis (aOR: 1.24; 95% CI: 1.13–1.37).

4. Discussion

In this study, HIV status was associated with lower levels of IL-17A and higher levels of IL-22 and sCD14, a marker related to increased gut permeability and endotoxin exposure. BMI emerged as the major risk factor for NAFLD, regardless of HIV status, with positive association with CAP measurements. With the exception of IL-1β, no statistically significant associations with CAP were found for the other biomarkers. Such a finding is provocative and may suggest a potential involvement of an inflammatory component in the risk of NAFLD irrespective of HIV status.

Microbial translocation from the gut has been described as an underlying cause of immune activation and a predictor of disease progression among HIV-infected patients [22–26]. Even in the presence of treatment with combination ART, intestinal barrier deficits and higher levels of bacterial translocation persist when compared to non-HIV controls [12,27,28]. The resulting immune activation and secretion of pro-inflammatory cytokines can lead to heightened systemic inflammation status seen in these patients. Because of the portal circulation, leakage of bacteria and/or endotoxin would also cause a significant pro-inflammatory environment in the liver, potentially contributing to the enhanced NAFLD risk and liver disease progression in HIV-infected patients. Based on such evidence, we hypothesize that there would be increased markers of intestinal damage and endotoxin exposure-mediated immune activation in HIV-infected individuals compared to non-infected controls. While we found elevated levels of sCD14, a marker of endotoxin exposure and immune activation [23] among HIV-infected group compared to control, other inflammatory cytokines or markers (e.g., IL-1β, IL-6, TNFα, IL-10, CRP) were not significantly different and no significant association was found with CAP.

Unlike the pro-inflammatory cytokines, our studies found an association of HIV-status with reduced IL-17A and increased IL-22. The reduction in the levels of IL-17A agrees with evidence that HIV infection has significant effects on Th17 CD4 + T cells, particularly at the level of the intestinal mucosa [29,30]. Deficits in Th17-producing cells and IL-17 levels have been reported to persist, even after ART treatment [30]. Thus, increased sCD14 and reduced IL-17A levels are consistent with a model involving compromised intestinal integrity and endotoxin exposure, which in turn may increase liver inflammation and promote steatosis.

Why there are increased IL-22 levels in the HIV-group remains unclear. In general IL-22 has beneficial effects of mucosal barriers, although increased levels have also been also associated with certain inflammatory conditions [31]. It is important to point out that IL-22 is not only made by Th17 cells, but it is also made by a variety of cells including Th22 CD4 + T cells as well as several other cell populations [32], so it is not surprising that IL-22 levels did not parallel those of IL-17 in this study. It is possible that increased IL-22 levels may represent an attempt by the immune system to repair the impaired intestinal mucosa.

Since most of the participants provided samples obtained at two different visits (separated by a range of six months to two years), we explored the effect of changes in biomarkers over time with CAP measurements. The only biomarker found to have a statistically significant effect on CAP measurement over time was IL-1β levels at the second visit. While the significance of this finding needs further investigation, IL-1β is related to inflammatory processes, suggesting that underlying systemic inflammation might play a role in the increased risk of steatosis, regardless of HIV status. It was a little surprising that no statistically significant association was found with IL-6, another pro-inflammatory cytokine. In contrast to IL-1β, IL-6 is known to be significantly expressed in adipose tissue [33], and thus its levels might have been more heavily influenced by the BMI.

The use of ART may also be one of the factors involved in increased BMI and liver steatosis. Our observations of an association of the use of II with CAP measurements, even if only at time point 1, are in line with reports of the association of II-based ART regimens with increased weight gain [34] and hepatic steatosis [35] in HIV-positive patients.

In conclusion, insulin resistance, type 2 diabetes, obesity and dyslipidemia are key risk factors associated with NAFLD and NASH [3,5,6]. Thus, it was not surprising that the BMI was by far, the factor most clearly associated with CAP measurement, independent of HIV-status. Further exploration is needed to determine if the altered intestinal permeability and underlying inflammation suggested by our studies play a role in the increased risk of NAFLD in individuals living with HIV.

One of the main limitations of our studies was that the participants were chosen based on the availability of CAP measurements and samples from two visits, without any previous knowledge of their BMI or Fibroscan data. Following analysis, the HIV-negative participants turned out to have had higher median BMI and CAP measurements compared to the HIV-positive participants. These findings are in line with a previous study reporting a lower liver fat fraction (measured by MRI and spectroscopy) in HIV-positive patients compared to uninfected women, potentially the result of differences in the way fat is stored in the liver in order to maintain subcutaneous fat in HIV-positive women [36]. However, the strong association of BMI with CAP measurements and the potential effect of BMI on several biomarkers makes comparisons between HIV -positive and -negative groups not straightforward. In fact, the higher BMI in the control group might explain the relatively higher levels of pro-inflammatory markers seen in this group, even though differences with the HIV-positive group were not statistically significant. Finally, this study was the relatively small sample size, making it underpowered to detect differences between the groups.

In conclusion, our studies confirmed that the B MI is a key risk factor for liver steatosis independent of the HIV status. While comparison of HIV-positive and –negative subject groups suggested differences in Th17-family cytokines and in intestinal barrier permeability, the contribution of these factors to the risk of NAFLD remains unclear.

Abbreviations:

ART

Anti-retroviral therapy

BMI

Body-mass Index

CAP

Controlled attenuation parameter

CRP

C-Reactive Protein

EI

Entry inhibitors

HIV

Human Immunodeficiency virus

I-FABP-B

Intestinal fatty acid binding protein-B

II

Integrase inhibitors

NAFLD

Non-Alcoholic Fatty Liver Disease

NNRTI

Non-nucleoside reverse transcriptase inhibitors

NRTI

Nucleoside reverse transcriptase inhibitors

PAI-1

Plasminogen Activator Inhibitor-1

PI

Protease inhibitors

uPA

Urokinase Plasminogen Activator

uPAR

Urokinase Plasminogen Activator Receptor

WHIS

Women’s Interagency HIV Study