Prevalence and Risk Factors of Moderate-to-Severe Hepatic Steatosis in Human Immunodeficiency Virus Infection: The Copenhagen Co-morbidity Liver Study

Abstract

Background

People with human immunodeficiency virus (PWH) may be at risk of nonalcoholic fatty liver disease. We compared the prevalence of moderate-to-severe hepatic steatosis (M-HS) in PWH with human immunodeficiency virus (HIV)-uninfected controls and determined risk factors for M-HS in PWH.

Methods

The Copenhagen Co-Morbidity in HIV Infection study included 453 participants, and the Copenhagen General Population Study included 765 participants. None had prior or current viral hepatitis or excessive alcohol intake. Moderate-to-severe hepatic steatosis was assessed by unenhanced computed tomography liver scan defined by liver attenuation ≤48 Hounsfield units. Adjusted odds ratios (aORs) were computed by adjusted logistic regression.

Results

The prevalence of M-HS was lower in PWH compared with uninfected controls (8.6% vs 14.2%, P < .01). In multivariable analyses, HIV (aOR, 0.44; P < .01), female sex (aOR, 0.08; P = .03), physical activity level (aOR, 0.09; very active vs inactive; P < .01), and alcohol (aOR, 0.89 per unit/week; P = .02) were protective factors, whereas body mass index (BMI) (aOR, 1.58 per 1 kg/m²; P < .01), alanine transaminase (ALT) (aOR, 1.76 per 10 U/L; P < .01), and exposure to integrase inhibitors (aOR, 1.28 per year; P = .02) were associated with higher odds of M-HS.

Conclusions

Moderate-to-severe hepatic steatosis is less common in PWH compared with demographically comparable uninfected controls. Besides BMI and ALT, integrase inhibitor exposure was associated with higher prevalence of steatosis in PWH.

comorbidity, fatty liver disease, human immunodeficiency virus, NAFL, NAFLD

In the Western World, nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease in adults with an estimated overall prevalence of 25% [1]. A high prevalence of NAFLD has been reported for people with human immunodeficiency virus (PWH), but with a wide range from 13% to 73% due to substantial differences in study populations and diagnostic methods used [2–6]. A recent meta-analysis found a prevalence of NAFLD in PWH without viral hepatitis of 35% based on imaging procedures [7].

Nonalcoholic fatty liver disease covers a wide spectrum of liver disease from hepatic steatosis with accumulation of fat within the hepatocytes to nonalcoholic steatohepatitis (NASH) with additional inflammation and injury of the hepatocytes to liver cirrhosis, liver failure, and hepatocellular carcinoma. Hepatic steatosis has been considered a benign condition, but a recent study of HIV-uninfected individuals with serial liver biopsies showed that 44% of individuals with baseline hepatic steatosis progressed to NASH and 22% progressed to advanced fibrosis [8]. Because liver fibrosis is the only histological feature of long-term prognosis in NAFLD [9], this is of major concern, and patients at risk of progression to NASH and liver fibrosis should be identified to prevent disease progression. Risk factors for NAFLD in HIV infection have differed in previous studies. A meta-analysis found that an increase in body mass index (BMI), waist circumference, type 2 diabetes mellitus, hypertension, and high levels of total cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides, fasting glucose, alanine transaminase (ALT), aspartate transaminase (AST), and CD4⁺ T-cell count were all associated with higher odds of NAFLD [7]. Furthermore, antiretroviral treatment (ART) may contribute to the development of NAFLD due to adverse metabolic effects with mitochondrial dysfunction [10–12].

The aim of this study was to determine whether the prevalence of hepatic steatosis was different between PWH and matched HIV-uninfected individuals. We hypothesized that PWH had a higher prevalence of hepatic steatosis compared with HIV-uninfected individuals. Factors associated with hepatic steatosis were assessed in PWH, and the influence of HIV infection was evaluated.

METHODS

Study Populations

The Copenhagen Co-Morbidity in HIV Infection (COCOMO) study has been described in detail elsewhere [13, 14]. In short, the COCOMO study is an observational, longitudinal cohort study designed to estimate prevalence and incidence of nonacquired immune deficiency syndrome comorbidity in PWH living in Copenhagen, Denmark. Adult PWH were recruited consecutively from the outpatient clinics of the Departments of Infectious Diseases at Rigshospitalet and Amager Hvidovre Hospital in Copenhagen, Denmark from March 2015 through November 2016. The comparator group was retrieved from the Copenhagen General Population Study (CGPS), a prospective cohort study of >100 000 randomly selected adult individuals from the area of Copenhagen initiated in 2003 [15–17]. The comparator group was enrolled from March 2011 through April 2014, except for 20 participants enrolled from February 2004 to September 2008. The comparator group was assumed to be HIV uninfected because the prevalence of HIV infection was estimated to be 0.1% in the Danish, adult population in 2016 [18].

Data Collection

The data collection has been described in detail elsewhere [13]. In short, comprehensive questionnaires were completed comprising >100 items with information on current health, dietary habits, and lifestyle. Information on food and beverages was collected by a semiquantitative food frequency questionnaire. Study participants were instructed to report current food and beverage intake. Information on alcoholic beverages included frequency information reported as “never or almost never”, “few times a month”, “few times a week”, or “daily or almost daily” and a quantitative information reported as “average number of alcoholic units per week.” Information on previous alcohol intake was not collected. Frequency information on current dietary habits was collected and reported as “times per week.” Human immunodeficiency virus-specific information and status of hepatitis B and C coinfection was retrieved from medical records. Data were >95% complete unless otherwise stated. All data were collected uniformly in the COCOMO study cohort and the CGPS study cohort with identical questionnaires, laboratory equipment, and physical examination techniques.

Computed Tomography Scan of Upper Abdomen

A computed tomography (CT) scan of the upper abdomen was performed on a Aquillion One scanner (Toshiba Medical Systems, Otawara-shi, Tochigi-ken, Japan) using identical scan protocols for the 2 cohorts [13,19]. Liver attenuation was measured for all CT scans using Vitrea 3.1 imaging software (Vital Images Inc., Minnetonka, MN). A region of interest (ROI) with an area of 1500 mm² (±100 mm²) was placed in Coinaud liver segments 5 and 6. The average liver attenuation was calculated from the 2 ROIs, and results were presented in Hounsfield Units (HUs). All analyses were performed by trained physicians blinded to clinical and biochemical details of the study participants. A pilot study of 20 participants demonstrated a high interrater correlation (R² = 0.98 and Spearman rho = 0.99) with no bias.

All participants in the COCOMO study were invited to a CT scan; 921 participants (84%) attended. Participants from the CGPS aged 40 years or above were randomly invited to a CT scan; 70% accepted the invitation [20].

Definitions of Outcome

The physiologic attenuation of the liver parenchyma ranges from 55 to 65 HUs by unenhanced CT of the liver [21]. Liver attenuation is inversely correlated with liver fat content, yielding lower HUs with increasing amounts of hepatic steatosis. In this study we defined moderate-to-severe hepatic steatosis as a CT liver attenuation <48 HUs with a specificity of 100%, sensitivity of 53.8%, positive predictive value of 100%, and negative predictive value of 93.9% [22]. Sensitivity analyses were conducted to test a threshold of CT liver attenuation ≤40 HUs, which has been used to exclude mild hepatic steatosis in previous literature [23, 24].

Ethics

The study was approved by the regional ethics committee of the Capital Region of Denmark (H-15 017 350; H-KF-01-144/01) and conducted in accordance with the declaration of Helsinki. All participants provided informed consent. The study has been registered at ClinicalTrials.gov (NTC02382822).

Statistical Analyses

People with human immunodeficiency virus and uninfected controls with a CT scan of the abdomen aged 40 years or older were matched on sex and 5-year age strata in a ratio of 1:2, except that men aged 40–55 were matched 1:1 due to availability (Supplementary Figure S1). Baseline clinical and demographic data of the 2 cohorts were compared by Fisher’s exact test and χ ² test (categorical variables), and Kruskal-Wallis and Mann-Whitney’s U test (continuous variables). Univariable and multivariable logistic regression models were conducted in PWH with moderate-to-severe hepatic steatosis as outcome. Two multivariable regression models were constructed with a priori selection of independent variables. Both models were adjusted for age (per decade), sex (female vs male), and Caucasian ethnicity (no vs yes). The metabolic model was further adjusted for the following: body mass index ([BMI] per 1 kg/m²), plasma total cholesterol (per 1 mM), plasma triglycerides (per 1 mM), diabetes (yes vs no), plasma glucose (per 1 mM), and plasma alanine aminotransferase ([ALT] per 10 IU/L). The lifestyle model was further adjusted by the following: smoking status (never smoker, current smoker, previous smoker), alcohol consumption (per 1 unit per week), and level of physical activity (inactive, moderate inactive, moderate active, very active). The association between a positive HIV status and moderate-to-severe hepatic steatosis was estimated in the total population. The association between HIV-specific variables including ART drug classes were estimated in univariable analyses and multivariable analyses after adjustment for sex, age, ethnicity, BMI, and duration of HIV infection. Results are presented as crude and adjusted odds ratios (aORs) with 95% confidence intervals (CIs). A P < .05 was considered statistically significant. The interaction between BMI (>25 kg/m²) and HIV status was tested to determine whether HIV modifies the effect of BMI on moderate-to-severe hepatic steatosis. We defined hepatitis B virus infection (HBV) as presence of hepatitis B surface antigen, hepatitis C virus infection (HCV) as presence of anti-HCV antibodies (anti-HCV), excessive alcohol intake as an average consumption of >14 alcoholic units per week for men and >7 alcoholic units per week for women, abdominal obesity as a waist-to-hip ratio of ≥0.90 for men and ≥0.85 for women according to the International Diabetes Federation [25], and metabolic syndrome as a minimum of 3 of the following 5 items: (1) waist circumference waist circumference of ≥94 cm for men and ≥80 cm for women; (2) systolic blood pressure ≥130 mmHg and/or antihypertensive treatment; (3) plasma HDL ≤1.036 mmol/L for men, and plasma HDL ≤1.295 mmol/L for women; (4) plasma triglycerides ≥1.693 mmol/L; and (5) self-reported diabetes mellitus and/or antidiabetic treatment and/or nonfasting plasma glucose ≥11.1 mmol/L [25]. All analyses were conducted in R version 3.4.1.

RESULTS

A total of 1099 participants were included in the COCOMO study. Participants were excluded due to age below 40 years (n = 191), CT scan unavailability (n = 143), HBV (n = 23), HCV (n = 82), excessive alcohol consumption (n = 174), or missing information on these parameters (n = 52). The final study population comprised 453 PWH. A total of 1192 participants from the CGPS were selected for the comparator group; participants were excluded due to excessive alcohol consumption (n = 415). The final control population comprised 765 individuals (Supplementary Figure S2).

Clinical and Demographic Characteristics

Clinical and demographic characteristics of PWH and uninfected controls are depicted in Table 1, and HIV-specific characteristics of PWH are in Table 2. In short, PWH were more likely males (86% vs 82%), of non-Scandinavian descent (25 vs 4%), with lower BMI (25 vs 26 kg/m²), less alcohol use (48 vs 72 grams/week), and higher physical activity level and educational level. The majority of PWH acquired HIV through sex between men (71%), received ART (99%), and were well treated with HIV ribonucleic acid (RNA) <50 copies/mL (97%) and a median CD4 T-cell count of 690 cells/µL (interquartile range [IQR], 520–884). Clinical and demographic characteristics stratified by presence of moderate-to-severe hepatic steatosis can be found in Supplementary Tables S2 and S3.

Table 1.

Clinical and Demographic Characteristics of People With HIV and Population Controls^a

Characteristics	PWH (n = 453)	Controls (n = 765)	P Value
Age (years), median (IQR)	52.4 (46.8–61.0)	53.4 (47.7–61.4)	.16
Sex (male), n (%)	388 (85.7)	625 (81.7)	.09
Ancestry, n (%)			<.01
Scandinavian	338 (75.6)	728 (96.0)
Other European	52 (11.6)	28 (3.7)
Middle East and Indian Subcontinent	4 (0.9)	0 (0.0)
Other	53 (11.9)	2 (0.3)
Educational level, n (%)			<.01
None	50 (11.5)	91 (11.9)
Short	104 (24.0)	209 (27.4)
Middle Length	177 (40.9)	425 (55.7)
University	102 (23.6)	38 (5.0)
Smoking, n (%)			<.01
Current smoker	116 (25.6)	72 (9.4)
Previous smoker	173 (38.2)	331 (43.3)
Never smoker	164 (36.2)	359 (46.9)
Alcohol (grams/week), median (IQR)	48.0 (0–108)	72.0 (36–108)	<.01
Physical activity, n (%)
Inactive	39 (8.9)	49 (6.4)
Moderate inactive	147 (33.5)	250 (32.8)
Moderate active	194 (44.2)	385 (50.5)
Very active	59 (13.4)	78 (10.2)	.07
Abdominal obesity, n (%)	314 (71.5)	469 (61.5)	<.01
Waist circumference (cm), median (IQR)	94.0 (87.0–104.0)	93.0 (86.0–101.0)	.04
Body mass index (kg/m2), median (IQR)	24.7 (22.4–27.5)	26.0 (23.7–28.4)	<.01
WHO BMI category, n (%)
Underweight, <18.4 kg/m2	10 (2.2)	2 (0.3)
Normal weight, 18.5–24.9 kg/m2	229 (50.8)	287 (37.5)
Overweight, 25–29.9 kg/m2	158 (35.0)	349 (45.6)
Obese ≥30 kg/m2	54 (12.0)	127 (16.6)	<.01
Diabetes, n (%)	33 (7.3)	31 (4.1)	.02
Metabolic syndrome, n (%)	183 (43.4)	261 (35.1)	.01
Lipid lowering treatment, n (%)	80 (18.2)	88 (11.5)
Antidiabetic treatment, n (%)	25 (5.5)	26 (3.4)	.10
Biochemistry, median (IQR)
Plasma ALT (IU/L)	26 (20–34)	22 (17–29)	<.01
Plasma total cholesterol (mM)	4.9 (4.2–5.7)	5.4 (4.8–6.1)	<.01
Plasma triglycerides (mM)	1.8 (1.3–2.8)	1.5 (1.0–2.2)	<.01
Plasma LDL (mM)	2.8 (2.2–3.5)	3.2 (2.7–3.9)	<.01

Abbreviations: ALT, alanine aminotransferase; BMI, body mass index; HIV, human immunodeficiency virus; IQR, interquartile range; LDL, low-density lipoprotein; PWH, people with HIV; WHO, World Health Organization.

^aMissing variables for Copenhagen Co-Morbidity in HIV Infection (COCOMO) study (Copenhagen General Population Study): Ancestry: 6 (7); educational level: 20 (2); physical activity: 14 (3); abdominal obesity: 14 (2); waist circumference: 14 (2); BMI 2 (0); metabolic syndrome 31 (21); lipid-lowering treatment 13 (1); ALT 30 (9); cholesterol 21 (9).

Table 2.

Characteristics of People With HIV (n = 453)

HIV-specific characteristics	PWH (n = 453)
Route of HIV Transmission, n (%)
MSM	316 (70.7)
HSX	101 (22.6)
IDU	2 (0.4)
Other	28 (6.3)
Blood CD4 T-cell count (cells/µL), median (IQR)	690 (520–884)
<200	4 (0.9)
200–349	25 (5.6)
350–500	71 (15.8)
>500	349 (77.7)
Blood CD4 nadir T-cell count (cells/ µL), median (IQR)	220 (110–320)
Plasma HIV RNA ≥50 copies/mL, n (%)	14 (3.1)
Duration of HIV infection (years), median (IQR)	16.0 (8.3–23.1)
cART, n (%)	445 (98.9)
ART exposure, n (%)
NRTI	441 (97.4)
NNRTI	353 (77.9)
Integrase inhibitors	141 (31.1)
Protease inhibitors	258 (57.0)
Didanosine	76 (16.8)
Thymidine analog	261 (57.6)
Duration of ART exposure (years), median (IQR)
NRTI	15.1 (7.2–22.4)
NNRTI	7.4 (3.6–11.5)
Integrase inhibitors	1.9 (0.9–5.2)
Protease inhibitors	10.6 (4.8–19.2)
Didanosine	2.5 (0.8–5.8)
Thymidine analog	6.2 (3.5–9.1)

Abbreviations: ART, antiretroviral therapy; cART, combination antiretroviral therapy; HIV, human immunodeficiency virus; HSX, heterosexual; IDU, injection drug use; IQR, interquartile range; MSM, men who have sex with men; NNRTI, nonnucleoside reverse-transcriptase inhibitor; NRTI, nucleoside reverse-transcriptase inhibitor; RNA, ribonucleic acid.

Prevalence of Moderate-To-Severe Hepatic Steatosis in People With Human Immunodeficiency Virus and Uninfected Controls

Thirty-nine (8.6%; 95% CI, 6.4%–11.6%) PWH had CT-defined moderate-to-severe hepatic steatosis compared with 109 (14.2%; 95% CI, 11.9%–16.9%) HIV-uninfected controls (P < .001). The distribution of liver attenuation in PWH and uninfected controls are depicted in the Supplementary Figure S3. The median CT liver attenuation was comparable in PWH and controls (61.3 HUs [IQR, 56.5–65.6] vs 61.6 HU [IQR, 53.9–66.1]; P = .56).

Human Immunodeficiency Virus Infection and Moderate-To-Severe Hepatic Steatosis

Compared with controls, PWH had lower odds of moderate-to-severe hepatic steatosis in unadjusted and adjusted analyses (Figure 1). The association between BMI and moderate-to-severe hepatic steatosis was not modified by HIV status (P = .91 for interaction). In PWH, neither current CD4 T-cell count, nadir CD4 T-cell count <200 cells/µL, plasma HIV RNA ≥50 copies/mL, nor duration of HIV infection were associated with moderate-to-severe hepatic steatosis (Supplementary Table S3). No association was found between moderate-to-severe hepatic steatosis and exposure to nucleoside reverse-transcriptase inhibitors, nonnucleoside reverse-transcriptase inhibitors, integrase inhibitors, protease inhibitors, didanosine, or thymidine analogs (stavudine and zidovudine) (Supplementary Table S3). However, the cumulative duration of exposure to an integrase inhibitor was associated with higher odds of moderate-to-severe hepatic steatosis in univariate analyses (OR = 1.19 [95% CI, 1.02–1.39], per year; P = .02), and the association increased after adjustment for age, sex, BMI, and duration of HIV infection (aOR = 1.28 [95% CI, 1.00–1.65], per year; P = .05), although it did not reach statistical significance. Cumulative duration of exposure to a thymidine analog was not associated with higher odds of moderate-to-severe steatosis in univariate analyses, but after adjustment for age, sex, BMI, and duration of HIV infection, a positive association was found (aOR = 1.19 [95% CI, 1.03–1.37], per year; P = .02) (Table 3).

Table 3.

Factors Associated With Moderate-to-Severe Hepatic Steatosis in People With HIV

Variable	Crude OR (95% CI)	P Value
Sex (female vs male)	0.30 (0.07–1.28)	.10
Age (per decade)	1.12 (0.80–1.57)	.51
Age groups
<50 years	Ref
50–60 years	1.12 (0.51–2.46)	.78
61–70 years	1.06 (0.43–2.64)	.89
>70 years	1.80 (0.55–5.90)	.33
Ancestry
Scandinavian	Ref
Other European	1.79 (0.74–4.36)	.20
Middle East and Indian Subcontinent	3.84 (0.39–38.18)	.25
Educational Level
None	Ref
Short	0.30 (0.11–0.79)	.02
Middle length	0.17 (0.06–0.44)	<.01
University	0.39 (0.15–0.98)	.05
Smoking
Never smoker	Ref
Current smoker	0.64 (0.27–1.54)	.32
Previous smoker	0.76 (0.36–1.60)	.47
Alcohol (per 1 unit/week)	0.89 (0.81–0.97)	.01
Physical Activity
Inactive	Ref
Moderate inactive	0.33 (0.13–0.81)	.02
Moderate active	0.16 (0.06–0.41)	<.01
Very active	0.10 (0.02–0.50)	<.01
Abdominal obesity (yes vs no)	3.79 (1.32–10.91)	.01
Waist circumference (per 1 cm)	1.16 (1.11–1.20)	<.01
BMI (per 5 kg/m2)	6.84 (4.11–11.40)	<.01
BMI ≥25 kg/m2 (yes vs no)	15.56 (4.71–51.39)	<.01
Diabetes (yes vs no)	4.86 (2.08–11.39)	<.01
Metabolic syndrome (yes vs no)	6.24 (2.67–14.60)	<.01
Biochemistry
Plasma ALT (per 10 IU/L)	1.73 (1.42–2.10)	<.01
Plasma AST (per 10 IU/L)	1.44 (1.16–1.78)	<.01
Plasma total cholesterol (per 1 mM)	1.31 (0.97–1.76)	.07
Plasma triglycerides (per 1 mM)	1.40 (1.18–1.65)	<.01
Plasma HDL (per 1 mM)	0.10 (0.03–0.31)	<.01
Plasma LDL (per 1 mM)	1.17 (0.83–1.64)	.38
Duration of ART Exposure (per year)
Integrase inhibitors	1.19 (1.02–1.39)	.02
Thymidine analog	1.07 (0.98–1.17)	.14
Stavudine	1.42 (1.08–1.88)	.01

Abbreviations: ALT, alanine aminotransferase; ART, antiretroviral therapy; AST, aspartate transaminase; BMI, body mass index; CI, confidence interval; HDL, high-density lipoprotein; HIV, human immunodeficiency virus; IQR, interquartile range; LDL, low-density lipoprotein; OR, odds ratio; Ref, reference.

Figure 1.

Association between human immunodeficiency virus infection and moderate-to-severe hepatic steatosis. Odds ratio and 95% confidence interval were obtained from univariate and multivariable logistic regression analyses with results shown on a log₁₀ scale. The metabolic model is adjusted for age, sex, ethnicity, body mass index, plasma total cholesterol, plasma triglycerides, diabetes, plasma glucose, and alanine aminotransferase. The lifestyle model is adjusted for age, sex, ethnicity, smoking status, weekly alcohol consumption, and physical activity level.

Factors Associated With Moderate-To-Severe Hepatic Steatosis in People With Human Immunodeficiency Virus

Factors associated with moderate-to-severe hepatic steatosis in PWH can be found in Table 3 and Figure 2. In PWH, abdominal obesity, diabetes, metabolic syndrome and higher BMI, waist circumference, plasma ALT, plasma AST, and plasma triglycerides were associated with higher odds of moderate-to-severe hepatic steatosis in unadjusted models (Table 3). Higher physical activity level, higher educational level, and higher plasma HDL concentration were associated with lower odds of moderate-to-severe hepatic steatosis. After adjusting for potential metabolic confounders, higher BMI and higher ALT were associated with higher odds of moderate-to-severe hepatic steatosis, whereas female sex was associated with lower odds. After adjusting for potential lifestyle confounders, female sex and higher physical activity level were associated with lower odds of moderate-to-severe hepatic steatosis (Figure 2).

Figure 2.

Factors associated with moderate-to-severe hepatic steatosis in people with human immunodeficiency virus. Odds ratio and 95% confidence interval were obtained from univariate logistic regression analyses with results shown on a log₁₀ scale. The metabolic model (A) is adjusted for age, sex, ethnicity, body mass index (BMI), plasma total cholesterol, plasma triglycerides, diabetes, plasma glucose, and alanine aminotransferase (ALT). The lifestyle model (B) is adjusted for age, sex, ethnicity, smoking status, weekly alcohol consumption, and physical activity level.

In PWH, a higher weekly alcohol consumption within the national recommendations for excessive alcohol intake was associated with lower odds of moderate-to-severe hepatic steatosis in univariate and multivariate analyses after adjusting for age, sex, ethnicity, BMI, and physical activity level (Table 3 and Figure 2). Although a weekly consumption of beer seemed protective of moderate-to-severe hepatic steatosis, no association was found with wine, liquor, sugar-sweetened beverages, coffee, fast food, and type of meat product were not associated with moderate-to-severe hepatic steatosis in adjusted analysis (Supplementary Table S4).

Sensitivity Analyses

In sensitivity analyses with a CT liver attenuation threshold of ≤40 HUs, the lower prevalence of moderate-to-severe hepatic steatosis in PWH compared with HIV uninfected controls persisted (3.5% [95% CI, 2.2%–5.7%] vs 6.4% [95% CI, 4.9–8.4], P = .04). Accordingly, a positive HIV status was associated with lower odds of hepatic steatosis in univariable analyses (OR = 0.54 [95% CI, 0.30–0.95], P = .03) and after adjusting for age and sex (aOR = 0.53 [95% CI, 0.30–0.95], P = .03).

DISCUSSION

In this study of 453 predominantly well treated PWH without chronic hepatitis and excessive alcohol use and 765 HIV-uninfected controls, PWH had a lower prevalence of CT-defined moderate-to-severe hepatic steatosis than HIV-uninfected controls. Human immunodeficiency virus infection was independently associated with lower odds of hepatic steatosis.

In our cohort of PWH without viral hepatitis or excessive alcohol intake, 8.6% had CT-evidence of moderate-to-severe hepatic steatosis, which was considerably lower than compared with a meta-analysis of NAFLD in PWH [7]. Several reasons may account for the discrepancy. First, time may play a role because first-generation antiretroviral drugs had more liver toxicity than currently used agents. Second, some studies included individuals with signs of liver disease (eg, persistently elevated liver enzymes) [4, 6, 25–27] or individuals with metabolic disorders [3, 28]. Third, the prevalence of hepatic steatosis varies globally [29, 30] due to increased adoption to a Western diet and sedentary lifestyle as well as genetic variation [20, 31, 32]. Fourth, the presence of steatosis may differ due to different diagnostic methodology used. Finally, PWH had higher rates of smoking, less alcohol use, higher physical activity, and higher education level and were more frequently on antidiabetic and lipid-lowering therapies compared with controls, which may be protective factors of hepatic steatosis. Our study supports the findings of Price et al [10] who found a lower prevalence of hepatic steatosis in unselected PWH compared with HIV-uninfected controls assessed by CT liver scans (13% vs 19%). Overall, our study and Price et al [10] question the proposed higher risk of moderate-to-severe hepatic steatosis for PWH compared with a demographically similar group of HIV-uninfected individuals. Future studies should explore this in more detail because our findings do not exclude the possibility of an increased risk of mild hepatic steatosis or of more progressive NAFLD in PWH. Our study design does not permit any distinction as to whether the difference in the proportion of hepatic steatoses between the 2 groups is related to HIV itself or factors associated with HIV infection. Finally, residual confounding of lifestyle, for example, cannot be precluded.

Few studies have been able to investigate the association between HIV infection and hepatic steatosis due to lack of a HIV-uninfected comparator group. A key finding of this study was that a positive HIV status independently was associated with lower odds of moderate-to-severe hepatic steatosis. The result was robust even after adjustment for age, sex, ethnicity, and potential metabolic and lifestyle confounders and when using a lower threshold for moderate-to-severe hepatic steatosis of ≤40 HUs. It is interesting to note that Price et al [10] reported that HIV was independently associated with lower odds of hepatic steatosis (OR = 0.44, P < .002), which is consistent with our findings. This may emphasize the complexity underlying the pathogenesis of hepatic steatosis in PWH [33] and warrants future studies.

Adipose tissue abnormalities leading to lipodystrophy and atrophy are associated with specific antiretroviral drugs, in particular with thymidine analogs [34]. Thymidine analogs and didanosine have hepatotoxic properties. Price et al [10] found an association between didanosine use and hepatic steatosis, but this was not reproduced in our study. A possible explanation may be that only 1 of 6 PWH in our study had been exposed to didadosine and that the exposure time was less in the COCOMO cohort compared with Multicenter AIDS Cohort Study (MACS) (2 vs 4 years) [10]. However, we did see an association between use of thymidine analogs and hepatic steatosis. It is interesting to note that use of thymidine analogs was discontinued approximately 1 decade before inclusion in COCOMO. Similarly, low visceral and subcutaneous adipose tissue density was associated with prior exposure to thymidine analog and/or didanosine exposure in the cohort [35]. Collectively, this suggests that the hepatotoxic effects of thymidine analogs may be long-lasting in terms of moderate-to-severe hepatic steatosis. Individuals exposed to thymidine analogs may require additional work-up for hepatic steatosis. Furthermore, we found an association between cumulative exposure to integrase inhibitor treatment and hepatic steatosis. Of note, use of integrase inhibitors has been associated with excess weight gain [36]. It is likely that there may be a direct link between weight gain and hepatic steatosis. Alternatively, integrase inhibitors may induce hepatic steatosis regardless of overall weight gain. Future studies are warranted to study whether specific integrase inhibitors may infer an increased risk of hepatic steatosis and fibrosis.

Male gender, higher BMI, and higher ALT were associated with higher odds of moderate-to-severe hepatic steatosis in PWH. These results are consistent with previous findings, and, in particular, the association between BMI, insulin resistance, and hepatic steatosis are well established [2, 3, 10, 36]. To our surprise, we did not find a significant association between diabetes and moderate-to-severe hepatic steatosis in PWH after adjusting for metabolic risk factors. In our study, PWH were more frequently on antidiabetic- and lipid-lowering treatment compared with controls, which may indicate more frequent physician encounters due to regular HIV care. One may speculate, that PWH initiate therapy for diabetes and dyslipidaemia at an earlier stage, which may cause a lower rate of hepatic fat accumulation. Furthermore, there could be a synergistic effect of diabetes and increasing BMI on the development of hepatic steatosis, because the comparator group had higher BMI and more overweight individuals. A synergistic effect of excessive alcohol intake and increased BMI on liver disease has been reported previously [37], and future studies should explore these possible synergistic effects in NAFLD. It is interesting to note that a moderate alcohol consumption seemed to be protective of moderate-to-severe hepatic steatosis. No information on previous alcohol consumption was collected, and PWH with a history of excessive alcohol intake and potential advanced fibrosis may have affected the results. Furthermore, previous studies have demonstrated possible protective effects of moderate beer and red wine consumption. A study by Padro et al [38] demonstrated a positive effect of beer on the function of HDL and its capacity to protect against LDL oxidation and to increase the efflux of cholesterol. Thus, a moderate beer consumption may have beneficial effects on the development of hepatic steatosis but needs to be explored in more detail. Furthermore, PWH reported a more regular consumption of red wine compared with white wine, beer, and liquor, suggesting that the high amount of polyphenols, for example, in red wine may not only have cardioprotective effects but also hepatoprotective effects with less inflammation, less insulin resistance, and improved lipid profile [39]. However, our data did not support this speculation and needs to be explored in more detail. Finally, no association was found between moderate-to-severe hepatic steatosis and wine, liquor, nonalcoholic beverages, fast food, or meat items in adjusted analyses. However, current international guidelines on treatment of NAFLD focus on changes in diet and lifestyle [40], and future studies should explore the role of different diets in randomized controlled trials.

To our knowledge, this is the largest study of moderate-to-severe hepatic steatosis in PWH with a comparable HIV-uninfected control group using identical methodologies. Our study is limited by a homogeneous population of PWH, which limits the generalizability to other settings. With a sensitivity of 54%, it cannot be precluded that PWH and uninfected controls may have been missed in the diagnosis of moderate-to-severe hepatic steatosis. Liver CT attenuation may not identify individuals with mild hepatic steatosis and individuals with advanced liver fibrosis. Sampling errors cannot be avoided despite the attempt to minimize this, no information on previous alcohol consumption, inflammatory markers, gut microbiota, or insulin resistance (eg, HOMA-IR) were available, and unmeasured residual confounding cannot be excluded. No testing for HIV, HBV, or HCV were available for the comparator group. Finally, causality cannot be inferred in a cross-sectional study.

CONCLUSIONS

In conclusion, the prevalence of moderate-to-severe hepatic steatosis in this cohort of well treated PWH was lower compared to a demographically comparable cohort of HIV-uninfected individuals, and HIV infection was independently associated with lower odds of moderate-to-severe hepatic steatosis. Male sex, higher BMI, and higher ALT were associated with higher odds of hepatic steatosis. Exposure to integrase inhibitor treatment was associated with moderate-to-severe hepatic steatosis and should be explored in more detail.

Supplementary Data

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

Notes

Acknowledgments. We thank all study participants for their participation. We also thank all medical staff at the Department of Infectious Diseases, Rigshospitalet and Amager Hvidovre Hospital for their support and participation.

Author contributions. D. M. K.-K., S. D. N., J. L., K. F. K., and T. B. contributed to study design. D. M. K.-K. contributed to data collection for the Copenhagen Co-Morbidity in HIV Infection (COCOMO) cohort, and K. F. K., A. F., P. E. S., J. L., S. S., L. K., and B. G. N. contributed to data collection for the Copenhagen General Population Study. D. M. K.-K., with supervision by T. B. and S. D. N. and statistical support from A. M., contributed to data analysis. D. M. K.-K. drafted the manuscript. All authors revised the manuscript and read and approved the final version.

Financial support. This work was funded by Simonsen Foundation, Novo Nordisk Foundation, Lundbeck Foundation, Rigshospitalet Research Council, Region Hovedstaden and Danish National Research Foundation (Grant 126).

Potential conflicts of interest. A. M. has received honoraria, consultancy fees, speaker fees, and travel support from ViiV, Gilead, and A. Craig Eiland. K. F. K. reports grants from the Danish Research Foundation during the conduct of the study, in addition to grants from the Research Council of Rigshospitalet, A. P. Moller og Hustru Chastine McKinney Mollers Fond, the Danish Heart Foundation, and Canon Medical Corporation, outside the submitted work. S. D. N. reports unrestricted grants from Novo Nordisk Foundation, Lundbeck Foundation, Augustinus Foundation, Rigshospitalet Research Council, travel grants from Gilead, MSD, BMS, and GSK/ViiV, and advisory board activity for Gilead and GSK/ViiV. T. B. reports grants from Pfizer, Novo Nordisk Foundation, Simonsen Foundation, and GSK and personal fees from Pfizer, outside the submitted work. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Presented in part: Conference on Retroviruses and Opportunistic Infections (CROI), Washington State Convention Center, Seattle, Washington State, 4–7 March 2019.