Assessment of the Association of HIV Infection with Hepatic Steatosis or Fibrosis: a Cross-sectional Case–Control Study

Human immunodeficiency virus (HIV) infection and antiretroviral therapy have been associated with non-alcoholic fatty liver disease (NAFLD), but few studies have evaluated whether HIV infection is an independent risk factor for the development of hepatic steatosis and advanced liver fibrosis. To study the prevalence and severity of hepatic steatosis and advanced fibrosis in people living with HIV and control outpatients. We conducted a cross-sectional analysis of relevant data from 875 pairs of individuals belonging to an HIV-dedicated outpatient clinic and an adult primary care clinic of an inner-city hospital. Hepatic Steatosis Index (HSI) and FIB-4 index were calculated as non-invasive measures of steatosis and fibrosis, respectively. A multivariate logistic regression analysis was performed to assess predictors of steatosis and advanced fibrosis. The prevalence of hepatic steatosis, determined by HSI ≥ 36, was higher in HIV-negative subjects (71.5% vs. 65.4%, p = 0.006). The prevalence of advanced fibrosis, determined by FIB-4 index ≥ 3.25, was higher in the HIV-positive group (7% vs. 1.7%, p < 0.001). Multivariable analysis did not identify HIV infection to be an independent risk factor for hepatic steatosis (p = 0.068) and advanced fibrosis. In this cohort, hepatic steatosis was more prevalent in non-HIV infected patients, while advanced fibrosis had a higher prevalence in people living with HIV. HIV infection was not found to be an independent risk factor for either hepatic steatosis or fibrosis.


Introduction
Hepatic steatosis is common in HIV infection and is associated with poor outcomes and increased mortality [1]. In the pre-antiretroviral therapy (ART) era, steatosis was the most common cause of abnormal liver function tests [2]. In the early ART era, hepatic steatosis was mainly related to the lipodystrophy syndrome with an increase in visceral fat content, and medication toxicity seen with thiamine analog nucleoside reverse transcriptase inhibitors (NRTIs) and specific protease inhibitors [3]. Advances in drug development have decreased the metabolic toxicity of antiretroviral therapy. Currently, it is uncertain whether chronic HIV infection confers an increased risk of hepatic steatosis.
Data on the prevalence and consequences of non-alcoholic fatty liver disease (NAFLD) in people living with HIV and on the risk factors for progression to fibrosis are limited. The reported prevalence of steatosis in HIV infection varies from 20 to 63% [2][3][4]. There is controversy over the This article is part of the Topical Collection on Medicine Paula Debroy, Sanjana Nagraj and Natalia Chamorro-Pareja contributed equally. prevalence compared to the general population, and only a few studies have compared HIV-infected and uninfected adults [5,6]. Obesity contributes to the development of steatosis, though most studies of NAFLD in people living with HIV do not stress this factor [5][6][7][8][9][10][11][12][13][14]. The Bronx has a high prevalence of obesity and diabetes when compared to the other New York City boroughs, both of which are wellknown risk factors for hepatic steatosis [15][16][17].
The objective of this study was to assess the association of HIV infection with hepatic steatosis by comparing the prevalence of hepatic steatosis in people living with HIV and non-HIV individuals attending outpatient clinics at an inner-city hospital in The Bronx, New York.

Study Population
This was a cross-sectional study of 875 adults living with HIV infection who were followed in the HIV-dedicated outpatient clinic at Jacobi Medical Center, Bronx, NY, from January 1 through December 31, 2014. Controls included adult patients who have been tested negative for HIV and were followed in the adult primary care clinic over the same period. The Institutional Review Board at the Albert Einstein College of Medicine and the respective committee of NYC Health + Hospitals approved this study. The study was conducted according to STROBE guidelines for reporting observational studies [18]. The STROBE statement checklist is presented in the supplementary material.

Data Collection
Clinical and laboratory data were obtained from electronic medical records. The following variables were collected: age, gender, weight, height, ethnicity, self-reported tobacco and alcohol use, history of diabetes mellitus (defined as ongoing use of anti-diabetic medication(s) or hemoglobin A1c (Hgb A1c) greater or equal to 6.5%), hypertension (defined as ongoing use of antihypertensive medication(s) or documented under the problem list of the electronic medical record), and dyslipidemia (defined as current use of any lipid-lowering medication or documentation in the problem list of the electronic medical record). The diagnosis of prediabetes was defined using established American Diabetes Association (ADA) criteria [19]. HIV-specific data included current use of ART, duration of ART use, and years since diagnosis. Virologic suppression was defined as HIV RNA viral load < 20/ml. Baseline laboratory investigations included aspartate aminotransferase (AST), alanine aminotransferase (ALT), total cholesterol, low-density lipoprotein cholesterol (LDL), high-density lipoprotein cholesterol (HDL), triglyceride, Hgb A1c, platelet count, CD4 T lymphocyte (CD4) count, and HIV-1 RNA (copies/ml). Hepatitis C serological status was determined by antibody testing (anti-HCV antibody). Hepatitis B serological status was determined by surface antigen testing (HBsAg).

Liver Steatosis and Fibrosis Assessment
Hepatic Steatosis Index (HSI) and FIB-4 index were calculated for all patients as they are non-invasive, inexpensive, and readily available methods that assess steatosis and fibrosis, respectively, and can be used as screening tools in innercity HIV clinics (refer to supplemental material for formulae) [20]. HSI < 30 has been found to exclude NAFLD with a sensitivity of 93.1% and hold a negative predictive value of 85.2%, while HSI > 36 has been shown to detect NAFLD with a specificity of 93.1% and have a positive predictive value of 86.7% [21]. Steatosis was defined as HSI > 36. HSI has also been validated in HIV monoinfected patients with an area under the receiver operating characteristic curve (AUROC) of 0.88 and an accuracy of 84.5% when compared to ultrasound and cross-validated against 1 H magnetic resonance spectroscopy [20,21].
Advanced fibrosis was defined as a FIB-4 index higher than 3.25. A FIB-4 index of lower than 1.45 has a negative predictive value of 94.7% to exclude a significant fibrosis (METAVIR F3-F4) with a sensitivity of 74.3%, while a FIB-4 higher than 3.25 offers a positive predictive value of 82.1% to identify extensive fibrosis (METAVIR F3-F4) with a specificity of 98.2% [22].

Statistical Analysis
Variables are expressed as median and interquartile range (IQR) for continuous and percentage for categorical variables. Comparisons between continuous variables were performed using the Mann-Whitney U test and between nominal variables using the Pearson's chi-squared test. The alpha value of statistical significance was set at 0.05. Multivariate logistic regression analysis was performed to assess predictors of steatosis with HSI ≥ 36 or advanced fibrosis. Only significant variables (p < 0.05) were then analyzed in a stepwise forward multivariate analysis model. Variables that were included in the FIB-4 and HSI equation were excluded from the analysis. HCV viral load was not included in the analysis due to the extent of missing data, particularly in the HIV-negative cohort. Odds ratios (OR) and 95% confidence intervals (CI) were reported for p values < 0.05. Data was analyzed using the statistical software SPSS (version 23.0, IBM, New York, NY, USA).

Baseline Characteristics
A total of 1750 (patients living with HIV 875) patient charts were reviewed. Baseline characteristics are presented in Table 1. The median patient age (IQR) was 51.7 years (41.5-58.8); 57% were female. The majority of patients were non-Caucasian. Patients living with HIV had higher rates of tobacco use (15.5% vs 33.9%; p < 0.001). Hepatitis C seropositivity was documented in 142 HIV-infected patients compared to 24 HIV-uninfected patients (p < 0.001). Among the HIV-infected cases, 94% were on ART, and 72% were virologically suppressed. The median (IQR) duration of antiretroviral therapy was 9 years (6-15). The median BMI was 29.1 kg/m 2 (IQR 24.7-34) in the HIV-negative group vs. 27.1 kg/m 2 (IQR 23.6-32) in the HIV group. A total of 141 (8.1%) individuals in the study population had class III obesity (BMI ≥ 40 kg/m 2 ) with a significantly higher percentage in the HIV-negative group (9.3% vs 6.9%, p < 0.005).
The prevalence of diabetes mellitus was significantly higher in the HIV-negative group (31.7% vs 15.4%, p < 0.001). There was no difference in the prevalence of pre-diabetes defined by Hgb A1c criteria (23.7% vs 25%, p = 0.504). The HIV-negative group had higher median Hgb A1c values, higher LDL levels, higher total cholesterol levels, and lower triglyceride levels.

Hepatic Steatosis
The prevalence of hepatic steatosis was 68% in the entire cohort and was higher in HIV-negative individuals (71.5% vs 65.4%, p = 0.006). Self-reported alcohol use did not differ between the two groups. On multivariate analysis, the presence of HIV infection was not found to be an independent risk factor for the presence of hepatic steatosis (p = 0.068) ( Table 2). Factors that were significantly associated with higher odds of steatosis were a diagnosis of hypertension and higher Hgb A1c levels. Hepatitis C infection had a lower odds ratio for steatosis, as did tobacco use. BMI and diabetes were excluded from the logistic regression analysis as they are incorporated in calculation of HSI.

Advanced Fibrosis
The overall prevalence of advanced fibrosis was 4.4% and was higher in people living with HIV (7% vs. 1.7%, p < 0.001). On multivariate analysis, independent factors associated with higher odds of advanced fibrosis were age (OR = 1.06, CI = 1.02-1.09, p < 0.001), HBV infection (OR = 2.82, CI = 1.02-7.85, p = 0.05), and HCV infection (OR = 8.35, CI = 4.80-14.54, p < 0.001) ( Table 3). HIV was not an independent risk factor for development of advanced fibrosis. ALT and AST were excluded from the logistic regression analysis as they are a part of the FIB-4 index calculation.

Discussion
In this cohort of patients living with HIV, we found a lower prevalence of hepatic steatosis compared to the control group, despite a higher prevalence of HCV seropositivity. The prevalence of hepatic steatosis in both cohorts was higher than anticipated (> 50% for both groups). Notably, HIV was not associated with an increased risk of steatosis or fibrosis in this study.
The overall prevalence of hepatic steatosis in patients with HIV in our cohort was 65.4% which is higher than that reported in other studies using imaging and invasive tests as diagnostic tools [4,7]. There are several factors that could potentially account for the higher prevalence rate of hepatic steatosis in our study. The median BMI for HIV-positive cases with steatosis in our study population was 30.2 kg/m 2 . Studies with a lower prevalence of steatosis had mean BMIs in the normal range [5,10], while studies reporting a high prevalence similar to ours, had mean BMIs in the overweight range [7]. Also, our study did not exclude patients with alcohol use or hepatitis C. The rates of self-reported alcohol use were not statistically different between both groups. Hepatitis C was not associated with hepatic steatosis on multivariate regression analysis. While BMI can explain the variation in HSI, it does not accurately predict hepatic adiposity, especially when compared to waist circumference and visceral fat content, both of which more closely reflect the presence of metabolic syndrome [23]. A prior study of liver biopsies in patients undergoing bariatric surgery showed that the presence of steatosis was significantly related to markers of insulin resistance rather than to BMI [24]. Also, hepatic fat content estimated by proton spectroscopy was a stronger predictor of insulin resistance by the homeostasis model assessment (HOMA) than was visceral fat content, as determined by magnetic resonance imaging [25].
The prevalence of advanced fibrosis in our HIV cohort was similar to a recent meta-analysis where the pooled prevalence was 21% [2,12]. The prevalence of hepatic fibrosis in our study was higher in the HIV cohort compared to non-HIV individuals. The key factor associated with the increased prevalence of advanced fibrosis noted in HIV-positive cases was a higher rate of HCV infection. HIV infection by itself was not related with an increased risk of advanced fibrosis. HCV infection in HIV-positive individuals has been shown to be associated with an increased risk of progression to fibrosis [26].
On multivariate analysis, age was found to be significantly associated with hepatic fibrosis (OR 1.05, p < 0.001). In our cohort, 94% of the patients living with HIV were on ART and 72% were virologically suppressed. The median (IQR) duration of antiretroviral therapy in this cohort was 9 years (6-15). Although, age was found to be a significant predictor of hepatic fibrosis, and a large portion of the cohort was on ART for a considerable duration, an accurate measure of years since HIV diagnosis and its association with hepatic fibrosis could not be studied. Varying duration of undiagnosed HIV infection, at times several years prior to serological diagnosis, varying duration of treatment with ART, patient migration, loss to follow-up in clinic, and medication non-compliance are some of the factors that made evaluating this association in this patient population challenging.
Our findings of lower prevalence rates of hepatic steatosis in patients living with HIV are compatible with the only other large study done with HIV-negative controls [5]. This may be in part due to currently used ART, which, unlike the previously used agents, is not implicated in medicationinduced hepatotoxicity and does not serve as a risk factor for increased incidence of steatosis in HIV-positive patients. In fact, data suggest that cumulative exposure to ART reduces the risk of developing hepatic steatosis [5,27].
Cross-sectional studies in HIV-positive patients have looked at the prevalence of steatosis determined by noninvasive methods, including ultrasound, CT scan, and elastography, with prevalence ranging from 13 to 55% [2][3][4]. Studies including HCV co-infected patients have similar prevalence rates [5,28,29]. These radiological techniques can be expensive and are not readily available to inner-city HIV clinics in low-income settings. There are limited liver biopsy assessment studies and they often include high-risk patients with advanced disease or unexplained elevation of transaminases [7,11]. These diagnostic modalities are also unlikely to be readily available in low-income settings. Several non-invasive serological methods for the diagnosis of hepatic steatosis have been validated in HIV-positive patients, including the liver fat score (LFS), the lipid accumulation product, and HSI [20,30]. Previous studies with serological and diagnostic modalities were found to significantly underestimate the prevalence of NAFLD [31], but the high prevalence of steatosis in our study with the use of HSI is similar to other prevalence studies which suggests that it is adequate for use in HIV clinics where metabolic abnormalities are prevalent.
We acknowledge the strengths and limitations of our study. This is the largest study conducted to date comparing the prevalence of hepatic steatosis in people living with HIV and HIV-negative controls. This study used accessible and noninvasive clinical markers to estimate hepatic steatosis so that these markers can be easily incorporated during evaluation of patients in HIV clinics. However, we recognize that the estimation of hepatic steatosis and fibrosis was based on surrogate markers. The HSI calculation includes BMI and the diagnosis of diabetes, both of which were common in our cohort. While transient elastography and other sophisticated measurements may be more sensitive tools to estimate hepatic steatosis in both HIV-negative and HIVpositive individuals they were not available at our institution at the time of data collection [28,32]. Information on the use of thymidine analog NRTIs or specific PIs was not recorded. HCV and HBV viral loads were not available for many patients, and treatment status was uncertain. Finally, Hgb A1c testing may underestimate serum glucose in HIVinfected patients with reduced red cell survival [33].

Conclusions
In summary, while hepatic steatosis was more prevalent in non-HIV infected patients and advanced fibrosis had a higher prevalence in the HIV-positive cohort, HIV infection itself did not seem to significantly influence the development of either steatosis or fibrosis. The ability of non-invasive clinical predictors to guide therapy and improve clinical outcomes remains to be determined.