Similar prevalence of hepatic steatosis among patients with chronic hepatitis C with and without HIV coinfection

Hepatic steatosis (HS) is frequently observed in HIV-infected patients. It is not known whether HIV infection is an independent risk factor for HS development. We aimed to analyze whether HIV coinfection was associated with a higher frequency of HS in patients with chronic hepatitis C. This was a retrospective cross-sectional study. 574 subjects with chronic hepatitis C virus (HCV) infection were included, 246 (43%) of them coinfected with HIV. All of them underwent transient elastography with controlled attenuation parameter (CAP) measurement. HS was defined as CAP ≥ 248 dB/m. 147 individuals (45%) showed HS in the HCV-monoinfected group and 100 (40.7%) in the HIV/HCV-coinfected group (p = 0.318). HS was associated with body mass index (BMI) [<25 Kg/m2 vs. ≥25 Kg/m2, 67 (23.5%) vs. 171 (62.9%); p = 0.001], with plasma HDL-cholesterol [<50 mg/dL vs. ≥50 mg/dL, 122 (48.6%) vs. 95 (37.5%), p = 0.012], with plasma triglycerides [<150 mg/dL vs. ≥150 mg/dL, 168 (40.2%) vs. 65 (52.4%); p = 0.016] and with plasma total cholesterol [<200 mg/dL vs. ≥200 mg/dL, 181 (41%) vs. 53 (52.5%); p = 0.035]. In the multivariate analysis, HS was associated with BMI [adjusted OR (AOR) = 1.264 (1.194–1.339); p = 0.001], age [AOR = 1.029 (1.001–1.058); p = 0.047] and HCV genotype 3 infection [AOR = 1.901 (1.081–2.594); p = 0.026]. HIV coinfection was not associated with HS [AOR = 1.166 (0.719–1.892); p = 0.534]. In conclusion, HIV coinfection is not related with an increased frequency of HS in HCV-infected patients.

The main causes of hepatic steatosis (HS) are metabolic factors 1 , alcohol use 2 and HCV infection, particularly HCV genotype 3 3,4 . Non-alcoholic fatty liver disease (NAFLD), HS in the absence of other causes than metabolic factors, is recognized as one of the most common liver diseases worldwide with an estimated prevalence of 25% among the general population in Western countries 5,6 . NAFLD has a spectrum of liver disease ranging from simple fatty liver to nonalcoholic steatohepatitis (NASH), fibrosis and cirrhosis [7][8][9][10] .
In HIV-uninfected patients, the prevalence and risk factors of NAFLD and its complications have been evaluated, specifically in industrialized countries 6,8,[11][12][13] . In HIV infection, studies carried out in unselected patients with or without HCV coinfection have reported a prevalence of HS, determined by the controlled attenuation parameter (CAP), of approximately 40% [14][15][16][17] . This rate nearly doubles that found of the general population in similar areas 5 and it is closer to the NAFLD prevalence observed in obesity and type 2 diabetes mellitus 8 . Despite this difference between HIV-infected patients and the general population, there are very few direct comparisons of HS frequency between individuals with and without HIV infection. Indeed, a study comparing men who have sex with men (MSM) with and without HIV infection found a lower prevalence of HS in HIV-infected MSM than in MSM without HIV infection 18 . On the contrary, two case-control studies conducted in small samples found higher frequencies of HS in HIV-infected individuals than in controls 19,20 . Because of this, the actual impact of HIV infection on the risk of HS still remains unknown.
Among HIV-infected patients with chronic hepatitis C, the frequency of HS estimated by liver biopsy showed a wide range from 23% to 72% 21 . In a metanalysis of HCV-infected patients who underwent liver biopsy, the rates of HS were similar in HIV/HCV-coinfected and HCV-monoinfected subjects 21 . However, patients undergoing a liver biopsy are not representative of the overall population with HCV infection. Thus, information provided by this study on the risk of HS inherent to HIV infection is very limited. Because of this, comparative studies of the HS prevalence conducted in unselected populations of HCV-infected patients with and without HIV coinfection are required. Non-invasive techniques may allow these studies, as the overall HCV-infected population, irrespective the estimated liver stage or the therapy perspectives, may be included 9,10 . This kind of studies may provide very valuable data on the risk of HS due to HIV infection minimizing the possible confounding effect of the HCV coinfection.
Because of these, we aimed to compare the prevalence of HS, evaluated using CAP, in patients with chronic HCV infection, with and without HIV coinfection, in order to appraise the effect of HIV infection on the HS presence in this setting. www.nature.com/scientificreports www.nature.com/scientificreports/ frequency of HS (CAP ≥ 248 dB/m) in HCV-monoinfected patients was 147 (44.8%) and in HIV/HCV-coinfected patients was 105 (40.7%) (p = 0.318). Eighty-three (25.3%) individuals without HIV infection showed severe steatosis compared to 51 (20.7%) among those with HIV coinfection (p = 0.2). There were no differences in the prevalence of steatosis between the group with and without HIV coinfection by BMI category (Fig. 2). In an analysis excluding patients with alcohol intake ≥50 g/d, there were no differences between the two study groups in the median CAP values nor in the prevalence of HS ( Supplementary Fig. S1).
factors associated with the presence of HS. In the univariate analysis, history of injecting drugs, alcohol intake, plasma triglycerides, BMI, plasma total cholesterol and plasma HDL cholesterol were related with HS, whereas HIV coinfection was not ( Table 2). In the multivariate logistic analysis, BMI, age and genotype 3 infection were independently associated with HS. HIV coinfection was not related with HS (Table 2).

Discussion
In our study, we found that the prevalence of HS in HIV/HCV-coinfected patients is similar to that observed in of HCV-monoinfected patients. These results suggest that HIV coinfection does not influence the development of HS in patients with chronic hepatitis C.
The prevalence of NAFLD in HIV infection, as estimated with CAP, has ranged from 39% to 41% across different independent reports in Western countries 14,15,22 . In the general population, NAFLD frequency is expected   www.nature.com/scientificreports www.nature.com/scientificreports/ to affect 25% individuals living in the USA 5 . Despite this sharp difference between the HIV-infected population and the general population, direct comparisons of large samples of unselected patients with and without HIV infection are lacking. In a recent retrospective study, the frequency of HS, evaluated by CAP, was higher among HCV-monoinfected individuals than HIV/HCV-coinfected patients 23 . The prevalence of CAP ≥ 238 dB/m was 29.5% for HIV/HCV-coinfected patients and 42.9% for HCV-infected patients 23 . However, a comparative analysis of risk factors for HS was not reported 23  In the present study, contrary to all expectations, the rates of HS, measured by CAP, among patients with HCV infection followed at a single tertiary care center were similar among those with and without HIV coinfection. www.nature.com/scientificreports www.nature.com/scientificreports/ However, there were a number of differences between both groups that could have influenced the frequency of HS. Notably, BMI and other metabolic factors were unevenly represented in both groups. The HCV-monoinfected group showed a greater proportion of overweight or obese patients, whereas those with HIV/HCV coinfection presented higher levels of plasma triglycerides and lower levels of plasma HDL-cholesterol. After adjustment for factors associated with HS, including those mentioned before, HIV infection was not a factor related with a higher likelihood of steatosis. Indeed, a stratified analysis by BMI, the strongest predictor of HS in the present study and in previous reports 14,15,17,22 , did not disclose any differences in the rates of steatosis by BMI category between patients with and without HIV infection. Finally, an analysis excluding patients with alcohol intake ≥50 g/day did not show either significant differences between both groups. Because of all the above reasons, the risk of NAFLD seems to be similar for patients with chronic hepatitis C with and without HIV infection.
The present study may have several limitations. First, this was a retrospective study and that design may involve lack of data unplanned to be gathered in clinical practice, as insulin resistance. However, all patients with HCV infection attended at our unit undergo the same protocol, including assessment of steatosis by CAP, at their initial clinical visit. CAP data was not available only among individuals in whom images could not be acquired. Second, alcohol intake was self-referred by patients during the clinical interview, and this could underestimate the true amount of alcohol consumption. Third, HS in the present study could represent a mixture of causes, from true NAFLD associated with metabolic factors to secondary steatosis related with HCV genotype 3 infection. Despite this, the main factors associated with HS in the present study were metabolic factors, those typically associated with NAFLD. The main strength of this study is the comparison of the prevalence of and factors associated with HS in a homogeneous population within the same unit, according to the same protocol and evaluated using a uniform technique throughout the study. A study of this kind has been claimed before by some experts 9,10 .
In conclusion, HS is very frequent in patients with chronic hepatitis C, with and without HIV coinfection. Among them, HS shows features of NAFLD, as it is mainly associated with components of metabolic syndrome, and is also related with HCV genotype 3 infection. Our findings indicate that HIV coinfection is not associated with a higher risk of HS in individuals with chronic hepatitis C background. Because of these, the management of HS in HIV/HCV-coinfected patients and HCV-monoinfected patients should be similar and aimed at controlling metabolic risk factors.

Methods patients and study design.
This was a retrospective cross-sectional study. All Spanish Caucasian patients who were attended at the Unit of Infectious Diseases of the Hospital Universitario Virgen de Valme, Seville (Spain), from November 2010 to March 2019, were selected if they had: (1) Chronic HCV infection, with persistent detection of plasma HCV RNA, with or without HIV coinfection; (2) A valid available hepatic elastography examination with evaluation of HS by CAP. Patients pretreated against HCV infection who did not achieve sustained virological response (SVR) were also included in the study.

Data collection.
Data from all patients were recorded following a pre-specified protocol before starting HCV therapy. At that date, electronic clinical records including demographics, self-referred alcohol intake by patients, anthropometry, blood test and hepatic transient elastometry with CAP were gathered. CAP and liver stiffness (LS) were measured by FibroScan (Echosens FibroScan 502, Paris). A cut-off of ≥248 dB/m and of ≥280 dB/m were selected to define the presence of mild HS (steatosis involving ≥10% of hepatocytes) and severe steatosis (≥66% steatotic hepatocytes), respectively 25 . All CAP and LS measurements were performed in fasting conditions by two trained operators. We had previously proven a high concordance between two trained operators in FibroScan measurements, to determine HS by CAP 26 or to evaluate LS 27 . All the measurements were the result of the evaluation of ten valid shots. For the present study, an hepatic transient elastometry was considered as valid if the interquartile range for liver stiffness was <30% of the median value and the success rate was ≥60% 14 .
Individuals with a body mass index (BMI) between 18 and 25 kg/m 2 , between 25 and 30, between 30 and 35, between 35 and 40 and >40 were considered as individuals with normal weight, pre-obesity, obesity class 1, class 2 and class 3 respectively in accordance with the WHO classification (http://www.euro.who.int/en/health-topics/ disease-prevention/nutrition/a-healthy-lifestyle/body-mass-index-bmi). We consider overweight a BMI up to 25 kg/m 2 . High alcohol intake was defined as ≥50 g/day 28 .

Statistical analysis.
For descriptive analysis, continuous variables were expressed as median (Q1-Q3) and categorical variables as frequencies (percentage). The χ 2 test or Fisher's exact test was used to compare the distribution of categorical variables between groups and Student's t-test or the Mann-Whitney U test was used for continuous variables. Binary logistic regression models were elaborated to assess the factors independently associated with the presence of HS. In those analyses, variables related to this condition with a univariate p value < 0.2, as well as age, sex and HIV infection, were included to obtain odds ratio (OR) values. Differences were considered significant for p values < 0.05.
A case-control study was carried out as secondary analysis. HIV/HCV-coinfected patients were considered cases. HCV-monoinfected patients were matched with cases by age, sex and BMI. Cases and controls were matched by BMI because it was the only independent predictor of HS in a previous study using CAP 14 . All analysis were carried out using the SPSS software 25.0 (IBM Corporation, Somers, New York, New York, USA).
ethics. This study was designed and performed according to the Helsinki declaration and was approved by the ethics committee of the Hospital Universitario Virgen de Valme (Seville, Spain). Informed consent was obtained from all individuals.