Hepatic steatosis among people living with HIV in Southern Brazil: prevalence and risk factors

Chronic liver disease is an important cause of morbidity and mortality among people living with human immunodeficiency virus (HIV) and is frequently related to non-alcoholic fatty liver disease (NAFLD). The objective is to estimate the prevalence and risk factors of hepatic steatosis among consecutive patients with stable HIV infection on antiretroviral therapy (ART). Also, the use of transient elastography (TE) as a mean to identify a subgroup at risk for non-alcoholic steatohepatitis (NASH) and/or liver fibrosis. HIV infected patients were enrolled between August2016 and February2017. Inclusion criteria: ≥18 years with undetectable HIV viral load. Exclusion criteria: pregnancy; alcohol intake ≥20 g/day and co-infection B or C viruses. Patients underwent ultrasound (US) to diagnose liver steatosis. Significant fibrosis (≥F2) was estimated if at least one of the following were present: APRI > 1.0, FIB4 > 3 and/or liver stiffness ≥7.1kPa. Subjects with TE ≥ 7.1kPa were proposed a liver biopsy and NAFLD Scoring System (NAS) ≥ 3 was considered as diagnosis of NASH. A total of 98 patients were included. Liver steatosis was diagnosed in 31 patients (31.6%) and was independently associated with male gender, BMI, ALT and total bilirubin levels. The prevalence of significant fibrosis assessed by TE, APRI and FIB4 was 26.9%, 6.4% and 3.2%, respectively. Seven patients had a TE result ≥7.1kPa. NASH was found in 5 (83.3%). Among HIV infected patients undergoing ART, almost one third have NAFLD. Neither TE, APRI or FIB4 were able to act as surrogates for significant liver fibrosis. Nevertheless, TE ≥ 7.1kPa was able to accurately select a subgroup of patients at risk for NASH.

Steatosis and fibrosis assessment. All patients underwent abdominal US with an experienced radiologist. Liver steatosis was diagnosed according to previously established criteria 21 . The following non-invasive fibrosis markers were calculated: AST to Platelet Ratio Index (APRI) as [AST (IU/L)/AST upper limit of normal (IU/L)/Platelets (10 9 /L) × 100] 22 and FIB4 score as [age (years) × AST (IU/L)]/[Platelets (10 9 /L) × ALT (IU/L)] 23 . Cut off values used for significant fibrosis (Metavir ≥ F2) were APRI > 1.0 and FIB-4 > 3.0. Transient elastography was performed by an experienced investigator after at least 4 hours of fasting using FibroScan ® (Echosens; Paris, France). The standard M probe was used in all except three patients that required the XL probe due to BMI ≥ 30 Kg/m 2 . A liver stiffness result was considered reliable if the interquartile range (IQR) was <30% and the success rate was ≥70%. Significant fibrosis (Metavir ≥ F2) was estimated using a liver stiffness result cut-off ≥7.1 kPa 7,13 . Steatosis was estimated using a controlled attenuation parameter (CAP) cut-off ≥238 dB/m 7,11 . Subjects with TE ≥ 7.1 kPa were proposed a liver biopsy. Histological evaluation was performed by an expert liver pathologist using the NAFLD Scoring System (NAS), which is a result of the unweighted sum of scores of steatosis (0-3), lobular inflammation (0-3), and hepatocellular ballooning (0-2) ranging between 0-8. A calculated NAS ≥ 3 was considered as diagnosis of NASH 24 .
Statistical analysis. Statistical tests were performed using The Statistical Package for Social Science version 18.0 (SPSS, Chicago, IL). Continuous variables were expressed as median (IQR), and categorical variables were presented as numbers (percentage). Mann-Whitney's U test was applied for comparisons of continuous variables between groups. Comparisons between categorical variables were made by the chi-square test or Fisher's exact test, when appropriate. Poisson regression model was used to identify factors associated with hepatic steatosis and to detect any potential confounders. This model was chosen since it is the most accurate in estimating the prevalence ratio (PR) in cross sectional studies 25 . Only variables with a p ≤ 0.10 at univariate analysis were entered in the multivariate model. Associations with a p ≤ 0.05 were considered statistically significant.

Results
A total of 98 HIV infected patients were included. Mean age was 49 ± 11 years and 53 (54.1%) were male. Patients characteristics such as demographics, comorbidities and laboratory results are summarized in Table 1. Fatty liver was detected by ultrasound in 31/98 (31.6%). The most common ART regimen used was Tenofovir (TDF) + Lamivudine (3TC) + Efavirenz (EFV) in 23.4% of the patients, followed by Zidovudine (ZDV) + 3TC + EFV in 20.4%. Triglyceride levels ≥150 mg/dL were detected in 54 (55.1%) patients, total cholesterol levels ≥200 mg/dL in 36 (36.7%), and ALT above the upper limit of normal (ULN) in 23 (23.5%). In univariate analysis, fatty liver was associated with male gender, BMI, IFG/diabetes, metabolic syndrome, triglycerides, LDL, total bilirubin and ALT levels, and non-invasive fibrosis scores (APRI and FIB-4). In multivariate analysis fatty liver was independently associated with male gender, BMI, triglycerides, total bilirubin and ALT levels (Tables 2 and 3).
Among the 31 patients with hepatic steatosis, prevalence of significant fibrosis (Metavir ≥ F2) assessed by serum biomarkers (APRI and FIB-4) was 6.4% and 3.2%, respectively. Transient elastography was performed in 26 of the 31 patients with hepatic steatosis, while five patients refused the procedure. Significant fibrosis (Metavir ≥ F2) assessed by TE was found in 7 (26.9%) patients and was associated with triglyceride levels, FIB-4 score and CAP values ( Table 4). Six of these patients agreed to undergo a liver biopsy, which was found compatible with NASH in 5 (83.3%) and with mild liver fibrosis without NASH in one. No biopsied patient had significant fibrosis ( www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
The present study investigated the prevalence and associated risk factors for liver steatosis among stable HIV infected outpatients in a tertiary care center in Southern Brazil. Furthermore, we explored the role of TE to identify a subgroup of patients at risk for NASH and/or liver fibrosis among this population. Using abdominal US, hepatic steatosis was detected in almost a third of our sample. This finding is similar to the prevalence reported in a recent systematic review and meta-analysis conducted among HIV monoinfected patients from several countries worldwide, which found NAFLD using mainly US imaging in 35.3% (95% CI 28.8-42.5) 26 . This prevalence seems to be at least 10% higher than what has been usually reported in the general population. Indeed, a recent meta-analytic assessment of the global NAFLD epidemiology among a large group of unselected individuals detected liver steatosis in only 25.2% (95% CI 22.1-28.6) 4 . Nevertheless, one could assume that the true prevalence of NAFLD is probably higher, since abdominal US is known to detect liver steatosis only when more than 20-30% of hepatocytes are affected 23 . In our study, hepatic steatosis was independently associated with male www.nature.com/scientificreports www.nature.com/scientificreports/ Among our group of 26 patients with steatosis that underwent TE, 7 (26.9%) had liver stiffness above 7.1kPa. This result is comparable with that found by other authors that used TE cut-offs between 7.0 and 7.4 kPa as a surrogate for significant fibrosis among HIV infected patients, and found 15% to 27.3% individuals within that range 5,7,23 . Remarkably, none of our six patients with liver stiffness above 7.1kPa that underwent liver biopsy had significant fibrosis. This finding suggests that TE may overestimate fibrosis staging in HIV patients with hepatic steatosis detected by US. On the other hand, all except one of our biopsied patients had NASH, indicating that TE was capable of identifying a subgroup at risk for this condition among HIV patients with liver steatosis. Future studies could explore different cut-off values for TE in this population in order to improve its usefulness as a tool to detect not only NASH, but also significant fibrosis.
Interestingly, there was discordance in our sample between TE and APRI and FIB-4. Indeed, the majority of the patients with TE ≥ 7.1kPa had normal FIB-4 and APRI values. This observation was also reported by other authors and highlights the difficulty of assessing correctly the liver fibrosis staging in this population without the use of an invasive method such as a liver biopsy 5,22 .
Our study has some limitations: it is a cross sectional study, so it was not possible to define the consecutive steps leading to the development of both steatosis and fibrosis in HIV infected patients; only a few of the NAFLD patients agreed to undergo liver biopsy, so we cannot rule out the occurrence of NASH and/or significant fibrosis among the remaining patients that refused the procedure; NAFLD fibrosis score (NFS) was not calculated due to the fact that some variables used for its calculation were not routinely collected for most of the patients that comprised this cohort. Finally, the small sample size among subjects submitted to TE prevented us to further explore multivariate analysis in this subgroup of patients.
In conclusion, it is important to realize that TE using a cut-off of 7.1 kPa was able to accurately detect patients with high risk for NASH among HIV monoinfected individuals with NAFLD. The significance of this finding should be validated in larger cohorts, preferably with long term follow-up. If the ability of TE to correctly identify NASH in this population is confirmed, invasive procedures such as liver biopsy could be reserved for identification of significant fibrosis, since TE was not accurate enough to identify this condition. Further studies should explore different TE cut-offs and/or combinations of non-invasive methods to better select patients at risk for significant fibrosis, because this individuals are the ones at higher risk for progressive liver disease and are the best candidates for more aggressive medical management to avoid deleterious outcomes.