The impact of cART and other risk factors on the development of liver fibrosis in patients infected with HIV has been subject to controversial debate over the past years. Several studies indicate that long-term cART may worsen liver fibrosis in patients with HIV/HCV co-infection and that this effect is mostly seen with NRTIs. However, effective control of HIV, especially through the introduction of PIs has recently been shown to be associated with slower liver fibrosis progression in patients with HIV/HCV co-infection who underwent paired liver biopsies [6
]. Moreover, little is known about risk factors in patients without hepatitis co-infection and few data exist on liver-related long-term complications of cART [3
In our prospective study of HIV infected individuals, mostly without hepatitis co-infection (76% of all patients), elevated ALT, AST and γ-GT levels were observed in 19%, 8% and 45.5% of those without viral hepatitis, suggestive of the presence of cART-related hepatotoxicity and/or metabolic disorders. However, only one patient fulfilled the Hy's rule criteria (ALT > 3xULN and total bilirubin > 2xULN) [23
]. Despite there being no generally accepted definition of drug-induced hepatotoxicity, elevations of liver transaminases (ALT and AST) are currently considered as the best indicator [5
When using TE for fibrosis assessment, 16% and 5% of patients were predicted to have significant liver fibrosis and cirrhosis, respectively. Interestingly, this was not exclusively attributable to HCV co-infection, as 42% and 30% of patients with liver stiffness suggestive of significant fibrosis and cirrhosis were indeed HCV RNA negative.
Recently, Castera and co-workers have proposed an algorithm for non-invasive fibrosis assessment that includes a combination of TE and FT (Bordeaux algorithm). When combining the two modalities, the presence or absence of significant liver fibrosis and cirrhosis was predicted with an accuracy of 87.7% and 95.7%, respectively [22
When we applied the above-mentioned criteria to our data, only 8% and 3% of our patients had significant liver fibrosis or cirrhosis. Moreover, in patients with HIV mono-infection, the proportion of patients with significant fibrosis and cirrhosis was even smaller (3% and 0.6%, respectively).
Thus, despite a high prevalence of chronic elevated ALT, AST or γ-GT levels, there was only evidence for liver fibrosis in a small proportion of HIV mono-infected patients. It should be noted however, that the Bordeaux algorithm has not yet been validated in HIV-infected individuals with or without HCV co-infection.
In a longitudinal analysis, liver stiffness values did not differ significantly over a median follow-up time of 24 months in patients with HIV mono-infection, indicating that patients on cART may not be at risk of progressive liver disease. However, the rather short follow-up period may impose a latency bias on this observation as the actual mean rate of liver fibrosis progression within one year has been estimated to be as low as 0.085-0.120 for fibrosis stages derived according to the METAVIR scoring system in patients with chronic HCV infection [25
] and it may be even lower in patients with drug-induced fibrosis or steatohepatitis.
Overall, in patients in whom liver fibrosis was predicted by TE, factors independently associated with significant fibrosis included HCV-coinfection, chronic AST and γ-GT elevation as well as exposure to dideoxynucleosides and these results are in line with previous observations [8
]. With regard to cART-related hepatotoxicity, several specific antiretroviral agents such as nevirapine, zidovudine, stavudine, and other nucleoside analogs have been associated with liver fibrosis progression, especially in patients with HIV/HCV co-infection [8
]. Nevirapine in particular has been associated with ALT and γ-GT elevation and this may account for the high prevalence of chronic γ-GT elevation in our study as almost half of those with γ-GT elevation were taking nevirapine at the time. This may also explain the higher percentage of significant liver fibrosis predicted by Fibrotest that combines the results of five blood serum tests, including γ-GT.
The potentially hepatotoxic effect of dideoxynucleoside-induced mitochondrial toxicity has been discussed previously [28
]. The reduction of mitochondrial beta-oxidation of fatty acids may lead to hepatic steatosis, which, over time, can lead to chronic inflammation and liver fibrosis [30
]. It must be noted, however, that dideoxynucleosides have been widely replaced by newer NRTIs with superior liver safety profiles and the use of dideoxynucleosides is rapidly diminishing. This may also explain why there was no increase in liver fibrosis observed in our longitudinal study. Moreover, when using both TE and FT, cumulative time on dideoxynucleosides was only associated with significant fibrosis in the univariate analysis.
While liver steatosis and steatohepatitis may develop as a direct consequence of NRTI-toxicity, these conditions can also be associated with cART-related metabolic disorders, including hyperglycemia, insulin resistance, dyslipidemia, and obesity. Indeed, high HOMA-IR and BMI were associated with predicted liver fibrosis when assessed by TE and/or FT. However, their impact did not remain significant in the multivariate models.
In several studies, excessive alcohol use was also shown to be associated with liver fibrosis in the HIV population [8
]. In the present study, we excluded patients with alcohol abuse (more than 2 standard drinks in men/one standard drink in women) in order to minimize the alcohol-related impact on liver fibrosis. Indeed, alcohol consumption below this threshold was not associated with increased liver stiffness in this study.
A limitation of the present study is the lack of fibrosis assessment by liver biopsy. However, the role of liver biopsy as "gold-standard" is subject to increasing debate, owing to possible side effects as well as frequently occurring sampling errors and variability of sample interpretation [33
]. Furthermore, liver biopsy is rarely recommended in patients with HIV mono-infection.