Clinical syndromes and consequences of antiretroviral-related hepatotoxicity5, 1, 3


  • Marina Núñez

    Corresponding author
    1. Wake Forest University Health Sciences, Winston Salem, NC
    • Department of Internal Medicine, Section on Infectious Diseases, Wake Forest University Health Sciences, Medical Center Boulevard, Winston Salem, NC 27157
    Search for more papers by this author
    • fax: 336-716-3825

  • 5, 1, 3

    Potential conflict of interest: Nothing to report.


Highly active antiretroviral therapy (HAART)-related hepatotoxicity complicates the management of patients infected with human immunodeficiency virus (HIV), increases medical costs, alters the prescription patterns, and affects the guideline recommendations. Among the clinical consequences derived from HAART-related liver toxicity, hypersensitivity reactions and lactic acidosis are recognized as acute events with potential to evolve into fatal cases, whereas there seems to be other syndromes not as well characterized but of equal concern as possible long-term liver complications. Belonging to the latter category of syndrome, HAART-related nonalcoholic steatohepatitis, liver fibrosis, portal hypertension, and nodular regenerative hyperplasia are discussed in this review. Updated information on liver toxicity of current antiretroviral drugs, including the most recently licensed, is provided. Management and prevention of liver toxicity among HIV-infected patients treated with HAART are reviewed as well. (HEPATOLOGY 2010;52:1143–1155)

Physicians treating human immunodeficiency virus (HIV)-infected patients often deal with aminotransferase elevations which have to be interpreted and managed. New hepatic problems which might be related to the use of highly active antiretroviral therapy (HAART) continue to be revealed. In addition, new antiretrovirals have been licensed for which information on liver safety is limited. I refer to past reviews for previous information on the subject.1-8 After those publications, prescription patterns and guideline recommendations have continued to evolve, and physicians treating HIV in 2010 manage new antiretroviral drugs and new aspects of the epidemics.9 This review focuses on the clinical consequences of liver toxicity associated with HAART, updates information on the subject, and includes liver safety data of most recently approved antiretroviral drugs. This article aims to help health care providers prevent and handle antiretroviral toxicity within contemporary management of patients with HIV.


ALT, alanine aminotransferase; d-drug, dideoxynucleoside drug; FDA, U.S. Food and Drug Administration; HAART, highly active antiretroviral therapy; HBV, hepatitis B virus; HCV, hepatitis C virus; HIV, human immunodeficiency virus; HLA, human leukocyte antigen; NASH, nonalcoholic steatohepatitis; NNRTI, non-nucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; ULN, upper limit of the normal range.


There is not an uniform and internationally accepted definition of drug hepatotoxicity or drug-induced liver injury, another term used to refer to the liver disturbances caused by drugs. Although alkaline phosphatase elevation can be also a marker of liver toxicity (and it is very prominent in cases with a mixed or cholestatic pattern), aminotransferase elevation reflecting hepatocellular injury is more commonly used as definition of hepatotoxicity.10 The AIDS Clinical Trials Group criteria11 grades it according to the following score system: grade 1 (1.25×-2.5× upper limit of the normal range [ULN]); grade 2 (2.6×-5× ULN); grade 3 (5.1×-10× ULN); and grade 4 (>10× ULN). Some authors have proposed to score HAART-related hepatotoxicity according to baseline levels in subjects having abnormal liver enzyme values at baseline: grade 1 (1.25×-2.5× baseline); grade 2 (2.6×-3.5× baseline); grade 3 (3.6×-5× baseline); and grade 4 (>5× baseline).12 The presence of jaundice along with high aminotransferase levels is associated with a poor prognosis (≥10% mortality), a phenomenon known as ”Hy's rule” in honor of the pioneer researcher Hyman Zimmerman. The validity of Hy's rule, first proposed in the 1970s has been confirmed by more recent studies.13

Challenges in Causality Assessment

There is difficulty discerning and dissecting out the number of individual factors that may contribute collectively to liver damage in patients receiving antiretroviral therapy. Several drugs are combined in a given HAART regimen making difficult the attribution of hepatotoxicity to a particular drug. Moreover, HIV-infected patients may be receiving concurrent medications with potential for liver toxicity as well, such as antimycobacterial drugs, lipid-lowering agents, antifungals, antibiotics, and anticonvulsants. It is also difficult to make comparisons among reported cohorts, because individuals often differ on the factors predisposing to elevations of liver enzymes, like the presence/absence of concurrent viral hepatitis. Biochemical, pharmacokinetic/dynamic, and pathological correlations of HAART hepatotoxicity have been poorly characterized, which makes it often difficult to determine the true incidence of drug-induced liver injury. In many instances, the hepatotoxic potential of a drug has been recognized only after post-marketing experience with the drug.

Risk Factors for HAART-Related Hepatotoxicity

Chronic viral hepatitis has been consistently reported to increase the risk of severe HAART hepatotoxicity (relative risk = 2.1).14 There is an estimated 2.7-fold to 5-fold increased risk of severe alanine aminotransferase (ALT) elevation on HAART with hepatitis C virus (HCV) coinfection.15-17 Chronic hepatitis B virus (HBV) infection appears to also carry a higher risk, with a 9.2 hazard risk of grade 4 liver enzyme elevations reported in one study.17 The same authors also observed that discontinuing lamivudine, an antiretroviral also active against HBV, was a factor associated with aminotransferase elevation in HIV/HBV-coinfected patients (hazard risk = 6.8 for grade 4 liver toxicity).17 The presence of underlying liver inflammation as reflected by elevated ALT at baseline has been also identified as a risk factor for HAART liver toxicity.16, 17 Isolated studies have identified additional host factors including older age, female sex, thrombocytopenia, renal insufficiency, high HIV RNA levels, increased body mass index, and non-black ethnicity.15-18

Aside from host factors, several individual antiretrovirals or classes have been independently associated with HAART hepatotoxicity, such as nevirapine, protease inhibitors, high doses of ritonavir (≥600 mg/day), and prolonged zidovudine or stavudine exposure.14, 17, 18 Alcohol use and concurrent hepatotoxic medications are additional factors identified.15, 16, 18 Lastly, an increase in CD4 cell counts of >50 cells/mm3 after HAART initiation was associated with almost two-fold increased risk of severe ALT elevation in one study.14 Other risk factors individual to each pathogenic mechanism are covered within its corresponding section.

Mechanisms of HAART-Related Hepatotoxicity

A major challenge for mechanistic classifications is that the pathogenesis of drug hepatotoxicity is poorly understood in many instances. It involves several mechanisms, regulatory systems, and risk factors with complex interactions.19 Hepatotoxicity events are more often idiosyncratic, that is, they are unpredictable and occur with variable latency and low incidence.10 Idiosyncratic drug-induced liver injury can be further classified as allergic and nonallergic.20 The pathogenesis of drug hepatotoxicity involves exposure to the toxic agent (the parent drug or most often a reactive metabolite), the amount of which depends on genetically determined metabolism of the agent by the liver. Following exposure, the toxic moiety induces some type of stress or functional disturbance, with mitochondrial injury being one of the most important targets recognized.21, 22 A number of adaptation mechanisms are then initiated to counteract the inflicted damage.23, 24 In addition, innate and adaptive immune responses are other factors of interest which determine the progression and severity of liver injury.25, 26 Detailed reviews focusing on pathogenesis and mechanisms of drug-induced liver injury are available elsewhere.10, 19, 20, 27

Liver toxicity caused by antiretroviral therapy can be inflicted through several mechanisms. The pathogenesis often remains enigmatic. Table 1 summarizes the mechanisms of HAART-related liver toxicity by antiretroviral class. Five categories are proposed: hypersensitivity reactions, direct mitochondrial inhibition, disturbances of lipid/sugar metabolism and steatosis, direct cell stress, and immune reconstitution in the presence of viral hepatitis coinfection. Despite the limitations of the classification, which ultimately is merely descriptive, it may be useful in clinical practice because it describes typical clinical characteristics of hepatotoxicity for specific antiretrovirals or classes and might give hints on the mechanism, ultimately helping the management.

Table 1. Mechanisms of liver toxicity by antiretroviral class
Antiretroviral ClassMechanisms of HAART-Related Liver Toxicity
Hypersensitivity ReactionsMitochondrial ToxicityLipid/Sugar Metabolism Disturbances and SteatosisDirect Liver Cell StressImmune Reconstitution in Viral Hepatitis Coinfection(s)
  1. D-drugs: dideoxinucleosides (didanosine, stavudine, zalcitabine); zalcitabine has been removed from the market.

Nucleos(t)ide reverse transcriptase inhibitorsAbacavir (associated with HLB5701)Possible for all D-drugs > the othersD-drugs likely Other?DidanosineAll
   Reported with:  
Non-nucleoside reverse transcriptase inhibitorsAll Possible for allNevirapine EfavirenzAll
Protease inhibitorsFosamprenavir Darunavir Possible for all Ritonavir-boosted PIs > non-ritonavir-boosted PIsRitonavir TipranavirAll
Fusion inhibitor    All
CCR5 blockerMaraviroc   All
Integrase inhibitor    All

As reflected in Table 1, some antiretrovirals or classes may be toxic for the liver through different pathways, a feature which is characteristic of drug-induced hepatotoxicity in general.19 Immune reconstitution in the setting of viral hepatitis is a mechanism of aminotransferase elevation shared by all antiretrovirals, just because is the result of an effective HAART.28 Disturbances in lipid and sugar metabolism which seem to be contributors to a not well-defined steatohepatitis syndrome can be caused by all or several members in three antiretroviral classes: nucleoside reverse transcriptase inhibitors (NRTIs), non-nucleoside reverse transcriptase inhibitors (NNRTIs), and protease inhibitors (PIs).29 Mitochondrial liver toxicity leading to steatosis and lactic acidosis, which is secondary to mitochondrial RNA depletion by NRTI use, is particular to that class.30 Hypersensitivity reactions with liver involvement are common to NNRTIs but are possible also for specific drugs in other classes.31-37 Direct liver cell stress, which is dose-dependent, seems to be the underlying mechanism of liver toxicity of ritonavir and tipranavir.8, 38, 39 Of note, higher boosting doses of ritonavir (400 mg/day) increase toxicity, whereas liver safety using ≤200 mg/day is comparable to that of nonboosted PIs.8, 38

Drug-to-drug interactions can increase the potential for hepatotoxicity as in the case of stavudine and didanosine.40, 41 In a similar manner, ribavirin, which is given for the treatment of HCV, should not be given concomitantly with stavudine because there is risk of liver decompensation, especially in cases of advanced liver disease.42

Liver Safety Profile of FDA-Approved Antiretroviral Drugs

Table 2 summarizes the information provided in the package inserts of all antiretroviral drugs approved to date by the U.S. Food and Drug Administration (FDA).31-37, 43-58 Unfortunately, the information is heterogeneous and often scattered, which makes it difficult to compare the potential for hepatotoxicity across antiretrovirals. Most recently marketed drugs tend to provide separate data for subjects coinfected with viral hepatitis, whereas prescribing information of older drugs includes information on liver side effects which only have been revealed after extensive use. The liver-related black box warnings (see note at the end of this article) are based on reports of lactic acidosis with liver damage in the case of NRTIs, on hypersensitivity reactions secondary to drugs from three different classes, and on direct toxicity for didanosine (NRTI) and tipranavir/ritonavir (PI). Mitochondrial liver toxicity has motivated a mandatory black box warning across the class, although the potential toxicity is not homogeneous for the various NRTIs.

Table 2. Information on Hepatotoxicity Reported in the Full U.S. Prescribing Information of FDA-Approved Antiretrovirals
Drug (Reference)Severe ALT Elevation* (%)Increased Risk in HCV/HBV CoinfectionOther Groups with High Risk of HepatotoxicityHepatic Failure/ Liver FatalitiesLiver-Related Black Box Warning
  • *

    Grade 3-4 AIDS Clinical Trials Group (ACTG) ALT elevation (>5× ULN) unless stated otherwise.

  • **

    Grades 2-4 ACTG ALT elevation (>2.5× ULN).

  • ***

    Defined as shifts from grade 0 at baseline to at least grade 3 (>5× ULN) or from grade 1 (1.25×-2.5× ULN) to grade 4 (>10× ULN).

  • Symptomatic events: 1%-11%.

  • ¶¶

    Ritonavir boosting doses most often 200 mg/day except with atazanavir (100 mg/day) and tipranavir (400 mg/day).

  • ¶¶¶

    Includes data on 400 mg/day dose of ritonavir.

  • 1

    Hypersensitivity reactions with liver involvement.

  • 2

    Lactic acidosis.

  • 3

    Increased risk of liver failure with/without fatal outcome in patients with underlying liver disease.

  • 4

    Risk of liver decompensation in patients treated for HCV with interferon and ribavirin.

  • 5

    Hepatic failure reported postmarketing.

  • 6

    Contraindicated in patients with moderate-severe liver impairment.

  • ALT, alanine aminotransferase; D4T, stavudine; DDI: didanosine; FDA, U.S. Food and Drug Administration; HU, hydroxyurea.

 Abacavir (10)6%Yes12Yes12
 Didanosine (11)6%Yes+HU,+D4T, pregnancyYes23Yes2
 Emtricitabine (12)2%-5%NoYes2
 Lamivudine (13)3.7%-3.8%Yes2Yes2
 Stavudine (14)6%-13%Yes+DDI ± D4T, pregnancyYes234Yes2
 Tenofovir (15)4%NoYes2
 Zidovudine (16)3.1%Yes2Yes2
 Delavirdine (17)4.1%-5.1%Yes5No
 Efavirenz (18)2%-8%20%+Other hepatotoxic drugsYes5No
 Etravirine (19)2.6%YesYes15No
 Nevirapine (20)5.3%-14%YesHigher CD4counts, F>MYes136Yes1
Protease inhibitors¶¶     
 Atazanavir/ritonavir (21)3%-9%15%NoNo
 Darunavir (22)5.6%-6.9%**YesYes3No
 Fosamprenavir/ritonavir (23)4%-8%PossiblyNon-HIV postexposure prophylaxisNoNo
 Indinavir (24)2.6%-4.9%YesNo
 Lopinavir/ritonavir (25)3%-11%¶¶¶PossiblyYes3No
 Nelfinavir (26)1%-2%***NoNo
 Ritonavir (27)5.3%-8.5%PossiblyYes3No
 Saquinavir (28)Not providedYes+RifampinUnclearNo
 Tipranavir/ritonavir (29)9.7%10.3-20.3%Yes36Yes
Fusion inhibitors     
 Enfuvirtide (30)5.3%-6.2%NoNo
CCR5 blockers     
 Maraviroc (31)2.4%Insufficient dataYes1Yes1
Integrase inhibitors     
 Raltegravir (32)4%27%**NoNo

Clinical Syndromes of HAART-Related Hepatotoxicity

The full extent of hepatic damage related to HAART use has not been fully elucidated. Although there are acute events which have been clearly recognized, other syndromes are less evident at this time. The most relevant issues are addressed in this section.

Acute Events

Hypersensitivity Reactions.

Hypersensitivity reactions are idiosyncratic reactions of the host, not related to the dose of the drug, and are immune-mediated. They involve the generation of neoantigens formed by the reaction of liver proteins with reactive drug metabolites.59 They usually occur within the first 4-6 weeks of treatment. These hypersensitivity reactions with liver involvement have resulted in black box warnings for three drugs: nevirapine, abacavir, and maraviroc. Acute hepatitis leading to liver failure with fatal outcome in the context of a hypersensitivity drug reaction has been reported with nevirapine and abacavir, both in HIV-infected patients and in subjects receiving prophylaxis after HIV exposure.60-64 Nevirapine discontinuation due to hypersensitivity-related skin rash occurs in 5%-7% of patients.65-68 It is unknown how many of those allergic reactions are accompanied by liver involvement; however, statistically significant association between the two events has been reported, with skin rash preceding liver toxicity.69 Abacavir discontinuations due to hypersensitivity reactions range between 5% and 8%.70, 71 With widespread human leukocyte antigen (HLA)-B*5701 testing, the incidence is expected to be currently much lower (see below). There has been one reported case of fatal hepatitis with allergic features possibly related to maraviroc in an HIV-1–negative volunteer.36 Cases of acute hepatitis and liver failure with characteristics of hypersensitivity reaction have also been reported with efavirenz.72, 73 Severe elevation of aminotransferases along with skin rash has been reported as well in HIV-negative individuals receiving ritonavir-boosted fosamprenavir as part of a postexposure prophylaxis regimen.74 Aminotransferase elevation in the setting of skin rash suggesting a hypersensitivity reaction has also been reported with darunavir, which has prompted a warning by the FDA.75

The risk of hypersensitivity-related severe hepatic toxicity with a nevirapine-based regimen, in some occasions with fatal outcome, is higher among patients naïve for antiretroviral therapy with CD4 counts >250 cells/mm3 for women and >400 cells/mm3 for men, but can also occur with any CD4 count.76 They occur almost exclusively within the first 6 weeks of nevirapine treatment. Low body mass index is an independent risk factor for this type of event.69 Individuals with HLA-DRB1*0101 and >25% CD4 cells have the greatest risk of developing nevirapine -induced hypersensitivity reactions.77 Other HLA haplotypes have been reported more recently to be associated with a higher risk of hypersensitivity reactions to nevirapine.78, 79 Some authors have suggested that these events do not occur in antiretroviral-experienced patients in whom nevirapine is initiated, although this seems to be the case only when HIV load is undetectable.65, 80, 81 Of note, life-threatening events of hypersensitivity hepatotoxic reactions in non–HIV-infected subjects who took nevirapine as part of postexposure prophylaxis regimens led to contraindicating the drug in that setting.82

A genetic predisposition has been identified for hypersensitivity reactions to abacavir, which can be identified through a pharmacogenetic test.83 Thus, HLA-B*5701 screening has been shown to reduce the risk of hypersensitivity reaction to abacavir, with a reported negative predictive value for immunologically confirmed hypersensitivity reaction of 100% in white and black individuals.70, 84 The positive predictive value of a positive HLA-B*5701 result is much lower, with 55% of those carrying the allele able to tolerate abacavir; however, it is not recommended to use abacavir in those at risk.70

Lactic Acidosis with Hepatic Steatosis.

There is an infrequent but distinctive type of severe hepatotoxicity involving mitochondrial damage, clinically defined by lactic acidosis and hepatic steatosis which evolves to acute liver failure and carries a high mortality. The main feature of the hepatic lesion is microvesicular or macrovesicular steatosis and mitochondrial depletion in liver cells. Focal necrosis, fibrosis, cholestasis, proliferation of biliary ducts, and Mallory bodies may appear if the process evolves.30, 59, 85, 86 Mitochondrial abnormalities are seen by electron microscopy.87 NRTIs are implicated as causal agents, which subsequently served as motivation for the generalized black box warning for all NRTIs regardless of each drug's potential for mitochondrial toxicity. The pathogenesis of this syndrome has not yet been completely elucidated. Severe mitochondrial injury of the hepatocytes secondary to NRTIs has been reported in asymptomatic patients with normal lactic acid levels and in the absence of steatohepatitis.88

The hepatic abnormalities in lactic acidosis secondary to NRTI toxicity have been described in a systematic review of cases reported in the literature which included 90 patients with lactic acidosis.85 Laboratory evidence of mild to moderate hepatic dysfunction was present in 41 of the 63 cases (65%) in whom information was given, with median (range) aminotransferase values between 1.5 and 2.5 (1.4-10.7) times above the ULN. Of 39 premortem or necropsy liver biopsies, 36 (92%) had hepatic steatosis: macrovesicular steatosis in 12 (31%), microvesicular steatosis in eight (21%), and with a mixed pattern in 16 (41%). The other three biopsies showed inflammation and hepatic fibrosis. Mortality was 48% in this review of cases.

Lactic acidosis has been reported in persons receiving both single-NRTI and dual-NRTI regimens including combinations of zidovudine or stavudine with didanosine, zalcitabine, or lamivudine.86 The role of each specific NRTI in the development of lactic acidosis is often difficult to determine because the patients might have been exposed to several NRTIs and frequent changes in medications are made. Nevertheless, it is known that the dideoxynucleosides (d-drugs) have a higher potential for mitochondrial toxicity with greater ability to inhibit mitochondrial DNA synthesis in vitro and in vivo.30, 85, 86, 89 Several cohorts suggest that the coadministration of stavudine and didanosine is associated with the greatest relative risk.40, 41 Of note, this drug combination is contraindicated by the guidelines due to high risk of lactic acidosis.9 Hydroxyurea, which was used in the past as adjuvant treatment with didanosine, increases its toxic effect due to the rise of intracellular levels of 5′-triphosphate products.90, 91 Cumulative exposure to NRTI is another factor believed to be important for the development of lactic acidosis.85, 92 In addition, lactic acidosis appears to be more common in women and the obese.85, 86, 93 An increased risk of lactic acidosis among pregnant women being treated with didanosine and stavudine has been also reported.43, 46 A contribution to the pathogenesis of lactic acidosis of HIV and HCV has been suggested but it has not been established.30, 85, 93

Progressive Liver Damage

Increased Liver Fibrosis in Patients with Chronic HCV?

In a retrospective and cross-sectional study in which liver biopsies from 152 HIV/HCV-coinfected patients were evaluated, associations between accelerated fibrosis progression and nevirapine, and between slower fibrosis progression and PI use were found.94 As recognized by the authors, this study has several limitations, including biases in the selection of the study population and lack of evidence of linear progression of HCV-related liver fibrosis. The number of patients in the nevirapine and efavirenz groups was low. In addition, the effect of NRTIs was not evaluated, and the variables exploring the effect of antiretroviral drugs on liver fibrosis were categorical, and therefore did not take into account the duration of exposure. Three other retrospective cross-sectional studies do not support those results.95-97 Therefore, based on the available data, we cannot affirm that nevirapine accelerates liver fibrosis progression in HIV/HCV-coinfected patients.

For the effect of antiretroviral therapy to be assessed, it is necessary to take into account additional factors which may have opposite effects on fibrosis progression rate. Thus, adequate control of HIV replication has been shown to be associated with lower necroinflammatory scores, slower liver disease progression, and decreased mortality, whereas alcohol intake contributes to more advanced fibrosis.96-99 Therefore, in order to determine a possible negative impact of antiretroviral drug(s) on the liver disease of HIV/HCV-coinfected patients, longitudinal studies with pathology information and inclusion of multiple factors in the analysis would be most valuable. The role of transient elastography as a noninvasive tool for monitoring of liver disease progression remains to be elucidated.

Portal Hypertension and Nodular Regenerative Hyperplasia.

Of more concern is the report by Spanish authors of nine cases of portal hypertension complicated by variceal bleeding, ascites, or hepatic encephalopathy without known underlying liver disease.100, 101 Five patients were thought likely to have fibrosis, either through liver biopsy or transient elastography. Of note, portal thrombosis occurred in six cases. All patients had maintained prolonged viral suppression under HAART. Through a case-control study, the researchers identified prolonged didanosine use as the only factor associated with these cases of cryptogenic liver disease.

In a separate report, French authors described eight HIV-infected patients who developed portal hypertension, and liver biopsy revealed nodular regenerative hyperplasia.102 As a result, three of the patients were included in a liver transplant list. Like in the Spanish cases, all patients had well-controlled HIV replication and had been exposed to didanosine. The authors discuss that nodular regenerative hyperplasia appears to have a vascular etiology, with occlusion of terminal branches of the hepatic arterioles and portal venules. They speculate that HIV infection and antiretroviral drugs, in particular didanosine, could contribute to the production of thrombotic intrahepatic phenomena leading to liver damage and portal hypertension.

The reports prompted other groups to communicate 23 additional cases of symptomatic liver disease which have been subsequently published.103-107 Common pathology findings for those cases included compressed liver cells, sinusoidal dilation (with or without nodular regenerative hyperplasia), and in some cases also fibrosis. In several cases, there was also portal thrombosis. More recently, focal fibrous obliteration of small portal veins has been recognized in some of these cases.108 Most of those cases had a history of didanosine use, often over 2 years. The growing cumulative data is of concern for a possible new form of liver toxicity of the drug. In a recent nested case-control study, the association of noncirrhotic portal hypertension with prolonged didanosine use was very robust.109 Interestingly, a report of ”fatal portal hypertension” had already been reported in 2001 by Australian investigators.110 The patient had been exposed to didanosine and stavudine for 14 months. Of concern, the subject had been off therapy for more than 2 years when he died due to variceal hemorrhage, suggesting that the changes in the liver leading to portal hypertension are irreversible. Further research is needed to sort out the link of didanosine (or dideoxynucleosides) with this complication, in particular, the question of which are the additional required predisposing factors. In this regard, several individuals in one of the case series of nodular regenerative hyperplasia had thrombophilic disorders.107

Nonalcoholic Steatohepatitis.

Nonalcoholic steatohepatitis (NASH) is the result of complex metabolic disturbances in which lipid and carbohydrate metabolysis pathways are altered.111 HAART has been shown to alter both metabolic systems.29 Although HAART-related NASH has thus far not been defined as a specific entity, there are data supporting the contribution of HAART to the development of liver steatosis which as a result can lead to inflammation and fibrosis.112

Steatosis in HIV-infected patients has been reported to be independently associated with the use of dideoxynucleosides and occasionally of other NRTIs.113-118 However, other studies have not found such an association.119, 120 NRTIs can cause mitochondrial toxicity and steatohepatitis in a condition reflective of diminished mitochondrial beta-oxidation of fatty acids.121, 122 In an in vitro study, incubation with high concentrations of stavudine can rapidly induce accumulation of lipids within rat hepatocytes.123 However, some authors have found no correlation between mitochondrial function or DNA and the presence of NASH.112 Steatosis may be part of a metabolic syndrome associated with HAART. Thus, hyperglycemia, overweight, and insulin resistance have been associated with liver steatosis in treatment-experienced HIV-infected patients.112, 114 Several studies assessing liver histopathology have found NASH in more than half the HAART-treated HIV-infected patients who underwent liver biopsy due to chronic unexplained aminotransferase elevation, some of them also with lipodystrophy.112, 124, 125 Significant liver fibrosis, and even cirrhosis, has been recognized in some of those patients.111 NASH can also worsen HCV-related liver disease.126-129 Thus, liver steatosis was reported to be present in 40%-61% of patients and associated with higher degrees of liver fibrosis in two HIV/HCV-coinfected cohorts.114, 119

Clinical Consequences of HAART-Related Hepatotoxicity

The clinical consequences of HAART-related hepatotoxicity are summarized in Table 3. Even if we were to assume that asymptomatic aminotransferase elevation is not clinically relevant for the patient, it at least increases costs due to additional tests and clinic visits, and medication changes. It also alters the prescription patterns and has an impact on the recommendations of antiretroviral treatment issued in official guidelines.

Table 3. Negative Clinical Implications of HAART-Related Hepatotoxicity
Clinical PracticeClinical Outcomes
Additional visits and tests for work-upAcute events:
Hospital admissions Symptomatic hepatitis
Increased costs Acute liver failure with or without death
Package inserts increase patient anxietyChronic syndromes
Changes of HAART regimens Portal hypertension/nodular regenerative hyperplasia
Changes in prescription patterns Liver fibrosis
Impact on treatment guidelines Hepatic steatosis/steatohepatitis

HAART hepatotoxicity may have devastating consequences. Although infrequent, symptomatic acute hepatitis can evolve into liver failure and result in death. As elaborated in previous sections, some ”chronic hepatotoxicity syndromes” are also of concern and can lead to severe liver complications and death.

Prevention and Management of HAART Liver Toxicity

General rules for the management of severe HAART hepatotoxicity (grades 3 and 4) are summarized in Fig. 1. The prevention and management strategies addressing specific HAART hepatotoxicity syndromes are outlined in the following sections.

Figure 1.

Algorithm for the management of severe HAART-related toxicity.

Hypersensitivity Reactions with Liver Involvement.

HLA-B*5701 screening is an effective way to prevent exposure to abacavir in susceptible subjects.70, 84 A close follow-up and selection of patients with lower CD4 counts in antiretroviral-naïve patients can minimize the risk of nevirapine-related idiosyncratic reactions with liver involvement. It is unclear if this also applies to treatment-experienced subjects, although it seems prudent to take the same precaution when there is not complete viral suppression.65, 80, 81 For other drugs able to cause liver hypersensitivity reactions, close follow-up is recommended during the first weeks of treatment, with liver enzymes tested if the patient develops an allergic rash.

Should a hypersensitivity reaction develop, the suspected drug and all the other components of HAART should be discontinued (Fig. 1). A new regimen can be restarted when symptoms resolve. If the patient develops hepatic failure, supportive treatment is recommended along with discontinuation of HAART and, if possible, other hepatotoxic drugs.9

Lactic Acidosis with Hepatic Steatosis.

Cases of lactic acidosis with acute hepatitis and hepatic steatosis are likely in decline because d-drugs have been displaced by NRTIs that are less toxic for the mitochondria. Thus, the preferred NRTI combination currently includes tenofovir, which does not affect mitochondrial DNA content or level of mitochondrial enzymes in liver cells, and emtricitabine, which has a low potential for mitochondrial toxicity.9, 30, 89, 130 Guidelines contraindicate the combination of two d-drugs.9 Individual use of those drugs should also be discouraged, considering the availability of many other compounds.

If severe lactic acidosis occurs in a clinical setting in which this syndrome is highly suspected, all antiretroviral drugs should be discontinued.9 However, it is important to note that the diagnosis is established by clinical correlations, drug history, and lactate level, and interpretation of high lactate level should be done only in the context of clinical findings. Management includes symptomatic support with fluid hydration, intravenous bicarbonate infusion, hemodialysis or hemofiltration, parenteral nutrition, or mechanical ventilation depending on the severity of the syndrome.30, 85, 86, 90, 92, 93 Intravenous administration of thiamine and/or riboflavin has been reported to rapidly resolve hyperlactatemia in isolated cases.131 After the acute phase, HAART can be resumed using NRTIs with less propensity for mitochondrial toxicity (e.g., tenofovir, lamivudine, emtricitabine, abacavir) or NRTI-sparing regimens.9 Close monitoring of serum lactate after restarting NRTIs has been recommended, although its interpretation has to be done in accordance with clinical status, because the meaning of elevated lactate levels in asymptomatic patients is unknown at this time.

Direct Liver Toxicity.

Antiretrovirals able to inflict direct liver cell stress can cause symptomatic hepatitis. HAART discontinuation is warranted (Fig. 1). However, other causes should be ruled out such as alcohol use, other hepatotoxic drugs, acute viral hepatitis, and in the presence of HBV coinfection, withdrawal of an active anti-HBV agent (i.e., lamivudine, emtricitabine or tenofovir) or development of HBV resistance. After symptoms subside and serum aminotransferases return to normal, a new antiretroviral regimen without the potential offending agent(s) can be constructed.

Whether asymptomatic patients with elevated aminotransferases in the presence of an agent with potential for direct hepatotoxicity should discontinue current HAART and start a new antiretroviral regimen without the offending agent is an undecided matter. Aminotransferase elevation >10× ULN even in the absence of symptoms is considered enough reason to stop the agent. However, although some physicians may consider discontinuing antiretrovirals if ALT level is >5×10× ULN, others may continue therapy with close monitoring unless direct bilirubin is also elevated.9 In selected cases, such as in the absence of other options due to extensive antiretroviral exposure and intolerance or resistance to other drugs, the latter option might be justified. In this era of availability of multiple antiretrovirals, maintaining a patient with chronic aminotransferase elevation on an intrinsically hepatotoxic antiretroviral is becoming less and less justified.

Patients with Concurrent Chronic Viral Hepatitis Infection(s).

Patients with concurrent HCV infection have higher risk of HAART-related aminotransferase elevation.1, 2, 5-7, 12 Although caution is recommended with NNRTIs in HCV-coinfected patients, the class should not be used in patients with cirrhosis, especially if Child-Pugh stage is B or C. Tipranavir, which is used with high doses of ritonavir for boosting, is contraindicated in patients with cirrhosis.9 Several other drugs should be used with caution in patients with underlying liver disease (Table 2). On the contrary, raltegravir has been shown to have an excellent liver safety profile in HCV/HIV-coinfected subjects.132

With HBV/HIV coinfection, a regimen which contains anti-HBV active drugs (tenofovir, emtricitabine, lamivudine) is recommended with the purpose of also controlling HBV replication.9 As long as that is achieved, patients should not have higher risk of HAART hepatotoxicity than those with HIV monoinfection. However, if cirrhosis is present, the same restrictions for tipranavir and the NNRTI class apply. In like manner, other drugs better suit HIV-infected subjects on concurrent treatment with drugs with high potential for hepatotoxicity (e.g., isoniazide).

Immune reconstitution that causes aminotransferase elevation in the presence of HBV-coinfection is a known phenomenon which results from increased T cell activation against viral particles.28 Elevated aminotransferases and high levels of HBV DNA at baseline seem to be predisposing factors.28 At present, there are no recommendations for the prevention of this type of event. However, because HBV DNA levels at week 4 of HAART treatment are higher in patients with hepatic flare-ups,105 achieving prompt and complete HBV suppression might be the best way to minimize these HAART-related hepatic flare-ups. That is more likely to be achievable with a regimen including tenofovir in patients with high HBV DNA levels. Should a hepatic flare occur in a HBV-coinfected patient, it is expected to spontaneously resolve while continuing on HAART, as long as control of HBV replication is achieved.


To prevent steatohepatitis, control of hyperglycemia, hyperinsulinemia, and hyperlipidemia should be pursued in patients with the metabolic syndrome. Certain antiretrovirals may help that purpose. At present, raltegravir is the HAART ”third agent” with the most benign lipid safety profile and should be strongly considered in patients with underlying obesity, insulin resistance, or lipid abnormalities. Unboosted atazanavir also has a good lipid safety profile, but its use without ritonavir places it in the category of ”acceptable regimen”, meaning that it may be selected for some patients but is a less satisfactory regimen.9 Alternatively, ritonavir-boosted atazanavir and ritonavir-boosted darunavir have the most favorable lipid safety profile among the boosted PIs.

The same recommendation applies to patients who already have developed NASH, in an attempt to minimize the hyperlipidemia. To date, NASH has proven to be a difficult disease to treat. Lifestyle modifications resulting in weight loss through decreased caloric intake and moderate exercise is generally believed to be beneficial in patients with NASH, but is often difficult to maintain in the long term.133 Given that insulin resistance plays a dominant role in the pathogenesis of NASH, many studies have examined the use of insulin sensitizers. Results with metformin (a biguanide) have been heterogeneous, most studies have been small and have flaws in the design, and its effect may be largely related to its ability to induce weight loss.134-138 Data with thiazolidinediones (pioglitazone, troglitazone, and rosiglitazone) for the treatment of NASH are more robust.139-142 However, it is unclear whether a thiazolidinedione-associated increase in adiposity and weight gain would ultimately limit its benefits.133 In addition, long-term toxicities of these agents include a potential for cardiovascular events and fracture risk. Therefore, we need more information regarding the efficacy and safety of these agents before recommendations for safe use can be made.

Portal Hypertension and Nodular Regenerative Hyperplasia.

Available information points toward didanosine as the antiretroviral agent linked to cases of noncirrhotic portal hypertension, which should discourage use of this agent.100, 109 With the continued decrease in its use, this complication should fade away and disappear over time. A conclusion which may be drawn from these cases of nodular regenerative hyperplasia is the need to obtain image studies, and in selected cases, also liver biopsy for diagnosis when HIV-infected patients have persistent and unexplained liver enzyme elevation while on HAART. Strategies for the management of noncirrhotic portal hypertension include placement of transjugular intrahepatic portosystemic shunt (TIPS) and liver transplant.143, 144 Anticoagulant therapy with low-molecular-weight heparin is a more specific treatment for this entity which has been recently reported.145


HAART hepatotoxicity complicates the management of HIV-infected patients, increases medical costs, alters the prescription patterns, and has an impact on official treatment recommendations. Several mechanisms of liver toxicity in patients receiving HAART have been recognized. Although infrequent, HAART-related liver damage may have devastating consequences. Among clinical syndromes of HAART liver toxicity, hypersensitivity reactions and lactic acidosis are recognized as acute events with potential to evolve into fatal cases, whereas there are other syndromes not as well characterized but of equal concern as possible long-term liver complications. Among the latter, HAART-related NASH, liver fibrosis, portal hypertension, and nodular regenerative hyperplasia are discussed. Prevention is the best strategy to minimize the cases of hepatotoxicity and includes recognition of antiretrovirals' liver safety profile and of susceptible hosts. Management of hepatotoxic events includes discontinuation of suspected culprits and changes in HAART regimens as well as identification of mechanisms involved and treatment of specific disorders.

NOTE: Definition of boxed warning as it is found in the Code of Federal Regulations Title 21, Volume 4 (chapter 1, subchapter c): (1) Boxed warning. Certain contraindications or serious warnings, particularly those that may lead to death or serious injury, may be required by the FDA to be presented in a box. The boxed warning ordinarily must be based on clinical data, but serious animal toxicity may also be the basis of a boxed warning in the absence of clinical data. The box must contain, in uppercase letters, a heading inside the box that includes the word ”WARNING” and conveys the general focus of the information in the box. The box must briefly explain the risk and refer to more detailed information in the ”Contraindications” or ”Warnings and Precautions” section, accompanied by the identifying number for the section or subsection containing the detailed information.


I thank James W. Johnson, Pharm.D., for kindly providing the ”black box warning” definition.