|Home | About | Journals | Submit | Contact Us | Français|
The treatment of HIV has improved over the last decade due in part to the success of treatment with combination-therapy drug “cocktails,” better referred to as highly active antiretroviral therapy. The safety and efficacy of these regimens is often complicated by the presence of infectious hepatitis and occurrence of drug-induced liver injury. In most patients, these complications manifest as mild laboratory abnormalities and exist without clinical consequence. However, acute and chronic hepatic injury may lead to increased morbidity and mortality. It is essential to identify all potential causes of liver injury to correctly implement steps to alleviate the condition while maintaining optimal control of HIV viremia.
The advent of highly active antiretroviral therapy (HAART) has dramatically reduced the clinical impact of infection with HIV.1,2 The incidence of AIDS and resultant opportunistic infections has declined among HIV-infected patients, and HIV disease is now regarded by many as a chronic disease process. Notwithstanding the successes in HIV management, there are a number of significant concerns associated with life-long HAART. The liver is a key organ required for normal homeostasis as well as for the metabolism of many drugs. HIV-infected patients frequently experience liver injury as a result of coinfection with one or more of the hepatitis viruses, substance and/or alcohol abuse, or concomitant medications including those used in HAART regimens. HAART-associated hepatotoxicity, arbitrarily defined as three- to fivefold or greater elevation in serum transaminases, is often asymptomatic and resolves without modification of therapy.3-5 In isolated instances, however, serious and potentially life-threatening conditions may arise. Discerning the role of HAART in hepatotoxic reactions of HIV patients may be difficult due to frequent preexisting liver pathology, such as that arising from infection with hepatitis B or C virus. Moreover, polypharmacy is common in HIV-infected individuals, and a very large number of medications are known to have effects on liver function and drug metabolism. Furthermore, although some HAART drugs have been linked to hepatotoxicity, the discontinuation of one or more antiretroviral medications for putative drug-related hepatotoxicity may limit future treatment options or exacerbate HIV viremia and mutant virus selection. Therefore, it is imperative to rule out other potential etiologies before discontinuing HAART drugs. Symptoms of hepatotoxicity may also resolve spontaneously without any changes to antiretroviral therapy or clinically important consequences. This review will consider some of the pathogenic mechanisms involved in the hepatotoxic effects of the different classes of HAART drugs, the risk factors associated with hepatotoxicity in treated patients, and potential interventions to alleviate the condition.
The liver plays an essential role both in normal homeostasis and drug metabolism. There are a great many commonly used prescription and over-the-counter medications, including “natural” and herbal remedies, that can cause liver injury.6,7 It is often difficult to discern the role of any one agent in hepatic injury; in some cases one agent can increase the toxicity of another (eg, coadministration of acetaminophen and alcohol).6
When a drug is considered to be directly hepatotoxic, the liver toxicity should be dose-related, have a post-exposure latency period, and all individuals should be susceptible to it.6 One such drug is acetaminophen, a major cause of acute hepatic failure in the United States.6 Idiosyncratic reactions are another mechanism of hepatotoxicity. These are often sporadic, non–dose-related, and characterized by hypersensitivity symptoms (eg, fever, eosinophilia). Such reactions may occur following use of the nonnucleoside reverse transcriptase inhibitor (NNRTI) class of antiretroviral medications (eg, rash and eosinophilia associated with the use of nevirapine and/or efavirenz).3,8 Other types of liver toxicity, including cholestatic injury typical of chlorpromazine and amoxicillin/clavulanic acid formations and microvesicular effects (“fatty liver”) seen with most nucleoside reverse transcriptase inhibitor (NRTI) drugs or with tetracycline, are well described.9,10 Lastly, drug-associated acute or chronic hepatitis has been described in patients taking HAART (see below), but the diagnosis of true drug-related hepatitis may often be obscured in HIV-positive individuals by the frequent coexistence (whether detectable or undetectable) of viral hepatitis.
Cytochrome P450 (CYP) is a multigene family encoding heme-containing enzymes, found in the endoplasmic reticulum.11 As one of the richest sources of CYP enzymes in the body, the liver plays a central role in the biotransformation of endogenous substances and xenobiotics.11 Cytochrome P450 is involved in the oxidation of ethanol by liver microsomes, and this may be one component of the increased susceptibility to toxicity of certain drugs among alcoholics.12 The CYP3A subfamily of enzymes is one of the most abundant in the body, and the ability of these enzymes to be affected by a large number of medications and nonmedications (such as grapefruit juice) can lead to problems in the first-pass metabolism of many medications.11 For example, the ability of one drug to inhibit the action of CYP3A may lead to increased levels of another drug and an associated enhancement of its toxicity. It is important to bear such interactions in mind when considering HIV-infected patients with hepatotoxicity, especially in view of the fact that polypharmacy is common in this population.
Reactive oxygen species such as superoxide and nitric oxide are known to play important roles in host defense systems but can also damage host tissues.13 Nitric oxide–stimulated production of inflammatory mediators such as interleukin-1 can inhibit CYP production and decrease resistance of the liver to trauma.14 On the other hand, in smaller quantities nitric oxide can decrease the levels of proinflammatory products and stimulate protein synthesis and DNA repair mechanisms in hepatocytes.14 Oxidative stress is associated with chronic viral infections such as viral hepatitis, and reactive oxygen species may contribute to hepatocellular carcinoma risk in patients infected with hepatitis B or C virus subsequent to ongoing inflammation of the liver.13 Stimulation of the microsomal pathway by inducers such as ethanol (as described above) can also result in the production of acetaldehyde, which in turn can affect oxidative processes in the liver (eg, glutathione depletion, increased toxicity of free radicals).12 The ability of HAART and other drugs to disrupt the normal production or elimination of reactive oxygen species in the liver may therefore be another factor relating to their potential to cause hepatotoxicity.
Mitochondria play an essential role in energy production (adenosine triphosphate generation) and are also important for the metabolism of cholesterol and ethanol.15 Mitochondria contain their own nonnuclear DNA (mtDNA) and require the action of a nuclear DNA-encoded DNA polymerase (DNA pol γ) to replicate. Because DNA pol γ is more error-prone and lacks the repair mechanisms of other DNA polymerases, errors may occur in mtDNA that can ultimately lead to mitochondrial dysfunction.15 If mitochondrial function is inhibited, as is suspected in the case of hepatotoxicity related to NRTIs and some hepatotoxicity related to protease inhibitors, fatty acid accumulation occurs, and there is increased metabolism of pyruvate to lactate.15 This is one potential mechanism underlying the complication of lactic acidosis, seen infrequently among patients treated with NRTIs. A lipodystrophy syndrome characterized by nausea and fatigue, lipoatrophy, lactic acidemia, and hepatic dysfunction has been described among patients treated with these agents.16 This may complicate the diagnosis of lipodystrophy, hyperlipidemia, and insulin resistance associated with the use of protease inhibitors (see below).
In view of their potential to affect these and other hepatic systems, a wide variety of drugs have been associated with nonalcoholic steatohepatitis (NASH), which is characterized by steatosis, hepatocellular injury, mixed lobular inflammation, and fibrosis in the perisinusoidal and perivenular regions.17 Associations between HAART and NASH may be explained in one of three possible ways: (1) a spurious (coincidental) association, (2) the drug may cause or exacerbate risk factors for NASH (eg, diabetes, central obesity, hypertriglyceridemia), or (3) the drug may cause NASH by a direct mechanism.17 All of these processes may play a role in liver injury associated with the use of HAART.
Dual NRTIs, in combination with either protease inhibitors or NNRTIs, remain the backbone of most antiretroviral therapy regimens.10 Currently approved NRTIs include zidovudine, didanosine, zalcitabine, stavudine, lamivudine, abacavir, and emtricitabine.18 In addition, tenofovir is a nucleotide analog that is now widely used in HIV therapy.18 Mitochondrial toxicity is a prevailing explanation for hepatotoxicity among patients treated with NRTIs, especially stavudine, didanosine, and zalcitabine.10 Mitochondrial toxicity results in lactic acidosis, which may present with or without hepatic microsteatosis.12 One recent case report describes a patient on stavudine and didanosine who was admitted for severe acute pancreatitis 18 months following the implementation of antiretroviral therapy.19 The patient subsequently developed lactic acidosis, hepatic failure, renal failure, and grade III encephalopathy. Liver biopsy findings were consistent with mitochondrial toxicity (microvascular steatosis and giant mitochondria).19 Although rare (<1 per 100 years of patient therapy), lactic acidosis has been observed in patients on either dual or single NRTI regimens. The symptoms in some of these cases may be insidious (eg, fatigue and weight loss).15 Elevations in lactate and lactate:pyruvate ratios are the characteristic biochemical abnormalities, and this may be associated with elevations in transaminase and creatine kinase levels.15
Other effects of NRTIs include HIV-associated lipodystrophy. The resemblance of this disorder to multiple symmetric lipomatosis, an inherited mitochondrial disorder, also suggests adverse drug effects on the function of these organelles.10 As noted above, NRTIs have been noted to have inhibitory effects on DNA pol γ, which is needed for efficient mtDNA synthesis. Based upon the assumption that mitochondrial toxicity relates mainly to the inhibition of DNA pol γ by NRTIs, an in vitro hierarchy of toxicity has been established whereby: zalcitabine ≥ didanosine > stavudine > lamivudine > zidovudine > abacavir.10 Accordingly, all NRTIs now carry a black box warning regarding risk of lactic acidosis and severe hepatomegaly with steatosis arising from treatment. Tenofovir, a nucleotide analog, may also precipitate mitochondrial injury in conjunction with other nucleosides, though the mechanism remains unclear.20 Also, the hypersensitivity reaction associated with abacavir may be accompanied by liver enzyme elevations or liver failure (Abacavir prescribing information). Because of their widespread use as a component of antiretroviral therapy regimens, the potential for hepatotoxicity with all NRTIs should be borne in mind when evaluating liver dysfunction in treated patients.
NNRTIs are also associated with hepatic injury. Several mechanisms have been suggested. These include hypersensitivity reaction, direct cholestatic injury or mediation of immune reconstitution syndromes.21 Although rare, the use of both efavirenz and nevirapine has been associated with the development of fulminant hepatic failure and/or toxic hepatitis.22-24 By comparison, the incidence of severe (grade 3/4) hepatotoxicity appears to be less with delavirdine.25 Some studies suggest a higher incidence of hepatotoxicity in patients treated with nevirapine as compared with efavirenz, but data are contradictory.23,26-28 In one study, transaminase abnormality was seen in 12.5% of nevirapine-treated patients (N=610).29 In this study, hepatotoxicity was related to the duration of prior exposure to antiretroviral therapy, hepatitis C virus coinfection, and higher baseline alanine aminotransferase (ALT). In other studies, the use of a protease inhibitor in a nevirapine regimen and hepatitis B virus infection were likely risk factors for hepatotoxicity,24,30 but a correlation between the plasma concentration of nevirapine and hepatotoxicity could not be established.30
In a large study of hepatic safety (nevirapine, n=1731; controls, n=1912), it was concluded that nevirapine hepatotoxicity is asymptomatic in a majority of patients and is easily treated with the temporary discontinuation of nevirapine until transaminase levels return to normal.31 Clinically relevant hepatic events occurred in most cases during the first 4–6 weeks. Controlled clinical studies also showed no excess in mortality secondary to hepatotoxicity versus controls.31 The hepatotoxicity observed with nevirapine use is associated with discrete risk factors including female gender, higher CD4+ cell counts (especially women with CD4+ cell counts >250/mm3), and coinfection with hepatitis B or C virus.32
Hepatotoxicity of efavirenz also appears to be related to the concurrent use of protease inhibitors and to viral hepatitis infection. In one study of HIV-infected patients, severe hepatotoxicity was observed in 15.3% (n=124) of those infected with hepatitis C virus, versus 3.2% (n=188) of non–hepatitis C infected efavirenz users (relative risk [RR], 4.8, 95% confidence interval [CI], 2.0–11.8).24 NNRTI-related idiosyncratic/hypersensitivity reactions (rash, eosinophilia) seen in patients treated with nevirapine or efavirenz do not appear to be associated with hepatotoxicity as measured by traditional enzyme markers.23,24 Notwithstanding the observed severe hepatotoxic reactions seen in isolated patients treated with NNRTIs and the accepted need for close monitoring, it is notable that as a class these drugs have been associated with the lowest hepatic-associated death rate (0.8% of all deaths) compared with other antiretroviral therapy drug classes in large retrospective analyses (N=9,003 patients).25 Thus, although both nevirapine and efavirenz have been associated with fulminant hepatic failure, their overall safety and tolerability profiles render both valuable antiretrovirals in the management of HIV.25,33
Hepatotoxic reactions have been observed in patients taking protease inhibitors, and risks appear to vary by medication. One study found a higher incidence of severe hepatotoxicity (grade 3 or 4) in patients on ritonavir (48% of all cases), and ritonavir use in this study was associated with severe hepatotoxicity in univariate logistic regression analysis (odd ratio, 6.2; 95% CI, 2.8–13.7).34 In contrast, no greater risk was found in patients taking protease inhibitors other than ritonavir versus two NRTIs alone. Severe hepatotoxicity was observed at equal frequency among ritonavir users who were or were not coinfected with hepatitis C virus (30%). The authors of this study suggest that hepatotoxicity in ritonavir-treated patients is likely to be an effect of the medication, as the incidence was unaffected by chronic viral hepatitis. Another study of 1,325 patients confirmed the association of indinavir use with severe hyperbilirubinemia at 6 months (P=.003), 12 months (P=.009), and 24 months of antiretroviral therapy (P=.019), although viral hepatitis coinfection had no significant effect on this complication.35 In this study, ritonavir was associated with a significantly higher incidence of severe hepatotoxicity versus other protease inhibitors in the first 6 months of therapy (P=.0001). A significant effect of hepatitis virus coinfection on severe hepatotoxicity was seen in patients on ritonavir (P=.0014), but only in the first 6 months of treatment. Although the exact mechanism is unknown, authors of both studies suggest the hepatotoxicity of ritonavir may relate in part to its effects as an inhibitor of the CYP system.34,35
Increased bilirubin levels seen in patients on indinavir and atazanavir does not represent hepatotoxicity and may not have negative clinical implications. One pivotal study demonstrated that indinivir is a direct inhibitor of UDP-glucuronyltransferase which is required for bilirubin conjugation. Inhibition leads to elevated levels of total and indirect bilirubin. Clinicians must be aware of this and should warn their patients that jaundice may occur. HIV-infected patients with the Gilbert polymorphism appear most susceptible.36
Protease inhibitors also may have other direct and indirect effects on the liver. These drugs suppress the breakdown of nuclear sterol-regulatory element binding proteins in liver and adipose tissues; this leads to an excess of fatty acid and cholesterol biosynthesis in the liver.37 They also suppress the breakdown of triglyceride-rich lipoproteins, resulting in their subsequent accumulation, and suppress glucose transporter inhibition in adipose tissue and muscle cells, resulting in peripheral insulin resistance.37 Symptoms associated with protease inhibitor-associated lipodystrophy (elevated cholesterol, triglycerides, glucose, and insulin) may differ from those of NRTI-associated HIV lipodystrophy.16 These metabolic effects of protease inhibitors may cause further insult to liver function, especially in the face of viral hepatitis coinfection. Metabolic disturbances associated with the use of protease inhibitors can also adversely affect cardiovascular risk in this population. Overall, some degree of liver injury has been associated with all six currently available protease inhibitors, and some drugs, such as ritonavir, are commonly administered in lower doses (≤200 mg bid) to enhance the function of other, less toxic protease inhibitors (eg, lopinavir).38
Several studies have identified factors associated with hepatotoxicity in patients on HAART. These factors include coexisting infection with hepatitis B or C virus and elevated baseline transaminases. It has been observed in some case reports that patients on HAART may experience viral hepatitis disease as a result of an “immune restoration” disease secondary to antiretroviral therapy. Thus, hepatic inflammation in these individuals appears to be mediated by, for example, previously impaired hepatitis C virus–specific immune responses.39 However, additional study is needed on the apparent association between CD4 count increase and ALT increase, as other factors may be involved. In this regard, it is important to note that in some cases, hepatitis C virus infection may only be recognized following seroconversion secondary to potent antiretroviral therapy.40 Previously hepatitis C virus–negative HIV-positive patients may therefore actually have had viral coinfection of the liver prior to the initiation of HAART. Such findings should be borne in mind when considering risk factors for hepatotoxicity as described below.
In a retrospective single-center survey of 394 patients treated with HAART, the risk of elevated liver enzymes (defined as five times the upper limit of normal and an absolute increase >100 U/L) was assessed by Cox proportional hazards analysis.5 The incidence of liver enzyme elevations was 37% in patients coinfected with hepatitis B or C virus, versus 12% in patients without coexisting viral hepatitis. After correcting for baseline elevated transaminase levels, seropositivity for hepatitis B or C virus was associated with a 2.78- and 2.46-fold increased risk for liver enzyme elevations (95% CI, 1.50–5.16 and 1.43–4.24, respectively). Another study examined risk factors for elevated liver enzymes in patients treated with ritonavir or saquinavir with or without stavudine.41 Hepatitis B virus seropositivity (RR, 8.8; 95% CI, 3.3–23.1) and stavudine use (RR, 4.9; 95% CI, 1.5–16.0) were significant risk factors for elevated liver enzymes. A number of other risk factors for grade 4 liver enzyme elevations, including coexisting viral hepatitis, have been identified using multivariate analysis in a retrospective cohort study of 560 patients treated with HAART (Table 1).42 Among the identified risk factors for grade 3 or 4 liver enzyme elevations in multivariate analysis, chronic coinfection with hepatitis B or C virus was most notable. The effect of elevated baseline ALT alone is difficult to assess because it is usually accompanied by coinfection; although baseline ALT was identified in this study as a risk factor for grade 3 or 4 liver enzyme elevations, the hazard ratio was only 1.03 (P=.11). Viral hepatitis was also an independent risk factor for the development of severe hepatotoxicity among patients taking NNRTI–based regimens, as was the use of nevirapine and concurrent protease inhibitor use (Table 2).24 For patients on protease inhibitor-based therapies, ritonavir use appears to be an important determinant of hepatotoxicity; however, a significant effect of CD4 cell recovery on severe hepatotoxicity was also observed, in support of the immune restoration hypothesis as described above (Table 3).34
Using samples from a parent trial comparing lopinavir/ritonavir versus nelfinavir in the setting of a stable nucleoside backbone, 57 hepatitis C/HIV coinfected subjects were analyzed in terms of hepatitis C viral load and ALT increases following initiation of therapy. Immune reconstitution, defined by CD4+ rebound, occurred in both treatment arms. Mean hepatitis C viral loads increased in both groups, and in some case, this increase was accompanied by an antiretroviral therapy flare. These data suggest that the mechanism for ALT flare following ART initiation may be related to a paradoxical increase in hepatitis C viral load.
Given the potential impact of hepatotoxic reactions on outcomes in patients undergoing treatment for HIV, the judicious monitoring of liver enzymes, especially during the initiation of HAART, appears warranted. A suggested regimen for liver enzyme monitoring is shown in Table 4.3 As noted above, clinical manifestations of hepatotoxicity such as elevated transaminases often will spontaneously resolve without interrupting antiretroviral therapy. In one study of patients undergoing HAART, only 17.1% (n=35) of patients with grade 4 liver enzyme elevations experienced clinical symptoms, and all patients had resolution to grade 2 or less regardless of whether they continued (n=23) or discontinued (n=12) antiretroviral therapy.42 Another study comparing outcomes in efavirenz- and nevirapine-treated patients found that, among patients experiencing severe hepatotoxicity, there was no significant difference in ALT levels at a median 8-month follow-up between patients who continued versus those who discontinued antiretroviral therapy.24
If, after thorough clinical assessment, one or more antiretroviral drugs is believed to be a causative factor in severe hepatotoxicity, discontinuation of the agent may be necessary. In this case, it may be advantageous to completely stop all antiretroviral therapy rather than just one or two agents, so as to minimize the potential for development of individual drug-resistant strains.3 Frequent assessment of disease progression (eg, CD4 count) is necessary in patients discontinued from antiretroviral therapy.
If a high risk of hepatotoxicity is suspected, additional precautions may be warranted when implementing therapy with selected drugs. About two thirds of nevirapine-associated hepatitis cases occur within the first 12 weeks of therapy.3 A lead-in period with 200 mg of nevirapine for 2 weeks has been proposed as a means of reducing the incidence of nevirapine-associated hepatotoxicity and skin rash, and it is not recommended that patients be rechallenged with nevirapine if severe hepatotoxic reactions occur.3 In contrast to nevirapine, protease inhibitor-associated hepatotoxicity may occur anytime during antiretroviral therapy.3
Establishing a balance between efficacy and toxicity is an essential consideration in the design of any drug regimen. Long-term HAART is necessary to prevent disease recurrence and/or exacerbation in HIV-infected individuals and is effective in prolonging survival. Life-long antiretroviral therapy, however, is limited by several significant tolerability issues, including hepatotoxicity. Frequent monitoring for drug-related side effects is necessary, especially in the early phases of treatment. All classes of antiretroviral drugs have the potential to cause hepatotoxic reactions through one or more mechanisms. Most hepatotoxicity associated with HAART is asymptomatic and limited to elevated liver enzymes, which may resolve without the interruption of therapy. Even grade 4 liver enzyme elevations may not be an immediate indication for the discontinuation of antiretroviral therapy.42 Coinfection with hepatitis B or C virus and baseline elevations in liver enzymes appear to be relatively consistent risk factors for antiretroviral therapy-associated hepatotoxicity in some (but not all) studies. However, antiretroviral therapy is not contraindicated in patients with one or more of these or other risk factors. Instead, close monitoring for clinical symptoms is warranted in such patients, especially during the initiation of new HAART. It is likely that both retrospective and prospective evaluations of clinical trials involving new and existing HAART regimens will help clinicians optimize and individualize regimens for patients in common clinical settings (eg, hepatitis C virus infection). An improved understanding of pharmacokinetic interactions, particularly in patients with underlying comorbid liver diseases, will also improve treatment safety.