SUPPLEMENT ARTICLE

Drug-Induced Liver Injury Associated

with the Use of Nonnucleoside
Reverse-Transcriptase Inhibitors
Douglas T. Dieterich, Patrick A. Robinson, James Love, and Jerry O. Stern
Department of Medicine, Mt. Sinai Medical Center, New York, New York

Human immunodeficiency virus (HIV)–infected patients frequently present with elevated levels of serum
transaminases (alanine aminotransferase [ALT] and/or aspartate aminotransferase [AST]). This has often been
attributed to the hepatic effects of antiretroviral (ARV) drugs, including nonnucleoside reverse-transcriptase
inhibitors (NNRTIs). A review of cohort studies investigating the incidence of hepatotoxicity among patients
receiving ARV therapy suggests that the overall rate of ALT and/or AST elevations is similar among all ARVs.
The rate of severe hepatotoxicity, ALT and/or ASTlevels 15 times the upper limit of normal (ULN), during
therapy with NNRTIs is relatively low but may be significantly higher in patients with concurrent chronic
viral hepatitis (hepatitis B or C). A comprehensive analysis of 17 randomized clinical trials of nevirapine
demonstrated that 10% of all nevirapine-treated patients developed elevated levels of ALT and/or AST 15 times
the ULN; however, almost two-thirds (6.3% of nevirapine-treated patients) of these elevations were asymp-
tomatic. Symptomatic hepatic events were seen in 4.9% (3.2%–8.9%) of nevirapine-treated patients.

The introduction of combination antiretroviral therapy
(ART) has led to significant reductions in morbidity
and mortality associated with HIV infections [1]. In-
creasingly, adverse effects caused by combination ART
are being reported and are emerging as a major safety
concern. In particular, hepatotoxicity is perceived as
being frequently associated with the use of nonnucleo-
side reverse-transcriptase inhibitors (NNRTIs), a com-
mon component of combination ART [2]. Drug-in-
duced hepatic injury is currently responsible for 150%
of cases of acute liver failure in the United States and
is the most frequent reason for withdrawing an ap-
proved drug from the market [3]. Most drugs, however,
cause liver injury infrequently. When therapeutic doses
are used, idiosyncratic hepatic reactions occur at rates
of 1 per 1000 to 1 per 100,000 population. The pattern
Reprints or correspondence: Dr. Douglas T. Dieterich, Dept. of Medicine, Mt.
Sinai Medical Center, 100th St. and Madison Ave., Annenberg Bldg., 23rd Fl., Rm.
23-90, New York, NY 10029 (douglas.dieterich@msnyuhealth.org).
Clinical Infectious Diseases 2004; 38(Suppl 2):S80–9
 2004 by the Infectious Diseases Society of America. All rights reserved.
1058-4838/2004/3805S2-0007$15.00
of hepatic injury is consistent for each drug and each
drug class. The present article outlines the evidence for
the association of hepatotoxicity with ART and reviews
the current data available from studies investigating the
incidence of NNRTI-associated hepatotoxicity.
HEPATOTOXICITY AND ART
There is an increasing focus on the toxicities and ad-
verse reactions associated with combination ART, such
as drug-induced liver injury, neuropathy, and pancre-
atitis. Recently, reports have linked the use of NNRTIs,
such as nevirapine (NVP) and efavirenz, with the de-
velopment of hepatotoxicity [4]. There is particular
concern about HIV-infected patients coinfected with
hepatitis B virus (HBV) or hepatitis C virus (HCV),
because the progression of liver disease is accelerated
in these patients [5].
The clinical presentations of hepatotoxicity in HIV-
infected patients range from asymptomatic elevations
of transaminase levels to fulminant hepatic failure,
which is a rare, severe decompensation of liver function
that may lead to death despite changes in or the dis-

S80 • CID 2004:38 (Suppl 2) • Dieterich et al.

Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331

by Universitas Diponegoro user
on 31 August 2018
 continuation of ART. It is associated with systemic symptoms,
jaundice, coagulopathy, and markedly elevated alanine ami-
notransferase (ALT) levels [6]. Less severe cases of ART-related
hepatotoxicity are typically reversible and are characterized by
abnormal serum ALT and/or aspartate aminotransferase (AST)
levels in the presence or absence of clinical symptoms of liver
injury [7]. Regardless of the severity of the hepatotoxicity, the
interplay of the effects of the ART drugs and of associated risk
factors, such as alcohol use, underlying diseases, and concom-
itant drugs, influences the patient’s susceptibility to the hep-
atotoxic effects of all drugs [8]. It is difficult to ascertain an
accurate picture of the incidence of ART-related hepatotoxicity,
because of limitations of the adverse drug-reaction reporting
system. For example, underreporting, data reliability, and un-
known denominators are often problematic. However, an anal-
ysis of the US Food and Drug Administration’s clinical adverse
reaction database suggested that the reporting rates of ART-
related hepatotoxicity and hepatic-related deaths are similar for
all ART drugs (figures 1 and 2) [9].
INCIDENCE OF NNRTI-ASSOCIATED
HEPATOTOXICITY
Clinical studies have indicated that grade 3 (ALT and/or AST
levels 15 times the ULN) and grade 4 (ALT and/or AST levels
110 times the ULN) hepatotoxicity is observed in ∼5%–10%
of HIV-positive patients treated with combination ART for 16
months [4, 10–12]. Although abnormal laboratory transami-
nase levels are common in these patients, there is a perception
that serious clinical hepatic events are more frequent during
treatment with NNRTI-based regimens than NRTI- or PI-based
regimes [4]. In particular, the use of NVP has been reported
to be associated with a number of severe clinical hepatic events
when administered to non–HIV-infected patients for the non-
approved indication of postexposure prophylaxis [13].
A retrospective cohort study determined the incidence of
NNRTI hepatotoxicity in a group of HIV-infected patients, pre-
dominantly homosexual white men from a New York City prac-
tice [2]. A total of 272 patients received NNRTIs: 40 (15%)
received delavirdine, 91 (33%) received efavirenz, and 141
(52%) received NVP. Of the patients with known viral hepatitis
status, 18 (9%) of 190 were coinfected with HBV, and 24 (12%)
of 205 were coinfected with HCV. Grade 3–4 elevations in ALT
and/or AST levels occurred in 3 (1.1%) of 272 patients; the
rate was similar among all 3 NNRTI treatment groups. Coin-
fection with HBV or HCV was not associated with a significant
increase in AST or ALT levels.
In another US cohort study, 312 and 256 patients, respec-
tively, were prescribed efavirenz and NVP [4]. Of these, HCV
and HBV were detected in 43% and 7.7% of patients, respec-
tively. The rate of occurrence of grade 3 or elevations in 4 ALT
Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331
by Universitas Diponegoro user
on 31 August 2018
and/or AST levels was significantly greater among patients with
chronic viral hepatitis (69%) and those who were prescribed
PIs (82%) [4]. However, 84% of patients with chronic HCV
or HBV did not experience severe hepatotoxicity. This study
concluded that severe hepatotoxicity occurs throughout the
course of NNRTI-based therapy and was more common among
patients prescribed NVP (occurring in 15.6%) than those pre-
scribed efavirenz (8.0%). Patients who were coinfected with
HBV or HCV and those who were coadministered PIs were at
increased risk of severe hepatotoxicity. Lebray et al. [14] have
criticized this study on the following grounds: use of a single
ALT determination as a baseline for the definition of severe
hepatotoxicity is inadequate, because spontaneous fluctuations
in AST and/or ALT levels are common in patients with viral
hepatitis; the study results do not support a casual relationship
between NNRTI administration and hepatotoxicity; and the
favorable outcome for the majority of patients who continued
to take ART despite severe hepatotoxicity suggests that the
NNRTI was unlikely to be the causative agent. The authors
accepted that increases in AST or ALT levels are imperfect
indicators of hepatotoxicity but stated that they are routinely
used in the clinic [15]. They also stated that, in their experience,
ALT fluctuations to the degree observed in this study are rare
in HCV-infected patients. Finally, they stressed that they re-
ported an association between NNRTI use and hepatotoxicity
but did not assign causality.
A retrospective study from Amsterdam (The Netherlands)
that used data from a subset of the ATHENA cohort investi-
gated whether the use of specific antiretroviral drugs was as-
sociated with a higher risk of developing grade 4 ALT and/or
AST levels in patients who were starting combination ART [12].
Grade 4 ALT and/or AST levels were defined as levels 110 times
the ULN and 1200 U greater than baseline levels. Fewer than
10% of patients were treated with zalcitabine, abacavir, saqui-
navir soft-gel capsules, amprenavir, NPV, or efavirenz; there-
fore, the study was not powered to detect the hepatotoxic effects
of these drugs. The incidence of grade 4 ALT and/or AST el-
evations was 6.3%. More than 80% of grade 4 ALT and/or AST
elevations were asymptomatic, and there were no hepatic-
event–ssociated deaths. Risk factors were identified as high
baseline ALT levels, chronic HBV or HCV coinfection, being
ART and receiving a first combination ART regimen, recent
start of a regimen that contained NVP or high-dose ritonavir,
and female sex. In HBV-coinfected patients, discontinuing la-
mivudine use was a risk factor. Although the hazards ratios for
grade 4 ALT and/or AST elevations were high (ritonavir, 4.9
and NVP, 9.6), the crude absolute risks were low (ritonavir,
3.2%; 95% CI, 1.7%–5.6% and NVP, 2.4%; 95% CI, 0.8%–
5.6%).
Similar results were reported by Reisler et al. [10] in a meta-
analysis of laboratory data and mortality from 21 prospective
Liver Injury and NNRTIs • CID 2004:38 (Suppl 2) • S81
 Figure 1. Reporting rates of antiretroviral-related hepatotoxicity for all antiretroviral drugs (US Food and Drug Administration database, 1996–2000)
[9]. Rates are events per 10,000 treated patients. NRTI, nucleoside reverse-transcriptase inhibitor; NNRTI, nonnucleoside reverse-transcriptase inhibitor;
PI, protease inhibitor.

AIDS Clinical Trial Group (ACTG) clinical studies conducted
between 1991 and 2000 that involved 10,611 patients. Patients
received a range of ART regimens: single nucleoside reverse-
transcriptase inhibitors (NRTIs), dual NRTIs, NRTI(s) plus
NNRTI(s), NRTIs plus protease inhibitors (PIs), and NRTIs
plus NNRTI plus PI regimens. The overall incidence of severe
hepatotoxicity, defined as increases in ALT or AST levels to 15
times the ULN, was 6.2% (95% CI, 5.7–6.7%). The frequency
of ALT and/or AST levels 15 times the ULN in patients who
received an NNRTI (NVP, efavirenz, or delavirdine) and 2
NRTIs was higher than the rate in patients who received reg-
imens that contained a PI and 2 NRTIs (8.2% vs. 5.0%; P p
.0063). Regimens composed of 2 NRTIs and either NVP or
efavirenz were associated with rates of ALT and/or AST levels

Figure 2. Reporting rates of antiretroviral (ARV)–related hepatic mortality for all antiretroviral drugs (US Food and Drug Administration database,
1996–2000) [9] Rates are events per 10,000 treated patients. NRTI, nucleoside reverse-transcriptase inhibitor; NNRTI, nonnucleoside reverse-transcriptase
inhibitor; PI, protease inhibitor.

S82 • CID 2004:38 (Suppl 2) • Dieterich et al.

Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331

by Universitas Diponegoro user
on 31 August 2018
 15 times the ULN (8.9% and 10.8%, respectively) that were
not significantly different. Nevertheless, different frequencies of
elevated ALT and/or AST levels have been found in some stud-
ies. For example, higher rates of asymptomatic cases of ALT
and/or AST levels 15 times the ULN during nevirapine treat-
ment (9.9%) were observed in clinical trial FTC-302 [14]. The
period of highest risk was during the first 4 weeks of treatment.
However, after 4 weeks, the rate of ALT and/or AST levels 15
times the ULN closely resembled that observed in other con-
trolled NVP trials. These results may be explained, in part, by
the stratification of patients with higher CD4 cell counts ac-
cording to whether they received NVP and also the greater
proportion of female patients (59%) enrolled in this trial.
Lesser degrees of ALT and/or AST level elevations (to ⭓3
times the ULN) were common (incidence, 13.1 cases per 100
person years) in one study of HIV-infected patients receiving
NVP-based regimens [16]. An elevation of ALT and/or AST
levels to ⭓3 times ULN developed in 76 (12.5%) of 610 patients
over a median period of follow-up of 8.7 months, which reflects
an incidence of 13.1 cases per 100 person-years. According to
Kaplan-Meier analysis of the incidence of ALT and/or AST
levels ⭓3 times ULN in this cohort was 3.7% at 3 months,
9.7% at 6 months, and 20.1% at 12 months. Prolonged ex-
posure to any ART drug, coinfection with HCV, and abnormal
baseline levels of ALT were associated with a higher risk of
asymptomatic hepatotoxicity. Clinical (symptomatic) hepato-
toxicity was found in 7 (1.1%) of 610 patients, and 4 of 7 had
chronic liver disease prior to receiving NVP therapy. All 7 pa-
tients had resolution of symptoms and improvement after the
discontinuation of therapy. In a univariate analysis, receiving
stavudine was associated with a relative hazard of 2.29 (95%
CI, 1.25–4.2) for experiencing a 3-fold increase in AST or ALT
levels while receiving NVP. The figure for zidovudine was 0.44
(95% CI, 0.24–0.82). These data may suggest that mitochon-
drial toxicity, which has been associated with stavudine use,
was partially responsible for the observed elevations in liver
enzyme levels.
Analyses of hepatic events from other cohorts, such as those
in the CHORUS (Collaborations in HIV Outcomes Research
US) and Target studies, have confirmed that the frequency and
risk of significant hepatic events with NVP- or efavirenz-based
therapies are similar to those associated with the use of other
ART drugs [17–20]. More important, the frequency of serious
hepatic events or hepatic-event–associated death associated
with ART, including NVP and efavirenz use, is very low, with
median rates ranging from 0.06% to 0.13% [10, 11].
The most comprehensive cohort study that has compared
the hepatic safety of NNRTIs was recently presented by
EuroSIDA [21]. The study included 9803 patients who were
followed-up prospectively at 72 centers in 26 countries across
Europe until January 2002. Rates of ALT and/or AST levels 15
Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331
by Universitas Diponegoro user
on 31 August 2018
times ULN and severe clinical liver failure were analyzed among
patients starting either NVP or efavirenz therapy for the first
time. A total of 2991 patients initiated combination therapy
with NNRTIs (NVP regimens, 1648 and efavirenz regimens,
1343). The prevalence of HCV antibody positivity was 33.2%
and 32.1%, respectively in the 2 treatment groups. Within 3975
person-years of follow-up, there was no significant difference
between the rates of ALT and/or AST levels 15 times ULN in
patients treated with NVP or efavirenz (2.1% and 1.7%, re-
spectively; P p .47). Liver failure occurred infrequently (n p
14); the incidence of liver failure associated with NVP therapy
was 0.3 cases per 100 person-years (95% CI, 0.1–0.6), and, with
efavirenz therapy, it was 0.4 cases per 100 person-years (95%
CI, 0.2–0.9). The majority of these patients (11 of 14) died,
and 8 of them were coinfected with HCV and/or HBV.
As discussed in the present supplement by Kaplowitz [22],
the majority of serious adverse drug-induced hepatic events are
unpredictable, not dose-related, and are either immune-me-
diated hypersensitivity or idiosyncratic reactions [3]. However,
the more frequent adverse hepatic events, such as asymptomatic
elevations in ALT and/or AST levels, may be associated with
higher levels of the parent drug or a metabolite. In one case-
control study, 70 patients receiving NVP-based therapy were
stratified between those with normal ALT and/or AST levels
(control subjects) and those who developed any degree of ALT
and/or AST level elevation. A correlation was observed between
ALT and/or AST level elevations, HCV coinfection, and higher
plasma levels of NVP, compared with the control group [23].
In patients with chronic HCV coinfection, NVP concentrations
of 16 mg/mL were associated with a 92% risk of hepatotoxicity.
The question remains whether elevated plasma NVP levels in
these patients represent a true cause-and-effect relationship or
merely an epiphenomenon associated with preexistent HCV
infection or other liver pathology. NVP is extensively biotrans-
formed via hepatic cytochrome P450 (oxidative) metabolism;
any pathological process that has the potential to raise hepatic
ALT and/or AST levels may also be expected to alter hepatic
nevirapine metabolism and, therefore, NVP plasma levels as
well. Clearly, further studies will be needed to assess whether
ART therapeutic drug monitoring will improve safety moni-
toring for serious hepatic events.
INCIDENCE OF NVP-ASSOCIATED
ASYMPTOMATIC AND SYMPTOMATIC
HEPATOTOXICITY IN CONTROLLED AND
NONCOMPARATIVE RANDOMIZED CLINICAL
TRIALS
To critically evaluate the occurrence of hepatic events and as-
sociated risk factors, an analysis of symptomatic and asymp-
tomatic hepatic events associated with specific ART in ran-
Liver Injury and NNRTIs • CID 2004:38 (Suppl 2) • S83
 domized clinical trials of NVP was performed. Such a detailed
analysis for NVP or other antiretroviral therapy drug has never
been previously undertaken. The risk of asymptomatic and
symptomatic hepatic events associated with NVP-based ART
was considered to be particularly relevant.
Data from 17 randomized clinical trials completed between
1991 and 2001 were compiled into a common data set (table
1). The objective of these analyses was primarily to identify
and characterize all potential hepatic events and determine re-
lated risk factors for NVP-associated hepatotoxicity. All hepatic
events were classified as either asymptomatic or symptomatic.
The methodology of these analyses has been described in detail
[11]. The patient populations were diverse, international, and
multiracial.
Data from 9 randomized controlled trials (n p 3642) were
analyzed; a total of 1731 patients were treated with NVP, and
1911 patients served as control subjects. In most cases, the
control group received placebo plus the same background ART
drugs as the NVP group. In some trials, an active ART drug
was substituted for NVP. Because the majority of patients came
from one trial, BI 1090 (n p 2249; NVP-treated patients, 1121;
control subjects, 1128), this trial was evaluated separately and
as part of the pooled data combined with the other 8 controlled
trials (n p 1393; NVP-treated patients, 610; control subjects,
783). The other 8 trials included the CHARM (Compact Highly
Active Antiretroviral Medication), Incas, and Atlantic trials. A
second analysis (n p 2545) included data from the 1731 pa-
tients from the randomized controlled trials who were treated
with NVP, and these results were combined with data from 814
patients who were treated with NVP in 8 noncomparative trials,
includeding FTC-302. In both the controlled and noncompar-
ative trials, background ART regimens usually consisted of 1
or 2 NRTIs, although 5 trials included PI therapy. Eighty per-
cent of all patients treated with NVP ART with received ⭓3
drugs.
To identify all potential symptomatic hepatic events, a set of
3 broad and inclusive case selection rules were used that iden-
tified all instances of potential clinical hepatic events: (1) all
cases for which any hepatic and/or biliary terms were men-
tioned as adverse events; (2) cases for which all nonhepatic or
systemic terms were used that were consistent with liver disease
or potential immune-mediated events observed within 14 days
before or after raised levels of ALT and/or AST (⭓3 times the
ULN); and (3) all patients who developed ALT and/or AST
levels 15 times the ULN while receiving treatment. Internal and
external experts, who were blinded to treatment status, reviewed

Table 1. Overview of controlled clinical trials and noncomparative clinical trials.

Type of study, number, and location(s) Study regimen Trial design/population

Controlled clinical trials
1100.1011, Australia, Italy, The Netherlands NVP/ZDV vs. ZDV; NVP/ZDV/ddI or ddC OL randomized comparative, exp./6 months OL
comparison of NVP vs. no NVP, followed by 6
months OL NVP given to both groups
1100.1037, United States NVP/ZDV vs. PBO/ZDV DB, PBO-control, exp./6 months DB; 6 months OL
NVP given to both groups
1100.1038, United States NVP/ZDV vs. PBO/ZDV or NVP vs. PBO in naive DB, PBO-control naive/6 months DB; 6 months OL
patients NVP given to both groups
1100.1046, Italy, The Netherlands, Australia, NVP/ZDV/ddI vs. PBO/ZDV/ddI vs. NVP/ZDV/PBO DB, PBO-control, naive/exp.; later, 32 PBO patients
Canada first started NVP after rolling into trial 1036.
1100.1090, Europe, S. Africa N. America, NVP/3TC vs. 3TC plus background ART in each DB, PBO-control, exp./naive
Argentina group
1100.1135, United States NVP/ddC/ZDV vs. ddC/ZDV vs. SQV/ddC/ZDV Modified DB, PBO-control, naive
1100.1136, United States NVP/ddI/ZDV vs. ddI/ZDV vs. 3TC/ddI/ZDV DB, PBO-control, naive
1100.1229 (Atlantic), Europe, N. America d4T/ddI/3TC vs. d4T/ddI/IDV vs. d4T/ddI/NVP OL randomized comparative, naive
1100.1289, Europe, South Africa ZDV/3TC/ABC  NVP  HU  prednisolone Factorial design: OL randomized comparative, naive
Noncomparative clinical trials
1100.834, United States 12.5, 50, 200, 400, 600 mg NVP and ZDV OL, staggered, rising dose cohorts, exp.
1100.947, United States NVP monotherapy OL, noncomparative, naive
1100.1009, United States NVP/ddI or NVP/ZDV/ddI or ddC OL, parallel group, exp.
1100.1010, Australia, Italy, The Netherlands NVP monotherapy; option to add ZDV, ddI, or ddC OL, noncomparative, naive
1100.1047, Italy NVP/ddI/ZDV vs. ZDV/ddI DB, PBO-control, exp.
1100.1204, United States NVP/indinavir, plus stable background nucleoside OL, single-arm, noncomparative, exp.
therapy
1100.1286, Argentina, United States, Canada NVP + background + prednisone 2 weeks vs. Randomized, OL naive/exp.
NVP + background
1100.1264 (FTC-302), S. Africa PBO/3TC/D4T/NVP and FTC/PBO/d4T/NVP Randomized, FTC vs. 3TC: DB, naive

NOTE. 3TC, lamivudine; ABC, abacavir; ART, antiretroviral therapy; d4T, stavudine; DB, double blind; ddC, zalcitabine; ddI, didanosine; exp., experienced;
FTC, emtricitabine; HU, hydroxyurea; OL, open label; NVP, nevirapine; PBO, placebo; SQV, saquinavir; ZDV, zidovudine; , with or without.

S84 • CID 2004:38 (Suppl 2) • Dieterich et al.

Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331

by Universitas Diponegoro user
on 31 August 2018
 Table 2. Rules used to select potential clinical cases of hepatic !6%, except in FTC-302, where the proportion was 9.9%.
events for review Among the controlled trials, the proportions remained consis-
tent across all trials for NVP. In controlled trial BI 1090, the
Rule and event
proportions was comparable for NVP and control treatments
1. Liver: An adverse event with a WHO preferred term from the
(5.8% and 7.3%, respectively; RR, 0.8; P 1 .50), whereas the
hepatic and biliary body system (excluding elevated LFT values,
cholelithiasis, and porphyria). proportions were lower for controls in the remaining controlled
2. Primary (probable): The adverse event must have had a WHO trials (5.9% and 2.9%, respectively; RR, 2.1; P ! .01). The onset
preferred term on the list of events expected to have a reason- of most asymptomatic hepatic events with ALT and/or AST
able likelihood of being associated with hepatic toxicity. levels 15 times the ULN (figure 3) occurred during the first 6
3. Secondary (possible): The adverse event must have had a weeks, whereas the risk declined by more than two-thirds af-
WHO preferred term on the list of events expected to be less
specific to hepatotoxicity. Additionally, the event must have had terward. There was variability in the risk of events across all
an overlapping onset and end date with another possible event. trials during the first 6 weeks of treatment. The top 2 panels
4. Fatal: Any event that was fatal. in figure 3 illustrate the differences, primarily between the con-
NOTE. For all patients with an adverse event with onset within 14 days
trol groups, in trial BI 1090 and the other 8 controlled trials.
after an alanine aminotransferase (ALT)/aspartate aminotransferase (AST) level After 18 weeks, the risk was similar for patients in NVP and
13 times the upper limit of normal (ALT and/or AST measured up to 14 days
after treatment end), rules 2, 3, and 4 were required. If the event started
control groups in the controlled trials. The risk after 18 weeks
before the elevated ALT or AST level was measured, it was required to have
been ongoing within the 14 days before the elevated liver function test (LFT)
result. Examples of preferred terms referring to hepatic events (not a complete Table 3. Demographic and clinical characteristics of patients
listing): hepatobiliary (e.g., hepatitis, hepatic failure, hepatic function abnormal, in randomized clinical trial analyses.
cirrhosis, hepatocellular damage, jaundice, liver fatty, hepatic steatosishepa-
torenal syndrome), primary events (arthritis, fever, chills, rash [any], conjunc-
Controlled trials
tivitis, eosinophilia, infection [bacterial, viral, fungal, or mycobacterial], gastro- All trials,
enteritis, flu, pruitus, abdominal pain), secondary events (anorexia, fatigue, Control NVP-treated NVP-treated
malaise, headache, nausea and /or vomiting). WHO, World Health Organization. Characteristic subjects patients patients
Total no. treated 1911 (100.0) 1731 (100.0) 2545 (100.0)
all potential cases identified by the screening process (table 2). Mean exposure, days 348 357 339
Two mutually exclusive categories were formed that consisted Age, years
of either asymptomatic hepatic events with ALT and/or AST Mean 37.2 37.2 37
levels 15 times the ULN or a symptomatic hepatic event. Symp- SD 8.5 8.9 8.6
tomatic events were further subcategorized by the presence or Sex
absence of rash. ORs were used to describe the effects of risk Female 331 (17.3) 339 (19.6) 633 (24.9)
factors, coupled with 1-sided Fisher’s exact tests of significance. Male 1580 (82.7) 1392 (80.4) 1912 (75.1)
Kaplan-Meier cumulative probability estimates illustrated pat- Race
terns of risk over time. Asian 14 (0.7) 10 (0.6) 18 (0.7)
The demographic and clinical characteristics of the cohort Black 323 (18.7) 323 (18.7) 683 (26.8)
are summarized in table 3. Among the 2545 NVP-treated pa- Hispanic 84 (4.4) 70 (4.0) 141 (5.5)
tients from the 17 controlled and noncomparative trials com- Other 27 (1.4) 32 (1.8) 39 (1.5)
bined (“all trials”), the mean duration of treatment exposure Unknown 300 (15.7) 206 (11.9) 206 (8.1)
was 48 weeks. The majority of patients in both analysis groups White 1172 (61.3) 1090 (63.0) 1458 (57.3)
were white men with baseline CD4+ cell counts of !200 cells/ Baseline CD4 cell count,
mm3. However, there were substantial numbers of women cells/mm3
(1600), black patients (1600), and patients with baseline CD4+ 0–100 604 (31.6) 629 (36.3) 662 (26.0)
cell counts of 1350 cells/mm3 (1700) in the combined database. 1100– 200 462 (24.2) 431 (24.9) 511 (20.1)
The risk of developing asymptomatic elevated ALT and/or 1200– 350 450 (23.5) 361 (20.9) 653 (25.7)
AST levels 15 times the ULN during NVP-based therapy was 1350– 500 280 (14.6) 232 (13.4) 481 (18.9)
variable during the first year and dependent on the observed 1500 114 (6.0) 78 (4.5) 237 (9.3)
population [11]. Ten percent of all patients treated with NVP Baseline ALT level
developed ALT and/or AST levels 15 times the ULN; however, Normal 1301 (68.1) 1173 (67.9) 1833 (72.0)
almost two-thirds of these cases of elevated levels were not 11–2⫻ ULN 428 (22.4) 388 (22.4) 493 (19.4)
associated with any clinical symptoms (table 4). The frequency 12–3⫻ ULN 98 ( 5.1) 105 (6.1) 133 (5.2)
of asymptomatic ALT and/or AST levels 15 times the ULN in 13⫻ ULN 56 (2.9) 43 (2.5) 57 (2.2)
all trials was 6.3%. The proportion of patients with asymptom- NOTE. Data are no. (%) of patients, unless indicated otherwise. ALT,
atic ALT and/or AST levels 15 times the ULN in all trials was alanine aminotransferase; NVP, nevirapine; ULN, upper limit of normal.

Liver Injury and NNRTIs • CID 2004:38 (Suppl 2) • S85

Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331

by Universitas Diponegoro user
on 31 August 2018
 Table 4. Risk of occurence of alanine aminotransferase and/or aspartate aminotrans-
ferase (ALT and/or AST) levels 15 times the upper limit of normal (ULN) among nevirapine
(NVP)–treated patients in controlled trials and in all trials (i.e., controlled and noncom-
parative trials).

Controlled trials
All trials,
NVP-treated Control NVP-treated
patients, subjects, patients,
Type of elevated ALT/AST event, % (n) % (n) Risk % (n)
trial group or trial (n p 1731) (n p 1911) ratio P (n p 2545)
All events in all trials 8.8 (153) 6.2 (118) 1.5 !.01 10.0 (254)
Asymptomatic events
In all trials 5.8 (101) 5.5 (105) 1.1 .35 6.3 (160)
In controlled trials
BI 1090 5.8 (65) 7.3 (82) 0.8 1.50 …
Others 5.9 (36) 2.9 (23) 2.1 !.01 …
In noncomparative trials
FTC-302 … … … … 9.9 (38)
Others … … … … 5.6 (122)

for NVP-treated patients was similar in all the trials, including
FTC-302.
An increased risk of asymptomatic ALT and/or AST levels
15 times the ULN was associated with an elevated baseline ALT
and/or AST level 12.5 times the ULN (RR, 4.3; P ! .01) and
coinfection with HBV (RR, 2.3; P ! .01) or HCV (RR, 5.2;
P ! .01). Sex and baseline CD4 cell count were inconsistent risk
factors, except in men with CD4 cell counts ⭓400 cells/mm3
(RR, 1.6; P ! .01). Overall, the risks of developing ALT and/or
AST levels 15 times the ULN with NVP-based therapies were
similar to those associated with other ART drugs, particularly
with regard to HBV or HCV coinfection [5, 11].
Approximately 5% of all patients treated with NVP devel-
oped symptomatic hepatic events (tables 5 and 6) [11]. Rash
and other possibly immune-mediated events occurred concur-
rently with hepatic events in 2.2% of NVP-treated patients,
whereas 2.7% of NVP-treated patients developed hepatic symp-
toms without rash (other hepatic events). Approximately one-
half (46%) of the symptomatic hepatic events were associated
with rash. For rash-associated hepatic events, the rates among
patients who received NVP ranged from !0.5% in the BI 1090
study to 6% in FTC-302, whereas the rates offor other hepatic
events were nearly identical in FTC-302 (2.9%) and BI 1090
(2.8%). Although substantial differences were observed among
trials for rash-associated hepatic events, there was consistency
across all clinical trials, including FTC-302, for the risk of other
hepatic events. Patients in BI 1090 had lower mean baseline
CD4 cell counts (192 cells/mm3), which suggests that the CD4
cell count may play a role in rash-associated events but may
not be a risk factor for other hepatic events.
In controlled trials, the NVP treatment group was at higher
risk for symptomatic hepatic events than were control subjects.
Rash-associated hepatic events were found to occur almost ex-
clusively among patients who received NVP, yielding an esti-
mated increase in risk of 110-fold (RR, 11.2; P ! .01). NVP was
also found to moderately increase the risk of other hepatic
events (RR, 2.7; P ! .01). After the first 6 weeks, the risk of a
rash-associated event dropped !30-fold (figure 4). The risk also
diminished after 6 weeks for other hepatic events.
The diagnoses of patients with rash-associated events fell into
categories determined primarily by whether or not specific he-
patic symptoms were reported and by the general intensity of
the event. The majority of patients (34 of 57) who developed
rash-associated hepatic events also experienced at least one pos-
sibly immune-related symptom in addition to rash, most often
fever. Most affected patients (41 of 57) discontinued treatment
with NVP permanently, sometimes after rechallenge.
For most patients whose consensus diagnosis was possible
drug-induced hepatitis with rash, the investigator reported spe-
cific hepatic symptoms, such as nausea, vomiting, abdominal
pain, or jaundice. However, this was not the case for patients
whose diagnosis was elevated liver function test (LFT) results
with rash, among whom only 5 of 20 had symptoms other than
elevated LFT results that were not specific for hepatic events,
such as fatigue or dyspepsia. Events diagnosed as possible drug-
induced hepatitis with rash were generally more intense, char-
acterized by higher elevations of ALT and/or AST levels and
coupled with more systemic symptoms in addition to rash.
Almost all patients with a diagnosis of possible drug-induced
hepatitis with rash discontinued NVP permanently. There were
20 patients whose hepatic event was an increase in LFT values
in temporal association with a rash, and the clinical presen-
tation of these events was generally mild. These events were
almost never categorized by the reporting investigator as serious

S86 • CID 2004:38 (Suppl 2) • Dieterich et al.

Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331

by Universitas Diponegoro user
on 31 August 2018
 (1 of 20 events). Although this is not recommended in clinical
practice, the investigator continued or reintroduced therapy
successfully in ∼60% of these 20 NVP-treated patients.
The diagnoses of patients who developed other hepatic events
formed 3 main categories (possible drug-induced hepatitis,
acute infectious hepatitis, and neoplastic hepatitis) determined
by whether there was underlying chronic liver disease or
whether an acute infection was suspected. Most patients with
other hepatic events either continued NVP treatment through-
out the hepatic event or were able to restart NVP without
symptoms recurring.
The key risk factors for developing rash-associated hepato-
toxicity were treatment with NVP, female sex, and higher base-
line CD4 cell count. An increased risk was observed in women
with a baseline CD4 cell count of ⭓250 cells/mm3 and in men
with CD4 counts of ⭓400 cells/mm3 [14]. The development
of a rash-associated hepatic event did not correlate with elevated
baseline AST and/or ALT levels. There were insufficient data
to determine whether rash-associated hepatic events had any
relationship with HBV or HCV coinfection.
The risk profile for developing other hepatic events was sim-
ilar to that associated with asymptomatic ALT and/or AST el-
evations 15 times the ULN. The risk of symptomatic hepato-
toxicity was increased if the patient was coinfected with HBV
(RR, 3.9; P ! .01) but was not increased with HCV coinfection
(RR, 1.2; P p .40). Having a baseline ALT and/or AST level
12.5 times the ULN was also a significant risk factor (RR, 3.2;
P ! .01). Sex and race were not risk factors.
Overall, a low incidence of hepatic events was seen among

Table 5. Frequencies of symptomatic hepatic events among all

nevirapine-treated patients in all clinical trials (controlled and
noncomparative trials).

Percentage (no.)
of patients
Type of hepatic event, trial(s) (n p 2545)
Symptomatic (rash-associated and all others)
All trials 4.9 (125)
Controlled trial BI 1090 3.2 (36)
Noncomparative trial FTC-302 8.9 (34)
Other trials 5.3 (55)
Rash-associated symptomatic
All trials 2.2 (57)
Controlled trial BI 1090 0.4 (5)
Noncomparative trial FTC-302 6.0 (23)
Other trials 2.8 (29)
Figure 3. Cumulative probability of asymptomatic alanine aminotrans-
Other hepatic symptomatic
ferase and/or aspartate aminotransferase (ALT/AST) levels 15 times the
upper limit of normal (ULN) in controlled and in all trials (controlled and All trials 2.7 (68)
noncomparative trials) The probability of asymptomatic ALT/AST levels Controlled trial BI 1090 2.8 (31)
15 times the ULN in BI 1090, controlled trials and all trials (controlled Noncomparative trial FTC-302 2.9 (11)
and noncomparative trials) is also shown. Con, control subjects; NVP, Other trials 2.5 (26)
neviparine-treated patients.

Liver Injury and NNRTIs • CID 2004:38 (Suppl 2) • S87

Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331

by Universitas Diponegoro user
on 31 August 2018
 Table 6. Frequency of symptomatic hepatic events in controlled
trials.
Nevirapine-
treated Control
patients, subjects,
% (n) % (n)
a
Type of hepatic event (n p 1731) (n p 1911) Risk ratio
All symptomatic 4.0 (70) 1.2 (23) 3.5
Rash-associated 1.2 (20) 0.1 (2) 11.2
Other 2.9 (50) 1.1 (21) 2.7
a
Significance level, P ! .01.
patients who receive NVP-based therapy. The majority of cases
of AST and/or ALT elevation were asymptomatic, and this event
did not occur at a rate substantially different from that seen
with other ART drugs [11, 24]. The risk of associated symp-
tomatic hepatic events was relatively low and was more com-
mon during the first 6 weeks of treatment. Almost one-half of
the events were associated with rash and were more frequent
in women and in patients with elevated CD4 cell counts at
baseline. The majority of symptomatic events were not serious
and improved after the withdrawal of study drug. Fulminant
hepatic failure related to NVP treatment was rare in clinical
trials and occurred at a rate (0.1%) similar to that found with
other ART drugs in the ACTG database [10].
DISCUSSION
The management of HIV infection has become increasingly
complex since the introduction of potent combination ART.
HIV-infected patients are exposed to an increasing array of
drugs to treat HIV, prevent opportunistic diseases and immune
dysfunction, and manage comorbidities. Hepatic complications
have become common; recently, these have been attributed to
the use of NNRTIs, in particular NVP.
Clinicians who treat patients with HIV infection frequently
face difficulties differentiating the etiology of abnormal liver
transaminase levels. The differential diagnosis will encompass
liver injury related to combination ART, other potentially hep-
atotoxic concomitant drugs and herbs, or underlying acute or
chronic liver disease [25]. Although all ART drugs are associated
with elevations of ALT and/or AST levels to varying degrees,
there are insufficient data to suggest that the risk of serious or
life-threatening clinical hepatotoxicity is correlated with small
differences in the rate of asymptomatic background elevated
ALT and/or AST levels found with one ART regimen compared
with another.
The data discussed in the present article suggest that there
is an apparent modest class effect of NNRTIs in terms of ab-
normal liver enzyme levels, but the rate of serious clinical
S88 • CID 2004:38 (Suppl 2) • Dieterich et al.
Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331
by Universitas Diponegoro user
on 31 August 2018
(symptomatic) hepatotoxicity in patients receiving these drugs
is low. Furthermore, analyses have demonstrated that asymp-
tomatic elevation of ALT and/or AST levels at baseline, together
with HBV and HCV coinfection, are significant predictors of
hepatotoxicity for all ART drugs. The incidence of asymptom-
atic elevated ALT and/or AST levels with NVP therapy appears
to be similar to that reported for other ART drugs. Symptomatic
hepatic events associated with NVP use are infrequent and may
not occur more often than they do with other ART drugs.
However, a unique hepatic event among patients treated with
NVP has been identified in clinical trials and is suggestive of
an immune-mediated phenomenon associated with rash. The
risk of this rash-associated hepatic event is almost exclusively
within the first 6 weeks of NVP treatment and is greater in
women with CD4 cell counts ⭓250 cell/mm3 and in men with
CD4 cell counts ⭓400 cell/mm3. The frequencies of hepatic
events, including ALT and/or AST levels 15 times the ULN,
serious clinical hepatic events, and hepatic-related mortality,
related to NVP treatment are similar and consistent between
randomized clinical trials and cohort studies, compared with
their frequencies in association with therapy with other ART
drugs.
It is difficult for physicians treating patients with HIV in-
fection to distinguish liver toxicities in individuals receiving
several drugs who may also have risk factors for hepatotoxicity
such as viral hepatitis infections, alcohol use, female sex, and
substance abuse. Additionally, many common antibiotics, an-
tifungals, and antivirals prescribed for HIV-infected patients
are independently associated with hepatotoxicity. There is,
therefore, a need to conduct further studies to evaluate the
hepatic effects of NNRTIs, to clarify both class and drug effects,
as well as to understand the hepatotoxic potential of antiret-
roviral and other drugs used in the management of the HIV-
Figure 4. Cumulative probability of rash-associated hepatic events
associated with nevirapine (NVP) treatment in all clinical trials: the prob-
ability of rash-associated hepatic events (%) in BI 1090, FTC-302, and
all other nevirapine trials over the first 3 months
 infected patient. As with any decision to prescribe a drug, a
careful evaluation of the potential risks and benefits of using
NVP must be made for each individual.
References
1. Department of Health and Human Services, Henry J. Kaiser Family
Foundation. Guidelines for the use of antiretroviral agents in HIV-
infected adults and adolescents. Available at: http://www.
hivatis.org, 2002.
2. Palmon R, Koo BC, Shoultz DA, Dieterich DT. Lack of hepatotoxicity
associated with nonnucleoside reverse transcriptase inhibitors. J Acquir
Immune Defic Syndr 2002; 29:340–5.
3. Lee W. Drug-induced hepatotoxicity. N Engl J Med 2003; 349:474–85.
4. Sulkowski MS, Thomas DL, Mehta SH, Chaisson RE, Moore RD. Hep-
atotoxicity associated with nevirapine or efavirenz–containing antiret-
roviral therapy: role of hepatitis C and B infections. Hepatology 2002;
35:182–9.
5. Sulkowski MS, Thomas DL, Chaisson RE, Moore RD. Hepatotoxicity
associated with antiretroviral therapy in adults infected with human
immunodeficiency virus and the role of hepatitis C or B virus infection.
JAMA 2000; 283:74–80.
6. Shakil A, Kramer D, Mazariegos G, Fung JJ, Rakela J. Acute liver failure:
clinical features, outcome analysis, and applicability of prognostic cri-
teria. Liver Transpl 2000; 6:163–9.
7. Zimmerman H. Drug-induced liver disease. In: Schiff E, Sorrell W,
Muldrey W, eds. Schiff’s diseases of the liver. 8th ed. Philadelphia:
Lippincott-Raven Publishers, 1999:973–1064.
8. DeLeve L, Kaplowitz N. Prevention and therapy of drug-induced he-
patic injury. In: Wolfe M, ed. Therapy of diagnostic disorders. Phila-
delphia: WB Saunders, Harcourt, Brace, 2000:334–48.
9. Lanes S. Reporting rates of adverse events for antiretroviral medica-
tions, 1996–2001. Ridgefield, CT: Boehringer Ingelheim Pharmaceu-
ticals, 2001.
10. Reisler R, Servoss JC, Sherman KE, et al. Incidence of hepatotoxicity
and mortality in 21 adult antiretroviral treatment trials: ACTG Liver
Diseases Focus Group [abstract 43]. In: Program and abstracts of the
1st International AIDS Society Conference on HIV Pathogenesis and
Treatment (Buenos Aires). 2001.
11. Stern J, Robinson P, Love J, Lanes S, Imperiale M, Mayers D. A com-
prehensive hepatic safety analysis of nevirapine in different populations
of HIV infected patients. J Acquir Immune Defic Syndr 2003; 34(Suppl
1):S21–33.
Downloaded from https://academic.oup.com/cid/article-abstract/38/Supplement_2/S80/331331
by Universitas Diponegoro user
on 31 August 2018
12. Wit FW, Weverling GJ, Weel J, Jurriaans S, Lange JM. Incidence of
and risk factors for severe hepatotoxicity associated with antiretroviral
combination therapy. J Infect Dis 2002; 186:23–31.
13. Anonymous. Serious adverse events attributed to nevirapine regimens
for postexposure prophylaxis after HIV exposures—worldwide,
1997–2000. MMWR Morb Mortal Wkly Rep 2001; 49:1153–6.
14. Lebray P, Benhamou Y, Pol S, Myers R, Poynard T, Di Martino V.
NNRTI-related or unrelated hepatotoxicity? [letter]. Hepatology
2002; 36(2):512–3.
15. Sulkowski MS, Thomas DL, Mehta SH, Chaisson RE, Moore RD. Reply
to “NNRTI-related or unrelated hepatotoxicity” [letter]. Hepatology
2002; 36:513–4.
16. Martinez E, Blanco JL, Arnaiz JA, et al. Hepatotoxicity in HIV-1–in-
fected patients receiving nevirapine-containing antiretroviral therapy.
AIDS 2001; 15:1261–8.
17. Piliero P, Purdy B. Nevirapine-induced hepatitis: a case series and
review of the literature. AIDS Reader 2001; 11:379–82.
18. Dieterich D, Becker S, Fusco J, et al. Low incidence of grade III/IV
hepatotoxicity in first HAART observations from 1,237 patients fol-
lowed for one year (poster 4534). In: Program and abstracts of the
XIV Conference of the International AIDS Society (Barcelona). 2002.
19. Becker S. Liver toxicity in epidemiological cohorts. Clin Infect Dis
2004; 38:S49–55 (in this issue).
20. Imperiale S, Lanes S, Stern J, Love J, Robinson P, Mayers D. TARGET:
incidence of elevated ALT/AST with HAART in a large observational
cohort [abstract P150]. In: Program and abstracts of the 6th Inter-
national Congress on Drug Therapy in HIV Infection (Glasgow). 2002:
150.
21. Friis-Møller N, Kirk O, Reiss P, et al. Safety of non-nucleoside reverse
transcriptase therapy: data from the EUROSIDA study [abstract 24].
In: Abstracts of the 5th International Workshop on Adverse Drug
Reactions and Lipodystrophy in HIV (Paris). Antivir Ther 2003; 8:L20.
22. Kaplowitz N. Drug-induced liver injury. Clin Infect Dis 2004; 38:S44–9
(in this issue).
23. Gonzalez de Reqena D, Nurez M, Jimenez-Nacher I, et al. Liver toxicity
by nevirapine. AIDS 2002; 16:290–1.
24. Stern J, Love J, Robinson P, Lanes S, Imperiale M, Mayers D. Hepatic
safety of nevirapine: results of the Boehringer Ingelheim Viramune
Hepatic Safety Project (abstract LBOr15). In: Program and abstracts
of the XIV Conferenece of the International AIDS Society (Barcelona).
2002.
25. Orenstein R, Tsago N. Looking beyond HAART: drug-related hepa-
totoxicity in patients with human immunodeficiency virus infection.
Pharmacotherapy 2002; 22:1468–78.
Liver Injury and NNRTIs • CID 2004:38 (Suppl 2) • S89