Full text of DOI:10.1592/phco.21.17.1345.34431

Retrospective Evaluation of the Effect of Valproate
Therapy on Transaminase Elevations in Patients with
Hepatitis C
Rex S. Lott, Pharm.D., Kristin M. Helmboldt, Pharm.D., and Karl J. Madaras-Kelly, Pharm.D.
Study Objective. To evaluate whether a relationship exists between valproate
treatment and transaminase elevations in patients who are positive for
hepatitis C virus (HCV).
Design. Retrospective medical record review.
Setting. Veterans affairs medical center.
Patients. A total of 214 HCV-positive patients; 28 were treated with valproate
(study group), 186 were not (control group).
Intervention. Demographic characteristics; valproate treatment histories;
plasma concentrations of alanine aminotransferase (ALT) and aspartate
aminotransferase (AST), albumin, and bilirubin; and exposure to other
potentially hepatotoxic drugs, including probable alcohol abuse, were
evaluated.
Measurements and Main Results. Hepatotoxicity was staged by comparing
maximum ALT and AST values against upper limits of normal or patients’
mean elevated baseline ALT and AST values. Data were analyzed using a
logistic regression model. Control patients and those with longer durations
of HCV were more likely (p<0.0171 and p<0.0142, respectively) to exhibit
higher stages of hepatotoxicity. More valproate-treated patients were
exposed to other potential hepatotoxins at the time of peak transaminase
elevations (50% vs 39%, p<0.005), whereas more control patients received
two or more potential hepatotoxins (13% vs 4%, p<0.00005).
Conclusion. Valproate treatment, either alone or in the presence of other
potential hepatotoxins, does not appear to be related to increased
transaminase elevations in patients with HCV.
(Pharmacotherapy 2001;21(11):1345–1351)
From the Boise Veterans Affairs Medical Center, and the
Department of Pharmacy Practice and Administrative
Sciences, College of Pharmacy, Idaho State University, Boise,
Idaho (all authors).
Supported in part by the Boise Veterans Affairs Medical
Center and an unrestricted educational grant from Abbott
Laboratories.
Presented in part at the annual meeting of the American
College of Clinical Pharmacy, Los Angeles, California,
November 6, 2000, and the Western States Residency
Conference, Pacific Grove, California, May 20, 2000.
Address reprint requests to Rex S. Lott, Pharm.D.,
Veterans Affairs Medical Center, 500 West Fort Street
(119A), Boise, ID 83702.
Infection with hepatitis C virus (HCV) is
common and often leads to severe liver
impairment and failure. An estimated 4 million
Americans are HCV positive,1, 2 and chronic
hepatitis C develops in 75–85% of these patients,
with persistent or fluctuating elevations of
alanine aminotransferase (ALT). Patients with
chronic hepatitis C have a 1–5% risk for
developing hepatocellular carcinoma, and
cirrhosis may develop in 20–30%.1, 2 In the
United States, hepatitis C causes an estimated
8000–10,000 deaths annually.3

1346
PHARMACOTHERAPY Volume 21, Number 11, 2001
Valproate (divalproex sodium, valproic acid)
between treatment with valproate and evidence
of liver damage. Our objective was to compare
the occurrence of increased transaminase levels
in HCV-positive patients treated with valproate
versus those not treated with the agent.
has been associated with idiosyncratic, severe
hepatotoxicity and with dose-related, usually
reversible increases in ALT and aspartate
aminotransferase (AST) levels.4, 5 Valproate
therapy is administered for various neurologic
and psychiatric disorders, including epilepsy,
migraine, and bipolar disorder. Patients with
hepatitis C are commonly treated with valproate
for coexisting mood disorders. The Centers for
Disease Control and Prevention provide no
recommendations regarding administration of
potentially hepatotoxic drugs other than alcohol
for patients with hepatitis C. Nevertheless,
clinicians often express concern that treatment
with potential hepatotoxins, such as valproate,
may increase risk of liver damage in HCV-
positive patients. To our knowledge, no
published studies have examined the effect of
valproate treatment on liver function in patients
with hepatitis C.
Methods
We retrospectively evaluated all patients
identified as HCV positive by enzyme immunoassay
between January 1992 (when this screening test
became available) and April 2000. Demographic
data, valproate treatment histories, and results of
plasma concentration determinations for ALT
and AST, bilirubin, and albumin were collected.
The study group consisted of HCV-positive
patients who started valproate therapy on or after
their date of diagnosis of hepatitis C. In these
patients, valproate was prescribed for treatment
of bipolar disorder or mood instability related to
posttraumatic stress disorder; none had a seizure
disorder. The control group consisted of patients
with hepatitis C who did not receive valproate
therapy. Administration of other prescribed
We evaluated all HCV-positive patients at the
Boise Veterans Affairs Medical Center, Boise,
Idaho, to determine whether a relationship exists
HCV-positive patients
439
Nonvalproate-treated
Valproate-treated patients
72
patients
367
Evaluable
study patients
n=28
Evaluable
control patients
n=186
Excluded study patients
Excluded control patients
No ALT/AST values
obtained before or after
valproate treatment
18
Valproate treatment
No ALT/AST values
No ALT/AST values
obtained before
hepatitis C diagnosis
162
before
hepatitis C diagnosis
26
obtained after
hepatitis C diagnosis
19
Figure 1. Selection and stratification of HCV-positive patients. ALT = alanine aminotransferase; AST = aspartate
aminotransferase.

VALPROATE AND TRANSAMINASE ELEVATIONS IN PATIENTS WITH HEPATITIS C Lott et al 1347
Table 1. Stages of Potential Hepatotoxicity
potentially hepatotoxic substances—acetaminophen,
ALT and AST Values
any 3-hydroxy-methylglutaryl coenzyme A
reductase inhibitor, any thiazolidinedione, or any
protease inhibitor—was evaluated in both groups.
Risk of hepatotoxicity with acetaminophen is
unknown in patients with hepatitis C; therefore,
any exposure to acetaminophen was considered
potentially hepatotoxic. Evidence of probable
alcohol abuse was also evaluated in both study
and control patients. Information regarding
alcohol abuse was often not readily obtainable by
review of patient medical records. Therefore,
patient prescription records were reviewed, and
treatment with disulfiram or naltrexone was used
as an indication of overt alcohol abuse.
Stage
Normal Baseline
Elevated Baseline
0
1
2
3
4
< 1.25 x ULN
1.25–2.5 x ULN
2.6–5 x ULN
5.1–10 x ULN
> 10 x ULN
< 1.25 x baseline
1.25–2.5 x baseline
2.6–3.5 x baseline
3.6–5 x baseline
> 5 x baseline
ALT = alanine aminotransferase; AST = aspartate aminotransferase;
ULN = upper limit of normal.
Adapted from reference 6.
Statistical Analyses
Statistical analyses included the t test or, when
appropriate, ␹ or Fisher exact test for
2
To be included in the control group, ALT and
AST determinations must have been recorded at
least 6 months before diagnosis of hepatitis C,
and these determinations must have been
repeated at least once after diagnosis. To be
included in the study group, ALT and AST
determinations must have been recorded within 6
months preceding the start of valproate therapy,
and these determinations must have been
repeated at least once during valproate therapy.
Patients were considered to have received
valproate therapy if either valproic acid or
divalproex sodium had been prescribed for at
least 1 month. Pharmacy prescription and refill
records were reviewed to assess valproate
administration. Figure 1 presents the selection
and stratification process for our population of
HCV-positive patients.
Hepatotoxicity was staged for each patient
using a modified published method.6 When a
patient’s baseline ALT and AST values were
within normal range (ALT < 31 U/L, AST < 35
U/L), maximum values were compared with
upper limits of normal. When a patient’s baseline
ALT and AST values were elevated, maximum
values were compared with the patient’s mean
baseline ALT and AST values. For the control
group, baseline ALT and AST values were the
mean of all available determinations before
diagnosis of hepatitis C. For the study group,
baseline ALT and AST values were the mean of
all available determinations after diagnosis of
hepatitis C and before valproate treatment.
Hepatotoxicity stages are defined in Table 1.6 If a
patient’s ALT or AST values did not fall in the
same staging category, the higher of the two was
used for classification. Alcohol abuse or
exposure to any other potentially hepatotoxic
substance occurring during the 6 months before
peak ALT and AST elevations was assessed.
comparison of demographic variables and
characteristics of both patient groups. The effect
of valproate on hepatotoxicity (indicated by
elevated transaminase levels) and the influence of
other possible predictors were evaluated using
logistic regression. The SAS release 8.02 (SAS
Institute, Cary, NC) was used for statistical
modeling. The study was approved by the local
institutional review board and the research and
development committee of the Boise Veterans
Administration Medical Center.
Results
A total of 214 HCV-positive patients met the
inclusion criteria (Figure 1). Of these, 28
(13.1%) received valproate and 186 (86.9%) did
not. Demographic and laboratory characteristics
are detailed in Table 2. The two groups did not
differ significantly with respect to age and
gender. Duration of HCV was significantly
longer in study group patients than in controls
(45.8 ± 25.1 vs 33.5 ± 20.3 mo). At the time of
peak ALT and AST elevations, plasma albumin
concentrations in the study and control groups
were not significantly different (3.90 ± 0.44 vs
3.98 ± 0.62 g/dl), whereas plasma bilirubin
concentrations were significantly lower in the
study group (0.52 ± 0.33 vs 1.33 ± 3.97 mg/dl).
Characteristics of valproate treatment for the
study group are shown in Table 3. Dosages were
375–3000 mg/day (mean 1031 ± 527 mg/day).
Body weight was unavailable for 4 of the 28
patients; dosages for the other 24 patients were
5.22–43.42 mg/kg/day (mean 13.9 ± 7.63
mg/kg/day). Plasma valproate concentrations
were available for 24 study patients. At the time
of peak elevations in ALT and AST, plasma
concentrations were 10–88 µg/ml (mean 52 ± 20
µg/ml). Duration of exposure to valproate before

1348
PHARMACOTHERAPY Volume 21, Number 11, 2001
Table 2. Patient Characteristics and Laboratory Parameters
Valproate-Treated
Patients
Control Patients
(n=186)
Variable
Age (yrs)
M/F
(n=28)
p Value
47.0 ± 5.4
26/2
45.8 ± 25.1
3.90 ± 0.435
0.516 ± 0.328
48.0 ± 8.8
177/9
33.5 ± 20.3
3.98 ±0.624
1.33 ± 3.97
> 0.4a
NSb
Hepatitis C diagnosis (mo)
Plasma albumin at peak ALT and AST levels (g/dl)
Plasma bilirubin at peak ALT and AST levels (mg/dl)
< 0.005a
> 0.05a
< 0.05a
ALT = alanine aminotransferase; AST = aspartate aminotransferase; NS = not significant.
Data are mean ± SD.
at test.
b
␹2.
Table 3. Characteristics of Valproate Treatment in Study
Patients (n=28)
versus 12 controls (6.5%) exhibited stage 4
hepatotoxicity.
Characteristic
Value
The effect of valproate on hepatotoxicity
(indicated by elevated transaminases) and the
influence of other possible predictors were
evaluated using four sets of logistic regression
models, one set for each of four outcomes: a
binary model comparing stage 0–1 versus stage
2–4 hepatotoxicity, a binary model comparing
stage 0–2 versus stage 3–4 hepatotoxicity, a
binary model comparing stage 0–3 versus stage 4
hepatotoxicity, and an ordinal model comparing
each of the 0–4 stages with the others. For each
set of models, the initial model considered all
potential predictors of hepatotoxicity: a binary
drug group variable for valproate exposure, a
binary variable for exposure to any other
hepatotoxin, a binary variable indicating whether
baseline transaminases were elevated, duration of
hepatitis C diagnosis (mo), age (yrs), a binary
variable indicating depressed levels of serum
albumin at the time of peak transaminase
elevation, a binary variable indicating elevated
levels of serum bilirubin at the time of peak
transaminase elevation, and interactions between
the drug group variable and each potential
confounder.
For each binary logistic regression model, the
only statistically significant predictor of
hepatotoxicity was the binary variable indicating
elevated plasma levels of bilirubin. Elevated
plasma bilirubin levels were associated with
higher stages of hepatotoxicity in each binary
model (p<0.0001; odds ratio estimates of 8.69
with 95% confidence interval [CI] 3.68–20.51 for
the first binary model described above, 8.30 with
95% CI 3.22–21.36 for the second, and 16.36
with 95% CI 4.55–58.80 for the third). Adjusting
for confounding, no statistically significant effect
of valproate on hepatotoxicity was noted with
Mean dosage (mg/day)
1031 ± 527
13.9 ± 7.63
Mean dosage (mg/kg/day; n=24)
Mean plasma valproate concentration
at peak ALT and AST levels (µg/ml; n=24)
Mean duration of valproate treatment
at peak ALT and AST levels (mo)
Mean total duration of valproate
treatment (mo)
52 ± 20
4.70 ± 8.36
9.79 ± 11.08
ALT = alanine aminotransferase; AST = aspartate aminotransferase.
Data are mean ± SD.
ALT and AST elevations peaked ranged from
0.3–43 months (mean 4.8 ± 8.3 mo). Total
duration of valproate exposure (until data
collection for this study was completed) was
1–43 months (mean 9.79 ± 11.1 mo). In nine
study patients, increases in ALT and AST levels
apparently led to discontinuation of valproate.
At the time of peak elevations in ALT and AST,
significantly more patients in the study group
than the control group were exposed to other
2
potential hepatotoxins (50% vs 39%, ␹
p<0.00005; Table 4). However, more control
patients than study patients were receiving two
2
or more potential hepatotoxins (13% vs 4%, ␹
p<0.00005). Because of the small number of
patients, exposure to specific other hepatotoxins,
including apparent alcohol abuse, was analyzed
by Fisher exact test. Differences between the
study and control groups were significant only
for exposure to protease inhibitors (p=0.046).
Higher stages of hepatotoxicity were found in
control patients (p<0.05; Table 5). One study
defined stage 3 or 4 as severe hepatotoxicity.6
Stage 3 or 4 hepatotoxicity was observed in two
(7.1%) valproate-treated patients and 21 (11.3%)
control patients. No valproate-treated patients

VALPROATE AND TRANSAMINASE ELEVATIONS IN PATIENTS WITH HEPATITIS C Lott et al 1349
Table 4. Exposure to Other Potential Hepatotoxins at Time of Peak ALT and AST Levels
No. (%) of
Valproate-Treated
Patients (n=28)
No. (%) of
Control Patients
(n=186)
p Value
Patients receiving other potential hepatotoxin
Acetaminophen
Apparent alcohol abuse
Protease inhibitors
HMG-CoA reductase inhibitors
Thiazolidinediones
14 (50)
4 (14)
8 (29)
2 (7)
2 (7)
0 (0)
73 (39)
42 (23)
49 (26)
1 (0.5)
5 (2.6)
1 (0.5)
48 (26)
25 (13)
< 0.005a
NSb
NSb
0.046b
NSb
NSb
Patients receiving one other potential hepatotoxin
Patients receiving two or more other potential hepatotoxins
13 (46)
1 (4)
< 0.03a
< 0.00005a
ALT = alanine aminotransferase; AST = aspartate aminotransferase; NS = not significant; HMG-CoA = 3-hydroxy-methylglutaryl coenzyme A.
a
␹2.
bFisher exact test.
Table 5. Hepatotoxicity Stages in Study and Control
Patients
suggests that hepatitis C itself is a better
predictor of hepatic toxicity than is exposure to
drugs such as valproate.
No. (%) of
Valproate-Treated
Patients (n=28)
No. (%) of
Control Patients
(n=186)
Severe or fatal valproate-induced hepatotoxicity
is a rare, idiosyncratic reaction. The etiology of
valproate-induced hepatic injury is not fully
understood. One study hypothesized that a
metabolite of valproate, 4-en-VPA, causes direct
hepatic damage (microvesicular steatosis).4 Risk
factors associated with increased incidence of
fatal hepatotoxicity include age younger than 2
years; polytherapy with enzyme-inducing
antiepileptic drugs; and conditions such as
mental retardation, developmental delay,
congenital metabolic abnormalities, and other
neurologic diseases. In high-risk patients, fatal
hepatotoxicity may occur in approximately 1/600
treated patients. In adults without risk factors,
valproate-induced fatal hepatotoxicity occurs in
less than 1/500,000 patients.4 Mild, usually
transient elevations in ALT or AST are commonly
observed during valproate treatment and are not
usually associated with clinically significant
hepatic damage. Most clinicians assume that
unless ALT and AST levels are elevated to more
than 3 times upper limits of normal, valproate
treatment can be safely continued. In the staging
system we used, either stage 3 or stage 4 was
originally defined as severe hepatotoxicity.6
Stages 3 and 4 correspond to ALT and AST
elevations at least 5.1 times the upper limits of
normal in patients with normal baseline enzyme
levels and 2.6 times baseline in patients with
elevated baseline enzyme levels. Logistic
regression analysis of our data should help clarify
this issue. Using this technique, we examined
the association between any stage of apparent
hepatotoxicity and potential influences on
hepatotoxicity.
Stage
0
1
2
3
20 (71)
5 (18)
1 (4)
2 (7)
0 (0)
85 (46)
70 (38)
10 (5)
9 (5)
4
12 (6)
p<0.05 ␹2.
these models. With the ordinal logistic model
(comparing each hepatotoxicity stage with the
others), three variables were statistically
significant predictors of hepatotoxicity stage:
elevated bilirubin plasma concentrations
(p<0.0001), longer duration of hepatitis C
diagnosis (p=0.0142), and being in the control
group (p=0.0171).
Discussion
To our knowledge, this is the first study
evaluating risk of liver damage (transaminase
elevations) secondary to valproate treatment in
HCV-positive patients. Our results seem to
indicate that valproate treatment is not
significantly associated with increased risk of
liver damage in these patients. In addition,
neither administration of the other potentially
hepatotoxic drugs reviewed in this study nor
probable concomitant alcohol abuse appeared to
increase the potential risk of hepatic damage
related to valproate therapy in our population of
HCV-positive patients. The significant statistical
relationship between duration of hepatitis C
diagnosis and stage of hepatotoxicity observed at
the time of peak ALT and AST concentrations

1350
PHARMACOTHERAPY Volume 21, Number 11, 2001
Our study had several limitations. Its
relatively conservative valproate dosages and
modest plasma valproate concentrations may be
indicative of this cautious approach to treatment.
In addition, it is possible that treatment with
valproate in the study patients is a marker for
more intensive medical-psychiatric attention and
that these patients were receiving more health
care services than the control patients. The
higher rate of exposure to other potential
hepatotoxins (including probable alcohol abuse)
in the control patients may partially account for
the higher stages of apparent hepatotoxicity in
this nonvalproate-treated group. Notably,
however, our logistic regression models did not
identify potential hepatotoxin exposure as a
significant predictor of hepatotoxicity staging. In
addition, study patients, who were apparently at
lower risk for hepatotoxicity, had a significantly
higher overall rate of exposure to potential
hepatotoxins and a higher rate of exposure to
protease inhibitors than control patients. The
numbers of patients exposed to protease
inhibitors were small in both groups, however,
and the clinical significance of this statistical
difference is unclear.
retrospective design limited our ability to assess
actual incidence of severe hepatotoxicity. We
used a surrogate marker for hepatotoxicity,6 but
ideally, hepatotoxicity would be better assessed
using histologic examination. Of interest was the
clear, significant association between elevations
in plasma bilirubin at the time of peak elevations
in ALT and AST and hepatotoxicity stage. This
association is expected; bilirubin elevations
would be expected in patients with severe
hepatotoxicity. Our ability to fully evaluate the
effect of other potential hepatotoxins combined
with valproate was also limited by the study’s
retrospective design and by our use of a surrogate
marker to identify alcohol abuse. Alcohol abuse
likely was more common in both the study and
control groups. Alcohol abuse is believed to be a
significant risk factor for increased liver damage
in patients with HCV.
Our small sample may have limited our ability
to detect a significantly different risk for
development of severe hepatotoxicity. We
assumed that all patients who were reactive on
testing for HCV antibody by enzyme-linked
immunosorbent assay had either acute or chronic
active infection. Confirmation of active infection
would require confirmatory testing for HCV
RNA.3 If our assumption was incorrect, patients
without hepatitis C may have been over-
represented in our sample, resulting in under-
estimation of the impact of valproate in patients
with this disease. Diagnosis of active hepatitis C
in our patients whose ALT and AST levels were
persistently in the normal range might be
questioned. However, patients with hepatitis C
may commonly have fluctuating ALT and AST
values despite the presence of active infection. In
addition, normal ALT and AST levels with
concomitant severe liver damage (e.g.,
hepatocellular carcinoma) or active infection is
not uncommon.2, 7 Therefore, we had difficulty
ascertaining whether the elevated ALT and AST
values resulted from valproate therapy or natural
course of the disease.
Conclusion
It is not yet known whether administration of
potentially hepatotoxic agents such as valproate
in patients with hepatitis C is associated with
worsening of this condition. Clinical caution is
probably warranted until prospective or large-
scale epidemiologic studies clarify this question.
Data from our study indicate a low or
nonexistent risk of increased hepatic damage in
patients with hepatitis C who receive valproate
treatment. Although our data do not support
routine intensive monitoring of liver function
tests in HCV-positive patients treated with
valproate, these patients should nevertheless be
monitored for deterioration of liver function
during treatment.
Acknowledgment
Although a clear statistical association between
treatment with valproate and lower degrees of
apparent hepatotoxicity was noted, this
association does not clearly represent a causal or
protective association. Our study patients were
known to be positive for HCV and were being
treated with a potentially hepatotoxic drug; they
may have been treated more cautiously and
monitored more closely than were HCV-positive
patients who did not receive valproate. The
The authors acknowledge Chris Johnson, M.P.H.,
for assistance with statistical analyses.
References
1. Centers for Disease Control and Prevention. Recommendations
for prevention and control of hepatitis C virus (HCV) infection and
HCV-related chronic disease. MMWR 1998;47(no. RR-19):1–15.
2. Lam NP. Hepatitis C: natural history, diagnosis, and
management. Am J Health-Syst Pharm 1999;56:961–76.
3. National Institute of Diabetes and Digestive and Kidney
Diseases. Chronic hepatitis C: current disease management.

VALPROATE AND TRANSAMINASE ELEVATIONS IN PATIENTS WITH HEPATITIS C Lott et al 1351
National Institutes of Health publication no. 99-4230, May,
1999.
4. Bryant III AE, Dreifuss FE. Valproic acid hepatic fatalities. III.
U.S. experience since 1986. Neurology 1996;46:465–9.
5. Willmore LJ, Wilder BJ, Bruni J, Villarreal HJ. Effect of valproic
acid on hepatic function. Neurology 1978;28:961–4.
6. 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.
7. Morton TA, Kelen GD. Hepatitis C. Ann Emerg Med
1998;31(3):381–90.