To the Editor: Sodium valproate has been associated with acute hepatitis, including hepatic necrosis and death (1, 2). Risk factors include an age of less than 2 years, polytherapy, and neurological or metabolic abnormalities (1). Inspection of all published fatalities and reviews due to hepatic failure with valproate in the English, French, and Spanish literature revealed none in which an adult without preexisting hepatic or neurological disease was receiving valproate monotherapy (1, 2). We report what is to our knowledge the first such case.
Mr. A was a 51-year-old Hispanic man who had had 3 days of nausea, vomiting, and "orange urine." Three months previously, he began taking valproate as divalproex sodium, 125 mg t.i.d., for bipolar disorder not otherwise specified, consistent with the recent trend for the increasing use of alternatives to lithium as a first-line treatment for this disorder. He had a history of alcohol abuse (severity and duration unknown but no social or medical sequelae) and combat-related posttraumatic stress disorder; both had been in full, sustained remission for over 20 years. Mr. A smoked marijuana several times per week and had three drinks per week until treatment but ceased drinking alcohol after entering treatment. He had had several prior minor surgical procedures, but he had no current medical complaints and a normal primary care physical examination. His only medication was albuterol for asthma, as needed. Although no information on prior herbal or over-the-counter remedies was available, Mr. A agreed not to take any nonprescribed substances at the beginning of treatment. He had received no psychotropic medications since the 1970s. The results of screening laboratory tests before the administration of divalproex included normal aspartate aminotransferase (25 U/liter), alanine aminotransferase (22 U/liter), and total bilirubin (0.5 mg/dl) levels and a normal CBC.
After 1 month of minimal response, Mr. A’s divalproex dose was increased to 250 mg t.i.d. with no ill effects; his psychiatrist endorsed continuation at this dose when his aspartate aminotransferase, alanine aminotransferase, bilirubin, γ-glutamyltransferase, and CBC were again normal and his divalproex level was 51.6 μg/ml. Five weeks later, Mr. A’s symptoms had almost completely remitted; his psychiatrist increased his divalproex dose to 500 mg b.i.d.
After 3 months of divalproex treatment, Mr. A was icteric but had no hepatomegaly or tenderness. His divalproex level was 89.7 μg/ml, with total/direct bilirubin levels of 9.1/4.9 mg/dl (normal ranges=0.2–1.2/0.0–0.2, total/indirect), an aspartate aminotransferase level of 3330 U/liter (normal range=10–45), an alanine aminotransferase level of 3208 U/liter (normal range=7–52), a γ-glutamyltransferase level of 226 U/liter (normal range=10–65), an alkaline phosphatase level of 172 U/liter (normal range=30–115), an ammonia level of 162 μmol/liter (normal range=0–35), a prothrombin time of 18.8 sec (normal range=11–14), a partial thromboplastin time of 39 sec (normal range=27–37), an albumin level of 3.0 gm/dl (normal range=3.5–5.0), a ferritin level of 11,124 ng/ml (normal range=20–300), an iron level of 238 μg/dl (normal range=30–180); normal total iron-binding capacity, amylase, and hepatitis A/B/C serologies; noncontributory electrolyte levels; and a normal CBC. His urine toxicology was positive for cannabinoids; his blood alcohol level was nil.
Over his 10-day hospital course, as transfer for a liver transplant was being arranged, Mr. A became delirious but improved with lactulose and supportive care. His liver function worsened (a maximum aspartate aminotransferase level of 3645 U/liter, an alanine aminotransferase level of 3500 U/liter, a total bilirubin level of 19.7 mg/dl, a prothrombin time of 28 sec, a partial thromboplastin time of 48 sec, and an albumin level of 1.6 gm/dl); a biopsy was deferred pending transfer. Mr. A was transferred to a regional liver transplant center and died 1 month later.
The clinical diagnosis was unequivocally acute hepatitis, with the time course consistent with valproate monotherapy as the precipitant. No other risk factors were evident, although it is possible that this rare event was subserved by silent risk factors, such as characteristics of the cytochrome P450 system (e.g., reference 3). Thus, the risk of fatality from valproate monotherapy must be considered in the risk-benefit assessment during treatment choice, even in the absence of other manifest risk factors.