Julie C. Oki

Troglitazone-Induced Hepatic Failure Leading to Liver Transplantation: A Case Report

Troglitazone (Rezulin, Parke-Davis, Morris Plains, New Jersey) is a new oral agent approved for treatment of patients with type 2 diabetes mellitus [1]. It inhibits hepatic gluconeogenesis and increases peripheral insulin sensitivity [2], possibly by binding to peroxisomal proliferator-activated receptors and altering insulin-dependent gene expression in the liver [3, 4]. Because troglitazone is the first available agent in a new class of drugs for treatment of a common disease, it has been widely prescribed since its approval by the U.S. Food and Drug Administration on 29 January 1997. In clinical trials, troglitazone-induced hepatotoxicity (alanine aminotransferase level > three times the upper limit of normal) was identified in 1.9% of 2510 patients; these abnormalities resolved with discontinuation of therapy with the drug [5]. In contrast to this experience in clinical trials, postmarketing reports to the U.S. Food and Drug Administration since 15 December 1997 include liver transplantation and death caused by troglitazone hepatotoxicity in 3 of 500 000 patients treated in the United States and 2 deaths out of 150 000 patients treated in Japan [6]. Because of these reports, monthly monitoring of aminotransferase levels in all patients during the first 6 months of therapy and every 2 months for the remainder of the first year is now recommended [7]. The clinical characteristics and histologic correlates of severe troglitazone hepatotoxicity have not been described in detail or published, and physician awareness of troglitazone hepatotoxicity is primarily the result of mailings by the manufacturer. A recent summary of data from troglitazone clinical trials did not identify the possibility of troglitazone-induced hepatic failure [5]. Furthermore, it has been stated that troglitazone-induced changes in liver enzymes are largely mild and reversible [8]. We describe one of the patients reported to the Food and Drug Administration with liver failure apparently caused by troglitazone who ultimately required liver transplantation. Case Report A 59-year-old obese woman had had type 2 diabetes mellitus for 9 years that was complicated by retinopathy, nephropathy and neuropathy. She was prescribed troglitazone, 400 mg/d, because of poor glycemic control despite use of insulin, 150 U/d. She had a history of medullary sponge kidney, surgical removal of kidney stones, and intermittent episodes of hematuria and pyuria Her baseline serum creatinine level was 133 mol/L (1.5 mg/dL). She had no history of autoimmune diseases and had undergone hysterectomy and cholecystectomy. She was taking no other prescription or over-the-counter medications, denied using alcohol or intravenous drugs, and had not received recent blood transfusion. She reported allergy to penicillin and cephalosporins. Troglitazone improved her glycemic control and diminished her insulin requirement. She noted the onset of nausea and vomiting within 2 weeks of starting troglitazone therapy and dark urine after 2 months but did not report these symptoms to her physicians. After 3 months, she presented with nausea, vomiting, malaise, and hematuria. Although trimethoprim-sulfamethoxazole was prescribed for presumed recurrent urinary tract infection and cisapride was prescribed for possible diabetic gastroparesis, neither medication was used for more than 2 days because of nausea and vomiting. After receiving troglitazone for 3.5 months, the patient learned of potential troglitazone hepatotoxicity through the Internet and discontinued therapy with the drug. Over the following 2 weeks, she developed jaundice and was hospitalized. On presentation, she was noted to be icteric but showed no signs of hepatic encephalopathy. Laboratory tests revealed a prothrombin time of 16.5 seconds (international normalized ratio, 1.82), a bilirubin level of 148 mol/L (8.7 mg/dL), an alkaline phosphatase level of 5.7 nkat/L (345 U/L), an aspartate aminotransferase level of 13.3 kat/L (798 U/L), and an alanine aminotransferase level of 6.75 kat/L (405 U/L). Results of tests for anti-hepatitis B core IgM, hepatitis B surface antigen, anti-hepatitis A virus IgM, anti-hepatitis C virus, hepatitis B virus DNA, and hepatitis C virus RNA (the latter by polymerase chain reaction) and results of serology for acute cytomegalovirus and Epstein-Barr virus infection were negative. Her anti-smooth-muscle antibody titer was 1:20, and tests for antinuclear and antimitochondrial antibodies had negative results. Levels of serum ceruloplasmin and 1-antitrypsin and results of iron studies were normal. One year earlier, the patient had had minimal elevations of the alkaline phosphatase level (2.6 nkat/L [156.0 U/L]) and aspartate aminotransferase level (0.67 kat/L [40.0 U/L]). The leukocyte count (4600 cells/mm3) and the peripheral eosinophil count (110 cells/mm3) were normal. Ultrasonography revealed a hyperechoic liver, and computed tomography revealed a mildly irregular liver contour. On the second hospital day, the patient developed stage 1 hepatic encephalopathy and was transferred to a liver transplantation center. Transjugular liver biopsy showed severe parenchymal extinction and replacement of 80% of the tissue by loosely organized scar and collapsed parenchyma. Because her encephalopathy worsened from stage I to stage II and her prothrombin time increased to 24 seconds (international normalized ratio, 2.05), the patient underwent orthotopic liver transplantation 3 weeks after stopping troglitazone therapy. During the week before transplantation, her total bilirubin level increased to 279 mol/L (16.3 mg/dL), her serum creatinine concentration increased from 124 mol/L (1.4 mg/dL) to 256 mol/L (2.9 mg/dL), her serum albumin level decreased from 27 g/L to 23 g/L, and her partial thromboplastin time increased from 36 seconds to 49 seconds (high normal, 31 seconds). Her venous ammonia concentration decreased from 77 mol/L (134 g/dL) to 38 mol/L (64 g/dL) with lactulose treatment despite progression of encephalopathy. Other precipitants of hepatic encephalopathy, such as infection or gastrointestinal hemorrhage, were not present. The explanted liver weighed 1950 g and had an irregular nodular surface with marked capsular wrinkling. Histologic examination of the explanted liver confirmed the initial biopsy findings (Figure 1). During 4 months of posttransplantation observation, the transplanted liver functioned well, and the patient regained full cognitive function. Difficulties with persistent hematuria necessitated percutaneous and transurethral removal of renal calculi. Figure 1. Liver histologic findings. Top. Middle. Bottom. Discussion This patient developed nonspecific symptoms during troglitazone treatment that in retrospect were probably caused by drug-induced liver injury. On presentation, she had significant hepatic synthetic dysfunction and elevated aminotransferase levels even though she had stopped taking the drug 1 week earlier. Over the subsequent 2 weeks, her liver function continued to deteriorate, and she ultimately required liver transplantation. This patient had no other known exposure to hepatotoxins, and evaluation did not reveal any evidence of viral, metabolic, or autoimmune liver disease. Liver histologic findings were consistent with drug-induced liver injury. Although other causes of acute liver disease were not identified and biopsy results did not suggest chronic liver disease, the patient had risk factors for chronic liver disease. Nonalcoholic steatohepatitis causes slowly progressive fibrosis in some obese diabetic patients, but histologic examination of the liver did not show steatosis, inflammation, or sinusoidal fibrosis, characteristics that define this syndrome. Medullary sponge kidney is also associated with hepatic fibrosis in rare cases, but our patient did not have the characteristic ductal plate malformations and broad portal fibrosis. Of note, the patient became aware of possible troglitazone-induced hepatotoxicity through information available on the Internet. Patients increasingly obtain information of varying accuracy from the Internet. In this case, the patient identified relevant drug information before her physicians had received new monitoring recommendations from the manufacturer. Unfortunately, it was still too late to affect her outcome. The cause of troglitazone hepatotoxicity remains unknown. Troglitazone is primarily metabolized to sulfate and glucuronide conjugates in the liver. It also induces cytochrome P450 3A4 activity, and a small fraction undergoes cytochrome P450 oxidation to a quinone derivative. The molecular structure of the drug is similar to that of vitamin E; thus, it may be subject to oxidation-reduction reactions. It is thought to have protective properties against oxidant stress [9, 10], but its ability to undergo single-electron reduction could also confer injurious prooxidant reactivity. Examples of toxic quinones include the reactive metabolite of acetaminophen, doxorubicin, and various compounds from tobacco smoke. In this patient, the use of troglitazone seems to have been associated with the development of hepatic failure. The basis for this idiosyncratic reaction is unknown and is therefore unpredictable. Patients treated with troglitazone should have their serum aminotransferase and bilirubin levels monitored in accordance with the manufacturers recommendations: at the start of therapy, monthly during the first 6 months of treatment, every 2 months for the remainder of the first year, and periodically thereafter or if symptoms develop. Therapy with the drug should be discontinued if clinically significant abnormalities are found. From St. Louis University School of Medicine, St. Louis, Missouri; and University of Missouri-Kansas City, Kansas City, Missouri. Drs. Phillips and Brunt: Department of Pathology, St. Louis University School of Medicine, 3635 Vista Avenue, St. Louis, MO 63110. Drs. Quiason, Oki, and Isley: Uni

Frequency and Impact of SMBG on Glycemic Control in Patients With NIDDM in an Urban Teaching Hospital Clinic

Few published reports have documented the value of SMBG on glycemic control in patients with non-insulin-dependent diabetes mellitus (NIDDM), and no reports have evaluated predominantly African American patients who are at high risk for NIDDM and associated complications. In this study a 13- item survey was given to 98 patients with NIDDM to assess the frequency of self-monitoring of blood glucose (SMBG) and its impact on glycemic control. Sixty- one potients performed SMBG and 37 did not. More SMBG testers were taking insulin compared with the nontesters. GHb was comparable between groups. Among the testers there was no difference in mean GHb values based on the frequency of SMBG. Most testers performed SMBG before meals (93%) and recorded their values (85%); many had difficulty obtaining a good blood sample (30%). The most common reason for not testing was cost of supplies (77%). Performance of SMBG in these NIDDM patients was not associated with better glycemic control. Cost was a prohibitive factor for the nontesters.

Rosiglitazone and liver failure.

TO THE EDITOR: We read with interest the recent reports of hepatic dysfunction associated with the use of rosiglitazone (1, 2). According to the classification scheme of Naranjo and colleagues (3), both reports would at most suggest a possible association with the administration of rosiglitazone because of other confounding factors. The liver abnormalities in the case presented by Forman and colleagues (1) may be related to hypotension, as hypothesized by Freid and coworkers in an accompanying letter (4). The case presented by Al-Salman and associates (2) is clouded by a history of alcohol abuse, acetaminophen use, and concomitant administration of zafirlukast. Zafirlukast has been associated with hepatotoxicity and inhibits one of the metabolic pathways for the clearance of rosiglitazone, the cytochrome P4502C9 pathway. Contrary to what Al-Salman and colleagues suggest, rosiglitazone-associated injury may not be similar to the liver injury proposed by Neuschwander-Tetri and colleagues (5) because rosiglitazone metabolism does not produce a quinone, nor does it have a vitamin E moiety as a side chain. If this idiosyncratic hepatotoxicity is a class effect, it is more likely to be related to the basic thiazolidinedione structure. We believe that liver monitoring is indicated when the newer thiazolidinediones are used in therapy, but whether these agents have hepatotoxicity similar to that associated with troglitazone (5) is still open to question.

Dyslipidemias in Patients With Diabetes Mellitus: Classification and Risks and Benefits of Therapy

To characterize the lipid and lipoprotein abnormalities in patients with diabetes mellitus and evaluate the risks and benefits of marketed pharmacologic therapies, a MEDLINE search of the National Library of Medicine data base was performed of studies published from January 1966 to March 1994. Clinical trials assessing effects on lipids and lipoproteins, and adverse effects of marketed lipid‐lowering agents were extracted. Reviews and other relevant articles were included if they provided information regarding lipid and lipoprotein metabolism or guidelines on the treatment of dyslipidemias in patients with diabetes mellitus. An extensive review of clofibrate was not included. The most common dyslipidemia in patients with poorly controlled insulin‐dependent diabetes mellitus (IDDM) is combined elevated triglyceride and cholesterol levels, with reduced high‐density lipoprotein (HDL) cholesterol (mixed hyperlipidemia). Hypertriglyceridemia combined with a reduced HDL cholesterol is the most common dyslipidemia in patients with noninsulin‐dependent diabetes mellitus, but essentially any pattern of dyslipidemia may be present. Small and dense low‐density lipoprotein (LDL), glycosylation of lipoproteins, and increased oxidized lipoproteins may be present in patients with diabetes mellitus; all contribute to accelerated atherosclerotic cardiovascular disease. Insulin therapy generally corrects quantitative lipid abnormalities in patients with IDDM, so drug treatment is seldom indicated. Diet, exercise, and insulin or oral sulfonylureas will improve hypertriglyceridemia and low HDL concentrations, but do not always return them to normal. Drug therapy is indicated when nonpharmacologic measures are inadequate. It is administered based on the effects of each agent on lipids and lipoproteins, patient age, adverse effect profile, patient tolerability, and drug‐disease and drug‐drug interactions. A fibric acid derivative is the drug of choice for marked hypertriglyceridemia in patients with diabetes mellitus. Niacin can worsen glycemic control, but it may be required in severe hypertriglyceridemia, hypercholesterolemia, or mixed hyperlipidemia. Bile‐acid binding resins may accentuate hypertriglyceridemia but may be useful in selected patients with marked hypercholesterolemia and normal triglycerides. Hydroxymethylglutaryl coenzyme A reductase inhibitors are preferred in patients with elevated LDL cholesterol and mild hypertriglyceridemia. Patients with marked lipid abnormalities or mixed hyperlipidemias may require carefully dosed combinations of lipid‐lowering drugs.

Hepatotoxicity of thiazolidinediones

Thiazolidinediones are insulin sensitisers now widely used for the treatment of Type 2 diabetes mellitus. The initial marketed drug in this class, troglitazone, was removed from the market worldwide after approximately 3 years of use due to rare but severe hepatotoxicity, which sometimes resulted in liver failure leading to the need for liver transplantation, or even death. The unpredictability of such liver toxicity made the use of troglitazone highly problematic. Fortunately, the two newer drugs in this class, rosiglitazone and pioglitazone, have a much larger margin of safety for liver toxicity. Very rare reports of liver toxicity, usually milder and reversible, have been seen with these drugs. Therefore, whilst pharmacovigilance for hepatotoxicity is probably still warranted, the practitioner and patient can be fairly confident that these drugs are safe from a liver standpoint. Finally, recent work would suggest that these agents may prove useful to reduce hepatic fat in patients with non-alcoholic steatohepatitis, and may possibly protect against adverse metabolic consequences and the ultimate development of cirrhosis in patients with fatty livers.

Can Response to Bedtime Insulin with Daytime Sulfonylurea Be Predicted in Patients with Type 2 Diabetes Mellitus

Objective: To determine characteristics predictive of response in patients with Type 2 diabetes mellitus (Type 2 DM) who demonstrate good or poor blood glucose control while receiving bedtime insulin with daytime sulfonylurea (BIDS) therapy. Methods: A retrospective chart review of patients with Type 2 DM receiving BIDS therapy was performed. The criterion for responders was the mean of two consecutively obtained glycosylated hemoglobin (Hb) concentrations being less than or equal to 10.2% (HbA1c ≤7.0%). Setting: A university-affiliated diabetes specialty clinic staffed consistently by a pharmacist diabetes educator, four endocrinologists, and a pharmacotherapy specialist. Patients: Thirty-one patients with Type 2 DM who were predominantly African-American and women who had documented regular follow-up examinations for more than 12 months while receiving BIDS therapy. Data Collection and Measurements: Gender, ethnicity, height, weight, calculated body mass index, age of onset of diabetes mellitus, duration of diabetes mellitus before BIDS therapy, pre-BIDS treatment regimen, dosages of NPH insulin and glyburide, glycosylated Hb concentration, fasting blood glucose concentration, and duration of BIDS therapy were recorded. Results: There were no differences in age of onset of diabetes mellitus, duration of diabetes mellitus before BIDS was initiated, duration of therapy with BIDS, or baseline glycosylated Hb concentration between responders (n = 15) and nonresponders (n = 16). Patients in the responder group weighed less, had a lower body mass index, required smaller dosages of both glyburide and NPH insulin, and achieved a lower fasting blood glucose concentration. Conclusions: In this study population, with the exception of body mass index, there was no difference in suggested clinical characteristics of response between responders and nonresponders.