- Email: [email protected]
Incidence of idiopathic acute liver failure and hospitalized liver injury in patients treated with troglitazone
THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 2003 by Am. Coll. of Gastroenterology Published by Elsevier Science Inc.
Vol. 98, No. 1, 2003 ISSN 0002-9270/03/$30.00 PII S0002-9270(02)05843-4
Incidence of Idiopathic Acute Liver Failure and Hospitalized Liver Injury in Patients Treated With Troglitazone David J. Graham, M.D., M.P.H., Carol R. Drinkard, Ph.D., and Deborah Shatin, M.A., Ph.D. Office of Drug Safety, Center for Drug Evaluation and Research, Food and Drug Administration, Rockville, Maryland; and Center for Health Care Policy and Research, UnitedHealth Group, Minnetonka, Minnesota
OBJECTIVE: Troglitazone, a thiazolidinedione antidiabetic agent, was withdrawn from the U.S. market in March, 2000, after 94 cases of acute liver failure (ALF) were reported with its use. Based on a literature review, the estimated background rate of hospitalization for idiopathic acute liver injury is 22 per million person-years and for idiopathic ALF, less than 1 per million person-years. This study was conducted to estimate the incidence rates of hospitalized idiopathic acute liver injury and ALF among troglitazonetreated patients. METHODS: An observational retrospective inception cohort of patients treated with troglitazone was assembled using claims data from a large multistate health care organization. Patients with at least 90 days of health plan enrollment before their first troglitazone prescription between April, 1997 and December, 1998 were enrolled. Hospitalized cases of potential troglitazone-induced acute liver injury or ALF were identified from claims data based on International Classification of Diseases, 9th Revision, coding. Primary medical records were reviewed for case validation, and incidence rates of acute liver injury were calculated using person-years of troglitazone exposure as the denominator. RESULTS: A total of 7568 patients contributed 4020 personyears of troglitazone exposure. Of these, five were hospitalized with acute liver injury attributed to the drug and not explained by other causes. Incidence rates (95% CI) per million person-years of acute idiopathic liver injury were as follows: hospitalization (n ⫽ 5), 1244 (404, 2900); hospitalized jaundice (n ⫽ 4), 995 (271, 2546); and ALF (n ⫽ 1), 240 (6.3, 1385). CONCLUSION: Troglitazone use was associated with a marked increase in risk of hospitalized acute idiopathic liver injury and ALF. (Am J Gastroenterol 2003;98:175–179. © 2003 by Am. Coll. of Gastroenterology)
INTRODUCTION The marketing of troglitazone, the first thiazolidinedione approved in the United States for treatment of type 2 diabetes, began in March, 1997. Its mechanism of action dif-
fered from that of other oral antidiabetic agents by increasing the response of muscle and adipose tissue to circulating insulin (1). Serum liver transaminase levels greater than three times the upper limit of normal were noted in 1.9% of troglitazone-treated patients during phase 3 clinical trials, with two patients developing jaundice (2). Recently, a third case of hospitalized jaundice from the original trials was uncovered by an investigative reporter (3). Soon after marketing began, cases of acute liver failure (ALF) in patients treated with troglitazone were reported to the Food and Drug Administration (FDA). Four “Dear Health Care Professional” letters were sent to U.S. physicians, warning of the risk of ALF and recommending periodic liver enzyme monitoring as a means of prevention (4). Reporting of cases to the FDA and in the literature continued (5–10). In March, 1999, a FDA advisory committee was convened to review data on 43 reported U.S. cases of ALF and to decide whether troglitazone should be withdrawn from the market (11, 12). The drug remained on the market for 2 more yr, by which time, 94 cases of ALF had been reported (13). This study was performed to estimate the incidence rates of idiopathic ALF and hospitalized acute liver injury in an inception cohort of troglitazone users.
MATERIALS AND METHODS UnitedHealth Group is a national health care company including 43 health plans across the United States. The company maintains a research database covering about 3 million persons from 12 health plans located in 10 different states. These plans follow an independent practice association model consisting of large physician networks representative of the medical communities in which they practice. Administrative claims data consist of separate files for health plan enrollment, outpatient prescription fills, physician services, and facility use. These can be linked longitudinally using unique encrypted identifiers, protecting patient confidentiality. Health plan members with at least one troglitazone prescription between April 1, 1997, and December 31, 1998,
176
Graham et al.
AJG – Vol. 98, No. 1, 2003
Table 1. Summary of Five Patients Hospitalized With Troglitazone-Induced Acute Liver Injury Duration of Other Age Troglitazone Diabetes (yr) Sex Use (mo) Therapy
Alanine Aminotransferase (IU/L)
Aspartate Aminotransferase (IU/L)
55*† F
1
167 (ⱕ40)
171 (5–40) 10.4 (ⱕ1.5) 315 (4–120)
57
M
9
67* 77‡ 85*
F M M
5 0.5 5
Insulin
Total Bilirubin (mg/dl)
Alkaline Phosphatase (IU/L)
Medical Procedures
Outcome
ERCP: normal Discharged Liver ultrasound: normal with jaundice Insulin 978 (10–60) 1266 (10–43) 12.5 (ⱕ1.0) 217 (42–121) CT of abdomen: no liver Recovered Metformin abnormality noted Insulin 62 (4–40) 97 (15–45) 3.1 (⬍1.5) 63 (37–117) CT of abdomen: normal Recovered Insulin 97 338 (5–45) Recovered Insulin 608 (10–45) 416 (10–45) 15.6 (ⱕ1.0) 144 (50–136) ERCP: normal Died Liver ultrasound: normal
Normal ranges are listed in parentheses beneath the laboratory values shown in the table. * Patients had documented normal serum transaminase levels before onset of liver injury. † Patient enrollment ended shortly after hospital discharge; eventual outcome unknown. ‡ Patient had “normal laboratory evaluation” before onset of liver injury.
were identified (n ⫽ 9369). Those with at least 90 days of continuous enrollment before their first (index) troglitazone prescription during the study period were selected for inclusion in a retrospective inception cohort. The prior enrollment criterion was required to ensure that the index prescription was the patient’s first prescription. Person-years of exposure in the cohort were calculated as the cumulative days’ supply of all troglitazone prescriptions filled during the study period. Automated claims files of cohort members were searched using International Classification of Diseases, 9th Revision, and current procedural terminology codes to identify hospitalizations indicating a primary or secondary discharge diagnosis of liver disease or procedures suggesting possible liver disease, such as liver biopsy or transplantation, occurring after the date of their index troglitazone prescription. Claims profiles for these hospitalized patients were reviewed for evidence of conditions that might explain the liver disorder, such as viral hepatitis, metastatic cancer, or chronic liver disease. Hospital medical records were reviewed for patients whose automated claims did not suggest another cause of liver disease or for whom the claims data were inconclusive. For a diagnosis of ALF, hepatic encephalopathy, liver transplantation, or death in the setting of acute severe liver injury was required (14, 15). Incidence rates and 95% CIs for ALF and hospitalized liver injury were calculated as Poisson statistics using Stata, version 5 (Stata, College Station, TX). In these calculations, the number of troglitazone-treated patients with documented hospitalization for idiopathic liver injury, jaundice, and ALF served as numerators and the total person-years of troglitazone exposure in the cohort served as the denominator.
RESULTS The inception cohort included 7568 patients who began troglitazone during the study period. Of these, 49.7% were women, 82.6% were age 45 yr or older, and their median duration of troglitazone use was 140 days. A total of 4020 person-years of exposure were observed.
Nineteen patients were identified with a liver-related hospitalization after their index prescription. Based on review of automated claims, 10 were excluded as having liver disease unrelated to troglitazone (primary or metastatic cancer—five, chronic hepatitis C—two, liver disease not otherwise specified preceding troglitazone use— one, liver disease not otherwise specified with continuation of troglitazone— one, drug-induced hepatitis 1 day after the first troglitazone prescription— one). Medical records were sought for nine patients. Of these, one represented a claims error and one other could not be obtained because it was a psychiatric admission. Hemachromatosis with cirrhosis was documented in one, and one other had hepatitis C. For the remaining five patients, acute liver injury was documented in the hospital records for which other explanations were not apparent (Table 1). In four of these, hospitalization was the direct result of troglitazone-induced hepatotoxicity, whereas in one, elevated serum transaminase and creatine kinase levels were noted after admission for a urinary tract infection and were diagnosed as troglitazone related. The ages of these patients ranged from 55 to 85 yr, two were women, and the length of troglitazone therapy before onset of liver injury ranged from 2 wk to 9 months. None had known prior liver disease and in three, normal serum transaminase levels were documented either before initiation, or at some time during troglitazone therapy, but before the onset of liver injury. One other patient was reported to have had a “normal laboratory evaluation” within the preceding 1–2 yr. Four had jaundice of which two were diagnosed as cholestatic and two as hepatocellular in nature. One of the latter developed hepatic encephalopathy and died. A liver biopsy from this patient showed submassive hepatocellular necrosis. This case was reported to the FDA. Of the remaining patients, three recovered and one was lost to follow-up because her health plan enrollment ended soon after discharge with a diagnosis of “drug-induced jaundice.” The incidence rates per million person-years of acute idiopathic liver injury (95% CI) were as follows: hospitalization
AJG – January, 2003
(n ⫽ 5), 1244 (404, 2900), hospitalized jaundice (n ⫽ 4), 995 (271, 2546), and ALF (n ⫽ 1), 240 (6.3, 1385).
DISCUSSION In this cohort of 7568 new troglitazone users, five cases of acute hospitalized liver injury, including one case of fatal ALF, were identified, for which an explanation other than troglitazone was not apparent after medical record review. The point estimates for the rates of hospitalization and ALF were 1244 and 240 per million person-years of troglitazone exposure, respectively. These rates are markedly increased above the estimated background rates for hospitalized acute idiopathic liver injury or ALF. Four epidemiological studies have estimated the background rate of hospitalized idiopathic hepatitis to range from 10 to 22 per million person-years (16 –19). The rate of hospitalization for idiopathic liver injury observed with troglitazone in this study was 55- to 124-fold greater than expected based on these studies. The background rate for idiopathic ALF in the general population has not been previously published. Two different approaches yielded estimates of less than one case per million person-years. Each year, about 2000 cases of ALF from all causes occur in the United States, with 10% from unexplained causes (14, 20). Using a U.S. census estimate of 281 million (21), an estimated rate of unexplained ALF of 0.7 per million person-years was obtained for the U.S. population. Using a second approach, four population-based epidemiological studies of liver disease were pooled for a total of 12 million person-years of observation (17–19, 22). Four cases of unexplained ALF were noted in these studies, yielding a rate estimate of 0.3 per million person-years (12, 13). By these metrics, the incidence of idiopathic ALF among troglitazone-exposed patients in this study was increased more than 240-fold compared with that expected in the general population. These estimates of the background rate of idiopathic ALF were based on general populations, not a population limited to type 2 diabetics. A recent epidemiological study from the United Kingdom followed type 2 diabetics treated with oral antidiabetic agents for a total of 114,000 person-years (23). There were no confirmed cases of ALF. However, one elderly patient died in the setting of jaundice. The medical record of this patient was not reviewed, and the causes of jaundice and death were not established. If it had been shown that other causes were absent, this would give an estimate of 8.8 per million person-years, still 27-fold below that observed in this cohort. Three unpublished studies, two of them randomized controlled clinical trials, suggest that the rate of ALF among troglitazone users was extremely elevated compared with that expected among type 2 diabetics (12, 13). In March, 1999, it was reported that one case of fatal ALF occurred in a company-sponsored postmarketing study of troglitazone, yielding an incidence rate of 1274 per million person-years
Idiopathic Acute Liver Failure and Troglitazone
177
(12, 13). In a National Institutes of Health-sponsored randomized clinical trial, the Diabetes Prevention Program, one of 585 patients in the troglitazone arm died after liver transplantation for ALF (580 person-years, rate 1724 per million person-years) (12, 13). Finally, in a randomized trial of 116 diabetic patients, contributing at most 58 personyears of observation, one patient treated with troglitazone died of ALF (rate ⬃17,000 per million person-years) (13). Several other lines of evidence suggest that the incidence rate of ALF with troglitazone observed in this study was not caused by type 2 diabetes itself. A review of the FDA’s adverse drug reaction database found that the reporting of ALF with troglitazone was 485-fold greater than with glipizide, a commonly used oral antidiabetic agent (13). If the high reporting with troglitazone was actually a result of the underlying diabetes being treated, one would expect roughly equivalent reporting for other antidiabetic drugs. In another study, data from the United Network for Organ Sharing, the organization responsible for coordinating solid-organ transplants in the United States, was reviewed for the years 1994 to 1998. During this period, three patients with ALF were registered for liver transplantation where troglitazone use was cited as the cause of liver injury. Based on the relative extent of troglitazone use compared with that of sulfonylurea agents and metformin, one would have expected 75 patients to be registered for transplantation with the United Network for Organ Sharing in which these latter drugs were listed as causative. However, during this interval, no patients were registered listing either of these drugs (2 [1 df] ⫽ 75, p ⬍ 0.001) (12, 13). An additional analysis of the United Network for Organ Sharing data for the years 1995 to 1997 found that the prevalence of diabetes among liver transplant registrants with acute hepatic necrosis was 4.8%, a level consistent with that of the general population (12, 13, 24). Finally, in the original phase 3 clinical trials, neither jaundice nor hospitalization for liver injury was observed in the diabetic control groups treated with drugs other than troglitazone (2). Together, these data suggest that the background incidence of unexplained ALF in type 2 diabetics is similar to that of the general population, and that the high rate observed with troglitazone in this study was not caused by type 2 diabetes. To help place these observations with troglitazone in context, a small number of epidemiological studies have been performed to estimate the rate of hospitalization for unexplained acute liver injury in association with the use of a variety of other drugs. For nonsteroidal anti-inflammatory drugs, liver injury hospitalization rates were estimated to range between 30 and 90 per million person-years in three separate studies (18, 19, 22). In another study, the rate of hospitalization associated with cimetidine use was 99 per million person-years (25). Hospitalization for acute liver injury has also been studied for several commonly used antibiotics. The rate of hospitalization with amoxicillin use was estimated at 17 and 25 per million exposed patients (26, 27). In these same studies, the acute liver injury hospital-
178
Graham et al.
ization rates for amoxicillin-clavulanic acid were 118 and 56 per million patients (26, 27). Erythromycin, an antibiotic with recognized hepatotoxicity was associated with hospitalized acute liver injury in two to 25 patients per million treated (17, 28, 29). Finally, one population-based study of 1052 ketoconzaole users observed no hospitalizations for acute liver injury, though 1.9% of patients were noted to develop liver-related problems as outpatients (30). In the present study, the rate of hospitalization with idiopathic acute liver injury based on patient counts was 661 per million patients treated. In the original clinical trials with troglitazone, the corresponding rate was 797 (2). Although idiopathic acute liver injury leading to hospitalization has been reported in epidemiological studies with a number of drugs, the risk of death from such injury has been low. Among the studies reviewed above, fatal liver injury was observed in only one, involving nonsteroidal anti-inflammatory drug use, yielding an incidence rate of 11 per million person-years (18). However, in another study of 625,307 nonsteroidal anti-inflammatory drug users, there were no fatalities, yielding a 95% CI for the incidence rate of fatal ALF of 0 to 10 per million person-years (22). There were no liver fatalities among 53,436 and 93,433 patients treated with amoxicillin-clavulanic acid (26, 27), and none among 392,201 patients treated with erythromycin (28). An important issue to consider is the potential role of nonalcoholic steatohepatitis (NASH) in acute liver injury and ALF with troglitazone. Obesity, frequently present in type 2 diabetics, and type 2 diabetes itself, are each associated with fatty liver and NASH (31–33). Insulin resistance is believed to be the underlying initiator of these processes (31, 32, 34, 35). Fatty liver has been reported to be present in 25–75% of diabetic patients, depending on their degree of accompanying obesity (32), and NASH was found in 12.2% and 18.5% of diabetics in two recent studies, 2.6- and 2.7-fold greater than in nondiabetic controls (36, 37). The question has been raised whether nonalcoholic fatty liver disease, and NASH in particular, might increase the susceptibility of affected patients to hepatocellular injury caused by alcohol or other xenobiotics (38, 39). Insulin resistance and associated hyperinsulinemia promote a shift in mitochondrial function from oxidation of glucose to -oxidation of free fatty acids, resulting in intracellular oxidative stress with its potential for hepatocellular injury (38 – 40). In such a setting, the generation of reactive intermediates during metabolism of certain drugs by the liver could cause hepatocellular injury or death. Hepatocellular death may result from direct cellular damage or from the activation of apoptosis (41). The same process could also occur in normal livers though the threshold for injury might be greater. Of note, a recent study in human hepatoma cells showed that troglitazone induced hepatocellular apoptosis, whereas two other thiazolidinediones did not under conditions of that experiment (42). Perhaps activation of intracellular apoptosis by reactive metabolites is the basis of ALF with troglitazone.
AJG – Vol. 98, No. 1, 2003
In the present study, it seems unlikely that NASH was responsible for or a contributing factor toward the liver injury observed in the cohort. First, four of five patients identified (Table 1) had documented normal serum transaminase levels at some point before the onset of hospitalized liver injury. Serum transaminase levels are typically elevated 2- to 4-fold above normal and remain persistently so in NASH (33, 39, 43, 44). Also, jaundice caused by NASH occurs only as an end-stage event once cirrhosis is well established and advanced (33, 43, 44). Finally, four of five patients had normal liver imaging (ultrasound—three, CT scan— one). Although the sensitivity of ultrasound and CT imaging in the detection of fatty liver is less than perfect, the absence of hepatic abnormalities with these modalities provides some evidence against the likelihood of significant fatty liver disease. This study relied upon automated claims data to identify troglitazone-exposed patients and cases of potential liver injury. Incomplete prescription claims could have led to underestimation of cohort size. This seems unlikely to have occurred because pharmacy reimbursement depends upon filing of prescription claims, providing an incentive for claims submission. It is also possible that some cases of hospitalized liver injury were missed, either because of miscoding or misdiagnosis by treating physicians, or because some patients for whom medical records were not sought actually had drug-induced disease. In retrospect, medical records probably should have been sought for two additional patients, the patient with a diagnosis of “liver disease not otherwise specified” who remained on troglitazone and the patient with a diagnosis of “drug-induced hepatitis” the day after his first troglitazone prescription. If either or both of these represented troglitazone-associated cases, the incidence rate for hospitalized liver injury would be higher than reported here. Likewise, if the patient who was lost to follow-up because her health plan enrollment ended actually progressed to liver failure, the rate of ALF reported here would be increased. It is also possible that some cases may have switched health plan coverage before developing liver injury, thereby escaping detection. We believe the likelihood of this to be low but cannot exclude the possibility. The results of this study suggest that troglitazone is a potent hepatotoxin, conferring a substantially increased risk of acute liver injury including ALF. The mechanism underlying troglitazone-induced ALF is unknown, but may involve metabolic generation of reactive intermediates that lead to either direct hepatocellular death or the activation of hepatocellular apoptosis. Reprint requests and correspondence: David J. Graham, M.D., M.P.H., 5600 Fishers Lane, HFD-400, Office of Drug Safety, Center for Drug Evaluation and Research, Food and Drug Administration, Rockville, MD 20857. Received Feb. 22, 2002; accepted June 26, 2002.
AJG – January, 2003
REFERENCES 1. Imura H. A novel antidiabetic drug, troglitazone—reason for hope and concern. N Engl J Med 1998;338:908 –9 (editorial). 2. Rezulin. In: Physicians’ desk reference. Montvale, NJ: Medical Economics Company, 1998:2218 –21. 3. Willman D. Fears grow over delay in removing Rezulin. Company’s conduct in development of the drug also comes under fire. Los Angeles Times 2000;10 Mar:A-18. 4. Graham DJ, Drinkard CR, Shatin D, et al. Liver enzyme monitoring in patients treated with troglitazone. JAMA 2001; 286:831–3. 5. Gitlin N, Julie NL, Spurr CL, et al. Two cases of severe clinical and histologic hepatoxicity associated with troglitazone. Ann Intern Med 1998;129:36 –8. 6. Neuschwander-Tetri BA, Isley WL, Oki JC, et al. Troglitazone-induced hepatic failure leading to liver transplantation. Ann Intern Med 1998;129:38 –41. 7. Shibuya A, Watanabe M, Fujita Y, et al. An autopsy case of troglitazone-induced fulminant hepatitis. Diabetes Care 1998; 21:2140 –3. 8. Herrine SK, Choudhary C. Severe hepatotoxicity associated with troglitazone. Ann Intern Med 1999;130:163–4 (letter). 9. Murphy EJ, Davern TJ, Shakil AO, et al. Troglitazone-induced fulminant hepatic failure. Dig Dis Sci 2000;45:549 –53. 10. Jagannath S, Rai R. Rapid-onset subfulminant liver failure associated with troglitazone. Ann Intern Med 2000;132:677 (letter). 11. Stolberg SG. Stricter rules are urged on use of a diabetic drug. The New York Times 1999;27 Mar:A-9. 12. Graham DJ, Green L. Epidemiology of hepatotoxicity with troglitazone. FDA Metabolic-Endocrine Drugs Advisory Committee Meeting, Bethesda, MD, March 26, 1999. Available from U.S. Food and Drug Administration via the Internet at: www.fda.gov/ohrms/dockets/ac/99/transcpt/3499t1a.pdf;/ 3499t1b.pdf;/3499t1c.pdf;/34991d.pdf;/34991e.pdf. Accessed January 8, 2002. 13. Graham DJ, Green L. Final report: Liver failure risk with troglitazone. Office of Postmarketing Drug Risk Assessment, Center for Drug Evaluation and Research, Food and Drug Administration, Rockville, MD, December 19, 2000. 14. Lee WH. Acute liver failure. N Engl J Med 1993;329:1862– 72. 15. Williams R. Classification, etiology, and consideration of outcome in acute liver failure. Semin Liver Dis 1996;16:343–8. 16. Almdal TP, Sorensen TIA. Incidence of parenchymal liver disease in Denmark, 1981 to 1985: Analysis of hospital registry data. Hepatology 1991;13:650 –5. 17. Beard K, Belic L, Aselton P, et al. Outpatient drug-induced parenchymal liver disease requiring hospitalization. J Clin Pharmacol 1986;26:633–7. 18. Garcia Rodriguez LA, Gutthann SP, Walker AM, et al. The role of nonsteroidal anti-inflammatory drugs in acute liver injury. Br Med J 1992;305:865–8. 19. Carson JL, Strom BL, Duff A, et al. Safety of nonsteroidal anti-inflammatory drugs with respect to acute liver disease. Arch Intern Med 1993;153:1331–6. 20. Schiodt FV, Atillasoy E, Shakil AO, et al. Etiology and outcome for 295 patients with acute liver failure in the United States. Liver Transplant Surg 1999;5:29 –34. 21. US Census Bureau. Available via the Internet at: http://www.census.gov/population/cen2000/tab02.pdf. Accessed January 28, 2002. 22. Garcia Rodriguez LA, Williams R, Derby LE, et al. Acute liver injury associated with nonsteroidal anti-inflammatory drugs and the role of risk factors. Arch Intern Med 1994;154: 311–5.
Idiopathic Acute Liver Failure and Troglitazone
179
23. Jick SS, Stender M, Myers MW. Frequency of liver disease in type 2 diabetic patients treated with oral antidiabetic agents. Diabetes Care 1999;22:2067–71. 24. Everhart JE. Digestive diseases and diabetes. In: Diabetes in America, 2nd ed. National Diabetes Data Group, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD. NIH Publication No. 95–1468, 1995:457– 84. 25. Garcia Rodriguez LA, Wallander MA, Stricker BH. The risk of acute liver injury associated with cimetidine and other acid-suppressing anti-ulcer drugs. Br J Clin Pharmacol 1997; 43:183–8. 26. Garcia Rodriguez LA, Stricker BH, Zimmerman HJ. Risk of acute liver injury associated with the combination of amoxicillin and clavulanic acid. Arch Intern Med 1996;156:1327– 32. 27. Maggini M, Raschetti R, Agostinis L, et al. Use of amoxicillin and amoxicillin-clavulanic acid and hospitalization for acute liver injury. Ann Ist Super Sanita 1999;35:429 –33. 28. Derby LE, Jick H, Henry DA, et al. Erythromycin-associated cholestatic hepatitis. Med J Aust 1993;158:600 –2. 29. Carson JL, Strom BL, Duff A, et al. Acute liver disease associated with erythromycin, sulfonamides and tetracyclines. Ann Intern Med 1993;119:576 –83. 30. Garcia Rodriguez LA, Duque A, Castellsagne J, et al. A cohort study of the risk of acute liver injury among users of ketoconazole and other anti-fungal drugs. Br J Clin Pharmacol 1999;48:1847–52. 31. Silverman JF, Pories WJ, Caro JF. Liver pathology in diabetes mellitus and morbid obesity: Clinical, pathological, and biochemical considerations. Pathol Annu 1989;24:275–302. 32. McCullough AJ. Update on nonalcoholic fatty liver disease. J Clin Gastroenterol 2002;34:255–62. 33. Youssef W, McCullough AJ. Diabetes mellitus, obesity, and hepatic steatosis. Semin Gastrointest Dis 2002;13:17–30. 34. Chitturi S, Farrell GC. Etiopathogenesis of nonalcoholic steatohepatitis. Semin Liver Dis 2001;21:27–41. 35. Pagano G, Pacini G, Musso G, et al. Nonalcoholic steatohepatitis, insulin resistance, and metabolic syndrome: Further evidence for an etiologic association. Hepatology 2002;35:367– 72. 36. Wanless IR, Leutz JS. Fatty liver hepatitis (steatohepatitis) and obesity: An autopsy study with analysis of risk factors. Hepatology 1990;12:1106 –10. 37. Bernal-Reyes R, Saenz-Lubra A, Bernardo-Escudero R. Prevalencia de la esteatohepatitis no alcoholica (EHNA). Estudio comparativo con diabeticos. [Prevalence of non-alcoholic steatohepatitis. Comparative study with diabetic patients]. Rev Gastroenterol Mex 2000;65:58 –62. 38. Angulo P, Lindor KD. Insulin resistance and mitochondrial abnormalities in NASH: A cool look into a burning issue. Gastroenterology 2001;120:1281–5 (editorial). 39. Clark JM, Brancatti FL, Diehl AM. Nonalcoholic fatty liver disease. Gastroenterology 2002;122:1649 –57. 40. Sanyal AJ, Campbell-Sargent C, Mirshahi F, et al. Nonalcoholic steatohepatitis: Association of insulin resistance and mitochondrial abnormalities. Gastroenterology 2001;120: 1183–92. 41. Jaeschke H, Gores GJ, Cederbaum AI, et al. Mechanisms of hepatotoxicity. Toxicol Sci 2002;65:166 –76. 42. Yamamoto Y, Nakajima M, Yamzaki H, et al. Cytotoxicity and apoptosis produced by troglitazone in human hepatoma cells. Life Sci 2001;70:471–82. 43. Reid AE. Nonalcoholic steatohepatitis. Gastroenterology 2001;121:710 –23. 44. Contos MJ, Sanyal AJ. The clinicopathologic spectrum and management of nonalcoholic fatty liver disease. Adv Anat Pathol 2002;9:37–51.
Vol. 98, No. 1, 2003 ISSN 0002-9270/03/$30.00 PII S0002-9270(02)05843-4
Incidence of Idiopathic Acute Liver Failure and Hospitalized Liver Injury in Patients Treated With Troglitazone David J. Graham, M.D., M.P.H., Carol R. Drinkard, Ph.D., and Deborah Shatin, M.A., Ph.D. Office of Drug Safety, Center for Drug Evaluation and Research, Food and Drug Administration, Rockville, Maryland; and Center for Health Care Policy and Research, UnitedHealth Group, Minnetonka, Minnesota
OBJECTIVE: Troglitazone, a thiazolidinedione antidiabetic agent, was withdrawn from the U.S. market in March, 2000, after 94 cases of acute liver failure (ALF) were reported with its use. Based on a literature review, the estimated background rate of hospitalization for idiopathic acute liver injury is 22 per million person-years and for idiopathic ALF, less than 1 per million person-years. This study was conducted to estimate the incidence rates of hospitalized idiopathic acute liver injury and ALF among troglitazonetreated patients. METHODS: An observational retrospective inception cohort of patients treated with troglitazone was assembled using claims data from a large multistate health care organization. Patients with at least 90 days of health plan enrollment before their first troglitazone prescription between April, 1997 and December, 1998 were enrolled. Hospitalized cases of potential troglitazone-induced acute liver injury or ALF were identified from claims data based on International Classification of Diseases, 9th Revision, coding. Primary medical records were reviewed for case validation, and incidence rates of acute liver injury were calculated using person-years of troglitazone exposure as the denominator. RESULTS: A total of 7568 patients contributed 4020 personyears of troglitazone exposure. Of these, five were hospitalized with acute liver injury attributed to the drug and not explained by other causes. Incidence rates (95% CI) per million person-years of acute idiopathic liver injury were as follows: hospitalization (n ⫽ 5), 1244 (404, 2900); hospitalized jaundice (n ⫽ 4), 995 (271, 2546); and ALF (n ⫽ 1), 240 (6.3, 1385). CONCLUSION: Troglitazone use was associated with a marked increase in risk of hospitalized acute idiopathic liver injury and ALF. (Am J Gastroenterol 2003;98:175–179. © 2003 by Am. Coll. of Gastroenterology)
INTRODUCTION The marketing of troglitazone, the first thiazolidinedione approved in the United States for treatment of type 2 diabetes, began in March, 1997. Its mechanism of action dif-
fered from that of other oral antidiabetic agents by increasing the response of muscle and adipose tissue to circulating insulin (1). Serum liver transaminase levels greater than three times the upper limit of normal were noted in 1.9% of troglitazone-treated patients during phase 3 clinical trials, with two patients developing jaundice (2). Recently, a third case of hospitalized jaundice from the original trials was uncovered by an investigative reporter (3). Soon after marketing began, cases of acute liver failure (ALF) in patients treated with troglitazone were reported to the Food and Drug Administration (FDA). Four “Dear Health Care Professional” letters were sent to U.S. physicians, warning of the risk of ALF and recommending periodic liver enzyme monitoring as a means of prevention (4). Reporting of cases to the FDA and in the literature continued (5–10). In March, 1999, a FDA advisory committee was convened to review data on 43 reported U.S. cases of ALF and to decide whether troglitazone should be withdrawn from the market (11, 12). The drug remained on the market for 2 more yr, by which time, 94 cases of ALF had been reported (13). This study was performed to estimate the incidence rates of idiopathic ALF and hospitalized acute liver injury in an inception cohort of troglitazone users.
MATERIALS AND METHODS UnitedHealth Group is a national health care company including 43 health plans across the United States. The company maintains a research database covering about 3 million persons from 12 health plans located in 10 different states. These plans follow an independent practice association model consisting of large physician networks representative of the medical communities in which they practice. Administrative claims data consist of separate files for health plan enrollment, outpatient prescription fills, physician services, and facility use. These can be linked longitudinally using unique encrypted identifiers, protecting patient confidentiality. Health plan members with at least one troglitazone prescription between April 1, 1997, and December 31, 1998,
176
Graham et al.
AJG – Vol. 98, No. 1, 2003
Table 1. Summary of Five Patients Hospitalized With Troglitazone-Induced Acute Liver Injury Duration of Other Age Troglitazone Diabetes (yr) Sex Use (mo) Therapy
Alanine Aminotransferase (IU/L)
Aspartate Aminotransferase (IU/L)
55*† F
1
167 (ⱕ40)
171 (5–40) 10.4 (ⱕ1.5) 315 (4–120)
57
M
9
67* 77‡ 85*
F M M
5 0.5 5
Insulin
Total Bilirubin (mg/dl)
Alkaline Phosphatase (IU/L)
Medical Procedures
Outcome
ERCP: normal Discharged Liver ultrasound: normal with jaundice Insulin 978 (10–60) 1266 (10–43) 12.5 (ⱕ1.0) 217 (42–121) CT of abdomen: no liver Recovered Metformin abnormality noted Insulin 62 (4–40) 97 (15–45) 3.1 (⬍1.5) 63 (37–117) CT of abdomen: normal Recovered Insulin 97 338 (5–45) Recovered Insulin 608 (10–45) 416 (10–45) 15.6 (ⱕ1.0) 144 (50–136) ERCP: normal Died Liver ultrasound: normal
Normal ranges are listed in parentheses beneath the laboratory values shown in the table. * Patients had documented normal serum transaminase levels before onset of liver injury. † Patient enrollment ended shortly after hospital discharge; eventual outcome unknown. ‡ Patient had “normal laboratory evaluation” before onset of liver injury.
were identified (n ⫽ 9369). Those with at least 90 days of continuous enrollment before their first (index) troglitazone prescription during the study period were selected for inclusion in a retrospective inception cohort. The prior enrollment criterion was required to ensure that the index prescription was the patient’s first prescription. Person-years of exposure in the cohort were calculated as the cumulative days’ supply of all troglitazone prescriptions filled during the study period. Automated claims files of cohort members were searched using International Classification of Diseases, 9th Revision, and current procedural terminology codes to identify hospitalizations indicating a primary or secondary discharge diagnosis of liver disease or procedures suggesting possible liver disease, such as liver biopsy or transplantation, occurring after the date of their index troglitazone prescription. Claims profiles for these hospitalized patients were reviewed for evidence of conditions that might explain the liver disorder, such as viral hepatitis, metastatic cancer, or chronic liver disease. Hospital medical records were reviewed for patients whose automated claims did not suggest another cause of liver disease or for whom the claims data were inconclusive. For a diagnosis of ALF, hepatic encephalopathy, liver transplantation, or death in the setting of acute severe liver injury was required (14, 15). Incidence rates and 95% CIs for ALF and hospitalized liver injury were calculated as Poisson statistics using Stata, version 5 (Stata, College Station, TX). In these calculations, the number of troglitazone-treated patients with documented hospitalization for idiopathic liver injury, jaundice, and ALF served as numerators and the total person-years of troglitazone exposure in the cohort served as the denominator.
RESULTS The inception cohort included 7568 patients who began troglitazone during the study period. Of these, 49.7% were women, 82.6% were age 45 yr or older, and their median duration of troglitazone use was 140 days. A total of 4020 person-years of exposure were observed.
Nineteen patients were identified with a liver-related hospitalization after their index prescription. Based on review of automated claims, 10 were excluded as having liver disease unrelated to troglitazone (primary or metastatic cancer—five, chronic hepatitis C—two, liver disease not otherwise specified preceding troglitazone use— one, liver disease not otherwise specified with continuation of troglitazone— one, drug-induced hepatitis 1 day after the first troglitazone prescription— one). Medical records were sought for nine patients. Of these, one represented a claims error and one other could not be obtained because it was a psychiatric admission. Hemachromatosis with cirrhosis was documented in one, and one other had hepatitis C. For the remaining five patients, acute liver injury was documented in the hospital records for which other explanations were not apparent (Table 1). In four of these, hospitalization was the direct result of troglitazone-induced hepatotoxicity, whereas in one, elevated serum transaminase and creatine kinase levels were noted after admission for a urinary tract infection and were diagnosed as troglitazone related. The ages of these patients ranged from 55 to 85 yr, two were women, and the length of troglitazone therapy before onset of liver injury ranged from 2 wk to 9 months. None had known prior liver disease and in three, normal serum transaminase levels were documented either before initiation, or at some time during troglitazone therapy, but before the onset of liver injury. One other patient was reported to have had a “normal laboratory evaluation” within the preceding 1–2 yr. Four had jaundice of which two were diagnosed as cholestatic and two as hepatocellular in nature. One of the latter developed hepatic encephalopathy and died. A liver biopsy from this patient showed submassive hepatocellular necrosis. This case was reported to the FDA. Of the remaining patients, three recovered and one was lost to follow-up because her health plan enrollment ended soon after discharge with a diagnosis of “drug-induced jaundice.” The incidence rates per million person-years of acute idiopathic liver injury (95% CI) were as follows: hospitalization
AJG – January, 2003
(n ⫽ 5), 1244 (404, 2900), hospitalized jaundice (n ⫽ 4), 995 (271, 2546), and ALF (n ⫽ 1), 240 (6.3, 1385).
DISCUSSION In this cohort of 7568 new troglitazone users, five cases of acute hospitalized liver injury, including one case of fatal ALF, were identified, for which an explanation other than troglitazone was not apparent after medical record review. The point estimates for the rates of hospitalization and ALF were 1244 and 240 per million person-years of troglitazone exposure, respectively. These rates are markedly increased above the estimated background rates for hospitalized acute idiopathic liver injury or ALF. Four epidemiological studies have estimated the background rate of hospitalized idiopathic hepatitis to range from 10 to 22 per million person-years (16 –19). The rate of hospitalization for idiopathic liver injury observed with troglitazone in this study was 55- to 124-fold greater than expected based on these studies. The background rate for idiopathic ALF in the general population has not been previously published. Two different approaches yielded estimates of less than one case per million person-years. Each year, about 2000 cases of ALF from all causes occur in the United States, with 10% from unexplained causes (14, 20). Using a U.S. census estimate of 281 million (21), an estimated rate of unexplained ALF of 0.7 per million person-years was obtained for the U.S. population. Using a second approach, four population-based epidemiological studies of liver disease were pooled for a total of 12 million person-years of observation (17–19, 22). Four cases of unexplained ALF were noted in these studies, yielding a rate estimate of 0.3 per million person-years (12, 13). By these metrics, the incidence of idiopathic ALF among troglitazone-exposed patients in this study was increased more than 240-fold compared with that expected in the general population. These estimates of the background rate of idiopathic ALF were based on general populations, not a population limited to type 2 diabetics. A recent epidemiological study from the United Kingdom followed type 2 diabetics treated with oral antidiabetic agents for a total of 114,000 person-years (23). There were no confirmed cases of ALF. However, one elderly patient died in the setting of jaundice. The medical record of this patient was not reviewed, and the causes of jaundice and death were not established. If it had been shown that other causes were absent, this would give an estimate of 8.8 per million person-years, still 27-fold below that observed in this cohort. Three unpublished studies, two of them randomized controlled clinical trials, suggest that the rate of ALF among troglitazone users was extremely elevated compared with that expected among type 2 diabetics (12, 13). In March, 1999, it was reported that one case of fatal ALF occurred in a company-sponsored postmarketing study of troglitazone, yielding an incidence rate of 1274 per million person-years
Idiopathic Acute Liver Failure and Troglitazone
177
(12, 13). In a National Institutes of Health-sponsored randomized clinical trial, the Diabetes Prevention Program, one of 585 patients in the troglitazone arm died after liver transplantation for ALF (580 person-years, rate 1724 per million person-years) (12, 13). Finally, in a randomized trial of 116 diabetic patients, contributing at most 58 personyears of observation, one patient treated with troglitazone died of ALF (rate ⬃17,000 per million person-years) (13). Several other lines of evidence suggest that the incidence rate of ALF with troglitazone observed in this study was not caused by type 2 diabetes itself. A review of the FDA’s adverse drug reaction database found that the reporting of ALF with troglitazone was 485-fold greater than with glipizide, a commonly used oral antidiabetic agent (13). If the high reporting with troglitazone was actually a result of the underlying diabetes being treated, one would expect roughly equivalent reporting for other antidiabetic drugs. In another study, data from the United Network for Organ Sharing, the organization responsible for coordinating solid-organ transplants in the United States, was reviewed for the years 1994 to 1998. During this period, three patients with ALF were registered for liver transplantation where troglitazone use was cited as the cause of liver injury. Based on the relative extent of troglitazone use compared with that of sulfonylurea agents and metformin, one would have expected 75 patients to be registered for transplantation with the United Network for Organ Sharing in which these latter drugs were listed as causative. However, during this interval, no patients were registered listing either of these drugs (2 [1 df] ⫽ 75, p ⬍ 0.001) (12, 13). An additional analysis of the United Network for Organ Sharing data for the years 1995 to 1997 found that the prevalence of diabetes among liver transplant registrants with acute hepatic necrosis was 4.8%, a level consistent with that of the general population (12, 13, 24). Finally, in the original phase 3 clinical trials, neither jaundice nor hospitalization for liver injury was observed in the diabetic control groups treated with drugs other than troglitazone (2). Together, these data suggest that the background incidence of unexplained ALF in type 2 diabetics is similar to that of the general population, and that the high rate observed with troglitazone in this study was not caused by type 2 diabetes. To help place these observations with troglitazone in context, a small number of epidemiological studies have been performed to estimate the rate of hospitalization for unexplained acute liver injury in association with the use of a variety of other drugs. For nonsteroidal anti-inflammatory drugs, liver injury hospitalization rates were estimated to range between 30 and 90 per million person-years in three separate studies (18, 19, 22). In another study, the rate of hospitalization associated with cimetidine use was 99 per million person-years (25). Hospitalization for acute liver injury has also been studied for several commonly used antibiotics. The rate of hospitalization with amoxicillin use was estimated at 17 and 25 per million exposed patients (26, 27). In these same studies, the acute liver injury hospital-
178
Graham et al.
ization rates for amoxicillin-clavulanic acid were 118 and 56 per million patients (26, 27). Erythromycin, an antibiotic with recognized hepatotoxicity was associated with hospitalized acute liver injury in two to 25 patients per million treated (17, 28, 29). Finally, one population-based study of 1052 ketoconzaole users observed no hospitalizations for acute liver injury, though 1.9% of patients were noted to develop liver-related problems as outpatients (30). In the present study, the rate of hospitalization with idiopathic acute liver injury based on patient counts was 661 per million patients treated. In the original clinical trials with troglitazone, the corresponding rate was 797 (2). Although idiopathic acute liver injury leading to hospitalization has been reported in epidemiological studies with a number of drugs, the risk of death from such injury has been low. Among the studies reviewed above, fatal liver injury was observed in only one, involving nonsteroidal anti-inflammatory drug use, yielding an incidence rate of 11 per million person-years (18). However, in another study of 625,307 nonsteroidal anti-inflammatory drug users, there were no fatalities, yielding a 95% CI for the incidence rate of fatal ALF of 0 to 10 per million person-years (22). There were no liver fatalities among 53,436 and 93,433 patients treated with amoxicillin-clavulanic acid (26, 27), and none among 392,201 patients treated with erythromycin (28). An important issue to consider is the potential role of nonalcoholic steatohepatitis (NASH) in acute liver injury and ALF with troglitazone. Obesity, frequently present in type 2 diabetics, and type 2 diabetes itself, are each associated with fatty liver and NASH (31–33). Insulin resistance is believed to be the underlying initiator of these processes (31, 32, 34, 35). Fatty liver has been reported to be present in 25–75% of diabetic patients, depending on their degree of accompanying obesity (32), and NASH was found in 12.2% and 18.5% of diabetics in two recent studies, 2.6- and 2.7-fold greater than in nondiabetic controls (36, 37). The question has been raised whether nonalcoholic fatty liver disease, and NASH in particular, might increase the susceptibility of affected patients to hepatocellular injury caused by alcohol or other xenobiotics (38, 39). Insulin resistance and associated hyperinsulinemia promote a shift in mitochondrial function from oxidation of glucose to -oxidation of free fatty acids, resulting in intracellular oxidative stress with its potential for hepatocellular injury (38 – 40). In such a setting, the generation of reactive intermediates during metabolism of certain drugs by the liver could cause hepatocellular injury or death. Hepatocellular death may result from direct cellular damage or from the activation of apoptosis (41). The same process could also occur in normal livers though the threshold for injury might be greater. Of note, a recent study in human hepatoma cells showed that troglitazone induced hepatocellular apoptosis, whereas two other thiazolidinediones did not under conditions of that experiment (42). Perhaps activation of intracellular apoptosis by reactive metabolites is the basis of ALF with troglitazone.
AJG – Vol. 98, No. 1, 2003
In the present study, it seems unlikely that NASH was responsible for or a contributing factor toward the liver injury observed in the cohort. First, four of five patients identified (Table 1) had documented normal serum transaminase levels at some point before the onset of hospitalized liver injury. Serum transaminase levels are typically elevated 2- to 4-fold above normal and remain persistently so in NASH (33, 39, 43, 44). Also, jaundice caused by NASH occurs only as an end-stage event once cirrhosis is well established and advanced (33, 43, 44). Finally, four of five patients had normal liver imaging (ultrasound—three, CT scan— one). Although the sensitivity of ultrasound and CT imaging in the detection of fatty liver is less than perfect, the absence of hepatic abnormalities with these modalities provides some evidence against the likelihood of significant fatty liver disease. This study relied upon automated claims data to identify troglitazone-exposed patients and cases of potential liver injury. Incomplete prescription claims could have led to underestimation of cohort size. This seems unlikely to have occurred because pharmacy reimbursement depends upon filing of prescription claims, providing an incentive for claims submission. It is also possible that some cases of hospitalized liver injury were missed, either because of miscoding or misdiagnosis by treating physicians, or because some patients for whom medical records were not sought actually had drug-induced disease. In retrospect, medical records probably should have been sought for two additional patients, the patient with a diagnosis of “liver disease not otherwise specified” who remained on troglitazone and the patient with a diagnosis of “drug-induced hepatitis” the day after his first troglitazone prescription. If either or both of these represented troglitazone-associated cases, the incidence rate for hospitalized liver injury would be higher than reported here. Likewise, if the patient who was lost to follow-up because her health plan enrollment ended actually progressed to liver failure, the rate of ALF reported here would be increased. It is also possible that some cases may have switched health plan coverage before developing liver injury, thereby escaping detection. We believe the likelihood of this to be low but cannot exclude the possibility. The results of this study suggest that troglitazone is a potent hepatotoxin, conferring a substantially increased risk of acute liver injury including ALF. The mechanism underlying troglitazone-induced ALF is unknown, but may involve metabolic generation of reactive intermediates that lead to either direct hepatocellular death or the activation of hepatocellular apoptosis. Reprint requests and correspondence: David J. Graham, M.D., M.P.H., 5600 Fishers Lane, HFD-400, Office of Drug Safety, Center for Drug Evaluation and Research, Food and Drug Administration, Rockville, MD 20857. Received Feb. 22, 2002; accepted June 26, 2002.
AJG – January, 2003
REFERENCES 1. Imura H. A novel antidiabetic drug, troglitazone—reason for hope and concern. N Engl J Med 1998;338:908 –9 (editorial). 2. Rezulin. In: Physicians’ desk reference. Montvale, NJ: Medical Economics Company, 1998:2218 –21. 3. Willman D. Fears grow over delay in removing Rezulin. Company’s conduct in development of the drug also comes under fire. Los Angeles Times 2000;10 Mar:A-18. 4. Graham DJ, Drinkard CR, Shatin D, et al. Liver enzyme monitoring in patients treated with troglitazone. JAMA 2001; 286:831–3. 5. Gitlin N, Julie NL, Spurr CL, et al. Two cases of severe clinical and histologic hepatoxicity associated with troglitazone. Ann Intern Med 1998;129:36 –8. 6. Neuschwander-Tetri BA, Isley WL, Oki JC, et al. Troglitazone-induced hepatic failure leading to liver transplantation. Ann Intern Med 1998;129:38 –41. 7. Shibuya A, Watanabe M, Fujita Y, et al. An autopsy case of troglitazone-induced fulminant hepatitis. Diabetes Care 1998; 21:2140 –3. 8. Herrine SK, Choudhary C. Severe hepatotoxicity associated with troglitazone. Ann Intern Med 1999;130:163–4 (letter). 9. Murphy EJ, Davern TJ, Shakil AO, et al. Troglitazone-induced fulminant hepatic failure. Dig Dis Sci 2000;45:549 –53. 10. Jagannath S, Rai R. Rapid-onset subfulminant liver failure associated with troglitazone. Ann Intern Med 2000;132:677 (letter). 11. Stolberg SG. Stricter rules are urged on use of a diabetic drug. The New York Times 1999;27 Mar:A-9. 12. Graham DJ, Green L. Epidemiology of hepatotoxicity with troglitazone. FDA Metabolic-Endocrine Drugs Advisory Committee Meeting, Bethesda, MD, March 26, 1999. Available from U.S. Food and Drug Administration via the Internet at: www.fda.gov/ohrms/dockets/ac/99/transcpt/3499t1a.pdf;/ 3499t1b.pdf;/3499t1c.pdf;/34991d.pdf;/34991e.pdf. Accessed January 8, 2002. 13. Graham DJ, Green L. Final report: Liver failure risk with troglitazone. Office of Postmarketing Drug Risk Assessment, Center for Drug Evaluation and Research, Food and Drug Administration, Rockville, MD, December 19, 2000. 14. Lee WH. Acute liver failure. N Engl J Med 1993;329:1862– 72. 15. Williams R. Classification, etiology, and consideration of outcome in acute liver failure. Semin Liver Dis 1996;16:343–8. 16. Almdal TP, Sorensen TIA. Incidence of parenchymal liver disease in Denmark, 1981 to 1985: Analysis of hospital registry data. Hepatology 1991;13:650 –5. 17. Beard K, Belic L, Aselton P, et al. Outpatient drug-induced parenchymal liver disease requiring hospitalization. J Clin Pharmacol 1986;26:633–7. 18. Garcia Rodriguez LA, Gutthann SP, Walker AM, et al. The role of nonsteroidal anti-inflammatory drugs in acute liver injury. Br Med J 1992;305:865–8. 19. Carson JL, Strom BL, Duff A, et al. Safety of nonsteroidal anti-inflammatory drugs with respect to acute liver disease. Arch Intern Med 1993;153:1331–6. 20. Schiodt FV, Atillasoy E, Shakil AO, et al. Etiology and outcome for 295 patients with acute liver failure in the United States. Liver Transplant Surg 1999;5:29 –34. 21. US Census Bureau. Available via the Internet at: http://www.census.gov/population/cen2000/tab02.pdf. Accessed January 28, 2002. 22. Garcia Rodriguez LA, Williams R, Derby LE, et al. Acute liver injury associated with nonsteroidal anti-inflammatory drugs and the role of risk factors. Arch Intern Med 1994;154: 311–5.
Idiopathic Acute Liver Failure and Troglitazone
179
23. Jick SS, Stender M, Myers MW. Frequency of liver disease in type 2 diabetic patients treated with oral antidiabetic agents. Diabetes Care 1999;22:2067–71. 24. Everhart JE. Digestive diseases and diabetes. In: Diabetes in America, 2nd ed. National Diabetes Data Group, National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD. NIH Publication No. 95–1468, 1995:457– 84. 25. Garcia Rodriguez LA, Wallander MA, Stricker BH. The risk of acute liver injury associated with cimetidine and other acid-suppressing anti-ulcer drugs. Br J Clin Pharmacol 1997; 43:183–8. 26. Garcia Rodriguez LA, Stricker BH, Zimmerman HJ. Risk of acute liver injury associated with the combination of amoxicillin and clavulanic acid. Arch Intern Med 1996;156:1327– 32. 27. Maggini M, Raschetti R, Agostinis L, et al. Use of amoxicillin and amoxicillin-clavulanic acid and hospitalization for acute liver injury. Ann Ist Super Sanita 1999;35:429 –33. 28. Derby LE, Jick H, Henry DA, et al. Erythromycin-associated cholestatic hepatitis. Med J Aust 1993;158:600 –2. 29. Carson JL, Strom BL, Duff A, et al. Acute liver disease associated with erythromycin, sulfonamides and tetracyclines. Ann Intern Med 1993;119:576 –83. 30. Garcia Rodriguez LA, Duque A, Castellsagne J, et al. A cohort study of the risk of acute liver injury among users of ketoconazole and other anti-fungal drugs. Br J Clin Pharmacol 1999;48:1847–52. 31. Silverman JF, Pories WJ, Caro JF. Liver pathology in diabetes mellitus and morbid obesity: Clinical, pathological, and biochemical considerations. Pathol Annu 1989;24:275–302. 32. McCullough AJ. Update on nonalcoholic fatty liver disease. J Clin Gastroenterol 2002;34:255–62. 33. Youssef W, McCullough AJ. Diabetes mellitus, obesity, and hepatic steatosis. Semin Gastrointest Dis 2002;13:17–30. 34. Chitturi S, Farrell GC. Etiopathogenesis of nonalcoholic steatohepatitis. Semin Liver Dis 2001;21:27–41. 35. Pagano G, Pacini G, Musso G, et al. Nonalcoholic steatohepatitis, insulin resistance, and metabolic syndrome: Further evidence for an etiologic association. Hepatology 2002;35:367– 72. 36. Wanless IR, Leutz JS. Fatty liver hepatitis (steatohepatitis) and obesity: An autopsy study with analysis of risk factors. Hepatology 1990;12:1106 –10. 37. Bernal-Reyes R, Saenz-Lubra A, Bernardo-Escudero R. Prevalencia de la esteatohepatitis no alcoholica (EHNA). Estudio comparativo con diabeticos. [Prevalence of non-alcoholic steatohepatitis. Comparative study with diabetic patients]. Rev Gastroenterol Mex 2000;65:58 –62. 38. Angulo P, Lindor KD. Insulin resistance and mitochondrial abnormalities in NASH: A cool look into a burning issue. Gastroenterology 2001;120:1281–5 (editorial). 39. Clark JM, Brancatti FL, Diehl AM. Nonalcoholic fatty liver disease. Gastroenterology 2002;122:1649 –57. 40. Sanyal AJ, Campbell-Sargent C, Mirshahi F, et al. Nonalcoholic steatohepatitis: Association of insulin resistance and mitochondrial abnormalities. Gastroenterology 2001;120: 1183–92. 41. Jaeschke H, Gores GJ, Cederbaum AI, et al. Mechanisms of hepatotoxicity. Toxicol Sci 2002;65:166 –76. 42. Yamamoto Y, Nakajima M, Yamzaki H, et al. Cytotoxicity and apoptosis produced by troglitazone in human hepatoma cells. Life Sci 2001;70:471–82. 43. Reid AE. Nonalcoholic steatohepatitis. Gastroenterology 2001;121:710 –23. 44. Contos MJ, Sanyal AJ. The clinicopathologic spectrum and management of nonalcoholic fatty liver disease. Adv Anat Pathol 2002;9:37–51.