Francis F. Fountain, MD; Elizabeth Tolley, PhD; Cary R. Chrisman, PharmD; and Timothy H. Self, PharmD

Transcription

1 Isoniazid Hepatotoxicity Associated With Treatment of Latent Tuberculosis Infection* A 7-Year Evaluation From a Public Health Tuberculosis Clinic Francis F. Fountain, MD; Elizabeth Tolley, PhD; Cary R. Chrisman, PharmD; and Timothy H. Self, PharmD Objectives: To determine the overall incidence of isoniazid (INH) hepatotoxicity in a public health tuberculosis clinic over a 7-year period, and to determine if systematic, limited aspartate aminotransferase (AST) monitoring would be of benefit in detecting INH hepatotoxicity. Methods: Evaluation of INH hepatotoxicity in adults aged > 25 years from a database maintained from fall 1996 to 2003 in a public health department clinic. Hepatotoxicity was defined as AST levels more than five times the upper limit of normal (ULN). Results: Among 3,377 patients started on INH therapy, 19 patients had AST levels more than five times the ULN, or a rate of 5.6 per 1,000 patients. Only 1 of 19 patients had prodromal symptoms associated with hepatotoxicity. After 1 month, 3 months, and 6 months of therapy, the numbers of hepatotoxic events per 1,000 patients were 2.75, 7.20, and The age-specific numbers of hepatotoxic events per 1,000 patients were 4.40 for those from 25 to 34 years of age, inclusive; 8.54 for those between 35 to 49 years of age, inclusive; and for those > 50 years old. Age > 49 years (p < 0.02) and baseline AST greater than ULN (p < ) were risk factors for hepatotoxicity. Conclusions: Consistent with earlier trials, INH hepatoxicity is age related. Our results suggest hepatotoxicity is also related to baseline AST greater than ULN. Moderate-to-severe hepatotoxicity frequently occurs without symptoms, suggesting the value of more widespread AST monitoring. (CHEST 2005; 128: ) Key words: aspartate aminotransferase; hepatotoxicity; isoniazid Abbreviations: AST aspartate aminotransferase; CI confidence interval; INH isoniazid; LTBI latent tuberculosis infection; OR odds ratio; TB tuberculosis; ULN upper limit of normal First introduced in the early 1950s, isoniazid (INH) is still an integral part of both the treatment of active and latent infections of Mycobacterium tuberculosis. For decades, it has been well documented that INH can cause adverse *From the Tuberculosis Clinic (Dr. Fountain), Memphis and Shelby County Health Department, Memphis; Department of Preventive Medicine, (Dr. Tolley), University of Tennessee Health Science Center, Memphis; Methodist Medical Center of Oak Ridge (Dr. Chrisman), Oak Ridge; and College of Pharmacy (Dr. Self), University of Tennessee Health Science Center, Memphis, TN. Manuscript received October 8, 2004; revision accepted December 24, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( org/misc/reprints.shtml). Correspondence to: Timothy Self, PharmD, Professor, College of Pharmacy, University of Tennessee Health Science Center, 910 Madison Ave, Room 308, Memphis, TN 38163; tself@ utmem.edu effects on the liver, ranging from mild transient elevations in aminotransferases (transaminases), which occur in approximately 10 to 20% of patients, to overt hepatitis, occurring much more rarely Studies 4,7,8,10 of INH hepatotoxicity have resulted in the identification of several risk factors, most importantly age 35 years. Other risk factors include female gender and concurrent use of ethanol. 4,7,8,10 Some data 6 9 suggest that African-American and Hispanic women may also be at greater risk of hepatotoxicity. Specific updated guidelines regarding the use of INH for latent tuberculosis infection (LTBI) were formulated in terms of age, baseline assessment, and clinical monitoring for hepatotoxicity. 11 Five studies 10,12 15 of INH hepatotoxicity in public health clinics have found incidences ranging from 0.1 to 4%, with some of the differences accounted for 116 Clinical Investigations

2 probably by the ages of the populations and the definitions of hepatotoxicity. Memphis was one of 21 cities included in the large Public Health Service Study conducted in Although no cases of hepatitis were associated with INH in the Memphis population, only 223 patients were enrolled in the study. 4 We sought to determine the overall incidence of INH hepatotoxicity in our public health tuberculosis (TB) clinic over an 8-year period. In addition, this study was initiated to determine if systematic limited aspartate aminotransferase (AST) monitoring would be of benefit in detecting INH hepatotoxicity. Another goal was to determine if black women had a greater incidence of hepatotoxicity in our population that is primarily composed of African Americans. was withdrawn by the health department TB physician. The hepatitis B surface antigen carrier state was not evaluated in this study. HIV testing was offered to all patients but was not recorded in this investigation. Statistical Methods Extracted variables included gender, age, race/ethnicity, year of treatment initiation, self-reported alcohol consumption, selfreported history of preexisting liver disease, and serum AST levels at baseline and follow-up at 1 month, 3 months, and 6 months. Age was categorized as 25 to 34 years, 35 to 49 years, and 50 years. Race/ethnicity was categorized as African American, white, Hispanic, Asian, and unknown. Preexisting liver disease was categorized as hepatitis (A, B, or C), cirrhosis, and other (including neonatal jaundice), trauma (gunshot wound or motor vehicle accident), alcoholism, thrombosis, jaundice, gallstones, and unknown etiology. Alcohol consumption was categorized as none, the equivalent of 1 to 7 drinks (a glass of wine, can of beer, or mixed drink) per week, 8 to 14 drinks per week, and 15 Materials and Methods Patients seen at the Memphis and Shelby County Health Department TB clinic receiving INH for the treatment of LTBI during the 7-year period from Fall 1996 through December 2003 were included in this investigation. Only adults aged 25 years were included in the study. Exclusion criteria were pregnancy, initiation of INH during the 3 months postpartum, initiation of INH if the baseline AST level was more than three times the upper limit of normal (ULN), or history of INH allergy. The Health Department TB physician (F.F.F.) has maintained a database for all patients started on INH for LTBI since January One other physician cooperated fully in the LTBI protocol until his retirement in To evaluate INH hepatotoxicity, an analysis of this database was performed in this university institutional review boardapproved study. Per current American Thoracic Society/Centers for Disease Control and Prevention/Infectious Diseases Society of America guidelines, we defined moderate hepatotoxicity as AST levels 5 to 10 times the ULN, and severe hepatotoxicity as AST levels 10 times the ULN. 11 Our goal in this study was to evaluate only moderate or severe hepatotoxicity. From 1996 to mid-1999, patients for whom LTBI treatment was indicated were offered 6 months of INH treatment. Since late 1999, patients were offered 9 months of treatment per updated American Thoracic Society/Centers for Disease Control and Prevention guidelines. INH dosages were 300 mg/d for patients weighing 60 kg and 5 mg/kg if weight was 60 kg. The protocol for AST testing was as follows: before treatment and at 1 month and 3 months if receiving 6 months of treatment; before treatment and at 1 month, 3 months, and 6 months if receiving 9 months of treatment. At the end of treatment, the AST was not measured since it would not change treatment. Additional AST tests were done at the discretion of the treating physician (eg, symptoms compatible with hepatotoxicity; AST values trending upward). Records from a total of 3,414 patients were identified. Of these, 35 patients were not started on therapy because their baseline AST levels exceeded four times the ULN (n 20), or were between three times and four times the ULN (n 15). Data from two other patients were omitted from the analysis. Data from one patient were excluded because all demographic characteristics were unknown. This individual completed at least 3 months of therapy without any adverse events. The other patient was excluded because active hepatitis C was diagnosed after 3 months of therapy with INH, at which time INH therapy Table 1 Demographic Characteristics Characteristics No. (%) Year of treatment initiation (0.47) (8.17) (13.00) (14.33) (13.50) (16.38) (18.42) (15.72) Age, yr ,533 (45.40) ,409 (41.72) (12.88) Gender Male 2,075 (61.45) Female 1,302 (38.55) Race/ethnicity African American 2,443 (72.34) White 403 (11.93) Hispanic 285 (8.44) Asian 229 (6.78) Unknown 17 (0.50) Alcohol consumption (self-reported), drinks/wk* None 2,474 (73.26) (19.69) (3.46) (3.58) History of liver disease None 3,220 (95.35) Hepatitis A, B, or C 109 (3.23) Cirrhosis 2 (0.06) Other 49 (1.45) *A drink is equivalent to a glass of wine, a can of beer, or a mixed beverage. Summation of numbers exceeds 3,377 and that of percentages exceeds 100% because three subjects had both hepatitis and another liver disease. Other liver diseases included neonatal jaundice, trauma (gunshot wound or motor vehicle accident), alcoholism, thrombosis, liver mass, jaundice, gallstones, and unknown etiology. CHEST / 128 / 1/ JULY,

3 drinks per week. When alcohol consumption was unavailable, the age/gender/race median consumption was assigned; for all subclasses, this imputed value was none. Sensitivity analysis indicated that results were essentially unchanged when subjects whose alcohol consumption was unavailable were excluded. If baseline AST levels were missing, and follow-up values were recorded, the baseline AST level was assumed to be less than the ULN. If more than one value for serum AST was available for a follow-up visit, the largest value was retained for analysis. Estimates of the incidence of hepatotoxicity and the noncompletion rates for the various subclasses were obtained using contingency tables. Comparisons among the various subclasses for incidence of hepatotoxicity and rate of noncompletion were made using logistic regression. Bivariate analyses were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for each putative risk factor. Dummy variables were used for variables with more than two levels, with the lowest level or the level with the largest frequency being designated as the reference level. Multivariable logistic regression was used to estimate ORs after controlling for effects of other variables in the model. Explanatory variables were included in the model building process, if bivariate analysis revealed an association with hepatotoxicity or with noncompletion at a level of significance In the final model, only those variables with significance levels 0.05 were retained. Results were robust because forward inclusion, backward elimination, and stepwise selection algorithms identified the same final models. Subsequently, sources of confounding among potential explanatory variables, which were excluded from the final model, were identified using bivariate analysis methods. Statistical analyses were performed with software (Version 9.00; SAS Institute; Cary, NC); p 0.05 was considered statistically significant. Results Records of 3,377 patients were available for this study. Demographic characteristics of these patients are depicted in Table 1. This sample was composed predominately of African-American men 50 years of age, who reported no current alcohol consumption or previous history of liver disease. A total of 19 patients had hepatotoxicity (Table 2), as indicated by elevated serum AST levels more than five times the ULN. INH was immediately discontinued in these patients, and additional liver function tests were done (alanine aminotransferase, alkaline phosphatase, total bilirubin). Only one of these patients had any prodromal symptoms associated with hepatotoxicity. The number of hepatotoxic events per 1,000 patients started on therapy was After 1 month, 3 months, and 6 months of therapy, the numbers of hepatotoxic events per 1,000 patients were 2.75, 7.20, and 4.10 (Table 3). The age-specific numbers of hepatotoxic events per 1,000 patients were 4.40 for those between 25 and 34 years of age, inclusive; 8.54 for those between 35 and 49 years of age, inclusive; and for those 50 years old (Table 4). Risk factors for experiencing elevated serum AST levels included age 49 years (p 0.02) and baseline Table 2 Characteristics of the Patients With Hepatotoxicity Patient No. Sex Race/Ethnicity Age, yr Date Treatment Started Alcohol Consumption per Week Baseline AST Peak AST/No. of Months After Start of INH* 5 10 ULN 1 M Black 36 04/23/2003 None /3 2 M Hispanic 34 02/11/2003 None /6 3 F Black 50 08/02/2000 None /1 4 M Black 57 05/22/2000 Unknown /1 5 M White 43 05/09/ quarts of beer /6 6 M Black 52 03/13/2000 None /1 7 M Black 59 09/30/ pint whiskey /3 8 M Black 29 04/24/1997 1/5 whiskey /3 9 M White 46 12/20/1996 Occasional /1 10 ULN 10 M Black 36 04/08/ drinks/wk (whiskey) /3 11 F White 50 04/19/2002 None /3 12 F Black 63 04/22/2002 None /3 13 F Black 41 06/08/ quarts of beer /1 14 M Black 38 06/30/2000 Unknown /6 15 F Black 34 09/01/1999 None /3 16 M Black 37 01/06/ pint whiskey /1 17 F Black 52 10/23/1997 None /3 18 F Black 38 10/01/ quarts of beer /3 19 F Hispanic 30 10/20/2003 None /3 *One of 19 patients had prodromal symptoms consistent with hepatotoxicity. No record of self-reported alcohol consumption was available from the patient s records. For the statistical analysis, unknown alcohol consumption was set to the median for the appropriate age-gender-race subclass, which were in all instances None. While patient reported no prodromal symptoms consistent with hepatotoxicity, patient was icteric (T. BIL. 5.8). Patient had malaise and nausea; T.BIL Clinical Investigations

4 Table 3 Compliance and Hepatotoxic Events Associated With Treatment of Latent Tuberculosis With INH (n 3,377) Completed Treatment No. (%) Hepatotoxic Events, No. At least 1 mo 2,182 (64.61) 6 At least 3 mo 1,389 (41.13) 10 At least 6 mo 732 (21.68) 3 serum AST level exceeding the ULN (p ) [Table 4]. Baseline AST data were missing in 48 patients. Elevated serum AST level was poorly correlated with self-reported alcohol consumption ( coefficient ; p ), with only 38.96% of those with elevated serum AST levels reporting any consumption of alcohol. Multivariable logistic regression analysis indicated that age and elevated AST levels were independent predictors of hepatotoxicity (Table 5). There was no association between hepatotoxicity and gender, race/ethnicity, self-reported alcohol consumption, or self-reported history of previous liver disease (Table 4). Furthermore, there was no interaction between gender and race/ethnicity (all p 0.6). Among African-American subjects, 8 of 896 men (0.89%) compared with 6 of 681 women (0.88%) experienced hepatotoxicity. Similarly, among white patients, 2 of 164 men (1.22%) and 1 of 102 women (0.98%) had elevated AST levels. Among Hispanics, only 1 of 118 men (0.85%) and 1 of 61 women (1.64%) had hepatotoxicity. Only 41.13% of the patients completed 3 months of isoniazid therapy, and 21.65% completed 6 months of therapy (Table 3). From 2000 through 2003, the rate of compliance with therapy was higher than during the 4 previous years (p ) [Table 6]. Factors associated with better compliance included being Hispanic (p ) and being at least 50 years old (p ). In contrast, having a history of hepatitis (A, B, or C) was associated with a Table 4 Bivariate Analysis of Factors Associated With Elevation of Transaminases by Greater Than Five Times the ULN (n 2,182) Characteristics No. With Elevation (%) OR (95% CI) p Value Age, yr (0.44) Reference (0.85) ( ) (2.08) ( ) 0.02 Gender Male 11 (0.87) Reference Female 8 (0.88) ( ) 0.98 Race/ethnicity* African American 14 (0.89) Reference White 3 (1.13) ( ) 0.71 Hispanic 2 (1.12) ( ) 0.76 Asian 0 (0.00) Unknown 0 (0.00) Alcohol consumption, drinks/wk None 12 (0.73) Reference (1.18) ( ) (1.64) ( ) (1.61) ( ) 0.45 Any (1 ) 7 (1.28) ( ) 0.24 History of liver disease None 18 (0.84) Reference Hepatitis (A, B, or C) 0 (0.00) Cirrhosis 0 (0.00) Other 1 (2.94) ( ) 0.21 Any previous history 1 (1.05) ( ) 0.85 Baseline transaminases ULN 40 No 12 (0.62) Reference Yes 7 (2.92) ( ) *For this analysis, patients whose race/ethnicity was Asian or unknown were deleted. A drink is equivalent to a glass of wine, a can of beer, or a mixed beverage. For this analysis, patients whose history of liver disease included only cirrhosis were deleted. Other liver diseases included neonatal jaundice, trauma (gunshot wound or motor vehicle accident), alcoholism, thrombosis, liver mass, jaundice, gallstones, and unknown etiology. CHEST / 128 / 1/ JULY,

5 Table 5 Results From Multivariable Logistic Regression Analysis of Risk Factors Associate With Elevation of Transaminases by Greater Than Five Times the ULN (n 2,182) Characteristics OR (95% CI) p Value Baseline AST ULN ( ) Age 50 yr ( ) higher risk of not completing at least 6 months of therapy (p 0.05). Although only two patients reported a history of cirrhosis, neither completed at least 6 months of therapy. After controlling for differences associated with year of treatment initiation (ie, before 2000 vs later), being Hispanic was not an independent predictor, because 94.4% of Hispanic patients were treated in 2000 or later (Table 7). Similarly, after controlling for age, neither being female nor having a baseline AST level greater than Table 7 Results From Multivariable Logistic Regression Analysis of Factors Associated With Voluntary Patient-Based Withdrawal Prior to Completing 6 Months of Treatment With INH (n 3,360) Characteristics OR (95% CI) p Value Treatment initiated ( ) Prior history of hepatitis ( ) 0.03 Age yr ( ) 0.02 Age 50 yr ( ) the ULN was an independent predictor of completing at least 6 months of therapy. Female subjects were on average older than male subjects. Subjects between 35 years and 49 years of age, inclusive, were more likely to have baseline AST levels that exceeded the ULN. Being between the ages of 35 years and 49 years (p 0.02) or being at least 50 years of Table 6 Bivariate Analysis of Factors Associated With Dropping Out Prior to Completing 6 Months of Treatment With INH (n 3,360)* Characteristics Dropouts, No. (%) OR (95% CI) p Value Year of treatment initiation ,155 (95.61) Reference ,474 (68.49) ( ) Age, yr ,229 (80.33) Reference ,091 (77.82) ( ) (72.20) ( ) Gender Male 1,634 (79.05) Reference Female 995 (76.89) ( ) 0.15 Race/ethnicity African American 1,924 (79.18) Reference White 318 (79.30) ( ) 0.96 Hispanic 194 (68.31) ( ) Asian 179 (78.17) ( ) 0.72 Unknown 14 (82.35) ( ) 0.75 Alcohol consumption, drinks/wk None 1,922 (78.03) Reference (77.27) ( ) (83.76) ( ) (81.82) ( ) 0.33 Any (1 ) 707 (78.73) ( ) 0.67 History of liver disease None 2,498 (77.97) Reference Hepatitis (A, B, or C) 94 (86.24) ( ) 0.05 Cirrhosis 2 (100.0) Other 38 (77.55) ( ) 0.88 Any previous history 131 (83.44) ( ) 0.11 Baseline transaminases ULN 40 No 2,356 (78.56) Reference Yes 273 (75.41) ( ) 0.18 *A total of 17 patients were omitted from these analyses. Their physicians withdrew them from treatment after their transaminase levels exceeded five times the ULN. After 1 month and 3 months of treatment 6 patients and 11 patients, respectively, were withdrawn from treatment by their physicians. A drink is equivalent to a glass of wine, a can of beer, or a mixed beverage. For this analysis patients whose history of liver disease included only cirrhosis were deleted. 120 Clinical Investigations

6 age (p ) was independently associated with better compliance (Table 7). In contrast, a history of hepatitis (A, B, or C) was independently associated with a higher risk of not completing at least 6 months of therapy. Discussion For many years, INH had been considered the optimal treatment of LTBI in the United States. Concerns regarding the percentage of patients who completed a full course of INH, which has been at various times recommended to be 6 months, 9 months, or 12 months, led to the investigation of treatment regimens of a shorter duration. A short course of pyrazinamide and rifampin administered for 8 weeks to a large international cohort of HIVpositive patients was found to be associated with a greater likelihood of completion of therapy as compared to 12 months of daily INH, with no statistically significant difference in the incidence of hepatotoxicity. 16 However, when these findings were applied widely, and to patients with different characteristics than those in the original studies, pyrazinamide/ rifampin was found to be associated with an increased incidence of hepatotoxicity, including fatalities, relative to INH alone. 17 Consequently, there has been renewed interest in the study of INH hepatotoxicity. INH is metabolized in the liver primarily by acetylation and hydrolysis, and it is these acetylated metabolites that are thought to be hepatotoxins. 18 Findings 19 in rats suggest that the hydrazine metabolite of INH and is subsequent effect on CYP2E1 induction is involved in the development of INHinduced hepatotoxicity. The recent Public Health Clinic studies by Nolan et al 10 and LoBue and Moser 12 had lower rates of INH hepatotoxicity (0.1% and 0.3%, respectively) than our rate of 0.56%. However, this difference is most likely due to the much younger overall population in these earlier studies, which included young children, adolescents, and patients in their early 20s, as well as middle-aged adults. In the LoBue and Moser study, 12 85% of the patients were 35 years of age, and 80% of patients in the study by Nolan et al 10 were 35 years old. In our investigation, 54.6% of patients were 35 years of age. As compared with other studies reporting approximately 60% completion rates for 6 months of treatment with INH, our completion rates were poor (only 41% of our patients completed 3 months of treatment). 10,12,13 Thus, the rate of hepatotoxicity in our population would likely have been higher if more patients had completed 6 months of treatment. However, INH hepatotoxicity frequently occurs within the first 3 months of treatment. For example 10 of 11 cases of INH hepatotoxicity reported by Nolan et al 10 occurred within the first 3 months of treatment. Reasons for high noncompletion rate in patients receiving INH treatment of LTBI include homelessness, substance abuse, and side effects. 12 Other possible reasons for nonadherence include lack of symptoms, length of treatment, inconvenience, and travel to clinic. The most common reason given for not returning to clinic by our patients was I am too busy. The reasons for better compliance from are not readily apparent. Our study provides further validation to age as a risk factor for INH hepatotoxicity, but we did not find that African-American women are at greater Table 8 Recent Public Health TB Clinic Studies Study Location Patient No. % Patients With Hepatotoxicity % Patients Age 35 yr Comments Observational studies Nolan et al 10 Seattle-King County, WA 11, Asymptomatic AST elevations not included LoBue and Moser 12 San Diego, CA 3, Hepatotoxicity defined as elevation in transaminases 5 times normal or 3 5 times normal with symptoms Current study Memphis, TN 3, Hepatotoxicity defined as AST 5 times ULN Clinical trials Jasmer et al 13 Atlanta, GA Boston, MA San Francisco, CA Hepatotoxicity defined as ALT 5 times normal McNeill et al 14 Greenville, NC Not given* Hepatotoxicity defined as ALT 160 Jasmer et al 15 San Francisco, CA Not given Hepatitis defined by presence of both symptoms and any abnormal transaminase *Mean age Mean age CHEST / 128 / 1/ JULY,

7 risk, as suggested by some earlier data. Our data further suggest that baseline elevation of AST is a risk factor for INH hepatotoxicity. American Thoracic Society/Centers for Disease Control and Prevention current recommendations for treatment of LTBI with INH state that routine baseline and follow-up laboratory monitoring can be eliminated in most persons, including older patients. 20 Exceptions include patients whose initial evaluation suggests a liver disorder. Baseline testing is also indicated for patients with HIV infection, pregnant women, and women in the immediate postpartum period (ie, within 3 months of delivery), persons with a history of chronic liver disease, persons who use alcohol regularly, and persons at risk for chronic liver disease. The American Thoracic Society/Centers for Disease Control and Prevention recommendations also state that Some experts recommend that isoniazid should be withheld if transaminase levels exceed three times the upper limit of normal if associated with symptoms and five times the upper limit of normal if the patient is asymptomatic. 20 Although some experts in the mid-1970s recommended monitoring liver function tests only in patients who had prodromal symptoms (malaise, anorexia, nausea and vomiting), Byrd et al 21 found that 73% of their patients had marked elevations in liver function tests (AST 5 times ULN) despite being asymptomatic. Other earlier reports 22,23 confirm that serious hepatic dysfunction can occur in asymptomatic patients. One report 13 further suggests that INH hepatotoxicity is rarely associated with symptoms. Our data are consistent with these reports, in that only one of our patients who had an increase in AST of either 5 to 10 times the ULN or 10 times the ULN was symptomatic. While it is possible that some of our patients would have become symptomatic had INH been continued, use of limited laboratory testing allowed for earlier discovery of INH hepatotoxicity. We believe that based on our results and those of other studies, a reassessment of the value of limited monitoring of AST in all patients 35 years of age, as well as those with other risk factors, receiving INH treatment for LTBI is appropriate. Weighing the benefits of limited laboratory monitoring vs the human and economic costs is of obvious importance. Five analyses 10,12 15 of INH hepatotoxicity in the public health setting have found incidences ranging from 0.1 to 4%, with some of the differences accounted most likely by the ages of the populations and the definitions of hepatotoxicity (Table 8). We have included our results in Table 8 for a comparison of these recent studies. ACKNOWLEDGMENT: A special thank you to Barbara Willingham, Administrative Aide, University of Tennessee, Department of Pharmacy. References 1 Randolph H, Joseph S. Toxic hepatitis with jaundice occurring in a patient treated with isoniazid. JAMA 1953; 152: Cohen R, Kalser MH, Thompson RV. Total hepatic necrosis secondary to isoniazid therapy. JAMA 1961; 176: Black M, Mitchell JR, Zimmerman HJ, et al. Isoniazidassociated hepatitis in 114 patients. Gastroenterology 1975; 69: Kopanoff DE, Snider DE Jr, Caras GJ. Isoniazid-related hepatitis: a U.S. Public Health Service cooperative surveillance study. Am Rev Respir Dis 1978; 117: Riska N. Hepatitis cases in isoniazid treated groups and in a control group. Bull Int Union Tuberc 1979; 54: Moulding TS, Redeker AG, Kanel GC. Twenty isoniazidassociated deaths in one state. Am Rev Respir Dis 1989; 140: Franks AL, Binkin NJ, Snider DE JR, et al. Isoniazid hepatitis among pregnant and postpartum Hispanic patients. Public Health Rep 1989; 104: Snider DE Jr, Caras GJ. Isoniazid-associated hepatitis deaths. Am Rev Respir Dis 1992; 145: Millard PS, Wiscosky TC, Reade-Christopher SJ, et al. Isoniazid-related fatal hepatitis. West J Med 1996; 164: Nolan CM, Goldberg SV, Buskin SE. Hepatotoxicity associated with isoniazid preventive therapy: a 7-year survey from a Public Health Tuberculosis Clinic. JAMA 1999; 281: ATS/CDC/IDSA joint statement. Treatment of tuberculosis. Am J Respir Crit Care Med 2003; 167: LoBue PA, Moser KS. Use of isoniazid for latent tuberculosis infection in a public health clinic. Am J Respir Crit Care Med 2003; 168: Jasmer RM, Saukkonen JJ, Henry MB, et al. Short-course rifampin and pyrazinamide compared with isoniazid for latent tuberculosis infection: a multicenter clinical trial. Ann Intern Med 2002; 137: McNeill L, Allen M, Estrada C, et al. Pyrazinamide and rifampin vs isoniazid for the treatment of latent tuberculosis. Chest 2003; 123: Jasmer RM, Snyder DC, Chin DP, et al. Twelve months of isoniazid compared with four months of isoniazid and rifampin for persons with radiographic evidence of previous tuberculosis. Am J Respir Crit Care Med 2000; 162: Gordin F, Chaisson RE, Matts JP, et al. Rifampin and pyrazinamide vs isoniazid for prevention of tuberculosis in HIV infected persons. JAMA 2000; 283: Centers for Disease Control and Prevention. Update adverse event data and revised American Thoracic Society/CDC recommendations against the use of rifampin and pyrazinamide for the treatment of latent tuberculosis infection. MMWR Morb Mortal Wkly Rep 2003; 52: Peretti E, Karlaganis G, Lauterburg BH. Acetylating of acetylhydrazine, the toxic metabolite of isoniazid, in humans: inhibition by concomitant administration of isoniazid. J Pharmacol Exp Ther 1987; 243: Yue J, Peng RX, Yang J, et al. CYP2E1 mediated isoniazidinduced hepatotoxicity in rats. Acta Pharmacol Sin 2004; 25: Clinical Investigations

8 20 Official Statement of the American Thoracic Society. Targeted tuberculin testing and treatment of latent tuberculosis infection. Am J Respir Crit Care Med 2000; 161: Byrd RB, Horn BR, Griggs GA, et al. Isoniazid chemoprophylaxis: association with detection and incidence of liver toxicity. Arch Intern Med 1977; 137: Scharer L, Smith JP. Serum transaminase elevations and their hepatic abnormalities in patients receiving isoniazid. Ann Intern Med 1969; 71: Bailey WC, Taylor SL, Dasacomb HI, et al. Disturbed hepatic function during isoniazid chemprophylaxis. Am Rev Respir Dis 1973; 107: CHEST / 128 / 1/ JULY,