Published ahead of print on April 3, 2008, doi:10.1164/rccm.200802-355OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200802-355OC
Hepatotoxicity of PyrazinamideCohort and Case-Control Analyses1 Tuberculosis and Chest Service, Centre for Health Protection, Department of Health, Hong Kong, China; and 2 Tuberculosis and Chest Unit, Grantham Hospital, Hospital Authority, Hong Kong, China Correspondence and requests for reprints should be addressed to Dr. Kwok Chiu Chang, M.B., M.Sc., Wanchai Chest Clinic, 1st Floor, Wanchai Polyclinic, 99 Kennedy Road, Wanchai, Hong Kong, China. E-mail: kc_chang{at}dh.gov.hk
Rationale: Relatively little is known about the hepatotoxicity of pyrazinamide. Objectives: We compared continuation-phase regimens incorporating pyrazinamide, isoniazid, and/or rifampin with those containing isoniazid and rifampin to evaluate the hepatotoxicity of pyrazinamide. Methods: Cohort and nested case-control analyses were conducted on a cohort of 3,007 patients with active tuberculosis (TB) managed at government chest clinics under a TB control program with treatment started from January 1 through June 30, 2001. Cases included all patients with probable hepatotoxicity from 12 or more weeks after starting treatment. Hepatotoxicity was considered probable when serum alanine transaminase exceeded three times the upper limit of normal. Each case was matched by sex and age with three control subjects selected randomly from the rest of the cohort. Treatment regimens of cases within 4 weeks preceding hepatotoxicity were compared with those of matched control subjects in comparable periods relative to the date of commencing treatment. Measurements and Main Results: Hepatotoxicity occurred in 150 (5.0%) patients at any time including 48 (1.6%) cases. From 12 or more weeks after starting treatment, the estimated risk of hepatotoxicity was 2.6% for regimens incorporating pyrazinamide, isoniazid, and/or rifampin, and 0.8% for standard regimens containing isoniazid and rifampin. Multivariable conditional logistic analysis showed a significant association between hepatotoxicity and, respectively, hepatitis B, previous hepatotoxicity, and treatment regimens. The adjusted odds ratio (95% confidence interval) of hepatotoxicity for regimens incorporating pyrazinamide, isoniazid, and/or rifampin relative to standard regimens was 2.8 (1.4–5.9). Conclusions: Adding pyrazinamide to isoniazid and rifampin increases the risk of hepatotoxicity appreciably.
Key Words: hepatotoxicity • pyrazinamide • tuberculosis
The introduction of pyrazinamide in the initial phase has further reduced the duration of rifampin-containing anti-tuberculosis (anti-TB) regimens from 9 months to 6 months (1). It has been shown that pyrazinamide acts predominantly in the initial 2 months in the modern regimen (2) due to need of an acidic microenvironment for killing tubercle bacilli (3). However, pyrazinamide may still be useful for containing treatment failure in case of a complicated continuation phase (4, 5). Clinical scenarios that may indicate the use of pyrazinamide in the continuation phase include the unavailability of drug susceptibility test results (6), intolerance of other first-line anti-TB drugs, initial resistance to isoniazid (2, 7, 8), and multidrug-resistant TB. Despite the important role of pyrazinamide, relatively little is known about its hepatotoxicity. The hepatotoxicity of pyrazinamide is both dose dependent (9) and idiosyncratic (10, 11). Initial use of pyrazinamide at a dosage of 3 g daily with isoniazid showed a high risk of hepatitis in comparison with isoniazid and para-aminosalicylic acid (12). Previous controlled trials that suggested nonsignificant hepatotoxicity for pyrazinamide at dosages of 1.5 to 2 g daily were neither designed nor sufficiently powered for evaluating hepatotoxicity (2, 13). Concern about the hepatotoxicity of pyrazinamide has recently been rekindled by studies on treatment of latent TB infection involving pyrazinamide and, respectively, rifampin (11), ofloxacin (14), levofloxacin (15), and ethambutol (16). A retrospective cohort study involving 430 patients has suggested that pyrazinamide is more hepatotoxic than isoniazid and rifampin (17), but their findings may be difficult to conclude due to important methodologic issues (18–20). The only pragmatic approach to address the hepatotoxicity of pyrazinamide may be by comparing pyrazinamide-containing regimens with standard regimens that do not contain pyrazinamide in comparable periods either in the initial or continuation phase. Because pyrazinamide would often be given in the initial phase unless the patient is at risk of hepatitis, it would be more difficult to identify comparable control subjects in the initial phase than it would be in the continuation phase. Thus, we conducted a case-control study nested in a cohort of patients with TB treated at government chest clinics in Hong Kong in 2001 to examine the hepatotoxicity of pyrazinamide given in the continuation phase. The case definition of TB was a clinically compatible illness confirmed by bacteriologic evidence or, in the absence of bacteriologic evidence, a case with compatible clinical, radiologic, and/or histologic findings, and showing appropriate response to treatment. In 2001, the number of TB notifications in Hong Kong was 7,262 (21). The proportions of pulmonary TB (with or without coexisting extrapulmonary TB) and extrapulmonary TB were 88% and 14.8%, respectively (21). Approximately 62% of cases with pulmonary TB were culture positive (21). This study was possible because it had been relatively common in Hong Kong to continue pyrazinamide with isoniazid and rifampin in the continuation phase for some patients with drug-susceptible and relatively extensive TB to reduce the risk of treatment failure; to give rifampin, ethambutol, and pyrazinamide in the continuation phase for patients with initial resistance to isoniazid (21); and to give pyrazinamide with isoniazid and ethambutol or levofloxacin for patients with drug-susceptible TB and intolerance of rifampin. Pyrazinamide might also be given in the continuation phase for patients with multidrug-resistant TB, which constituted about 1% of all patients.
To evaluate the hepatotoxicity of pyrazinamide, cohort and nested case-control analyses were conducted on a cohort of 3,007 patients with active TB managed at government chest clinics under a TB control program with treatment started from January 1 through June 30, 2001. Cases included all patients with probable hepatotoxicity after starting treatment for 12 or more weeks. Hepatotoxicity was considered probable when serum alanine transaminase (ALT) exceeded three times the upper limit of normal (ULN) in the absence of alternative clinical diagnoses (21, 22). In case of multiple episodes of hepatotoxicity, only the first episode after starting treatment for 12 or more weeks was selected for comparison. Each case was compared with three sex- and age-matched control subjects selected at random from the rest of the cohort. Treatment regimens received by each case within 4 weeks preceding hepatotoxicity were compared with those of matched control subjects within a comparable 4-week period relative to the date of commencing treatment. To improve comparability of regimens, control subjects were included only if they had received at least 3 weeks of treatment within the comparable 4-week period. Treatment regimens were classified into the following groups based on the predominance (>50% in that period) of any of the different combinations of isoniazid, rifampin, and pyrazinamide, with or without streptomycin, ethambutol, ofloxacin, or levofloxacin, and regardless of dosing frequency: (1) standard continuation-phase regimens comprising isoniazid and rifampin; (2) pyrazinamide-containing regimens comprising isoniazid and/or rifampin; (3) others.
Cohort Analysis
Nested Case-Control Analysis SPSS version 10 (SPSS, Inc., Chicago, IL) was used for statistical analysis. Approval was obtained from the institutional review board of the Department of Health of Hong Kong for conducting the study. Patient consent was not necessary because our study was an observational study.
Descriptive From a cohort of 3,007 patients who commenced treatment from January 1 to June 30, 2001, hepatotoxicity occurred in 150 (5.0%) patients at any time including the 48 (1.6%) cases. Table 1 summarizes the demographic characteristics.
The 48 cases comprised 10 females and 38 males. The time distribution of hepatotoxicity was positively skewed with a median of 20.6 weeks, an interquartile range from 15.6 to 28.6 weeks, and a range from 12.1 to 40.0 weeks. ALT levels were distributed as follows: more than three times and up to five times the ULN in 22 cases (of whom 10 had hepatitis symptoms); more than five times and up to 10 times the ULN in 14 cases; and more than 10 times the ULN in 12 cases. Total bilirubin levels were elevated more than two times the ULN in seven cases, of whom six had serum ALT that was elevated more than five times the ULN. Two out of 12 cases with ALT exceeding 10 times the ULN died of liver failure, whereas all of the 36 cases with lower ALT survived. One fatal case was a 67-year-old Chinese male with chronic hepatitis B and hepatotoxicity preceded by treatment with pyrazinamide, isoniazid, and rifampin. The other fatal case was a 75-year-old Chinese male with unknown status for chronic hepatitis B and C and hepatotoxicity preceded by treatment with isoniazid and rifampin. A total of 16 cases received regimens containing predominantly pyrazinamide, isoniazid, and/or rifampin. The status of hepatitis symptoms was unknown for one case with serum ALT less than 10 times the ULN. Assuming this case was symptomatic, hepatitis symptoms were present in all (100%) of six cases with serum ALT exceeding 10 times the ULN, and only 2 (20%) of 10 cases with lower serum ALT. Treatment was suspended in all except two cases who continued treatment nonstop but without pyrazinamide. One male patient resumed treatment with the same pyrazinamide-containing regimen, whereas 12 cases resumed treatment without pyrazinamide. A total of three cases experienced hepatotoxicity again, all with ALT exceeding five times the ULN: two with chronic hepatitis B after resuming treatment without pyrazinamide, and one with unknown status for chronic hepatitis B and C after resuming treatment with the same pyrazinamide-containing regimen.
Cohort Analysis The case fatality rate of hepatotoxicity from 12 or more weeks after starting treatment was 1 (0.16%) out of 617 patients taking pyrazinamide, isoniazid, and/or rifampin, and 1 (0.04%) out of 2,327 patients taking isoniazid and rifampin. Both fatal cases were males older than 49 years. Restricting the analysis to males older than 49, the corresponding risks were 0.42% and 0.12%. Neither difference was significant statistically. Ordinary logistic regression analysis showed that hepatotoxicity from 12 or more weeks after starting treatment was significantly associated with males but not with age. The adjusted odds ratios (95% confidence interval) were 1.00 (0.98–1.01) for age and 2.1 (1.0–4.4) for sex.
Nested Case-Control Study
This is probably the first case-control study that has evaluated hepatotoxicity of pyrazinamide by comparing continuation-phase regimens containing pyrazinamide, isoniazid, and/or rifampin, with standard continuation-phase regimens comprising isoniazid and rifampin. By examining a large cohort of 3,007 subjects in a nested case-control design in the continuation phase, the estimated risk of hepatotoxicity from 12 or more weeks after starting treatment was 2.6% for pyrazinamide-containing regime\ns with an adjusted best-estimated odds of hepatotoxicity about 2.5 to 2.8 times that of standard continuation-phase regimens. By using incidence proportion instead of incidence rate, and pyrazinamide-containing regimens instead of arbitrary apportioning of risk to pyrazinamide in a multidrug regimen, our findings have corroborated the hepatotoxicity of pyrazinamide demonstrated by a cohort study (17) but without committing two major methodologic pitfalls that might overestimate the toxicity of pyrazinamide (18–20). It might be argued that hepatotoxicity of pyrazinamide from 12 or more weeks after starting treatment might not be extrapolated to the initial phase. This argument may not stand if the premise is confounding due to timing, which has already been addressed by examining treatment in comparable periods with reference to the date of starting treatment. Thus, not only do our findings have clinical implications for the management of some patients for whom pyrazinamide may be indicated in the continuation phase, but they may also explain the predominance of hepatotoxicity in the initial phase (17), when pyrazinamide is used much more frequently. This reinforces the need for prudence in rechallenging patients with pyrazinamide in the wake of severe drug-induced hepatitis (23). A potential problem with the interpretation of the results of any observational study that compares treatment regimens is the clinician's selection bias. Although clinicians would continue to prescribe pyrazinamide in the continuation phase in some patients for various reasons as described in the introduction, they would probably have done so more frequently among patients without previous hepatotoxicity, which might also trigger a search for other potential risk factors, such as hepatitis B and history of habitual drinking. Thus, the clinician's selection bias could cause an imbalanced distribution of potential risk factors of hepatotoxicity across patients receiving different continuation-phase regimens, and negatively confound the association between pyrazinamide-containing regimens and hepatotoxicity. This could be adjusted by either including previous hepatotoxicity as a covariate in regression analysis or excluding subjects with previous hepatotoxicity. Consistent findings by both approaches have substantiated the hepatotoxicity of pyrazinamide. The definition of hepatotoxicity is arbitrary and different definitions have been used (24). The British Thoracic Society and the American Thoracic Society/Centers for Disease Control and Prevention/Infectious Diseases Society of America have recommended that potentially hepatotoxic medications be stopped if serum ALT exceeds either five times the ULN (with or without symptoms) (6, 23) or three times the ULN with jaundice and/or hepatitis symptoms (23). Close monitoring of liver function is recommended when ALT is two or more times greater than normal (6). We defined hepatotoxicity by ALT elevation exceeding three times the ULN with no reference to symptoms, partly because symptoms would often be insensitive and nonspecific, and partly because of the importance of early identification of hepatotoxicity (22). More stringent definitions for hepatotoxicity should attenuate the strength of association between hepatotoxicity and pyrazinamide, but we still found appreciable odds of hepatotoxicity for pyrazinamide. Although we showed a significant association between hepatotoxicity and, respectively, chronic hepatitis B and C, the low testing rates plus possible diagnostic suspicion bias necessitated caution in interpretation. The unadjusted odds of hepatotoxicity among our study patients with hepatitis B relative to hepatitis B virus noncarriers were comparable with the strength of association found by another local study (25), which demonstrated a significantly higher incidence of liver dysfunction during anti-TB treatment among patients with hepatitis B in comparison with those without (34.9% vs. 9.4%, P < 0.001). The adjusted strength of association between hepatotoxicity and hepatitis C among our patients was comparable to the findings of another study, which demonstrated a fivefold risk of transaminase elevation of at least 120 U/L among hepatitis C–infected patients (26). Although a significant association between advanced age and hepatitis during anti-TB treatment has been well reported (17, 25, 27–29), this was largely demonstrated in studies that examined the entire treatment period. Our finding of a nonsignificant association between hepatotoxicity from 12 or more weeks after starting treatment and age suggests that the risk of hepatotoxicity of anti-TB treatment might not increase with age when treatment consists predominantly of isoniazid and rifampin. This is corroborated by the generally low risks of hepatotoxicity in different age groups treated with standard continuation-phase regimens. The association between hepatotoxicity and sex is contentious. Although several reports suggested an increased risk of hepatotoxicity among females (29–32), a number of studies showed no significant sex difference for hepatotoxicity (17, 28, 33, 34). We found the odds of hepatotoxicity among males were 2.1 times that of females after adjusting for age. This was corroborated by a considerable difference in the unadjusted risk of hepatotoxicity for pyrazinamide-containing regimens between males and females (3.2% vs. 0% among subjects aged 35–49 yr, and 5.0% vs. 0.6% among subjects > 49 yr, respectively). Human immunodeficiency virus infection, which was not included in the analysis, was unlikely to be a confounding factor because unlinked anonymous screening showed a low incidence of below 0.5% among patients receiving anti-TB treatment in Hong Kong (21). It was impossible for our study to examine the effect of dosing schedules on the hepatotoxicity of pyrazinamide-containing continuation-phase regimens because almost all study subjects received pyrazinamide-containing regimens daily. A randomized controlled trial, which involved intermittent and daily regimens with identical dosages of pyrazinamide, showed that initial-phase intermittent regimens containing isoniazid, rifampin, and pyrazinamide were significantly less hepatotoxic than their daily counterparts (35). It is uncertain whether intermittent regimens with substandard dosages of pyrazinamide may be less hepatotoxic at the possible expense of treatment efficacy. Death in 2 out of 12 cases with serum ALT exceeding 10 times the ULN contrasted with the absence of case fatalities among 36 cases with lower serum ALT. Furthermore, both fatal cases were older than 65 years and one had chronic hepatitis B treated with pyrazinamide-containing regimens preceding hepatotoxicity. Despite small numbers and lack of statistical significance, our findings may suggest caution against delay in the diagnosis of hepatotoxicity. If early diagnosis of hepatotoxicity may reduce the case fatality rate, and the sensitivity of hepatitis symptoms is in the order of only 20% for serum ALT below 10 times the ULN, treatment with pyrazinamide may necessitate periodic monitoring of liver biochemistry in the presence of other risk factors such as chronic hepatitis B or C, and possibly advanced age and male sex. In conclusion, adding pyrazinamide to isoniazid and rifampin increases the risk of hepatotoxicity appreciably.
The authors would like to thank all colleagues in the Hong Kong Tuberculosis and Chest Service (Hong Kong, China) for their contribution to the computerized tuberculosis registry.
Originally Published in Press as DOI: 10.1164/rccm.200802-355OC on April 3, 2008 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form February 28, 2008; accepted in final form April 2, 2008
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
2 test of trend). The overall risks of hepatotoxicity were 0.8% for standard regimens and 2.6% for pyrazinamide-containing regimens. 
lu A, Acican T. Side effects due to primary antituberculosis drugs during the initial phase of therapy in 1149 hospitalized patients for tuberculosis. Respir Med 2006;100:1834–1842.
