ORIGINAL ARTICLE RiskFactorsforLiverInjurywithanElevatedSerum Bilirubin Concentration Caused by Antituberculous Drugs Hideaki Kato 1, Nobuyuki Horita 1, Naoki Miyazawa 2, Takashi Yoshiyama 3, Atsuhisa Ueda 1 and Yoshiaki Ishigatsubo 1 Abstract Objective No studies have so far sufficiently investigated the risk factors for drug-induced liver injury (DILI) with an elevated serum bilirubin concentration. Methods We conducted a historical cohort study observing inpatients admitted to two hospitals in Japan. A decreased level of activities of daily living (ADL) was defined as a Barthel Index score of <80. The patients were treated with standard regimens under a directly observed treatment short-course strategy. Results The cohort of 356 patients comprised 244 men (68.5%) and 112 women (31.5%), with a mean age of 63.8±20.2 years. Compared with the patients who did not experience DILI with a bilirubin level of 2.0 mg/dl, the patients who experienced DILI with a bilirubin level of 2.0 mg/dl more often had a decreased level of ADLs, were more likely to suffer from chronic cardiac disease, had lower serum albumin levels and were less often treated with four-drug regimens involving pyrazinamide (PZA). In a logistic regression analysis in which these five factors acted as independent variables, a decreased level of ADLs was the strongest predictor for DILI with a bilirubin level of 2.0 mg/dl, with an odds ratio of 16.5 (95%CI: 1.7-159; p= 0.015), followed by chronic cardiac disease, with an odds ratio of 4.0 (95%CI: 1.2-12.6; p=0.020). Conclusion A decreased level of ADLs and chronic cardiac disease are strong risk factors for DILI with a bilirubin level of 2.0 mg/dl resulting from the use of antituberculous drugs. Physicians should pay close attention to the possibility of DILI with a bilirubin level of 2.0 mg/dl when treating tuberculosis patients with a decreased level of ADLs and/or chronic heart disease. Key words: adverse effects, antibacterial agents, bile conjugation, performance status, pulmonary tuberculosis (Intern Med 52: 2209-2214, 2013) () Introduction Although the global incidence of tuberculosis peaked around 2003 and now appears to be in slow decline, the worldwide incidence of new cases is estimated to be nine million every year and the disease is responsible for 1.5 million deaths each year (1). The current standard treatment regimens comprising three or four drugs cure most tuberculosis patients and facilitate reduction of transmission to others in the community (2). Unfortunately, however, antituberculous medications have various adverse effects, one of which is drug-induced liver injury (DILI). DILI is usually related to changes in treatment regimens, the use of less effective second-line treatments and prolonged hospitalization (3). The presence of an elevated serum bilirubin concentration or bile conjugation in patients with acute liver disease often provides clinically meaningful information. This is because an increased serum bilirubin concentration reflects greater histologic evidence of hepatocellular damage (4), a longer course of the disease (4) and a poorer prognosis (5-7). Many Department of Internal Medicine and Clinical Immunology, Yokohama City University Graduate School of Medicine, Japan, Department of Respiratory Medicine, Saiseikai Yokohamashi Nanbu Hospital, Japan and Division of Respiratory Medicine, Japan-Anti-Tuberculosis Association Fukujuji Hospital, Japan Received for publication March 9, 2013; Accepted for publication April 21, 2013 Correspondence to Dr. Nobuyuki Horita, nobuyuki_horita@yahoo.co.jp 2209
previous studies have evaluated the risk factors for DILI and how to identify patients at risk of developing DILI. However, to the best of our knowledge, no studies have investigated the specific risk factors for DILI with an elevated serum bilirubin concentration. Therefore, in this study, we researched the risk factors for DILI with an elevated serum bilirubin concentration. Ethics statement Materials and Methods We conducted a historical cohort study reviewing inpatients admitted to two hospitals familiar with tuberculosis in different prefectures. This study was approved by the Institutional Review Board of Yokohama City University Hospital and Institutional Review Board of Fukujuji Hospital. These boards waived the requirement for informed consent due to the observational nature of the study design and the protection of the patients anonymity. Patients The medical charts of the patients at the two hospitals [Yokohama City University Hospital, Yokohama, Japan (638 beds, including 16 isolation beds for tuberculosis) and Fukujuji Hospital, Tokyo, Japan, a community-based teaching hospital (339 beds, including 60 isolation beds for tuberculosis)] were reviewed. All consecutive patients admitted to the tuberculosis isolation wards of the two hospitals with a primary diagnosis of lung tuberculosis were screened using the following inclusion criteria. The patients treated at the Yokohama City University Hospital were reviewed between January 2008 and November 2011, whereas those treated at Fukujuji Hospital were reviewed between January 2009 and December 2009, on an admission date basis. The inclusion criteria were as follows: (i) lung tuberculosis confirmed with positive sputum smears and cultures, (ii) newly diagnosed lung tuberculosis (i.e., patients who had already started treatment at another hospital were excluded), (iii) an age 20 years and (iv) treatment with one of the two standard regimens mentioned below with full susceptibility. Measurements Baseline patient data were obtained as of the date of admission. The presence of bilateral infiltration was evaluated on X-rays according to the agreement of at least four pulmonologists who were in charge of each patient, independent of this study. The existence of bilaterality was always noted, according to the recommendations of the Japanese guidelines for the assessment of bilateral infiltration on X- rays. Habitual alcohol drinkers were defined as patients who consumed alcohol 3 days a week. The level of activities of daily living (ADL) was measured based on the Barthel Index (range: 0-100 points), which comprises the following variables: the ability to feed oneself; the ability to get out of a wheelchair; the ability to perform activities related to personal hygiene, such as going to the toilet and bathing; the ability to walk on level ground; the ability to climb stairs; the ability to dress oneself; and the ability to maintain bowel and bladder control (8). In this study, a decreased level of ADL was defined as a Barthel Index score of <80, while a normal level of ADLs was defined as a Barthel Index score of 80 (9). We used this cutoff value based on the findings of a previous article (8). We also evaluated the cutoff value using a receiver operating characteristic curve. Chronic heart disease, renal disease, cerebrovascular disease and musculoskeletal disease (compression fractures of the spine followed by fractures of the extremities were most often observed) are not usually considered risk factors for DILI. However, we evaluated these factors in this study, as these diseases are often related to a decreased level of ADLs. Treatments A directly observed treatment short-course strategy was provided immediately after admission to the two tuberculosis isolation wards reviewed in this study. One of the following two daily combined regimens was used for the first two months: (i) isoniazid (INH; 5 mg/kg/day; max 300 mg/ day), rifampicin (RFP; 10 mg/kg/day; max 600 mg/day), pyrazinamide (PZA; 25 mg/kg/day; max 1,500 mg/day) and ethambutol (15-20 mg/kg/day; max 1,000 mg/day); or (ii) INH, RFP and ethambutol at the same dosages. The fourdrug regimen with PZA was used as a first-line regimen, while the three-drug regimen without PZA was primarily prescribed for patients with possible decreased drug tolerance. The treatment choice was made by the physicians treating each patient (2). Patients treated with standard regimens including streptomycin were excluded from this study because streptomycin regimens were usually avoided in order to remove the burden of intramuscular injection and were used in a very limited number of patients in our hospital. Outcomes We reviewed the treatment course of each patient for 60 days because PZA and ethambutol were withheld after two of months of treatment and because previous studies have shown that most cases of DILI resulting from antituberculous drugs occur within the first two months (10-13). DILI was defined as the change or withdrawal of treatment due to liver injury caused by antituberculous agents judged according to the agreement of at least four pulmonologists in charge of each patient. The judgment was principally based on criteria described in the guidelines (3): a serum transaminase concentration of more than three times the upper limit of the normal range (120 IU/L) with jaundice and/or symptoms of hepatitis; a serum transaminase concentration of more than five times the upper limit of the normal range (200 IU/L); or a total bilirubin level of 2.0 mg/dl, regardless of the level of transaminase. Elevation of the transaminase level that was not accompanied by a change in or 2210
Table 1. Patient Characteristics on Admission n = 356 Age (year) 63.8 ± 20.2 Sex (female) 112 (31.5%) Normal activities of daily living 215 (60.4%) Barthel Index 100 186 (52.2%) Barthel Index 80 95 29 (8.1%) Decreased activities of daily living 141 (39.6%) Barthel Index 60 75 19 (5.3%) Barthel Index 40 55 20 (5.6%) Barthel Index 20 35 11 (3.1%) Barthel Index 0 15 91 (25.6%) Chronic cardiac disease 37 (10.4%) Chronic pulmonary disease 26 (7.3%) Chronic liver disease 17 (4.8%) Hepatitis B virus infection 4 (1.1%) Hepatitis C virus infection 16 (4.5%) Chronic renal disease 26 (7.3%) Diabetes 81 (22.8%) Active malignancy 23 (6.4%) Human immunodeficiency virus infection 2 (0.6%) Cerebrovascular disease 73 (20.5%) Musculoskeletal disease 56 (15.7%) Habitual alcohol drinker 87 (24.0%) Use of immunosuppressant 27 (7.6%) Number of systemic medications 2.7 ± 2.7 Aspartate transaminase (AST) (IU/L) 35.6 ± 41.5 Alanine transaminase (ALT) (IU/L) 26.8 ± 32.8 Albumin (g/dl) 3.1 ± 0.8 Bilateral infiltration on X-ray 245 (68.8%) Four-drug regimen with pyrazinamide 289 (81.2%) Three drugs regimen without pyrazinamide 67 (18.8%) withdrawal of treatment was not regarded as DILI for the purpose of this study. Although elevated serum transaminase concentrations can theoretically be explained by other causes, such as biliary calculus or recent alcohol consumption, these possibilities were discounted by checking the patients charts. We classified DILI into DILI with a bilirubin level of 2.0 mg/dl and DILI with a bilirubin level of < 2.0 mg/dl using a cutoff value of a serum bilirubin concentration of 2.0 mg/dl measured on the day of DILI diagnosis (14). The observation was censored if one of following events occurred: (i) withdrawal or change of the treatment regimen due to a cause other than DILI, (ii) discharge of the patient alive and (iii) death during admission. Statistical analyses Figure 1. Receiver operating characteristic curve predicting drug-induced liver injury (DILI) with a bilirubin level of 2.0 mg/dl according to the Barthel Index. Area under the curve: 0.759. The Barthel Index is a scale used to assess activities of daily living with 5-point intervals, wherein 100 indicates the best level of activities of daily living. The arrows with numbers indicate the cutoff values. A cutoff value between 75 and 80 yielded the best Youden s index (sensitivity + specificity - 1) of 0.57, which indicates that this cutoff is the most useful. For the univariate analysis, we used the Wilcoxon ranksum test for continuous variables and the chi-square test for contingency (with the Yates correction, if needed). A logistic regression analysis was used to predict bivariate variables. Receiver operating characteristics curves and Youden s Index were used to evaluate the cutoff values. A rejection region of 0.05 was adopted. Continuous variables are presented as the arithmetic mean ± SD. The data analyses were performed using the Excel Toukei (SSRI, Tokyo, Japan) and GraphPad Prism ver. 5.0 (GraphPad Software Inc., San Diego, CA, USA) software programs. Results The cohort of 356 patients comprised 166 patients (46.6%) from Yokohama City University Hospital and 190 patients (53.4%) from Fukujuji Hospital. There were 244 men (68.5%) and 112 women (31.5%), with a mean age of 63.8±20.2 years. Table 1 provides a summary of the patient characteristics. Of the 356 patients, 215 (60.4%) had a normal level of ADLs (Barthel Index 80) and 141 (39.6%) had a decreased level of ADLs (Barthel Index <80) (Table 1). A cutoff value between 75 and 80 yielded the best Youden s index (sensitivity + specificity-1) of 0.57, which indicates that this cutoff is the most useful (Fig. 1). Age and the incidence rates of chronic heart disease, chronic respiratory disease, chronic renal disease, cerebrovascular disease and musculoskeletal disease were significantly different between the patients with a normal level of ADLs and those with a decreased level of ADLs. In our cohort of 356 patients, 316 (88.8%) did not experience DILI, 24 (6.7%) experienced DILI with a bilirubin level of <2.0 mg/dl and 16 (4.5%) experienced DILI with a bilirubin level of 2.0 mg/dl. The average duration from admission to occurrence of DILI with bile conjugation in the 16 patients was 16±11 days after admission. Of these 16 patients, eight were discharged alive after negative infectivity was confirmed and eight died during admission for the following reasons according to the death certificates: six due to tuberculosis, one due to liver cirrhosis and one due to pneumonia. The duration between the onset of DILI and death in the eight patients who died was 42±43 days on average. Compared with the 24 patients who experienced DILI 2211
Table 2. Comparison of Baseline Characteristics between Patients who Did and Did Not Experience Drug-induced Liver Injury (DILI) with Bilirubin 2.0 mg/dl DILI with bilirubi 2.0 mg/dl (+) (-) Comparison n = 16 n = 340 p Age (year) 70.1 ± 17.6 63.5 ± 20.3 0.245 Sex (female) 6 (37.5%) 112 (32.9%) 0.705 Decreased activities of daily living 15 (93.8%) 141 (41.5%) < 0.001 Chronic cardiac disease 7 (43.8%) 37 (10.9%) < 0.001 Chronic pulmonary disease 1 (6.3%) 26 (7.7%) 0.782 Chronic liver disease 3 (18.8%) 17 (5.0%) 0.075 Hepatitis B virus infection 0 (0.0%) 4 (1.2%) 0.437 Hepatitis C virus infection 2 (12.5%) 16 (4.7%) 0.420 Chronic renal disease 2 (12.5%) 26 (7.7%) 0.818 Diabetes 2 (12.5%) 81 (23.8%) 0.457 Active malignancy 0 (0.0%) 23 (6.8%) 0.579 Human immunodeficiency virus infection 0 (0.0%) 2 (0.6%) 0.160 Cerebrovascular disease 5 (31.3%) 73 (21.5%) 0.539 Musculoskeletal disease 2 (12.5%) 56 (16.5%) 0.941 Habitual alcohol drinker 2 (12.5%) 87 (25.6%) 0.376 Use of immunosuppressant 3 (18.8%) 27 (7.9%) 0.289 Number of systemic medications 3.3 ± 2.9 2.7 ± 2.6 0.343 Aspartate transaminase (IU/L) 45.1 ± 25.7 35.1 ± 42.1 0.002 Alanine transaminase (IU/L) 30.4 ± 27.4 26.6 ± 33.0 0.231 Albumin (g/dl) 2.5 ± 0.5 3.1 ± 0.8 < 0.001 Bilateral infiltration on X-ray 14 (87.5%) 245 (72.1%) 0.285 Four-drug regimen with pyrazinamide 9 (56.3%) 289 (85.0%) 0.007 Wilcoxon rank sum test and chi-square test were used. Table 3. Logistic Regression for Predicting Drug-induced Liver Injury (DILI) with Bilirubin 2.0 mg/dl Odds ratio [95%CI] p Decreased activities of daily living 16.5 [1.7 159] 0.015 Chronic cardiac disease 4.0 [1.2 12.6] 0.020 Aspartate transaminase (AST) (10 IU/L increase) 0.99 [0.89 1.1] 0.890 Albumin (1 g/dl increase) 0.90 [0.36 2.3] 0.828 Four-drugs regimen with pyrazinamide 0.76 [0.24 2.4] 0.646 We performed logistic regression analysis to predict DILI with bilirubin mg/dl. Parameters which were significantly different between patients who experienced and those who did not experience DILI with bil 0 mg/dl (Table 3) are included in this model. with a bilirubin level of <2.0 mg/dl, the 16 patients who experienced DILI with a bilirubin level of 2.0 mg/dl exhibited a higher rate of a decreased level of ADLs (93.4% vs 54.2%, p=0.020) and low serum albumin levels (2.5±0.5 mg/dl vs. 3.1±0.9 mg/dl, p=0.037). Other baseline characteristics were not significantly different between the two groups. The average duration from the start of treatment with antituberculous agents to DILI diagnosis was 16±11 days among the patients who experienced DILI with a bilirubin level of 2.0 mg/dl and 22±22 days among those who experienced DILI with a bilirubin level of <2.0 mg/dl, which was not significantly different. Compared with the 340 patients who did not experience DILI with a bilirubin level of 2.0 mg/dl, the 16 patients who experienced DILI with a bilirubin level of 2.0 mg/dl more often had a decreased level of ADLs, were more likely to suffer from chronic cardiac disease, had higher aspartate transaminase (AST) and lower serum albumin levels and were less often treated with four-drug regimens involving PZA (Table 2). In a logistic regression analysis in which these five factors acted as independent variables, a decreased level of ADLs was found to be the strongest predictor of DILI with a bilirubin level of 2.0 mg/dl, with an odds ratio of 16.5 (95%CI: 1.7-159; p=0.015), followed by chronic cardiac disease, with an odds ratio of 4.0 (95%CI: 1.2-12.6; p=0.020) (Table 3). The distribution chart of the levels of bilirubin and AST on diagnosis of DILI is shown in Fig. 2. This figure shows that a patient with a normal level of ADLs rarely experienced DILI with a bilirubin level of 2.0 mg/dl and that a very limited number of patients experienced elevation of both AST and bilirubin (very few plots are located in the 2212
Figure 2. Distribution chart of the levels of bilirubin and aspartate transaminase (AST) on diagnosis of drug-induced liver injury (DILI). This figure shows that the patients with a normal level of ADLs rarely experienced DILI with a bilirubin level of 2.0 mg/dl and that a very limited number of patients experienced elevation of both AST and bilirubin. upper right part of the figure). No parameters listed in Table 3 differed between the 332 patients who did not experience DILI with a bilirubin level of <2.0 mg/dl and the 24 patients who experienced DILI with a bilirubin level of <2.0 mg/dl. Discussion This is the first study to investigate the risk factors for DILI with an elevated serum bilirubin concentration caused by antituberculous drugs. We confirmed that a decreased level of ADLs and chronic cardiac disease are risk factors for DILI with a bilirubin level of 2.0 mg/dl. Several risk factors are known to increase the probability of DILI resulting from antituberculous agents, including senility, a female gender, chronic alcohol use, an abnormal baseline transaminase level, malnutrition, extensive disease on X-rays, PZA-containing regimens, chronic liver disease, a decreased level of ADLs, coinfection with hepatitis B and C virus and the human immunodeficiency virus (3, 8). In this study, we found that a decreased level of ADLs and chronic cardiac disease are strongly related to DILI with a bilirubin level of 2.0 mg/dl. INH and RFP are two key drugs used in the treatment of tuberculosis and both often cause DILI. The bilirubin concentration sometimes provides information regarding which antituberculous drug has induced DILI. INH and RFP cause DILI via different mechanisms, although the mechanisms underlying the liver damage resulting from these two drugs remain unexplained. Direct INH toxicity or toxicity from an INH metabolite is most likely responsible for hepatocyte death, which results in elevation of the level of transaminase. The histopathologic appearance of DILI caused by INH toxicity closely resembles that of viral hepatitis (14, 15). RFP is metabolized in the liver and excreted through the bile ducts. RFP is known to interact with the metabolism of other drugs (e.g., oral contraceptives), especially via enterohepatic circulation (16). In an animal model, the effects of RFP on the bile flow were biphasic: low-dose RFP increased the bile duct flow, while high-dose RFP decreased the bile duct flow (17). In this scenario, high-dose RFP increases the serum concentrations of RFP and serum bilirubin, which may result in a vicious cycle. In fact, DILI caused by RFP is often accompanied by an elevated serum bilirubin concentration (14). Given this, we hypothesized that cytotoxic DILI, primarily caused by INH, and bile congestion DILI, primarily caused by RFP, have different risk factors. In fact, DILI with and without bilirubin elevation had different risk factors in the current study. The impact of a decreased level of ADLs on DILI has been previously reported (8). However, it is as yet to be determined why DILI with a bilirubin level of 2.0 mg/dl occurs almost exclusively in patients with a decreased level of ADLs. Patients with a decreased level of ADLs may have either an impaired catabolism or may be excreting RFP, while they may also be vulnerable to RFP at the same serum concentration. In addition, patients with a decreased level of ADLs may have increased re-uptake of RFP via the enterohepatic circulation. However, there is, as yet, insufficient evidence to support these hypotheses. Chronic cardiac disease is not usually regarded to be a risk factor for DILI caused by antituberculous drugs. However, any cause of right-sided heart failure can result in hepatic congestion, including constrictive pericarditis, mitral stenosis, tricuspid regurgitation and cor pulmonale. Some studies have reported that right-sided heart failure leads to jaundice and biliary obstruction (18-20). These mechanisms may be associated with DILI with a bilirubin level of 2.0 mg/dl. The strength of our study is that it detected the relationship between cardiac disease and DILI supported by the senility of the cohort and the mean age of 63.8±20.2 years. More than 10% of our patients had chronic heart disease and were possibly at risk of right-sided heart failure. Although we know that RFP, a decreased level of ADLs and chronic cardiac disease are often associated with DILI with a bilirubin level of 2.0 mg/dl, clinicians cannot avoid using RFP in the treatment of tuberculosis patients with a decreased level of ADLs and/or chronic cardiac disease because RFP is regarded as being one of the most effective antituberculous agents (2). Therefore, clinicians must be on the lookout for DILI with a bilirubin level of 2.0 mg/dl when treating patients with a decreased level of ADLs and/ or chronic cardiac disease. The current study is associated with some limitations. First, this study was designed as a historical cohort study. Nonetheless, we believe that the study s strengths reside in the clear presentation of the inclusion criteria and outcomes. Second, our cohort did not include a sufficient number of patients with established risk factors for DILI, such as chronic liver disease, chronic renal disease and coinfection with human immunodeficiency virus and the hepatitis B and C viruses (Table 1). These risk factors were not associated with DILI with a bilirubin level of 2.0 mg/dl in this study. Nonetheless, we cannot deny the impact of these factors on 2213
DILI with a bilirubin level of 2.0 mg/dl in our study due to the insufficient power resulting from the small number of patients with these factors. Similarly, no parameters listed in Table 3 differed between the 332 patients who did not experience DILI with a bilirubin level of <2.0 mg/dl and the 24 patients who experienced DILI with a bilirubin level of <2.0 mg/dl. This is most likely due to the small sample size. In conclusion, a decreased level of ADLs and chronic cardiac disease are strong risk factors for the development of DILI with a bilirubin level of 2.0 mg/dl caused by the use of antituberculous drugs. Physicians should pay close attention to the possibility of DILI with a bilirubin level of 2.0 mg/dl when treating tuberculosis patients with a decreased level of ADLs and/or chronic heart disease. The authors state that they have no Conflict of Interest (COI). Acknowledgement We appreciate the valuable assistance of Dr. Izuki Amano in the data collection. References 1. Organization. Global tuberculosis control 2011. avairable from: http://www.who.int/tb/publications/global_report/en/ 2011. 2. Neff M, ATS, CDC, IDSA. ATS, CDC, and IDSA update recommendations on the treatment of tuberculosis. Am Fam Physician 68: 1854, 1857-1858, 1861-1862, 2003. 3. Saukkonen JJ, Cohn DL, Jasmer RM, et al. An official ATS statement: hepatotoxicity of antituberculosis therapy. Am J Respir Crit Care Med 174: 935-952, 2006. 4. Sherlock S. Biochemical investigations in liver disease; some correlations with hepatic histology. J Pathol Bacteriol 58: 523-544, 1946. 5. Hardison WG, Lee FI. Prognosis in acute liver disease of the alcoholic patient. N Engl J Med 275: 61-66, 1966. 6. Kim WR, Wiesner RH, Therneau TM, et al. Optimal timing of liver transplantation for primary biliary cirrhosis. Hepatology 28: 33-38, 1998. 7. Devarbhavi H, Singh R, Patil M, Sheth K, C KA, Balaraju G. Outcome and determinants of mortality in 269 patients with combination anti-tuberculosis drug-induced liver injury. J Gastroenterol Hepatol 28: 161-167, 2013. 8. Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J 14: 61-65, 1965. 9. Horita N, Miyazawa N, Yoshiyama T, et al. Decreased activities of daily living is a strong risk factor for liver injury by antituberculosis drugs. Respirology 18 (3): 474-479, 2013. 10. Ormerod LP, Horsfield N. Frequency and type of reactions to antituberculosis drugs: observations in routine treatment. Tuber Lung Dis 77: 37-42, 1996. 11. Yee D, Valiquette C, Pelletier M, Parisien I, Rocher I, Menzies D. Incidence of serious side effects from first-line antituberculosis drugs among patients treated for active tuberculosis. Am J Respir Crit Care Med 167: 1472-1477, 2003. 12. Teleman MD, Chee CB, Earnest A, Wang YT. Hepatotoxicity of tuberculosis chemotherapy under general programme conditions in Singapore. Int J Tuberc Lung Dis 6: 699-705, 2002. 13. Clinical trial of three 6-month regimens of chemotherapy given intermittently in the continuation phase in the treatment of pulmonary tuberculosis. Singapore Tuberculosis Service/British Medical Research Council. Am Rev Respir Dis 132: 374-378, 1985. 14. Thompson NP, Caplin ME, Hamilton MI, et al. Anti-tuberculosis medication and the liver: dangers and recommendations in management. Eur Respir J 8: 1384-1388, 1995. 15. Black M, Mitchell JR, Zimmerman HJ, Ishak KG, Epler GR. Isoniazid-associated hepatitis in 114 patients. Gastroenterology 69: 289-302, 1975. 16. Bolt HM. Interactions between clinically used drugs and oral contraceptives. Environ Health Perspect 102 Suppl 9: 35-38, 1994. 17. Oliven A, Bassan HM. Effects of rifampin and isoniazid on the isolated perfused rat liver. Chemotherapy 32: 159-165, 1986. 18. Cohen JA, Kaplan MM. Left-sided heart failure presenting as hepatitis. Gastroenterology 74: 583-587, 1978. 19. Logan RG, Mowry FM, Judge RD. Cardiac failure simulating viral hepatitis. Three cases with serum transaminase levels above 1,000. Ann Intern Med 56: 784-788, 1962. 20. Moussavian SN, Dincsoy HP, Goodman S, Helm RA, Bozian RC. Severe hyperbilirubinemia and coma in chronic congestive heart failure. Dig Dis Sci 27: 175-180, 1982. 2013 The Japanese Society of Internal Medicine http://www.naika.or.jp/imonline/index.html 2214