A total of 1,487 consecutive patients (> 18 years of age) who commenced antituberculous therapy at Gyeongsang National University Hospital between February 2009 to January 2013 were enrolled. Most patients received a 2-month treatment regimen of isoniazid, rifampicin, ethambutol, and pyrazinamide followed by isoniazid, rifampicin, and ethambutol for 4 months. All patients had received a fixed dose of first-line antituberculous drug daily according to World Health Organization recommendations based on weight [16]. Demographic and clinical data including age, sex, body mass index (BMI), concomitant diseases, infected organs with tuberculosis, previous history of tuberculosis, and alcohol consumption were retrospectively obtained from a review of medical records. Exclusion criteria were: (1) level of aspartate aminotransferase (AST) or ALT at baseline > 2 times upper limit of normal (ULN; n = 7); (2) loss to follow-up within 2 months of treatment initiation (n = 220); (3) inadequate date of laboratory testing at baseline or 2 to 4 weeks after treatment initiation (n = 86); (4) diagnosis of latent tuberculosis, nontuberculous mycobacterium infection, multidrug resistant tuberculosis, or other disease (n = 127); and (5) discontinuation due to other adverse effects except hepatotoxicity (n = 16). Finally, 1,031 patients were recruited. The Institutional Review Board of Gyeongsang National University Hospital reviewed and approved this study.

ADIH was defined as follows based on previous studies [3,17]. Other causes of elevated liver function tests were excluded. In cases with normal level of baseline AST and ALT before treatment initiation, liver transaminase levels were exceeded 3 times of upper normal range with symptoms of hepatitis or exceeded 5 times of upper normal range without symptoms. In cases with abnormal level of basal AST or ALT before treatment initiation, the level(s) after initiation of treatment was double the basal level. Level of total bilirubin was 2 times higher than ULN at the normal baseline or basal level at the abnormal baseline. Patients developing hepatotoxicity were categorized based on the timing of occurrence of ADIH (≤ 30-day group, ADIH detected within 30 days; > 30-day group, ADIH detected beyond 30 days). In subgroup analysis, patients developing hepatotoxicity were categorized according to occurrence within 7 days (≤ 7-day group) and beyond 7 days (> 7-day group).

Liver function was ascertained at baseline and every 2 to 4 weeks after treatment initiation. Unplanned visits and laboratory tests due to adverse events were performed according to patient need. Elevated AST or ALT values prompted more frequent visits. In cases of ADIH, all drugs were stopped and a step-wise rechallenge was begun after complete resolution of hepatotoxicity. During the presentation of hepatotoxicity, highest AST or ALT level and time of peak level were corrected. The prognosis of patients who successfully recommenced first-line drugs without further development of hepatitis, who began alternative regimens, or who terminated treatment were determined. Mean time to recovery was defined as the time interval between the onset of hepatitis and complete normalization.

All statistical analyses were performed with SPSS version 18.0 (SPSS Inc., Chicago, IL, USA). For between-group comparisons, the Mann-Whitney U test was used for continuous variable, with the chi-square test or Fisher exact test (for situations with small frequencies) was used for categorical variables. Risk factors associated with ADIH were determined by logistic regression analysis. The risk was expressed by calculating the odds ratio (OR) and 95% confidential interval (CI). A p < 0.05 indicated statistical significance.

Mean age ± standard deviation of the 1,031 patients (608 males, 59.0%) was 55.5 ± 18.3 years (Table 1). One hundred and eight patients (10.5%) developed ADIH a mean of 39.6 ± 43.7 days after treatment initiation. Baseline levels of AST and ALT in patients with ADIH were significantly higher than those in patients without ADIH. Baseline level of total bilirubin in patients with ADIH was significantly lower than that in patients without ADIH. Incidence of concomitant hepatitis C virus infection was significantly increased in patients with ADIH. However, the other variables were not significantly different. The most common infected site was the lung (68.9%), followed by pleural space (10.7%), lymph node (5.2%), gastrointestinal tract (4.1%), central nervous system (1.9%), vertebral joint (1.4%), and pericardium (1.2%) (Table 2). Thirty-one patients were simultaneously diagnosed with pulmonary and extrapulmonary tuberculosis.

Among 108 patients who developed ADIH, 73 patients (67.5%) developed hepatotoxicity within 30 days (Fig. 1). The mean onset time of hepatotoxicity of the ≤ 30-day group and > 30-day group was 15.6 ± 10.1 days and 89.7 ± 44.4 days, respectively (p < 0.001). The mean peak AST, ALT, and bilirubin level was 301.4 ± 350.0 U/L, 308.2 ± 319.9 U/L, and 1.5 ± 1.7 mg/mL, respectively, in the ≤ 30-day group and 241.5 ± 393.5 U/L, 211.5 ± 230.1 U/L, and 2.4 ± 3.4 mg/mL, respectively, in the > 30-day group. Except for the peak ALT level, there were no significant differences in the other variables between the two groups (Table 3).

Twenty-eight patients (25.9%) developed hepatotoxicity within 7 days. The mean hepatitis onset time of the ≤ 7-day group and > 7-day group was 5.7 ± 2.1 days and 51.4 ± 45.1 days, respectively (p < 0.001). In the ≤ 7-day group, rates of extrapulmonary tuberculosis were significantly higher than those of the > 7-day group (53.6% vs. 31.3%, p = 0.035) (Table 4). Among the 40 patients with extrapulmonary tuberculosis, 15 patients (37.5%) developed hepatotoxicity within 7 days. There were no significant differences of peak AST and ALT levels between the ≤ 7- and > 7-day groups. But, the mean AST and ALT levels at baseline were significantly higher in the ≤ 7-day group compared to the > 7-day group (34.1 ± 14.4 vs. 28.9 ± 14.0, p = 0.036; 30.1 ± 17.2 vs. 21.7 ± 12.4, p = 0.013, respectively) (Table 3). However, there were no significant differences in age, sex, BMI, pyrazinamide (PZA) combination, total bilirubin level at baseline and peak time, concomitant hepatitis B virus and hepatitis C virus infection, alcohol consumption, diabetes, and treatment strategy between the two groups.

As shown in Table 5 serum ALT > 40 IU/L (OR, 4.442; 95% CI, 2.594 to 7.604; p < 0.001), serum AST > 40 IU/L (OR, 2.609; 95% CI, 1.361 to 5.003; p = 0.004), serum bilirubin > 1.2 mg/dL (OR, 2.270; 95% CI, 1.060 to 4.861), and presence of anti-hepatitis C virus (OR, 5.701; 95% CI, 2.571 to 12.640; p < 0.001) were significantly associated with development of ADIH in univariate analysis. In multivariate analysis, serum ALT > 40 IU/L (OR, 2.995; 95% CI, 1.580 to 5.680; p = 0.001) and presence of anti-hepatitis C virus (OR, 4.204; 95% CI, 1.822 to 9.700; p = 0.001) were independent risk factors for development of ADIH. There was no significant difference in time to recovery between the ≤ 30- and the > 30-day groups (40.8 ± 34.4 and 47.4 ± 49.3 days, p = 0.755). In the ≤ 30-day group, 50 patients (68.1%) successfully completed treatment with first-line drugs without further hepatotoxicity development, and 23 patients (31.9%) commenced treatment with alternative drugs or terminated treatment. In the > 30-day group, 20 patients (57.1%) successfully completed treatment with first-line drugs, and 15 patients (42.9%) commenced alternative drug treatment or terminated treatment (Table 3). There was a significant difference in modes of management in both groups (p = 0.033). In the ≤ 7-day group, the proportion of patients who maintained first-line antituberculous drugs without withdrawal or termination was higher than that with patients in whom ADIH occurred later (81.5% vs. 58.7%, p = 0.033) (Table 4) and ADIH occurred between 8 and 14 days (81.5% vs. 18.5%, p = 0.036, data not shown). There were no cases of acute liver failure requiring liver transplantation or death.