Characteristics and risk factors for antituberculosis drug-induced liver injury in a cohort of patients with cirrhosis in a tertiary referral university teaching hospital in Thailand

An initial medical record review of 221 patients potentially eligible for the study was performed. We excluded 88 patients who had insufficient evidence supporting a diagnosis of cirrhosis, i.e., no radiological imaging results available for review, and 19 patients who lacked sufficient proof for a diagnosis of tuberculosis. The medical records of 114 cirrhotic patients with tuberculosis were subsequently reviewed carefully, and 50 further patients were excluded for a variety of reasons: 7 had died before starting ATDs; 5 had septic shock during the first 30 days after ATD initiation; 5 were given ATD regimen not containing isoniazid, rifampin, or pyrazinamide; 2 were exposed to ATDs for less than 3 days. Another 31 patients who had insufficient information in their medical records were also excluded (Figure 1). The final cohort comprised 64 cirrhotic patients with tuberculosis, of whom 50 and 14 were diagnosed with tuberculosis by microbiological tests and combined clinical presentation and radiological imaging, respectively (Table 1).

Figure 1

Of the 64 cirrhotic patients included in the analysis, 45 (70%) were male with a mean (SD) age of 54.4 (13.2) years. The diagnosis of cirrhosis was made by the characteristics of cirrhosis and portal hypertension found in either ultrasound (n = 32, 50%), CT (n = 19, 30%), or MR (n = 13, 2%) images. Sixteen (27%), 32 (53%), and 12 (20%) patients had CTP class A, B, and C cirrhosis, respectively; 4 patients lacked sufficient information to determine the CTP class. The most common cause of cirrhosis was alcohol (n = 21, 33%), followed by HBV and viral hepatitis C infection (HCV) (n = 13, 20%). The most common site of tuberculosis infection was lung and/or pleura (n = 40, 64%), followed by disseminated tuberculosis with 2 or more infected organs (n = 17, 27%), and extrapulmonary tuberculosis (n = 6, 10%; 1 patient with missing data on the site of tuberculosis infection). No patients received concomitant cancer chemotherapy or immunosuppressive drugs. Initial ATDs were prescribed to each individual patient based on the judgment of the physician. The ATD regimen comprised 3 hepatotoxic drugs (isoniazid, rifampin, and pyrazinamide) was given to 8/16 (50%) of the CTP patients in class A, 22/32 (69%) of patients in class B, and 5/12 (42%) of patients in class C patients (Table 2). The numbers of cirrhotic patients in CTP classes A, B, and C who received the regimen with 2 hepatotoxic drugs (isoniazid and rifampin) and 1 hepatotoxic drug (either isoniazid or rifampin) are summarized in Table 2.

There were 6 (9.4%) cirrhotic patients who developed ATDILI (Tables 3 and 4). The median duration from the initiation of ATDs to the development of ATDILI was 14 days (range: 6–66). All 6 patients had elevated AST or ALT levels with the median AST/ALT level of 47/24 at baseline and 205/112 when ATDILI was detected. The detailed information for each patient developing ATDILI is summarized in Table 5. Interestingly, there were more patients with HIV infection in the ATDILI group than in the non-ATDILI group (50% vs. 9%; P = 0.02). In addition, white blood cell count was significantly higher in the ATDILI group, compared with the non-ATDILI group, with median (range) of 12,010 (7,010–23,310)/mL vs. 6,375 (1,610–18,900)/mL (P = 0.01). All patients who developed ATDILI had CTP class B cirrhosis. However, there was no significant difference in the CTP score between the ATDILI group and the non-ATDILI group, with a median (range) of 8 (7–9) and 7 (5–13), respectively (P = 0.66). All patients (6/6, 100%) in the ATDILI group received ATD regimen containing 3 hepatotoxic drugs (isoniazid, rifampicin, and pyrazinamide), whereas only 31/58 (53%) patients in the non-ATDILI group received 3 hepatotoxic ATDs and 4 (7%) and 23 (40%) patients in the non-ATDILI group received 1 and 2 hepatotoxic ATDs, respectively, P = 0.09.

There were no significant differences in sex, age, etiology of cirrhosis, or median score of CTP and MELD between the 2 groups. Similarly, baseline laboratory results including serum aspartate aminotransferase (AST), alanine transaminase (ALT), tuberculosis, hemoglobin, albumin, and international normalized ratio (INR) were not different between the groups.

Among the 6 patients (4 non-HIV and 2 HIV-infected) who developed ATDILI, the concomitant medications used by the 4 non-HIV patients included insulin Mixtard, tamsulosin, tenofovir, lamivudine, and levothyroxine. These drugs are not considered as hepatotoxic [24]. However, the other 2 HIV-infected patients took nevirapine, stavudine, and efavirenz, which are known to be hepatotoxic.

The present study sought to determine risk factors for ATDILI in cirrhotic patients. Our finding suggests that the 3 hepatotoxic ATD regimens containing a pyrazinamide increase the risk for ATDILI in patients with cirrhosis. HIV infection and elevated white blood cell count also appear to be associated with an increased risk of developing ATDILI in patients with cirrhosis.

Making an accurate diagnosis of ATDILI in cirrhotics is challenging, because diagnostic criteria for ATDILI specifically for cirrhotic patients have not yet been established. The diagnostic criteria for ATDILI currently stated in most guidelines are based on “normal” liver enzyme levels at baseline [24, 25]. However, this is not always applicable to cirrhotic patients, because they tend to have alternating and elevated liver enzyme levels. Accordingly, Park et al. recently proposed diagnostic criteria of ATDILI for patients with chronic liver disease and cirrhosis [19]. In brief, ATDILI was diagnosed if liver enzyme levels increased ≥ 3 times the ULN for cirrhotic patients with normal liver enzyme levels at baseline or if liver enzyme levels increased ≥ 1.5 times of baseline values for those with elevated liver enzyme levels at baseline.

Overall, the incidence of ATDILI in the present study was 9% and increased to 16% in patients who received an initial ATD regimen containing 3 hepatotoxic drugs, i.e., isoniazid, rifampicin, and pyrazinamide. The incidence was concordant with previously reported incidence of 8%–27% among cirrhotics [19, 20, 21]. Patients with cirrhosis have a tendency to have liver injury due to impaired baseline liver function. Also, the half-life of ATDs is lengthened in patients with cirrhosis due to a decreased ability to metabolize these drugs [21]. Interestingly, all patients who developed ATDILI in this study had CTP class B cirrhosis, while none of CTP class C cirrhotics developed ATDILI. This finding is counterintuitive, but can be explained by the concern of physicians for ATDILI in patients with advanced liver disease, and patients with CTP class C cirrhosis were likely to receive a lesser number of hepatotoxic ATDs [22]. In the present study, only 42% of patients with CTP class C cirrhosis received an ATD regimen containing 3 hepatotoxic drugs, while up to 69% of patients with CTP class B cirrhosis received a regimen with 3 hepatotoxic drugs.

In the present study, ATDILI occurred exclusively in patients who received 3 hepatotoxic ATDs including isoniazid, rifampicin, and pyrazinamide. As consistent with previous studies, increasing the number of hepatotoxic drugs in the ATD regimen was shown to be associated with a greater chance of developing ATDILI [19, 27]. Park et al. found that adding 1 hepatotoxic drug to the ATD regimen increased ATDILI risk by 2.37-fold [19]. Steele et al. reported that adding rifampicin to the ATD regimen containing isoniazid significantly increased ATDILI incidence from 1.6% to 2.6% [27].

The finding of HIV infection as a potential risk factor for ATDILI in patients with cirrhosis is a new observation. To our knowledge, no previous study has mentioned this finding in cirrhotics. Previous study suggested that HIV infection increased the risk of ATDILI in patients with no underlying chronic liver diseases [28]. The risk is possibly due to an alteration of the oxidative pathway in patients with HIV infection [29]. Another possible cause is concomitant hepatotoxic antiretroviral therapy and other concomitant medications for opportunistic infection (e.g., antifungal agents) [30]. In the present study, of the 3 cirrhotic patients with HIV infection who developed ATDILI, 2 received concurrent antiretrovirals known to be hepatotoxic, i.e., nevirapine, stavudine, and efavirenz. However, the patients had received antiretroviral drugs without changing the regimen for 4 and 16 years before the initiation of ATDs; thus, it was unlikely that antiretroviral drugs caused liver injury. In the main analysis of the data, we found that cirrhotic patients with HIV infection had a higher proportion of those who developed ATDILI, i.e., 50% vs. 9%. Likewise, the subgroup analysis of patients receiving 3 hepatotoxic drugs found that 4 (50%) of cirrhotic patients with HIV infection and 3 (13%) of cirrhotic patients without HIV infection developed ATDLI; however, the difference was not quite significant (P = 0.07), probably the result of the small number of patients in the subgroup analysis lacking significant power to detect a difference. This finding suggests that cirrhotic patients with HIV infection may have a higher chance of developing ATDILI than those without HIV infection. Similarly, we found that patients in the ATDILI group had a greater number of white blood cells than those in the non-ATDILI group. The high white blood cell counts are likely a reaction secondary to liver injury rather than a factor that increases the chance of developing ATDILI, e.g., alcoholic hepatitis. No patients with ATDILI were actively drinking alcohol at the time of ATDILI diagnosis; however, 1 patient with cirrhosis caused by alcohol did not have available information on the status of alcohol drinking in their medical record.

The earliest ATDILI occurrence observed in this study was 6 days. To avoid ATDILI-induced liver failure, liver enzyme levels should therefore be monitored within 1 week after ATD initiation [31] and repeated periodically for at least 3 months in all cirrhotic patients. Concordantly, the current American Thoracic Society/Centers for Disease Control and Prevention/Infectious Diseases Society of America Clinical Practice Guidelines state that a liver function test should be conducted every 1–4 weeks for at least 2–3 months after drug initiation [25]. Based on pharmacokinetic studies of ATDs thus far, there has been no evidence supporting that a reduction of rifampin and pyrazinamide dosage would decrease the chance of ATDILI. The standard doses of rifampin and pyrazinamide are 10 and 25 mg/kg, respectively. Doses up to 35 and 60 mg/kg of rifampin and pyrazinamide, respectively, do not increase the risk of hepatotoxicity, while the effectiveness of tuberculosis treatment might increase [32, 33]. There were no prior drug pharmacokinetics or dose comparison studies in patients with cirrhosis. There are limited data for the dosage of isoniazid and its effect.

A strength of this study was that we were able to identify factors associated with ATDILI specifically for cirrhotic patients, because the study cohort comprised cirrhotic patients exclusively, while the cohorts in previous studies included patients with chronic liver diseases both without cirrhosis and with cirrhosis and other studies included cirrhotic and noncirrhotic patients as cases and controls, respectively [19, 20]. However, the present study also has some limitations. First, due to the retrospective nature of the study, laboratory blood testing schedules varied from physician to physician. As a result, some patients who had asymptomatic liver injury might have been underdetected. Second, the ATD regimen given to each individual patient was prescribed primarily based on the decision of the physicians. The risk of developing ATDILI might therefore have varied, depending on the initial regimen. To address this limitation, we conducted subgroup analysis of cirrhotics with 3 hepatotoxic ATDs and shown the similar trend of increased ATDILI risk in HIV-infected patients. Third, this study was conducted in a national tertiary care hospital. Thus, patients at our center tended to have more complicated underlying conditions, which may increase the chance of ATDILI. Caution should be exercised when applying our results in the context of general hospitals. A relatively large number of patients were excluded from the study cohort due to unavailable imaging to confirm the diagnosis of cirrhosis. Given the small numbers of patients included in this study, a future prospective, multicenter study is warranted to validate these results and determine its generalizability.