[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address Please contact the publisher to request reinstatement.
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
Purchase Options:
[Skip to Content Landing]
Citations 0
From the Centers for Disease Control and Prevention
April 21, 2010

Severe Isoniazid-Associated Liver Injuries Among Persons Being Treated for Latent Tuberculosis Infection—United States, 2004-2008

JAMA. 2010;303(15):1471-1473. doi:

MMWR. 2010;59:224-229

3 tables omitted

Since the 1960s, 6 to 9 months of isoniazid (INH*) has been the mainstay of treatment for latent tuberculosis infection (LTBI), but its application has been limited by concerns about the toxicity of INH and the long duration of treatment. To quantify the frequency of severe adverse events (SAEs) associated with LTBI treatment and to characterize the clinical features of affected patients, in January 2004 CDC began a national project to monitor SAEs associated with treatment for LTBI. State health departments were encouraged to report SAEs associated with any LTBI treatment regimen to a passive surveillance system. This report summarizes the results for 2004-2008, when 17 SAEs in 15 adults and two children (aged 11 and 14 years) were reported. All patients had received INH therapy and had experienced severe liver injury. Five patients, including one child, underwent liver transplantation. Five adults died, including one liver transplant recipient. These findings underscore the risk for an idiosyncratic drug-induced reaction in patients of any age treated with INH, including those with or without a putative predictor for INH-associated liver injury. Patients receiving INH for LTBI therapy should be monitored according to American Thoracic Society (ATS)/CDC recommendations because of the risk for drug-induced hepatoxicity.1,2 Providers should counsel patients to terminate INH therapy promptly and seek medical attention if they experience signs and symptoms of illness.

An SAE was defined as any drug-associated reaction resulting in a patient's hospitalization or death after at least 1 treatment dose for LTBI. Public and private health-care providers notified local health departments of SAEs. Local health departments then submitted standardized reports to CDC through their state health departments. Standardized reports included demographic information, LTBI treatment regimen, dates of treatment initiation and cessation, dates of hospitalization, results of testing for antibodies to viral hepatitis, clinical outcome, and dates of liver transplantation or death. Although the surveillance system was passive, CDC was available upon invitation to conduct extended onsite investigations. Investigations included medical record reviews and interviews of patients or their proxies and medical providers.

During 2004-2008, CDC received 21 reports of LTBI treatment—associated adverse events; however, four did not meet the SAE surveillance definition and were excluded from this analysis. All 17 patients with events meeting the SAE definition had received INH therapy and experienced liver injury. Of the 17 patients, two were children aged <15 years. For the 15 affected adults, the median age was 39 years (range: 19-63 years). The SAEs were diagnosed between the second and ninth month of therapy, with the exception of one adult whose regimen spanned 17 months because of repeated treatment interruptions and who was diagnosed with an SAE in the seventeenth month. Sixteen patients tested negative for antibodies to hepatitis A (IgM anti-HAV), hepatitis B (antibody to hepatitis B core antigen), and hepatitis C (anti-HCV); one adult had pretreatment coinfection with hepatitis C virus (HCV) and human immunodeficiency virus (HIV). Of 17 patients, five underwent liver transplantation, including one child. Five of 15 adults died, including a liver transplant recipient.

Onsite Clinical Investigations

For 10 SAEs, state health departments invited CDC personnel to conduct onsite investigations. All 10 affected patients had indications for LTBI treatment, were prescribed INH within the recommended dosage range, and took the medication as prescribed. Prescribers followed ATS/CDC guidelines for monthly clinical monitoring of all 10 patients.†1,2 Pretreatment serum aminotransferase concentrations were normal for five adults who underwent baseline testing. Monthly aminotransferase monitoring was scheduled for two adults: one with HCV/HIV coinfection and another patient aged >35 years.

SAE symptoms began in the 10 patients 1-7 months after INH initiation; for all patients, SAE diagnosis was based on symptoms rather than laboratory abnormalities. Seven patients initially experienced excess fatigue, nausea, or abdominal pain, but waited until the onset of jaundice before seeking medical attention. All patients had developed jaundice and markedly abnormal aminotransferase concentrations by the time of clinical evaluation. One patient had markedly abnormal aminotransferase concentrations 2 months before symptom onset, but the laboratory abnormalities were discovered incidentally during routine care by a provider who was unaware of LTBI treatment, and treatment continued until symptom onset. For seven of 10 patients, a provider other than the one who had prescribed the INH detected the SAE.

For two patients, treatment was discontinued within 3 days of symptom onset. Of the remaining eight patients, all discontinued INH at least 1 week after symptom onset. No patient discontinued INH until specifically instructed by a medical provider. All 10 patients underwent testing to exclude viral infections and other potential causes of liver injury. Liver biopsy or explanted liver histopathologic examination was performed for five patients; results from each revealed the presence of nonspecific changes consistent with drug-induced liver injury.3

Seven of 10 patients had a putative predictor‡ for INH-associated liver injury. Of the three patients without a putative risk factor, two had ingested acetaminophen-containing medications during INH therapy; however, the two had taken standard doses for less than 1 week.

Reported by:

State health departments; T Harrington, MD, L Manangan, MPH, J Jereb, MD, T Navin, MD, Div of Tuberculosis Elimination, National Center for HIV/AIDS, Viral Hepatitis, STDs, and Tuberculosis Prevention; K Powell, MD, EIS Officer, CDC.

CDC Editorial Note:

Approximately 4% of the U.S. population has latent tuberculosis infection (LTBI).4 Because LTBI can progress to active disease, CDC recommends testing and treatment of LTBI for persons in certain groups.1 The findings in this report underscore the importance of following ATS/CDC recommendations (see Box) regarding selection of candidates for LTBI treatment and for following recommendations for sustained clinical monitoring throughout LTBI treatment to detect rare, but severe, adverse events among patients of any age.

Box. American Thoracic Society/CDC recommendations for targeted testing and isoniazid treatment for latent tuberculosis infection (LTBI) and monitoring during treatment

  • Existing recommendations emphasize the careful selection of candidates for LTBI testing and treatment based on risk for infection. Persons who are not at risk for TB infection should not undergo testing for LTBI.

  • Monthly clinical monitoring, including a brief physical examination, for the signs and symptoms of LTBI treatment–associated adverse events is recommended for all patients.

  • Patients who have human immunodeficiency virus (HIV) infection, patients who have chronic liver disease, pregnant women, women in the immediate postpartum period (≤3 months after delivery), and patients who use alcohol regularly should be considered for baseline laboratory hepatic testing.

  • Although baseline laboratory testing is not routinely indicated in older persons, it may be considered on an individual basis, especially for patients who are taking medications for chronic medical conditions.

  • Routine laboratory testing is indicated for patients whose baseline testing is abnormal and other persons at risk for hepatic disease.

  • An evaluation including laboratory testing should be obtained upon the first sign or symptom of a possible adverse event. Providers should educate patients to discontinue treatment immediately, even before an evaluation is conducted.

  • In the absence of symptoms, isoniazid should be discontinued if aminotransferase values are five times the upper limit of normal.

  • In the presence of symptoms, isoniazid should be discontinued if aminotransferase values are three times the upper limit of normal.

Sources: CDC. Targeted tuberculin skin testing and treatment of latent tuberculosis infection. MMWR 2000;49(No. RR-6).

American Thoracic Society. An official ATS statement: hepatotoxicity of antituberculosis therapy. Am J Respir Crit Care Med 2006;174:935--52.

The finding that seven of 10 SAEs were diagnosed by medical providers other than the ones that prescribed INH indicates the importance of provider-to-provider and provider-to-patient communication for the safe administration of INH therapy. In this series, a diagnostic delay occurred for at least one patient who sought care from a provider other than the INH prescriber. Also, eight patients continued taking the medication while developing symptoms, a practice that has been noted in other published reports.5 Medical providers should emphasize to patients that INH treatment should be stopped immediately upon the earliest onset of symptoms (e.g., excess fatigue, nausea, vomiting, abdominal pain, or jaundice), even before a clinical evaluation has been conducted, and that initial symptoms can be subtle and might not include jaundice.

Two of the 17 patients in this series were children. Although the condition is thought to be rarer in children than in adults, INH-associated liver injury has been reported previously in children,6 and both clinicians and patients should be aware that SAEs can occur among patients of all ages. Nine of the 17 SAEs occurred beyond the third month of therapy, indicating that INH-associated liver injury is possible anytime during the treatment course. This finding was in contrast to an earlier study that found 10 of 11 episodes of INH-induced hepatotoxicity occurred during the first 3 months of therapy.7

In this case series, all patients were monitored according to current guidelines (i.e., monthly clinical evaluation, including symptom screening and physical examination),1,2 and two patients were selected for additional laboratory monitoring. However, despite adherence to current guidelines for monitoring, liver injury occurred, and SAE diagnosis was prompted by symptoms, not laboratory values. Additionally, three patients had no putative predictors of liver injury, indicating that careful monitoring is needed regardless of the patient's risk factor profile. Although all 10 patients in this series were symptomatic, INH-associated liver injury can occur even in the absence of symptoms.

INH-associated liver injury is an idiosyncratic reaction, independent of dosing, and is a diagnosis of exclusion.2 Historically the incidence has been estimated at 1 per 1,000 patients who begin treatment,1,2 but the lack of specific diagnostic criteria and heterogeneous definitions complicate comparisons across studies. The SAE surveillance system is the only national system that collects relevant public health data regarding the appropriateness of testing and treatment for LTBI and monitoring during treatment. However, as with all surveillance systems, underreporting is common in the SAE surveillance system, and LTBI is not reportable in most jurisdictions. In addition, calculation of INH-associated SAE rates is made difficult by the absence of reliable denominators for the number of persons initiating INH treatment, which has been estimated at 291,000 to 433,000 per year.8 Because the demographic characteristics of the patients who begin LTBI treatment with INH remain unknown, the risk factors for INH-associated liver injury cannot be determined conclusively.

LTBI treatment remains a key component of the TB elimination strategy in the United States. One study estimated that LTBI treatment prevented 4,000-11,000 TB cases in 2002 in the United States, substantially reducing the burden of TB.8 In the United States, 9 months of INH therapy is the standard LTBI treatment regimen. Efficacy and safety have not been established for other treatment regimens, such as 4 or 6 months of rifampin,9 3 months of INH and rifampin (the preferred regimen in the United Kingdom10), or 3 months of once-weekly INH and rifapentine, a regimen currently under investigation (CDC, unpublished data, 2010).

Until an equally effective, better-tolerated regimen is developed, 9 months of INH therapy remains the mainstay of LTBI treatment. CDC encourages optimal use of INH by targeting LTBI testing to those patients most likely to benefit from treatment of LTBI.1 No more than a 1-month supply of INH at a time should be prescribed, and treatment should be combined with careful clinical monitoring.1,2 Alcohol consumption, underlying liver disease, and the concurrent use of medications that are metabolized in the liver can increase the occurrence or severity of liver injuries among INH recipients.

Local providers should report possible INH-associated SAEs to their respective health departments and to the Food and Drug Administration's MedWatch (https://www.accessdata.fda.gov/scripts/medwatch). State health departments should report these events to CDC's Division of Tuberculosis Elimination (e-mail: LTBIdrugevents@cdc.gov).

What is already known on this topic?

Since the 1960s, 6 to 9 months of isoniazid (INH) has been the mainstay of treatment for latent tuberculosis infection (LTBI), but its application has been limited by concerns about the toxicity of isoniazid and the long duration of treatment.

What is added by this report?

During 2004-2008, a total of 17 serious liver injuries were reported in patients receiving INH therapy; five patients underwent liver transplantation, and five died, including one liver transplant recipient.

What are the public health implications for public health practice?

Patients receiving INH therapy for LTBI should be told categorically by medical providers to stop taking their medication immediately if they have symptoms such as nausea, vomiting, abdominal discomfort, or unexplained fatigue and to contact their providers for further evaluation.


10 Available.


†Monthly clinical monitoring (including a physical examination) for the signs and symptoms of adverse events is recommended by ATS and CDC for all LTBI treatment patients.1,2 Existing guidelines suggest that patients who have HIV infection, patients who have chronic liver disease, pregnant women, women in the immediate postpartum period (≤3 months of delivery), and patients who use alcohol regularly should be considered for baseline laboratory hepatic testing. Routine laboratory testing is indicated for patients whose baseline testing is abnormal and other persons at risk for hepatic disease.1,2

‡Predictors of INH-associated liver injury include preexisting liver disease, HIV infection, injection-drug use, concurrent alcohol consumption, pregnancy or the immediate postpartum period (≤3 months of delivery), concomitant administration of medications with hepatotoxic potential, and older age.1,2