Traditionally, routine biochemical monitoring of liver function has not been recommended in patients treated with first-line antituberculosis agents, unless baseline liver function is abnormal, or factors suggesting susceptibility to hepatotoxicity are present.1,2 Instead, regular monitoring for clinical signs of intolerance (eg, nausea, vomiting, anorexia, and abdominal pain) is recommended.
The Philadelphia Department of Public Health routinely evaluates all patients undergoing treatment for tuberculosis and latent tuberculosis, both biochemically and clinically. This paper presents the results of a 15-month retrospective analysis of hepatotoxicity in this patient cohort.
MATERIALS AND METHODS
We monitor symptoms and liver function tests (LFTs) monthly in all adult patients (age ≥18 years) undergoing therapy for active or latent tuberculosis. This monitoring takes place under a physician's supervision in a Philadelphia Health Department clinic. At the initial visit, patients are interviewed for alcohol use, pregnancy, over-the-counter and prescribed medications, chronic illness, and signs and symptoms of hepatotoxicity using a standardized form. Patients are educated about the signs and symptoms of hepatotoxicity and instructed to report any adverse event by telephone to clinical staff. During monthly clinic monitoring visits, these educational messages are reinforced.
Over a period of 15 months, from March 2005 to June 2006, 267 patients were treated and monitored (Fig. 1). Of these 267 patients, 246 charts were available for review, and the serial LFTs of 229 patients were accessible for analysis. Data were extracted from charts, placed in a Microsoft Excel database, and analyzed using SAS version 9.1 (SAS Institute Inc., Cary, NC). Baseline aspartate aminotransferase (AST) and alanine aminotransferase (ALT) measurements were taken before the onset of treatment. When a patient's ALT or AST level became elevated at least 5 times the upper limit of normal, all medication was discontinued. After transaminase levels returned to less than twice the upper limit of normal, individual drugs were reintroduced one at a time, beginning with rifampin (RIF).
Potential risk factors for hepatotoxicity such as sex, age, race/ethnicity, history of hepatitis B or hepatitis C, and self-reported alcohol consumption were analyzed. Incidence rates were calculated using contingency tables, and denominator data, included the total number of patients who began isoniazid (INH) or multidrug treatment during the same period and for whom charts were available. Bivariate analysis was used to determine the association between potential risk factors with the outcome of interest-high transaminase levels. Ethics approval was not required.
Ten patients (4.4%) developed transaminase elevations requiring discontinuation of therapy. All 10 of these patients developed transaminase elevations at least 5 times the upper limit of normal, and 4 patients had elevations greater than 10 times the upper limit of normal, considered "severe" by the World Health Organization.1 However, none of the patients was ever symptomatic.
Laboratory data and regimens used are presented in Table 1: 4 patients had serological evidence of chronic hepatitis B or C but asymptomatic. None were admitted to alcohol abuse; thus, 6 of our 10 patients (2.7% of the cohort studied) had no identifiable risk factors for hepatotoxicity at the initiation of therapy. Baseline liver functions were normal in 8 patients and slightly elevated in 2 patients, one of whom was subsequently determined to abuse alcohol.
As summarized in Table 1, specific regimens at the time toxicity was detected were INH alone (4 patients); INH, RIF, pyrazinamide (PZA), and ethambutol (5 patients); and INH plus RIF (1 patient). Most peak elevations occurred in the fourth to fifth month, and notably, 1 patient (patient G) developed hepatitis in the eighth month of therapy. Table 2 summarizes the cases of hepatitis with regard to risk factors and the administration of single versus multiple drug regimens: one half of the patients with hepatitis had no risk factors, and the documented cases of hepatitis were nearly equal among those receiving INH alone compared with those receiving multiple drugs.
Table 3 summarizes the demographic data. The strongest associations were with male sex, age, and underlying hepatitis; however, none of the associations reached statistical significance.
Our experience is notable both for the incidence of hepatitis (4.4%) in our patients and for the absence of symptoms despite significant hepatotoxicity. A recent review of 3377 patients on INH therapy found a rate of AST elevations greater than 5 times the upper limit of normal in only 19 patients (0.56%); 18 of these 19 patients were asymptomatic.3 Whereas the experience of some observers reflects a similarly low rate,4,5 others have found rates comparable to ours (eg, 6.4% with transaminase elevations requiring discontinuation of therapy, most of whom were asymptomatic).6 Published recommendations suggest monitoring only for symptoms in patients without known risk factors, but this would have missed more than half of our patients; although 1 patient abused alcohol, this was not a known risk at the inception of therapy, as many patients deny alcohol abuse. Outcome data for large cohorts4,5 have reinforced the impression of relative safety in administering INH. However, some authors strike a more cautionary note, describing patients with asymptomatic yet significant hepatotoxicity and suggesting reconsideration of more routine laboratory monitoring.6-10
Two patients were begun on antituberculous therapy with elevated baseline transaminases. The first, patient E, was started on therapy in the hospital and later referred to the Department of Health. He had no risk factors, and initial AST and ALT were 69 and 92, respectively. After 3 weeks, the values were 483 and 576, and the peak values of 1238 and 863 were recorded 1 week later. He was asymptomatic throughout.
The second patient (Patient J) started on therapy with elevated baseline transaminases had no apparent risk factors for hepatotoxicity but eventually admitted to covert alcohol abuse; he had begun therapy in the hospital with slight elevations in transaminases, ALT = 57 and AST = 106. These values remained relatively stable until he had received 4 months of therapy, when they rose to 176 (ALT) and 248 (AST). When confronted with the possibility of alcohol abuse, he readily admitted recent heavy drinking. Medications were discontinued, and both RIF and then INH were successfully reintroduced after transaminases had returned to normal. The patient abstained from further alcohol use, and subsequent transaminase determinations were normal. An AST that is significantly greater than the ALT is considered characteristic of alcoholic liver injury, and this clue was helpful in this patient.
Rifampin was successfully reintroduced in all patients. Except for the patient with alcoholic liver injury, INH was not readministered after the successful reintroduction of RIF, and therapy was completed with an RIF-based regimen that excluded INH and PZA. It may have been possible to reintroduce INH in patients whose hepatitis resulted from PZA, but it was considered safer for the patients to avoid the possibility of recurrent hepatitis. Although it is generally felt that hepatotoxicity appears early in the course of anti-TB therapy, our patients frequently developed toxicity in the fourth or fifth month, and one had it in the eighth month of therapy.
There are several weaknesses in this study: the number of patients is small, and it is retrospective. Although RIF was successfully reintroduced in all patients, we cannot be certain which drug caused hepatitis in the 5 patients who had also received INH and PZA; in those who received INH monotherapy, we consider the cause to have been the INH. Although standard monitoring may not have identified these cases of severe yet asymptomatic hepatotoxicity, it is not clear how progressive the liver damage would have been if the medications had been continued. If such monitoring were adopted on a widespread basis, the cost-effectiveness might be prohibitive, especially in resource-limited locales. Serological studies for acute hepatitis A, B, or C were not performed. Our demographic data showed associations with men, age, and history of hepatitis B or C, but these were not significant. This may reflect the small sample size; however, age and baseline transaminase levels have been noted by others to correlate positively with INH hepatotoxicity.3 Although interviewing for medications and medical status is routinely performed in our clinic, it is possible that concomitant medications or other chronic illnesses unknown to us contributed to some of the patients' hepatotoxicity. It is possible that the incidence of hepatitis in our patients resulted from laboratory error or selection bias. However, all laboratory studies were conducted in an accredited laboratory, and we are unaware of any factors possibly accounting for selection bias; in fact, we were able to review 92% of our patients' charts, 93% of which had serial LFTs performed which were the basis of this report.
Asymptomatic treatment-related hepatitis clearly occurs in adult patients receiving antituberculous chemotherapy; in this small study, the most common offending agent was INH, with a possible contribution from PZA. Current recommendations for monitoring for hepatotoxicity in these patients deserve further study, and alternative therapy for latent tuberculous infection should be considered, as suggested in recent publications.11,12
1. Centers for Disease Control and Prevention. Treatment of Tuberculosis, American Thoracic Society, CDC, and Infectious Diseases Society of America. MMWR Morb Mortal Wkly Rep
2. Saukkonen JJ, Cohn DL, Jasmer RM, et al. An official ATS statement: hepatotoxicity of antituberculous therapy. Am J Respir Crit Care Med
3. Fountain FF, Tolley E, Chrisman CR, et al. Isoniazid hepatotoxicity associated with treatment of latent tuberculosis infection: a 7-year evaluation from a public health tuberculosis clinic. Chest
4. Nolan CM, Goldberg SV, Buskin SE. Hepatotoxicity associated with isoniazid preventive therapy: a 7-year survey from a public health tuberculosis clinic. JAMA
5. LoBue PA, Moser KS. Use of isoniazid for latent tuberculosis infection in a public health clinic. Am J Respir Crit Care Med
. 2003;168(4):443-447. (Epub 2003 May 13).
6. Byrd RB, Horn BR, Solomon DA, et al. Toxic effects of isoniazid in tuberculosis chemoprophylaxis. Role of biochemical monitoring in 1,000 patients. JAMA
7. Millard PS, Wilcosky TC, Reade-Christopher SJ, et al. Isoniazid-related fatal hepatitis. West J Med
8. Moulding TS, Redeker AG, Kanel GC. Twenty isoniazid-associated deaths in one state. Am Rev Respir Dis
9. Agal S, Baijal R, Pramanik S, et al. Monitoring and management of antituberculosis drug induced hepatotoxicity. Gastroenterol Hepatol
10. Ahlawat K, Lewis JH. Tuberculosis of the liver, biliary tract and pancreas. In: Schlossberg D, ed. Tuberculosis and Nontuberculous Mycobacterial Infections
. New York: McGraw-Hill; 2006.
11. Lardizabal A, Passannante M, Kojakali F, et al. Enhancement of treatment for latent tuberculosis infection with 4 months of rifampin. Chest
12. Ashkin D, Julien J, Lauzardo M, et al. Consider rifampin BUT be cautious (Editorial). Chest