Isoniazid Preventive Therapy, Hepatitis C Virus Infection, and Hepatotoxicity among Injection Drug Users Infected with Mycobacterium tuberculosis

Abstract

Treatment of latent Mycobacterium tuberculosis infection with isoniazid can cause hepatotoxicity, but the risk of isoniazid-associated hepatotoxicity among persons coinfected with hepatitis C virus (HCV) is unknown. We conducted a prospective study among 146 injection drug users with M. tuberculosis infection and normal baseline hepatic transaminase values who were treated with isoniazid. Of 146 participants, 138 (95%) were HCV-seropositive. Thirty-seven participants (25%) were human immunodeficiency virus (HIV)—seropositive. Thirty-two (22%; 95% confidence interval [CI], 16%–30%) of 146 participants developed transaminase value elevations to >3 times the upper limit of normal. Transaminase value elevation was associated with concurrent alcohol use but not with race, age, presence of hepatitis B surface antigen, HIV-1 infection, or current injection drug use. Isoniazid was withdrawn from 11 participants (8%; 95% CI, 4%–13%). Of 8 deaths during follow-up, none were attributed to isoniazid-associated hepatotoxicity. The risk of transaminase value elevation and drug discontinuation for HCV-infected persons receiving isoniazid was within the range reported for populations with lower HCV prevalence.

Injection drug users are at increased risk of developing active tuberculosis if they are infected with Mycobacterium tuberculosis [1-4]. Isoniazid is effective in preventing progression to active tuberculosis [5] but can cause hepatotoxicity [6, 7]. The risk of isoniazid-associated hepatotoxicity increases with age [8, 9] and daily alcohol consumption [10]. Fatal hepatotoxicity is rare but has been reported [11, 12]. In persons receiving treatment for active tuberculosis, HIV and hepatitis C virus (HCV) infection have been associated with elevated hepatic transaminase values [13]. However, the impact of HCV infection on hepatic transaminase values among persons receiving isoniazid preventive therapy (treatment of latent tuberculosis infection) has not been described, nor has the impact of such transaminase value elevations on discontinuation of isoniazid therapy. This is particularly important among injection drug users (IDUs), of whom up to 90% are infected with HCV [7]. We therefore sought to describe the risk of hepatotoxicity and associated isoniazid withdrawal in a cohort of IDUs who were treated with isoniazid for latent M. tuberculosis infection.

PATIENTS AND METHODS

Patient population

Study participants were identified from the AIDS Linked to Intravenous Experience (ALIVE) natural history study of HIV-1 infection in IDUs [14]. Enrollment in ALIVE occurred from February 1988 through March 1989. Persons enrolled in ALIVE were ≥18 years old and AIDS-free and had injected drugs at least once since 1977. Tuberculin and anergy testing were offered at semiannual ALIVE study visits from March 1990 through January 1996. Tuberculin skin testing was performed with the Mantoux technique. A positive tuber-culin skin test (TST) was defined by induration of ≥5 mm in HIV-1—seropositive persons and ≥10 mm in HIV-1—seronegative persons. All persons with a positive TST were offered a clinical evaluation for evidence of active tuberculosis, including chest radiography and measurements of serum alanine transaminase and aspartate transaminase.

Participants who were TST-positive were offered isoniazid preventive therapy for 12 months if they were HIV-1—sero-positive and for 6 months if HIV-1—seronegative [15]. Inclusion criteria for the current study were TST-positivity without evidence of active tuberculosis, initiation of isoniazid preventive therapy between March 1990 and January 1996, and completion of at least 1 week of isoniazid therapy. Follow-up occurred through July 1996. Exclusion criteria included prior completion of a course of isoniazid preventive therapy, a history of isoniazid allergy, and baseline hepatic transaminase values >3 times the upper limit of normal.

Demographic data, medical history, physical examination findings, and baseline laboratory data (including HIV-1 antibody, HCV antibody, and hepatitis B surface antigen assay results) were obtained for participants at enrollment in the ALIVE study. Follow-up visits for the ALIVE cohort were conducted at 6-month intervals, during which serological testing for HIV-1 was repeated for HIV-1—seronegative individuals and information on antiretroviral therapy was obtained. Patient self-reports of injection drug and alcohol use were ascertained at the ALIVE visit preceding the initiation of isoniazid preventive therapy. At monthly follow-up visits after the initiation of isoniazid therapy, liver function tests were performed and information was obtained regarding adverse effects, adherence to preventive therapy, and symptoms of hepatotoxicity (at least 2 of the following: jaundice, dark urine, loss of appetite, nausea, vomiting, pruritis, and fatigue). The reason for discontinuation of isoniazid preventive therapy was documented. Data on deaths were obtained from the Maryland Bureau of Vital Records (Baltimore), family reports, other institutions, and the computerized National Death Index.

Isoniazid was administered either daily (5 mg/kg; 300 mg maximum) or twice weekly (15 mg/kg; 900 mg maximum). Vitamin B₆ (50 mg) was given concomitantly with isoniazid.

Definitions

Completion of therapy was defined as receipt of 52 weeks of treatment for HIV-1—infected participants and 26 weeks for HIV-1—seronegative participants. Elevated liver function test values were categorized as follows. Values >3 times the upper limit of normal were ≥120 U/L for serum aspartate transaminase and ≥160 U/L for alanine transaminase. Values >5 times the upper limit of normal were >200 U/L for serum aspartate transaminase and >270 U/L for alanine transaminase.

Alcohol consumption was graded according to frequency of ingestion (number of drinks per week). Immunoassays for hepatitis B surface antigen, antibody to hepatitis B surface antigen, and antibody to hepatitis B core antigen were performed on blood specimens with use of AUSZYME, AUSAB, and CORZYME, respectively (Abbott Laboratories) [16]. Antibody to hepatitis C virus was sought with a second-generation HCV 2.0 enzyme immunoassay (Ortho Diagnostic Systems) [17]. Specimens were screened for HIV-1 with an enzyme immunoassay (Genetic Systems), and findings were confirmed by Western blotting (Biotech HIV-1; Du Pont).

Statistical analysis

The χ² or Fisher’s 2-tailed exact test was used to compare categorical variables. The Kruskal-Wallis test was used for comparison of continuous variables, and 95% CIs were calculated for proportions. For comparisons of elevated liver function test values, follow-up extended through 26 weeks of treatment, regardless of HIV status. Statistical analyses were performed with SAS software version 6.12 (SAS Institute), STATA statistical software release 6.0 (Stata Corporation), and Epi Info version 6 (Centers for Disease Control and Prevention).

RESULTS

A TST was performed for 1357 individuals during the study period. Of these, 1238 (91%) had the test read, and 327 (26%) were positive. Among the 327 persons with a positive TST, 181 were excluded because of previous preventive therapy or loss to follow-up before therapy could be initiated (n = 142); missing/abnormal baseline or follow-up liver function test values (n = 30), or completion of <1 week of therapy (n = 9). Thus, 146 individuals were available for study.

The demographic characteristics of participants are shown in table 1. The mean age of participants studied was 42 years (SD, ±7 years). The composition of the study sample was similar to that of the ALIVE cohort with regard to age and race. However, the proportion of female participants was lower than in the parent cohort (14% vs. 23%; P = .01). Of the 146 participants, 138 (95%) had antibody to HCV and 37 (25%) were HIV-1—seropositive. Of the 37 HIV-1—infected individuals, information on antiretroviral therapy was available for 34 (92%). Thirteen (38%) of the 34 HIV-1—positive participants reported use of a nucleoside reverse transcriptase inhibitor during follow-up; none reported use of an HIV-1 protease inhibitor or nonnucleoside reverse transcriptase inhibitor.

Table 1.

Demographic characteristics of 146 persons receiving isoniazid as preventive therapy.

Characteristic	No. (%) of persons
Age, years
<35	20 (14)
≥35	126 (86)
Sex
Male	126 (86)
Female	20 (14)
Race/ethnicity
Black	140 (96)
Other	6 (4)
HIV infection
Negative	109 (75)
Positive	37 (25)
Hepatitis C virus infection
Negative	6 (4)
Positive	138 (96)

Of the 146 participants studied, 32 (22%; 95% CI, 16%–30%) had elevated hepatic transaminase values (>3 times the upper limit of normal) at least once during follow-up. Of these 32, 13 (41% [9% of study population]; 95% CI, 5%–15%) had values >5 times the upper limit of normal. Symptoms of hepatotoxicity after initiation of therapy occurred in 8 (25%) of the 32 whose transaminase values were >3 times the upper limit of normal, versus 18 (16%) of 110 individuals whose transaminase values were not elevated (P = .27; information missing for 4 patients). Four (31%) of 13 individuals with transaminase levels >5 times the upper limit of normal were symptomatic. Current alcohol use was associated with a >3-fold transaminase elevation (P = .004). This risk increased with the amount of alcohol intake and was greatest for people who had ≥7 drinks per week (table 2). Development of a >3-fold elevation in transaminase values was not associated with race, age >35 years, hepatitis B surface antigen status, hepatitis C infection, HIV-1 infection, or current injection drug use.

Table 2.

Demographic and clinical characteristics at enrollment, in relation to development of elevated hepatic transaminase levels during isoniazid preventive therapy.

Characteristic	No. (%) of persons with elevated valuea
Age, years
<35	5 (25)
≥35	27 (21)
Sex
Male	30 (24)
Female	2 (10)
Ethnicity
Black	29 (21)
Other	3 (50)
Hepatitis C virus infection
Negative	0
Positive	31 (22)
HBsAg
Negative	27 (23)
Positive	1 (6)
HIV-1 infection
Negative	24 (22)
Positive	8 (22)
Current injection drug use
None	8 (17)
<1/day	13 (24)
≥1/day	11 (24)
Current alcohol use (drinks/week)
<1	5 (9)
1–6	16 (28)
≥7	11 (37)
Total	32 (22)

NOTE. χ²values not significant except for current alcohol use (P = .004). HBsAg, hepatitis B surface antigen (information not available for 12 individuals).

^aValue >3 × normal.

Seventy-five of 109 HIV-1—seronegative individuals (69%) and 27 of 37 HIV-1—infected persons (73%) completed 6 months of treatment (P = .63). Of the 146 participants, 24 (16%) were noncompliant. Eleven persons (8%; 95% CI, 4%–13%) permanently discontinued therapy because of elevated transaminase values, 5 (3%) discontinued therapy for reasons other than hepatotoxicity, and 4 (3%) withdrew voluntarily. Two of 11 individuals (18%) who discontinued therapy because of elevated transaminase values were symptomatic. Of the 13 individuals whose transaminase values were elevated >5-fold, 6 (46%) discontinued taking isoniazid, compared with 5 (26%) of 19 whose values were elevated 3–5-fold (P = .28).

There were 8 deaths during the study period, of which 4 were HIV-1—infected participants; 2 of the 8 had liver enzyme values >3 times the upper limit of normal. The causes of death were AIDS (2), ventricular tachycardia (1), renal failure (1), illicit drug overdose (3), and hepatorenal failure with sepsis (1). The individual who developed hepatorenal failure with sepsis prior to death had transaminase elevations of >5 times the upper limit of normal recorded twice, but these elevations occurred >1 year before death. The patient had a history of alcohol abuse; he was HCV-positive, HIV-1—seronegative, and hepatitis B surface antigen—negative but was positive for antibody to hepatitis B surface antigen and core antigen.

This individual was treated with isoniazid over a span of 70 weeks, for a total period of 48 weeks. Transaminase elevations occurred intermittently during the course of therapy, with the highest elevation occurring 19 weeks after initiation of therapy; consequently, there were periods of interruption of isoniazid therapy. There was an interval of 20 weeks between the last dose of isoniazid and death. The cause of death for the other individual with a history of transaminase value elevation was illicit drug overdose.

DISCUSSION

Given the high prevalence of hepatitis C virus infection among IDUs, the association between HCV infection and elevated hepatic transaminase values [18], and the potential for isoniazid to cause hepatotoxicity, the risk of hepatotoxicity among IDUs receiving isoniazid for treatment of latent M. tuberculosis infection needs to be established. The impact of transaminase value elevation on the discontinuation of isoniazid therapy in this population also needs to be assessed.

In this study, an increase in hepatic transaminase values to >3 times the upper limit of normal occurred in 22% of IDUs (95% CI, 16%–30%) who received at least 1 week of isoniazid preventive therapy. This is within the range reported in previous studies in which transaminase values were monitored regularly, regardless of symptoms; in those studies, 10%–22% of participants had elevated transaminase values at least once during the course of therapy [19-25]. The prevalence of HCV infection in these populations was likely much lower than in our study because the study participants did not all have the high risk for HCV that the participants in our study had. However, the prevalence of HCV in these studies was unknown because testing for HCV was not available.

In one study in which 17% of participants had transaminase value elevations, none had evidence of hepatitis A, B, or C [25]. As expected, the rates of transaminase elevation seen in this and other studies were higher than those seen when liver function tests were performed only for patients who were symptomatic [8]. Most of the participants in this study whose hepatic transaminase values were elevated were asymptomatic and did not discontinue isoniazid therapy.

Consistent with findings in previous studies, the risk of hepatotoxicity in this study was associated with increased alcohol intake. However, there was no statistically significant difference in hepatotoxicity according to age, race, HBV status, or HIV status. The lack of statistical significance was due primarily to the small sample size. Although 22% of HCV-infected participants had elevated hepatic transaminase values but none of the HCV-negative participants did, this difference was not statistically significant because of the low number of HCV-negative participants.

From a clinical standpoint, the most important aspect is the incidence of discontinuation of isoniazid therapy due to hepatotoxicity. In contrast to the 22% of participants who developed transaminase value elevations, 8% of the study population (95% CI, 4%–13%) discontinued isoniazid therapy because of hepatotoxicity. In previous studies of isoniazid preventive therapy (none of which were conducted among solely IDUs or persons coinfected with HCV), the rates of discontinuation of such therapy because of hepatotoxicity ranged from 0.1% to 10% [5, 6, 8, 10, 22, 24-27].

Of the 8 participants who died during the study period, transaminase values were >3 times the upper limit of normal for one and >5 times the upper limit for another. The cause of death for 1 individual was hepatorenal failure and sepsis; this individual was HCV-positive and actively abused alcohol. Because of these comorbid conditions and the 20-week period between the last dose of isoniazid and death, it is unlikely that the death was attributable to isoniazid. For the other participant who had elevated hepatic transaminase values and subsequently died, the cause of death (illicit drug overdose) was not related to isoniazid use.

There are several limitations of this study. First, because of the longer duration of treatment in the HIV-1—infected individuals, there was the potential for ascertainment bias of elevated transaminase values. However, we accounted for this potential bias by comparing individuals who completed 26 weeks of treatment, regardless of HIV-1 status. Second, other potentially hepatotoxic exposures, such as acetaminophen and nucleoside reverse transcriptase inhibitors—as well as infection with Epstein-Barr virus, cytomegalovirus, or hepatitis A virus—were not quantified. Third, because of the low number of HCV-seronegative participants, a direct comparison of transaminase value elevation and therapy discontinuation among HCV-seropositive and HCV-seronegative individuals could not be made.

With the above limitations noted, the rates of isoniazid-associated hepatotoxicity and isoniazid-associated hepatotoxicity requiring drug withdrawal in this study are consistent with those in previous studies that were conducted among populations with a lower prevalence of HCV infection. Although these findings must be confirmed in larger studies of HCV-seropositive and HCV-seronegative individuals, the results of this study suggest that preventive therapy with isoniazid can be safely given to HCV-infected persons as long as liver function test values are closely monitored.