Abstract
Background:
Hepatotoxicity is frequent in the standard-of-care regimen for drug-susceptible tuberculosis, yet the association of each companion drug with hepatotoxicity has not been fully characterized. We examined the association between hepatotoxicity and the pharmacokinetics of rifampin and its companion drugs in standard therapy, as well as N-acetyltransferase 2 (NAT2) genotype.
Methods:
We evaluated HIRIF trial participants who were randomized to receive rifampin 10, 15, or 20 mg/kg/day during the intensive phase of tuberculosis treatment. We fitted Cox proportional hazards models to identify risk factors for grade 2 or higher (grade 2+) alanine transaminase (ALT) or aspartate transaminase (AST) elevation, and considered pharmacokinetic exposure of each antituberculosis drug and NAT2 genotype as potential covariates.
Results:
Among 168 participants with pharmacokinetic data, neither rifampin dose nor exposure were associated with the risk of grade 2+ ALT or AST elevation. Higher pyrazinamide exposure (hazard ratio [HR] 1.85 for every 50 mg*h/L AUC0–6h increase), higher isoniazid exposure (HR 1.40 for every 5 mg*h/L AUC0–6h increase), and among a subset with known NAT2 genotype, slow NAT2 acetylator status (HR 9.32 relative to fast, n=90) were associated with grade 2+ ALT or AST elevation in univariable analysis. In multivariable analysis, only pyrazinamide exposure was associated with hepatotoxicity.
Conclusions:
While higher pyrazinamide and isoniazid exposures and slow NAT2 acetylator status were each associated with increased hepatotoxicity, only pyrazinamide exposure was associated when considering pharmacokinetic variables together. Rifampin exposure was not associated with hepatotoxicity, supporting further evaluation of doses up to 20 mg/kg/day in a larger trial.
Keywords: Drug-susceptible tuberculosis, hepatotoxicity, rifampin, isoniazid, pyrazinamide, NAT2 acetylator
INTRODUCTION
Tuberculosis (TB) remains the leading infectious cause of death [1]. Despite the availability of highly efficacious treatment, morbidity and mortality among TB patients remain substantial, partly due to adverse events induced by antituberculosis drugs, which can lead to treatment discontinuation or interruption, increasing the risk of treatment failure, relapse, and drug resistance [1,2].
The 6-month regimen comprising isoniazid, rifampin, pyrazinamide, and ethambutol has been the standard therapy for drug-susceptible TB (DS-TB) since the 1980s [3,4]. Hepatotoxicity is one of the most frequent and severe adverse events occurring with this regimen, with reported incidence ranging from 2% to 28% [5]. While asymptomatic transaminase elevations are common during treatment, hepatotoxicity can be fatal if not recognized early and managed appropriately through timely treatment interruption [5]. Given that treatment interruption significantly erodes treatment efficacy [6], preventing drug-induced hepatotoxicity is critical to ensuring treatment success and patient safety during TB therapy.
Among the first-line antituberculosis drugs for DS-TB, isoniazid, rifampin, and pyrazinamide are known to have hepatotoxic properties, but their individual contributions to hepatotoxicity remain unclear due to their use in combination [7]. N-acetyltransferase 2 (NAT2) genotype, which primarily affects the extent of isoniazid metabolism and its resulting exposure, has also been linked to anti-TB drug-induced hepatotoxicity [8–10]. Furthermore, with growing interest in shortening the treatment duration for DS-TB by increasing rifampin doses [11–14], isolating the hepatotoxic impact of higher rifampin exposure is essential to ensuring its safe clinical use. To date, however, the quantitative contribution of each component drug to hepatotoxicity within standard therapy for DS-TB, particularly in the context of increased rifampin doses, has not been well characterized.
The Phase 2B dose-ranging trial, HIRIF (NCT01408914) [14], previously showed the potential of higher rifampin doses to shorten standard therapy without added toxicity. In the present analysis, we used data from the HIRIF trial to evaluate the association between hepatotoxicity and the pharmacokinetics (PK) of rifampin and its companion drugs in the standard of care regimen for DS-TB. We also explored the potential contribution of NAT2 genotype to hepatotoxicity.
METHODS
Patient population
We evaluated adults with newly diagnosed, smear-positive, pulmonary DS-TB enrolled in the Phase 2B HIRIF trial (NCT01408914). Participants were randomized to receive rifampin 10, 15, or 20 mg/kg/day during the intensive phase of treatment in a 1:1:1 ratio. The detailed trial design and participant characteristics have been previously described [14,15].
Pharmacokinetic data
The following previously reported PK parameters of isoniazid, rifampin, pyrazinamide, and ethambutol were utilized in this study: area under the plasma concentration-time curve (AUC) from 0 to 6 h post-dose (AUC0–6h) and maximum plasma concentration (Cmax) [16]. For participants whose isoniazid AUC0–6h was not calculable, AUC from 0 to last measurable time point (AUClast) values were used as imputed substitutes. The detailed design and methodology for the PK analysis have been previously described [16].
Pharmacogenomic data
Genotyping data for NAT2 was available in this study, as previously described [17]. NAT2 genotype was categorized into slow, intermediate, and fast acetylator status based on the single NAT2 tag nucleotide polymorphism (rs1495741).
Outcome
We defined hepatotoxicity as the time to grade 2 or higher alanine aminotransferase (ALT) or aspartate aminotransferase (AST) elevation during the intensive phase and up to 4 weeks thereafter (i.e., a total of 12 weeks), based on the primary safety endpoint used in the HIRIF trial [14].
Statistical analysis
All data processing, statistical analyses, and visualizations were conducted using R version 4.2.3 (R Foundation for Statistical Computing, Vienna, Austria), and a p-value < 0.05 was considered statistically significant. We compared the PK parameters for each drug between participants with and without the hepatotoxicity outcome using an independent two-sample t-test. We explored the effects of randomization or baseline characteristics on the PK parameters using an independent two-sample t-test, one-way analysis of variance (ANOVA), or linear regression, as applicable.
We performed Cox proportional hazards analysis to identify risk factors for hepatotoxicity, with PK parameters and NAT2 genotype as predictors of interest. We generated Kaplan-Meier estimates stratified by each drug’s AUC0–6h dichotomized at the population median and by NAT2 genotype. In the univariable Cox analysis, we tested PK parameters and NAT2 genotype as potential risk factors, and assessed other demographic and baseline clinical factors as potential confounders. We evaluated the proportional hazards assumption for all variables. For the multivariable Cox analysis, we included only PK parameters due to limited availability of NAT2 genotype and performed complete case analysis without imputation. We selected candidate risk factors for the final multivariable model using a stepwise procedure, testing linear relationships through forward inclusion (p < 0.05) and backward exclusion (p < 0.01).
We developed longitudinal models for ALT and AST using a nonlinear mixed-effects modeling approach with NONMEM version 7.5.1 (ICON plc, Gaithersburg, Maryland) to investigate whether the velocity (slope) of elevation was associated with exposure to individual drugs. We censored longitudinal trajectories of ALT and AST at the first grade 2 or higher elevation event, respectively. We explored structural models using linear or turnover models, including a linear or maximum effect (Emax) relationship with time. We described the two subpopulations representing low and high elevations of ALT and AST using a probabilistic mixture model that assigned individuals to each subpopulation based on their observed data, enabling the estimation of distinct slope profiles within the population [18]. First, we compared the PK parameters for each drug between the mixture model-assigned subpopulations of ALT and AST elevation using an independent two-sample t-test. Second, we evaluated the influence of drug exposure on the slopes of ALT and AST elevation using linear regression and a stepwise approach with forward selection (p < 0.05) and backward elimination (p < 0.01). We performed model selection based on numerical and visual assessments, including significance-based selection on log-likelihood testing, the precision of parameter estimates, goodness-of-fit plots, and visual predictive checks.
RESULTS
Participant characteristics
Among the 180 adults randomized in the HIRIF trial, 168 participants with available PK data and 90 participants with available NAT2 genotype regardless of PK data availability were included in the present analysis. Baseline characteristics, including weight and transaminase levels, and observed PK parameters for isoniazid, pyrazinamide, and ethambutol were similar across randomized rifampin dose groups (Table 1). In the 168 participants with available PK data, pyrazinamide exposure was positively correlated with isoniazid, rifampin, and ethambutol exposures, and ethambutol exposure was positively correlated with rifampin exposure (Supplementary Figure 1). Of the 90 participants with available NAT2 genotype, 32 (17.8%) were classified as slow, 42 (23.3%) as intermediate, and 16 (8.9%) as fast acetylator status. Isoniazid exposures increased across fast, intermediate, and slow acetylator status (Supplementary Figure 2).
Table 1.
Characteristics of HIRIF trial participants by randomized rifampin dose group.
| Randomized Rifampin Dose |
||||
|---|---|---|---|---|
| 10 mg/kg (n=58) | 15 mg/kg (n=57) | 20 mg/kg (n=53) | Total (n=168) | |
|
| ||||
| Baseline characteristics | ||||
| Age (years) | 24 (18 – 60) | 25 (18 – 59) | 26 (18 – 59) | 25 (18 – 60) |
| Weight (kg) | 53.3 (36.1 – 90.5) | 55.2 (36.4 – 88.5) | 55.5 (36.0 – 88.2) | 54.0 (36.0 – 90.5) |
| Body mass index (kg/m2) | 21.2 (15.7 – 34.9) | 21.2 (15.9 – 29.4) | 21.4 (15.4 – 36.7) | 21.2 (15.4 – 36.7) |
| Sex | ||||
| Male | 37 (63.8%) | 39 (68.4%) | 31 (58.5%) | 107 (63.7%) |
| Female | 21 (36.2%) | 18 (31.6%) | 22 (41.5%) | 61 (36.3%) |
| Baseline ALT (IU/L) | 24 (6 – 79) | 20 (7 – 63) | 22 (6 – 59) | 22 (6 – 79) |
| Baseline AST (IU/L) | 22 (12 – 77) | 20 (12 – 64) | 19 (11 – 85) | 20 (11 – 85) |
| Rifampin dose (mg) | 450 (300 – 750) | 900 (450 – 1,200) | 1,200 (600 – 1,500) | 600 (300 – 1,500) |
|
| ||||
| Pharmacokinetics | ||||
| Pyrazinamide AUC0-6h (h∙mg/L) | 167.8 (74.1 – 277.9) | 164.4 (114.7 – 294.0) | 163.9 (77.4 – 266.2) | 166.3 (74.1 – 294.0) |
| Pyrazinamide Cmax (mg/L) | 35.2 (21.1 – 66.4) | 33.4 (19.6 – 61.9) | 34.4 (18.9 – 54.3) | 34.6 (18.9 – 66.4) |
| Isoniazid AUC0-6h (h∙mg/L) | 5.6 (1.4 – 18.8) | 8.2 (1.6 – 20.6) | 6.5 (1.3 – 19.3) | 6.9 (1.3 – 20.6) |
| Isoniazid Cmax (mg/L) | 2.4 (0.5 – 6.3) | 2.5 (0.5 – 6.0) | 2.2 (0.6 – 6.7) | 2.3 (0.5 – 6.7) |
| Rifampin AUC0-6h (h∙mg/L) | 24.9 (2.6 – 57.9) | 43.1 (3.5 – 111.6) | 55.5 (13.6 – 132.9) | 36.2 (2.6 – 132.9) |
| Rifampin Cmax (mg/L) | 6.2 (0.6 – 12.5) | 10.2 (0.6 – 27) | 13.3 (3.9 – 39.2) | 9.2 (0.6 – 39.2) |
| Ethambutol AUC0-6h (h∙mg/L) | 7.2 (3.7 – 17.2) | 7.9 (3.8 – 16.8) | 7.8 (3.2 – 17.8) | 7.4 (3.2 – 17.8) |
| Ethambutol Cmax (mg/L) | 1.6 (0.7 – 5.2) | 2.0 (0.7 – 4.5) | 1.8 (0.9 – 5.6) | 1.7 (0.7 – 5.6) |
Data are expressed as median (range) except for sex which is expressed as number of participants (%).
Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; AUC0-6h, area under the concentration-time curve from 0 to 6 hours; Cmax, maximum plasma concentration.
Risk Factors for Grade 2 or Higher ALT or AST Elevation
Grade 2 or higher ALT or AST elevation was observed in 39 (23%) of the 168 participants with available PK data, with a median time to event of 27 (range: 13–68) days, and in 24 (27%) of the 90 participants with available NAT2 genotype, with a median time to event of 27 (range: 12–56) days. Time to grade 2 or higher ALT or AST elevation did not show any apparent trend across randomized rifampin dose groups (Figure 1). Pyrazinamide and isoniazid exposures were significantly higher in participants with grade 2 or higher ALT or AST elevation compared with those without elevation, while rifampin and ethambutol exposures were similar regardless of elevation (Figure 2).
Figure 1.
Longitudinal profiles of (a) ALT grade and (b) AST grade over time since treatment initiation for individual participants, stratified by randomized rifampin dose. The bar colors represent severity grade of ALT or AST elevation. The open circle and filled triangle represent the onset and resolution of grade 2 or higher elevation, respectively. Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase.
Figure 2.
Comparison of (a) pyrazinamide, (b) isoniazid, (c) rifampin, and (d) ethambutol AUC0–6h between participants with and without grade 2 or higher ALT or AST elevation. The filled circles represent the observed values. The lower and upper hinges represent the 25th and 75th percentiles, and the middle line represents the median. The whiskers extend from the hinge to the smallest or largest value no further than 1.5 times the interquartile range from the hinge. Significance was evaluated using an independent two-sample t-test. Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; AUC0–6h, area under the concentration-time curve from 0 to 6 hours.
In univariable Cox analysis, higher exposures to pyrazinamide and isoniazid, and slow NAT2 acetylator status among the subset with known NAT2 genotype, were associated with an increased hazard of grade 2 or higher ALT or AST elevation: hazard ratio [HR] 1.85 (95% CI, 1.23–2.81) for every 50 h∙mg /L increase in pyrazinamide AUC0–6h; HR 1.40 (95% CI, 1.04–1.87) for every 5 h∙mg/L increase in isoniazid AUC0–6h; HR 9.32 (95% CI, 1.22–70.92) for slow relative to fast NAT2 acetylator status (Table 2, Figure 3). Neither rifampin dose nor exposure, nor ethambutol exposure, or other demographic or clinical factors were associated with the hazard of grade 2 or higher ALT or AST elevation (Table 2, Supplementary Table 1). A sensitivity analysis of the Cox analyses described above adjusted for randomized rifampin dose yielded the same independent risk factors as the unadjusted analyses (Supplementary Table 2).
Table 2.
Cox proportional hazard ratios for grade 2 or higher ALT or AST elevation in univariable, full, and final models.
| Variable | Participant (n) | Event (n) | Univariable model | Full modela | Final model |
|---|---|---|---|---|---|
|
| |||||
| HR (95% CI) | HR (95% CI) | HR (95% CI) | |||
|
| |||||
| Pyrazinamide AUC0-6h (per 50 h∙mg/L increase) | 168 | 39 | 1.85 (1.23 – 2.81)** | 1.76 (0.99 – 3.15) | 1.85 (1.23 – 2.81)** |
| Pyrazinamide Cmax (per 10 mg/L increase) | 168 | 39 | 1.56 (1.10 – 2.22)* | - | - |
| Isoniazid AUC0-6h (per 5 h∙mg/L increase) | 168 | 39 | 1.40 (1.04 – 1.87)* | 1.35 (0.97 – 1.87) | - |
| Isoniazid Cmax (per 1 mg/L increase) | 168 | 39 | 1.29 (1.03 – 1.61)* | - | - |
| Rifampin AUC0-6h (per 50 h∙mg/L increase) | 168 | 39 | 1.07 (0.58 – 1.96) | 0.63 (0.31 – 1.29) | - |
| Rifampin Cmax (per 10 mg/L increase) | 168 | 39 | 0.97 (0.58 – 1.63) | - | - |
| Ethambutol AUC0-6h (per 5 h∙mg/L increase) | 168 | 39 | 1.46 (0.91 – 2.34) | 1.52 (0.88 – 2.63) | - |
| Ethambutol Cmax (per 1 mg/L increase) | 168 | 39 | 1.15 (0.84 – 1.57) | - | - |
| Rifampin dose (per 150 mg increase) | 180 | 46 | 1.05 (0.91 – 1.22) | - | - |
| Rifampin 15 mg/kg/day (relative to 10 mg/kg/day) | 60 (vs. 60) |
16 (vs. 16) |
0.99 (0.49 – 1.97) | - | - |
| Rifampin 20 mg/kg/day (relative to 10 mg/kg/day) | 60 (vs. 60) |
14 (vs. 16) |
0.84 (0.41 – 1.71) | - | - |
| NAT2 intermediate acetylator (relative to NAT2 fast acetylator) | 42 (vs. 16) |
9 (vs. 1) |
3.63 (0.46 – 28.62) | - | - |
| NAT2 slow acetylator (relative to NAT2 fast acetylator) | 32 (vs. 16) |
14 (vs. 1) |
9.32 (1.22 – 70.92)* | - | - |
| NAT2 slow acetylator (relative to NAT2 intermediate acetylator) | 32 (vs. 42) |
14 (vs. 9) |
2.57 (1.11 – 5.94)* | - | - |
The full multivariable model includes all prespecified predictors of interest (excluding Cmax parameters and NAT2 acetylator status) and potential confounders, regardless of statistical significance.
p < 0.05;
p < 0.01
Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; AUC0-6h, area under the concentration-time curve from 0 to 6 hours; CI, confidence interval; Cmax, maximum plasma concentration; HR, hazard ratio; NAT2, N-acetyltransferase 2.
Figure 3.
Kaplan-Meier curves for first grade 2 or higher ALT or AST elevation, stratified by median exposure to (a) pyrazinamide, (b) isoniazid, (c) rifampin, and (d) ethambutol as defined by AUC0–6h, and by (e) NAT2 acetylator status. Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; AUC0–6h, area under the concentration-time curve from 0 to 6 hours; NAT2, N-acetyltransferase 2.
In the final Cox model, only higher pyrazinamide exposure remained significantly associated with an increased hazard of grade 2 or higher ALT or AST elevation (Table 2). All other candidate risk factors did not meet statistical criteria for inclusion in the model after adjusting for pyrazinamide exposure. A sensitivity analysis in the subset of 84 participants with both available PK data and NAT2 genotype showed that higher isoniazid exposure and slow NAT2 acetylator status retained a nonsignificant trend towards increased hazard of grade 2 or higher ALT or AST elevation (Supplementary Table 3).
Longitudinal ALT/AST analysis
A total of 891 ALT and 958 AST values from 168 participants were included in the longitudinal analysis (Supplementary Figure 3). The longitudinal ALT and AST profiles were well described by linear models, combined with a two-subpopulation mixture model representing low and high elevations (Supplementary Figure 4, Supplementary Figure 5, Supplementary Table 4). While exposures to pyrazinamide and isoniazid were significantly higher in the model-assigned high ALT and AST elevation groups compared to the low elevation groups, exposure to rifampin was similar across the model-assigned elevation groups (Supplementary Figure 6), consistent with the findings from survival analysis. The model-predicted slopes of ALT and AST elevation were significantly associated with exposures to isoniazid, pyrazinamide, and ethambutol, but not rifampin (Figure 4). Drug exposure parameters, however, did not meet the prespecified statistical criteria for inclusion and thus were not included in the final longitudinal model (Supplementary Table 4).
Figure 4.
Association between observed (a) pyrazinamide, (b) isoniazid, (c) rifampin, and (d) ethambutol AUC0–6h and the longitudinal model-predicted slope of ALT and AST elevation. The blue and red filled circles represent the observed values in the model-assigned low and high ALT and AST elevation groups, respectively. The lines represent the predicted mean lines from linear regression, and the shaded areas represent the 95% confidence intervals for the regression lines. Significance was evaluated using linear regression. Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; AUC0–6h, area under the concentration-time curve from 0 to 6 hours.
DISCUSSION
We found that rifampin exposures achieved with doses up to 20 mg/kg/day were not associated with hepatotoxicity, further supporting the topline results of the HIRIF trial, which observed no increased risk of hepatic adverse events with higher rifampin doses [14]. Although our dataset was moderately sized, we did not observe any clear trends in the hazard of hepatotoxicity with higher rifampin exposure in either univariable or full multivariable models, indicating no apparent association. The current standard dose of rifampin (10 mg/kg) is widely considered subtherapeutic and has been associated with unfavorable treatment outcomes [19,20]. Accumulating evidence from both preclinical and clinical studies has shown the potential of higher rifampin doses to enhance therapeutic efficacy without a significant increase in safety concerns [21–27], making it a promising strategy to shorten treatment duration of the standard regimen for DS-TB. While prior studies primarily linked rifampin dose levels to safety outcomes, our study directly evaluated the association between continuous rifampin PK, which more precisely reflect drug exposure in the body, and hepatotoxicity risk. Our results provide robust rationale that higher rifampin doses up to 20 mg/kg pose little concern for hepatotoxicity compared to the current standard, thereby supporting its clinical use from a safety perspective.
Higher pyrazinamide exposure was the strongest predictor of hepatotoxicity in this study. Our findings align with the largest pyrazinamide PK-safety analysis to date. Data from the Study 31/ACTG A5349 (S31/A3549) Phase 3 trial showed that increased pyrazinamide exposure in the 6-month standard regimen was associated with a higher risk of ALT or AST ≥5 times the upper limit of the normal range [28]. While the precise mechanisms underlying pyrazinamide-induced hepatotoxicity are not fully understood, increased pyrazinamide metabolites have been reported to be highly correlated with the extent of hepatotoxicity [29], underscoring the importance of optimizing pyrazinamide exposure.
Since treatment interruption can often resolve hepatotoxicity but may also result in suboptimal drug exposure and potential treatment failure and drug resistance [5], it is crucial to proactively minimize pyrazinamide overexposure. Therapeutic drug monitoring (TDM) of pyrazinamide could help optimize exposure, however, the current capacity for TDM in high-burden clinical settings is limited. Pending increased capacity, flat dosing of pyrazinamide could be a feasible strategy to mitigate the risk of hepatotoxicity in the standard regimen for DS-TB. The S31/A3549 pyrazinamide analysis highlighted that flat dosing of pyrazinamide 1000 mg ensures a higher proportion of participants remain within the therapeutic window compared with weight-banded dosing, particularly by reducing the risk of high pyrazinamide exposure—and toxicity—in higher weight bands [28].
We found that higher isoniazid exposure and slow NAT2 acetylator status were significant predictors of hepatotoxicity in univariable analysis. NAT2 genotype was not included in the multivariable analysis due to limited availability, and in the final model, the contribution of higher isoniazid exposure to hepatotoxicity was eclipsed by the outsize contribution of pyrazinamide exposure. Our findings indicate that optimizing isoniazid exposure may still be important for minimizing hepatotoxicity in the standard DS-TB regimen. Isoniazid is predominantly metabolized by the highly polymorphic NAT2 enzyme, and isoniazid-induced hepatotoxicity is known to be primarily attributed to its toxic metabolites, the extent of which depends on NAT2 enzyme activity [5]. Accordingly, NAT2 genotype associated with slow acetylation has been reported to increase the likelihood of hepatotoxicity due to the accumulation of toxic metabolites [7–10]. This supports the notion that slow NAT2 acetylator status, identified as a risk factor in our analysis, is an underlying determinant of isoniazid-related hepatotoxicity risk.
NAT2 genotype-guided isoniazid dose adjustment has been proposed to reduce the risk of isoniazid-induced hepatotoxicity while improving clinical efficacy [30–35]. Furthermore, decreased acetyl-isoniazid levels have been reported to be associated with hepatotoxicity, suggesting the potential for acetyl-isoniazid levels as a biomarker for isoniazid-related hepatotoxicity risk [36,37]. Both genotype- and PK-guided approaches could be beneficial for optimizing isoniazid doses, thereby minimizing the risk of hepatotoxicity in the standard therapy for DS-TB, although practical challenges remain.
In our longitudinal analysis of ALT and AST trajectories, we found that the model-predicted slopes of ALT and AST elevation were significantly and independently associated with exposures to isoniazid, pyrazinamide, and ethambutol, but not rifampin. However, these PK parameter associations did not meet prespecified statistical criteria for inclusion in the final longitudinal model. These findings align with the results from the survival analysis, further suggesting that pyrazinamide and isoniazid, rather than rifampin, are contributors to hepatotoxicity.
Only results for ethambutol were divergent between the two analyses. Although ethambutol is generally not considered a major contributor to hepatotoxicity in standard therapy for DS-TB [7], we noted a significant correlation between its AUC0–6h and transaminase elevation slopes in longitudinal analysis. Given that weight-based dosing was used in the HIRIF trial and that ethambutol exposure correlated with pyrazinamide exposure, the strongest predictor of hepatotoxicity (Supplementary Figure 1), this association does not necessarily imply that ethambutol exposure itself is a direct cause of hepatotoxicity. Rather, it may reflect the influence of weight-based dosing strategies and the interplay between pyrazinamide and ethambutol exposures. One case series reported a high incidence of hepatotoxicity (58%) with combined ethambutol and pyrazinamide therapy for latent TB infection [38], but the independent hepatotoxic potential of ethambutol remains unclear due to the complex nature of multidrug treatment involving pyrazinamide co-administration. Ethambutol has been linked to hepatotoxicity in only a few case reports. One described cholestatic jaundice occurring two months after initiation of ethambutol, isoniazid, and streptomycin, and recurring upon rechallenge with ethambutol alone [39]. This provided convincing evidence of ethambutol’s role, although the cholestatic pattern is atypical in anti-TB drug-induced hepatotoxicity.
In conclusion, higher exposures to pyrazinamide and isoniazid, and slow NAT2 acetylator status, were each associated with increased hepatotoxicity, and only pyrazinamide exposure was associated when considering pharmacokinetic variables together. Rifampin exposure as achieved by doses of up to 20 mg/kg/day was not associated with hepatotoxicity. Because our analysis focused specifically on hepatotoxicity, one of the most common adverse events in the standard regimen for DS-TB, it does not necessarily reflect overall tolerability. Nevertheless, our findings support further evaluation of rifampin at doses up to 20 mg/kg/day in a larger clinical trial and its potential clinical use from a safety perspective, providing valuable insights into the hepatotoxicity risk associated with drugs within this regimen.
Supplementary Material
Data Availability:
The data are available upon request.
Acknowledgment:
E.Y., P.V.B., and G.E.V. contributed to the analysis plan. E.Y. and P.V.B. prepared the data for analysis. C.A.P. contributed to the pharmacokinetic data, and E.A.M., G.P., and G.R.D. contributed to the pharmacogenomic data. E.Y. performed the analysis and wrote the first draft of the manuscript with support from G.E.V. E.Y., P.V.B., R.M.S., G.R.D., C.D.M., and G.E.V. interpreted the data. C.A.P., M.B.B., J.M.C., D.V.V., E.S.G., R.I.C., J.J., K.T., D.G., E.O., L.L., G.R.D., and C.D.M. were members of the study team who executed the trial and collected the data. All authors critically revised the manuscript for important intellectual content and approved the final manuscript for publication.
Funding:
This work was supported by the National Institute of Allergy and Infectious Diseases (NIAID) at the National Institutes of Health (NIH) (grant number U01 AI091429 to C.D.M. and G.R.D. and K08 AI141740 to G.E.V.).
Footnotes
Prior Presentations: The work presented in this manuscript was presented in part at the Union World Conference on Lung Health in Paris, France, in 2023.
Conflict of Interest: Rifampin for the HIRIF trial was donated by Sanofi-Aventis. There are no other potential conflicts of interest to declare.
References
- 1.Global tuberculosis report 2024. Geneva: World Health Organization, 2024. Available at: https://www.who.int/publications/i/item/9789240101531. Accessed 11 July 2025. [Google Scholar]
- 2.Yee D, Valiquette C, Pelletier M, Parisien I, Rocher I, Menzies D. Incidence of serious side effects from first-line antituberculosis drugs among patients treated for active tuberculosis. Am J Respir Crit Care Med. 2003. Jun 1;167(11):1472–1477. [DOI] [PubMed] [Google Scholar]
- 3.Saukkonen JJ, Duarte R, Munsiff SS, et al. Updates on the treatment of drug-susceptible and drug-resistant tuberculosis: an official ATS/CDC/ERS/IDSA clinical practice guideline. Am J Respir Crit Care Med. 2025. Jan;211(1):15–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.WHO consolidated guidelines on tuberculosis: module 4: treatment and care. 2025. Available at: https://www.who.int/publications/i/item/9789240107243. Accessed 11 July 2025. [PubMed]
- 5.Tostmann A, Boeree MJ, Aarnoutse RE, de Lange WCM, van der Ven AJAM, Dekhuijzen R. Antituberculosis drug-induced hepatotoxicity: concise up-to-date review. J Gastroenterol Hepatol. 2008. Feb;23(2):192–202. [DOI] [PubMed] [Google Scholar]
- 6.Bastard M, Sanchez-Padilla E, Hewison C, et al. Effects of treatment interruption patterns on treatment success among patients with multidrug-resistant tuberculosis in Armenia and Abkhazia. J Infect Dis. 2015. May 15;211(10):1607–1615. [DOI] [PubMed] [Google Scholar]
- 7.Ramappa V, Aithal GP. Hepatotoxicity related to anti-tuberculosis drugs: mechanisms and management. J Clin Exp Hepatol. 2013. Mar;3(1):37–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cheli S, Torre A, Schiuma M, et al. NAT2 slow acetylator phenotype as a significant risk factor for hepatotoxicity caused by antituberculosis drugs: results from a multiethnic nested case-control study. Clin Infect Dis 2025. Aug 1;81(1):145–152. [DOI] [PubMed] [Google Scholar]
- 9.Wang P-Y, Xie S-Y, Hao Q, Zhang C, Jiang B-F. NAT2 polymorphisms and susceptibility to antituberculosis drug-induced liver injury: a meta-analysis. Int J Tuberc Lung Dis. 2012. May;16(5):589–595. [DOI] [PubMed] [Google Scholar]
- 10.Khan S, Mandal RK, Elasbali AM, et al. Pharmacogenetic association between NAT2 gene polymorphisms and isoniazid induced hepatotoxicity: trial sequence meta-analysis as evidence. Biosci Rep. 2019. Jan 15;39(1):BSR20180845. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Jindani A, Atwine D, Grint D, et al. Four-month high-dose rifampicin regimens for pulmonary tuberculosis. NEJM Evidence. 2023. Sep;2(9):EVIDoa2300054. [DOI] [PubMed] [Google Scholar]
- 12.Boeree MJ, Heinrich N, Aarnoutse R, et al. High-dose rifampicin, moxifloxacin, and SQ109 for treating tuberculosis: a multi-arm, multi-stage randomised controlled trial. Lancet Infect Dis. 2017. Jan;17(1):39–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Paton NI, Cousins C, Suresh C, et al. Treatment strategy for rifampin-susceptible tuberculosis. N Engl J Med. 2023. Mar 9;388(10):873–887. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Velásquez GE, Brooks MB, Coit JM, et al. Efficacy and safety of high-dose rifampin in pulmonary tuberculosis. A randomized controlled trial. Am J Respir Crit Care Med. 2018. Sep 1;198(5):657–666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Milstein M, Lecca L, Peloquin C, et al. Evaluation of high-dose rifampin in patients with new, smear-positive tuberculosis (HIRIF): study protocol for a randomized controlled trial. BMC Infect Dis. 2016. Aug 27;16(1):453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Peloquin CA, Velásquez GE, Lecca L, et al. Pharmacokinetic evidence from the HIRIF trial to support increased doses of rifampin for tuberculosis. Antimicrob Agents Chemother. 2017. Jul 25;61(8):e00038–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mackay EA, Platt G, Peloquin CA, et al. Impact of pharmacogenetics on pharmacokinetics of first-line anti-tuberculosis drugs in the HIRIF trial. J Infect Dis. 2025. Aug 14;232(2):e258–e265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Carlsson KC, Savić RM, Hooker AC, Karlsson MO. Modeling subpopulations with the $MIXTURE subroutine in NONMEM: finding the individual probability of belonging to a subpopulation for the use in model analysis and improved decision making. AAPS J. 2009. Mar;11(1):148–154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ramachandran G, Chandrasekaran P, Gaikwad S, et al. Subtherapeutic rifampicin concentration Is associated with unfavorable tuberculosis treatment outcomes. Clin Infect Dis. 2020. Mar 17;70(7):1463–1470. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Park JS, Lee J-Y, Lee YJ, et al. Serum levels of antituberculosis drugs and their effect on tuberculosis treatment outcome. Antimicrob Agents Chemother. 2015. Oct 12;60(1):92–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Espinosa-Pereiro J, Aguiar A, Nara E, et al. Safety, efficacy, and pharmacokinetics of daily optimized doses of rifampicin for the treatment of tuberculosis: a systematic review and Bayesian network meta-analysis. Clin Infect Dis. 2025. Aug 1;81(1):129–142. [DOI] [PubMed] [Google Scholar]
- 22.Svensson EM, Svensson RJ, Te Brake LHM, et al. The potential for treatment shortening with higher rifampicin doses: relating drug exposure to treatment response in patients with pulmonary tuberculosis. Clin Infect Dis. 2018. Jun 18;67(1):34–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Gumbo T, Louie A, Deziel MR, et al. Concentration-dependent Mycobacterium tuberculosis killing and prevention of resistance by rifampin. Antimicrob Agents Chemother. 2007. Nov;51(11):3781–3788. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ruslami R, Nijland HMJ, Alisjahbana B, Parwati I, van Crevel R, Aarnoutse RE. Pharmacokinetics and tolerability of a higher rifampin dose versus the standard dose in pulmonary tuberculosis patients. Antimicrob Agents Chemother. 2007. Jul;51(7):2546–2551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Boeree MJ, Diacon AH, Dawson R, et al. A dose-ranging trial to optimize the dose of rifampin in the treatment of tuberculosis. Am J Respir Crit Care Med. 2015. May 1;191(9):1058–1065. [DOI] [PubMed] [Google Scholar]
- 26.Diacon AH, Patientia RF, Venter A, et al. Early bactericidal activity of high-dose rifampin in patients with pulmonary tuberculosis evidenced by positive sputum smears. Antimicrob Agents Chemother. 2007. Aug;51(8):2994–2996. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Boeree MJ, Heinrich N, Aarnoutse R, et al. High-dose rifampicin, moxifloxacin, and SQ109 for treating tuberculosis: a multi-arm, multi-stage randomised controlled trial. Lancet Infect Dis. 2017. Jan;17(1):39–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Xu AY, Velásquez GE, Zhang N, et al. Pyrazinamide safety, efficacy, and dosing for treating drug-susceptible pulmonary tuberculosis: a phase 3, randomized, controlled clinical trial. Am J Respir Crit Care Med. 2024. Dec 1;210(11):1358–1369. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Hussain Z, Zhu J, Ma X. Metabolism and hepatotoxicity of pyrazinamide, an antituberculosis drug. Drug Metab Dispos. 2021. Aug;49(8):679–682. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Jing W, Zong Z, Tang B, et al. Population pharmacokinetic analysis of isoniazid among pulmonary tuberculosis patients from China. Antimicrob Agents Chemother. 2020. Feb 21;64(3):e01736–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Donald PR, Parkin DP, Seifart HI, et al. The influence of dose and N-acetyltransferase-2 (NAT2) genotype and phenotype on the pharmacokinetics and pharmacodynamics of isoniazid. Eur J Clin Pharmacol. 2007. Jul;63(7):633–639. [DOI] [PubMed] [Google Scholar]
- 32.Cho Y-S, Jang TW, Kim H-J, et al. Isoniazid population pharmacokinetics and dose recommendation for Korean patients with tuberculosis based on target attainment analysis. J Clin Pharmacol. 2021. Dec;61(12):1567–1578. [DOI] [PubMed] [Google Scholar]
- 33.Hein DW, Millner LM. Arylamine N-acetyltransferase acetylation polymorphisms: paradigm for pharmacogenomic-guided therapy- a focused review. Expert Opin Drug Metab Toxicol. 2021. Jan;17(1):9–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Khan N, Das A. Time for isoniazid pharmacogenomic-guided therapy of tuberculosis based on NAT2 acetylation profiles in India. Eur J Drug Metab Pharmacokinet. 2022. Jul;47(4):443–447. [DOI] [PubMed] [Google Scholar]
- 35.Azuma J, Ohno M, Kubota R, et al. NAT2 genotype guided regimen reduces isoniazid-induced liver injury and early treatment failure in the 6-month four-drug standard treatment of tuberculosis: a randomized controlled trial for pharmacogenetics-based therapy. Eur J Clin Pharmacol. 2013. May;69(5):1091–1101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Ulanova V, Kivrane A, Viksna A, et al. Effect of NAT2, GSTM1 and CYP2E1 genetic polymorphisms on plasma concentration of isoniazid and its metabolites in patients with tuberculosis, and the assessment of exposure-response relationships. Front Pharmacol. 2024. Mar 22;15:1332752. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Jeong I, Park J-S, Cho Y-J, et al. Drug-induced hepatotoxicity of anti-tuberculosis drugs and their serum levels. J Korean Med Sci. 2015. Feb;30(2):167–172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Younossian AB, Rochat T, Ketterer J-P, Wacker J, Janssens J-P. High hepatotoxicity of pyrazinamide and ethambutol for treatment of latent tuberculosis. Eur Respir J. 2005. Sep;26(3):462–464. [DOI] [PubMed] [Google Scholar]
- 39.Gulliford M, Mackay AD, Prowse K. Cholestatic jaundice caused by ethambutol. Br Med J (Clin Res Ed). 1986. Mar 29;292 (6524):866. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data are available upon request.