Optimizing Moxifloxacin Dose in MDR-TB Participants with or without Efavirenz Coadministration Using Population Pharmacokinetic Modeling

ABSTRACT

Moxifloxacin is included in some treatment regimens for drug-sensitive tuberculosis (TB) and multidrug-resistant TB (MDR-TB). Aiming to optimize dosing, we described moxifloxacin pharmacokinetic and MIC distribution in participants with MDR-TB. Participants enrolled at two TB hospitals in South Africa underwent intensive pharmacokinetic sampling approximately 1 to 6 weeks after treatment initiation. Plasma drug concentrations and clinical data were analyzed using nonlinear mixed-effects modeling with simulations to evaluate doses for different scenarios. We enrolled 131 participants (54 females), with median age of 35.7 (interquartile range, 28.5 to 43.5) years, median weight of 47 (42.0 to 54.0) kg, and median fat-free mass of 40.1 (32.3 to 44.7) kg; 79 were HIV positive, 29 of whom were on efavirenz-based antiretroviral therapy. Moxifloxacin pharmacokinetics were described with a 2-compartment model, transit absorption, and elimination via a liver compartment. We included allometry based on fat-free mass to estimate disposition parameters. We estimated an oral clearance for a typical patient to be 17.6 L/h. Participants treated with efavirenz had increased clearance, resulting in a 44% reduction in moxifloxacin exposure. Simulations predicted that, even at a median MIC of 0.25 (0.06 to 16) mg/L, the standard daily dose of 400 mg has a low probability of attaining the ratio of the area under the unbound concentration-time curve from 0 to 24 h to the MIC (fAUC_0–24)/MIC target of >53, particularly in heavier participants. The high-dose WHO regimen (600 to 800 mg) yielded higher, more balanced exposures across the weight ranges, with better target attainment. When coadministered with efavirenz, moxifloxacin doses of up to 1,000 mg are needed to match these exposures. The safety of higher moxifloxacin doses in clinical settings should be confirmed.

INTRODUCTION

Moxifloxacin is a bactericidal drug against Mycobacterium tuberculosis and is a key treatment option for both short and long regimens for multidrug- and rifampin-resistant tuberculosis (MDR TB and RR-TB, respectively) (1, 2). The World Health Organization (WHO) recommends either a flat fixed dose of 400 mg regardless of body weight for patients without risk factors for toxicity or a high-dose regimen of 400 to 800 mg daily based on prescribed weight bands (2). Moxifloxacin is readily absorbed from the gastrointestinal tract after oral administration and metabolized to N-sulfate (M1) and acyl glucuronide (M2) conjugates (3, 4) by UDP-glucuronosyltransferase (UGT). Induction of UGT by efavirenz probably explains the 30% reduction in moxifloxacin exposure when efavirenz-based antiretroviral therapy (ART) is dosed concomitantly in patients with drug-susceptible TB (5). Moxifloxacin is a P-glycoprotein substrate, and increased exposures have been associated with the rs4149015 single nucleotide polymorphism (SNP) of SLCO1B1 (6, 7). Approximately 50% of moxifloxacin is bound to plasma proteins (8).

Although moxifloxacin tolerability and efficacy have not been well defined at higher doses, a daily dose of 400 mg is generally well tolerated and thought to be appropriate for patients with susceptible Mycobacterium tuberculosis isolates (MIC < 0.5 mg/L). Conversely, for strains characterized by lower or unknown susceptibility, higher doses may be necessary (9). However, moxifloxacin causes QT interval prolongation, which is more marked at higher moxifloxacin concentrations (10). The efficacy of moxifloxacin has been linked to the ratio of the area under the unbound concentration-time curve from 0 to 24 h to the MIC (fAUC_0–24/MIC) (11, 12). A minimum fAUC_0–24/MIC target ratio of 53 has been proposed based on suppression of moxifloxacin-resistant Mycobacterium tuberculosis mutants in hollow fiber system (HFS) experiments (11). Compared with other fluoroquinolones, moxifloxacin has excellent penetration in diseased lung tissues, including cavitary lesions in MDR-TB participants (13, 14).

We report the PK and MIC distribution of moxifloxacin in participants with MDR-TB and RR-TB, including the effect of efavirenz-based ART. We then apply Monte Carlo methods to predict the steady-state moxifloxacin exposures achieved with currently recommended dosing regimens and compare them against known pharmacokinetic/pharmacodynamic (PK/PD) target values associated with efficacy.

RESULTS

Participant characteristics and PK data.

Moxifloxacin PK data were collected for 131 MDR-TB participants, contributing to 149 PK profiles since 24 participants contributed 2 full PK profiles to investigate the effect of tablet crushing. A total of 900 plasma concentration measurements were available and included in the analysis. Only 19 (2%) of moxifloxacin measurements were below the lower limit of quantification (LLOQ), of which 17 were predose measurements. Table 1 shows the baseline participant characteristics. More than half of the participants (60.3%) were HIV positive, of whom 36.7% were taking efavirenz-based ART at the time of PK sampling.

TABLE 1.

Baseline characteristics of participants^b

Characteristic	Value
No. of participants	131
Female (no. [%])	54 (41.2)
Age (median [IQR] [yrs])	35.7 (28.5–43.5)
Weight (median [IQR] [kg])	47.0 (42.0–54.0)
Height (median [IQR] [m])	1.66 (1.60–1.72)
Fat-free mass (median [IQR] [kg])	40.1 (32.3–44.7)
Body mass index (median [IQR] [kg/m2])	17.1 (16.4–18.3)
ALT (median [IQR] [U/L])	11.0 (7.0–18.0)
Creatinine clearance (median [IQR] mL/min])a	97.3 (86.0–196)
HIV positive (no. [%])	79 (60.3)
Efavirenz coadministration (no. [%])	29 (36.7)

^aComputed using the Cockroft-Gault (50) formula.

^bIQR, interquartile range; ALT, alanine aminotransferase.

Pharmacokinetic model.

Moxifloxacin PK was best described by a two-compartment disposition model, elimination via a liver compartment estimating hepatic extraction, and transit compartment absorption. The two-compartment model proved to be significantly better than a one-compartment model (ΔOFV = 157, 2 degrees of freedom [df], P < 0.001). Inclusion of the liver compartment improved the fit of the model with respect to simple first-order elimination from the central compartment (ΔOFV = 5) and allowed the addition of the effect of changes in clearance on first-pass elimination. The typical value of intrinsic clearance was 35.2 L/h, which corresponds to a hepatic extraction of 20% (based on hepatic blood flow of 70.7 L/h for a typical 41.1-kg of fat-free mass [FFM] of subject and an unbound fraction of 50%) and oral clearance of 17.6 L/h. Moxifloxacin absorption was described by a chain of transit compartments which showed a 25-point improvement in OFV compared to an absorption lag time (1 df) in addition to the 332-point drop in OFV versus a model without a delay in absorption (2 df, P < 0.001). The estimation of the number of transit (NN) compartments could not be robustly assessed, so to stabilize the model, we fixed this value to 18.54 compartments obtained during the development process.

Allometric scaling, which was included in the initial stage of model building, improved the model fit. We found FFM to be the best predictor of PK size (ΔOFV = 67, no additional parameters estimated) compared to total body weight (ΔOFV = 51). Crushing tablets prior to administration did not affect the PK of moxifloxacin. Coadministration of moxifloxacin and efavirenz-based ART resulted in a 79.0% (56.9 to 105) increment in moxifloxacin intrinsic clearance (ΔOFV = 55, 1 df, P < 0.001), corresponding to a 44% (36 to 51) reduction in AUC_0–24.

Between-subject variability (BSV) was supported on intrinsic clearance and bioavailability, while between-visit variability (BVV) was supported on intrinsic clearance only. The model also included between-occasion variability (BOV) on bioavailability, absorption rate, and mean transit time. Table 2 shows the parameter estimates from the final model and the 95% sampling importance resampling (SIR) confidence intervals (CIs). The visual predictive check (VPC) in Fig. 1 shows that the developed model adequately fits the observed concentration-time moxifloxacin data, supporting its use in performing simulations and exploring the probability of target attainment of the PK/PD target.

TABLE 2.

Parameter estimates from the final model

Parameter	Estimate (95% CI)c	Variabilityd (95% CI)c
Intrinsic clearance (L/h)a,b	35.2 (32.8–37.6)	BSV: 27.2 (21.0–33.0); BVV: 17.7 (12.0–24.3)
Central volume (L)b	104 (97.4–110)
Peripheral volume (L)b	49.4 (38.5–72.0)
Intercompartmental clearance (L/h)b	5.48 (4.24–7.12)
Mean transit time (h)	0.393 (0.215–0.592)	BOV: 86.5 (60.0–126)
Transit compartments (NN)	18.54 (fixed)
Absorption rate constant (1/h)	1.25 (1.07–1.53)	BOV: 51.6 (42.6–63.1)
Prehepatic bioavailability	1 (fixed)	BSV: 11.8 (2.94–18.0); BOV: 17.5 (13.6–20.9)
Efavirenz effect on intrinsic clearance (%)	+79.0 (56.9–105)
QH (L/h)e	70.7 (fixed)
fu (%)e	50 (fixed)
RUV proportional error (%)	5.80 (5.11–6.36)
RUV additive error (mg/L)	0.0581 (0.0510–0.0643)

^aClearance (CL/F) estimated in 17.6 L/h by multiplying intrinsic clearance × fraction unbound considering a protein binding of 50%.

^bDisposition parameter are scaled to an individual with a fat-free mass (FFM) of 40.1 kg.

^cCI, confidence interval obtained using sampling importance resampling (SIR) technique.

^dBetween-subject variability (BSV), between-visit variability (BVV), and between-occasion variability (BOV) parameters are reported as approximate coefficient of variation (%). RUV, residual unexplained variability.

^eQ_H, liver blood flow; f_u, unbound fraction in plasma.

FIG 1.

Visual predictive check stratified by efavirenz coadministration. The middle (solid) line is the 50th percentile of the observed concentrations. The lower and upper (dashed) lines are the 2.5th and the 97.5th percentiles of the observed concentrations, respectively. The shaded area is the 95% confidence interval for each percentile. The administered moxifloxacin dose was 400 mg for all participants.

MICs and target attainment.

The MIC was determined in 101 participants, and Fig. 2a shows their distribution; 72% were 0.25 mg/L or below at the start of treatment. Based on individual parameter estimates, the overall median AUC_0–24 on the PK sampling day was 17.2 (range, 2.66 to 41.3) mg·h/L, and 64% (76% among participants on efavirenz-based ART) had an fAUC/MIC_0-24, below the suggested threshold of 53 (Fig. 2b). The subplot in Fig. 2 shows the distribution of AUC_0–24 in the full cohort of participants, with lower values observed in those participants on efavirenz-based ART.

FIG 2.

(a) Distribution of MICs in the cohort. (b) fAUC/MIC_0–24. The ratio fAUC/MIC of 53 is represented as a vertical gray dashed line. The subplot in the right panel shows the moxifloxacin AUC_0–24 for participants with and without coadministration of efavirenz (blue dashed and pink solid bars, respectively).

Simulations.

Figure 3 shows the predicted moxifloxacin AUC_0–24 using several weight band dosing scenarios, the standard flat-dose regimen of 400 mg daily, the high-dose regimen of 600 to 800 mg daily (recommended by WHO for patients not at increased risk of toxicity), and a proposed dosing regimen to compensate for the effect of efavirenz on moxifloxacin exposure.

FIG 3.

Moxifloxacin AUC_0–24 stratified by WHO weight band, dosing regimen, and coadministration of efavirenz (600 mg once a day [QD]). From left to right, the plots show the two regimens currently recommended by WHO (with and without efavirenz coadministration) and the regimen we propose to overcome the efavirenz induction. The boxes show median and interquartile range, and the whiskers show the 5th and 95th percentiles. An AUC_0–24 of 26.5 mg·h/L necessary to achieve a PK/PD of 53 with the most prevalent MIC of 0.25 mg/L is represented by the horizontal gray dashed line.

The simulations of the WHO's flat dose of 400 mg of moxifloxacin show that patients with higher weight achieve lower moxifloxacin exposures. The high-dose WHO regimen (600 to 800 mg daily) achieves larger and more balanced exposures across weight bands, but with efavirenz coadministration, moxifloxacin exposure decreases to levels similar to or lower than those achieved with the 400-mg dose without efavirenz coadministration. To mitigate the induction effect of efavirenz, we propose higher doses of moxifloxacin, increasing the dose to 800 mg for weights <46 kg and 1,000 mg for weights ≥46 kg. Table 3 shows a comparison of moxifloxacin exposures achieved with current WHO regimens and the proposed moxifloxacin dosing schedule, which considers the effect of efavirenz.

TABLE 3.

Proposed moxifloxacin daily dose (mg) by weight considering the effect of efavirenz-based ART administration^b

Weight (kg)	Moxifloxacin daily dose (mg):
	WHO regimens		Proposed regimen for participants on efavirenz-based ART
	Standard dose	High dose
30–35	400	600a	800
36–45	400	600	800
46–55	400	800a	1,000
56–70	400	800	1,000
>70	400	800	1,000

^aSimulated doses from the higher end of the range proposed by WHO.

^bART, antiretroviral therapy.

Figure 4 shows that, by simulating the WHO high-dose regimen for moxifloxacin without efavirenz-based ART, the proportion of simulated patients achieving a target fAUC_0–24/MIC ratio of >53 is above 90% at MICs ≤0.25 mg/L. At the same MIC, the moxifloxacin flat dose of 400 mg has a probability of target attainment lower than 40%, which is similar to exposures achieved with the high-dose moxifloxacin regimen if patients are concomitantly treated with efavirenz. When efavirenz is administered concomitantly, moxifloxacin doses of up to 1,000 mg are needed to achieve therapeutic exposures.

FIG 4.

Probability of target attainment by dosing scenario with or without efavirenz coadministration. The horizontal gray dashed line represents the 90% of probability of target attainment. ART, antiretroviral therapy.

DISCUSSION

We developed a semimechanistic model describing the population PK of moxifloxacin in MDR-TB participants. We report that efavirenz-based ART reduces moxifloxacin exposure by approximately 50%. The current WHO recommendation of a flat 400-mg dose results in lower drug concentrations among heavier participants, while the high-dose regimen achieves more balanced exposures across weight bands. Doses at the higher end of the proposed range may be appropriate, especially against less susceptible Mycobacterium tuberculosis strains. In patients receiving efavirenz-based ART, higher moxifloxacin doses may be required to overcome the drug-drug interaction with efavirenz.

We observed increased intrinsic clearance of moxifloxacin in participants on efavirenz-based ART. The reduction in moxifloxacin exposure is likely a result of the induction of UGT by efavirenz (15) and is consistent with a previous report (5). However, a study by Naidoo et al., (5) where participants with drug-susceptible TB were treated with rifampin and moxifloxacin, ascribed a smaller increase in moxifloxacin clearance in patients treated with efavirenz (42% versus 79% in the current study). Rifampin is a potent inducer of various drug-metabolizing enzymes, including UGT (16); therefore, the effect of efavirenz-based ART on moxifloxacin exposure may have been partially masked by the coadministration of rifampin. The increased intrinsic clearance of moxifloxacin associated with efavirenz use in our study resulted in a 44% reduction in moxifloxacin AUC_0–24, suggesting that a dose adjustment may be required. Guidelines recommending a 400-mg dose of moxifloxacin do not consider the relationship between body size and moxifloxacin clearance. Moreover, the standard and high-dose regimens recommended by WHO do not account for the effect of cotreatment with efavirenz impacting the clearance of moxifloxacin and consequently diminishing exposures, which may lead to the development or amplification of resistance in addition to suboptimal efficacy (11, 17).

Moxifloxacin is considered a key drug in long- and short-term treatment regimens for MDR-TB and has recently been recommended for inclusion together with rifapentine in first-line treatment regimens for drug-susceptible TB (18). Efavirenz is still commonly used in people with TB/HIV coinfection, particularly in some African countries where the proportion of TB patients coinfected with HIV exceeds 50% (19). Efavirenz is considered a preferred nonnucleoside reverse transcriptase inhibitor (NNRTI) over other NNRTIs owing to its better compatibility with some TB drugs (20). The extrapolation of our results to a different population or treatment scenario should be done carefully since factors such as body composition or comedication (e.g., ART or rifapentine) may affect pharmacokinetics.

Our model captures the nonlinear increase in moxifloxacin intrinsic clearance with body size, which explains the low exposures in participants with higher weight bands observed with the standard 400-mg dose. We applied Monte Carlo simulation techniques on the distribution of MICs and AUC_0–24 to predict the doses of moxifloxacin, with and without efavirenz coadministration, required to achieve exposures across the weight bands attaining an fAUC_0–24/MIC of >53 in approximately 65% of participants with MDR-TB.

We included variability in drug resistance patterns via the use of MICs together with the unbound fraction of moxifloxacin to evaluate the probability of target attainment. The PK/PD target based on an HFS for moxifloxacin monotherapy is limited by gaps in its translation to participants, with insufficient evidence regarding free-drug concentrations, drug activity at the microbe-drug interface, and the activity of moxifloxacin as part of a multidrug regimen. Nonetheless, in the absence of clinical PK/PD data, the AUC/MIC ratio remains an important measure of activity.

The MICs in our cohort ranged from 0.06 to 16 mg/L, with the most common being 0.25 (46%), 0.12 (20%), and 0.5 (19%) mg/L; this distribution is similar to that reported previously in a mixed study population that included participants from Africa, Asia, Europe, and South America (21). In a similar population, the median MIC was 0.25 (0.125 to 0.5) mg/L (22). Moreover, limited evidence suggests that among fluoroquinolone-susceptible TB patients, higher MICs are associated with poor outcomes (23). Therefore, increased moxifloxacin doses in patients at risk of lower exposures is justified, provided that safety is not compromised. Another potential benefit of increasing moxifloxacin doses relates to its penetration into lung cavities. Although moxifloxacin accumulates in the lung cavity wall, concentrations in the cavity center are lower than those in blood, so the dose increase may help achieve better exposures in this hard-to-reach site of Mycobacterium tuberculosis infection (24).

With increased doses of moxifloxacin and significant variability in protein binding (25), regular monitoring of the QT interval duration is recommended particularly for patients receiving other QT-prolonging drugs (26, 27). Our simulations show that the AUC_0–24 values achieved with the proposed higher doses are below the range of exposures associated with an increased chance of QT prolongation (28). Nevertheless, the safety of the proposed doses in combination with efavirenz-based ART requires evaluation. Alternative antiretrovirals with fewer drug-drug interactions (e.g., dolutegravir) should be considered for use with moxifloxacin if available. However, the South African national guidelines recommend initiating ART with an efavirenz-containing regimen or using this regimen when starting TB treatment (29).

One limitation of our study is that simulations performed to evaluate the probability of target attainment are based on a PK/PD target derived in vitro and therefore may not necessarily be applicable in all clinical contexts. Second, a clinically validated PK/PD target for moxifloxacin exposure has not been defined (30); therefore, we may have under- or overestimated the potential consequence of the exposures we simulated. Nevertheless, the dose-exposure relationship should be studied further to evaluate the plausibility of regimens, including high doses of moxifloxacin.

Conclusions.

We characterized the PK of moxifloxacin in a South African population treated for MDR-TB and showed that efavirenz reduces moxifloxacin exposure by approximately 50%. Optimized moxifloxacin dosing is required, including the effect of body size on drug disposition, particularly in patients cotreated with efavirenz. Further clinical trials are necessary to validate the PK/PD targets generated from HFS and establish the safety of higher doses of moxifloxacin.

MATERIALS AND METHODS

Participant recruitment.

Participants 18 years or older with MDR-TB or RR-TB were recruited at two TB hospitals in Cape Town, South Africa, between July 2015 and September 2017. Eligibility criteria included pulmonary MDR-TB, either eligible for treatment or initiated on therapy in the month prior, with an available baseline sputum culture. Pregnant women were included if all other eligibility criteria were satisfied. Patients were excluded from the study if critically ill, medically unstable, or unwilling to participate. The standard treatment for MDR-TB at the time of the study consisted of moxifloxacin, terizidone, kanamycin, pyrazinamide, ethionamide, and/or variable doses of isoniazid (5 to 15 mg/kg), depending on the presence of katG and inhA isoniazid mutations identified in the baseline sputum culture (31). The moxifloxacin dose administered to all participants was 400 mg once daily. Ethambutol was added to the treatment regimen if the risk of resistance to ethambutol was considered low and the patient had not been exposed to ethambutol in the month prior to study recruitment.

Ethics.

Ethics approval for the study was granted by the University of Cape Town Human Rights Ethics Committee (HREC 065/2015). Written informed consent was taken from eligible participants prior to participation in a language of their choice (English, Afrikaans, or isiXhosa).

PK sampling.

PK blood sampling was performed on a single occasion approximately 1 to 6 weeks post-treatment initiation. After an overnight fast, participants were dosed with the prescribed antituberculosis drugs, including moxifloxacin. Blood specimens were collected predose and at 2, 4, 6, 8, and 10 h postdose. The plasma was separated by centrifugation within 30 min of sampling and stored at −70°C until analysis. Twenty-four participants were resampled approximately 14 days after the first PK visit, receiving the same study drugs crushed, rather than tablets, mixed with 200 mL of water to investigate the effect of crushing on the exposure of drugs used to treat MDR-TB (32).

Drug concentration analysis.

Moxifloxacin plasma concentrations were determined using liquid chromatography-tandem mass spectrometry performed in the Division of Clinical Pharmacology, University of Cape Town, South Africa. The method was linear over the range of 0.0629 to 12.9 mg/L. The interday accuracy of the assay ranged from 100.6% to 102.2%, and the coefficient of variation (%CV) of the precision ranged from 4.6% to 6.3%. Details of the assay have been described previously (33).

MICs.

Sputum samples were collected prior to treatment initiation for MIC determination. Following decontamination, Mycobacterium tuberculosis bacilli were cultured in MGIT medium; growth was monitored in a Bactec 960 instrument. The MICs were determined using a commercially available Sensititre MycoTB plate (Trek Diagnostics, Cleveland, OH, USA) following the manufacturer’s instructions. The MIC test range for moxifloxacin was 0.06 to 16 mg/L. The MIC values were used to derive the MIC-adjusted drug exposures achieved with selected dosing strategies for moxifloxacin, including scenarios where patients were treated with concurrent efavirenz-based ART.

Data analysis.

Concentration-time data were interpreted using nonlinear mixed-effects modeling in the software NONMEM version 7.4.3 (algorithm FOCE-I) and ancillary software as outlined by Keizer et al. (34). R software version 3.5.2 (35) was used for postprocessing of NONMEM output and generating graphics via RStudio version 1.2.1237 (36). A one- and two-compartment disposition model was explored together with first-order absorption and elimination. Lag time and a chain of transit compartments were investigated to describe the delay in appearance of moxifloxacin in the systemic circulation (37). The well-stirred liver model (38) was investigated to capture hepatic clearance, including the first-pass effect. Oral clearance was estimated based on intrinsic clearance multiplied by unbound fraction. Random-effect variability was assumed to follow a lognormal distribution and was included on several levels. Between-subject variability (BSV) and between-visit variability (BVV) were explored on all disposition parameters, while between-occasion variability (BOV) (39) was considered for all absorption parameters. BVV was included to account for possible changes in PK parameters in patients that were sampled twice, at approximately 14 days apart, to evaluate the effect of administering crushed tablets. To evaluate BOV, each dose and following samples were considered a separate occasion, so the dose before the sampling visit and the predose concentration were treated as separate from the dose administered during the PK visit and the following concentrations. The residual error model included additive and proportional components to describe the deviation of the predicted drug concentration from the observed. Concentrations below the lower limit of quantification (LLOQ = 0.0629 mg/L) were censored and handled with the M6 method proposed by Beal (40), i.e., LLOQ/2 was imputed, and the additive error component for the imputed values was inflated by LLOQ/2. The effect of body size on disposition parameters was described using allometric scaling (either total body weight or fat-free mass [FFM] [41]) with the exponents fixed to 0.75 and 1 for clearance and volume of distribution, respectively (including the parameters for the liver compartment). The effect of other covariates on moxifloxacin exposure was explored, including efavirenz coadministration and HIV status.

Model adequacy was evaluated using the change in objective function value (OFV) to compare nested models, goodness-of-fit plots, and visual predictive checks as advocated by Mould and Upton (42). The parameter uncertainty estimation was determined using the sampling importance resampling (SIR) (43) method.

Monte Carlo simulations based on the final model were used to evaluate the required dose in participants with and without efavirenz to achieve a PK/PD target for moxifloxacin of fAUC_0–24/MIC of >53 (11) across adapted weights bands defined by WHO (25.00 to 35.49, 35.50 to 45.49, 45.50 to 55.49, 55.50 to 69.49, and >69.5 kg) with the MICs observed in this study (38). The in silico population used in the simulations was obtained by pooling data from our study with the demographic data of 1,225 participants with drug-susceptible or -resistant TB from previous studies (38, 44–49). The individual weights ranged from 28 to 111 kg. Simulations were also guided by the currently available tablet strength using increments of 200 or 400 mg. Data will be available upon request.