Impact of Clofazimine Dosing on Treatment Shortening of the First-Line Regimen in a Mouse Model of Tuberculosis

ABSTRACT

The antileprosy drug clofazimine was recently repurposed as part of a newly endorsed short-course regimen for multidrug-resistant tuberculosis. It also enables significant treatment shortening when added to the first-line regimen for drug-susceptible tuberculosis in a mouse model. However, clofazimine causes dose- and duration-dependent skin discoloration in patients, and the optimal clofazimine dosing strategy in the context of the first-line regimen is unknown. We utilized a well-established mouse model to systematically address the impacts of duration, dose, and companion drugs on the treatment-shortening activity of clofazimine in the first-line regimen. In all studies, the primary outcome was relapse-free cure (culture-negative lungs) 6 months after stopping treatment, and the secondary outcome was bactericidal activity, i.e., the decline in the lung bacterial burden during treatment. Our findings indicate that clofazimine activity is most potent when coadministered with first-line drugs continuously throughout treatment and that equivalent treatment-shortening results are obtained with half the dose commonly used in mice. However, our studies also suggest that clofazimine at low exposures may have negative impacts on treatment outcomes, an effect that was evident only after the first 3 months of treatment. These data provide a sound evidence base to inform clofazimine dosing strategies to optimize the antituberculosis effect while minimizing skin discoloration. The results also underscore the importance of conducting long-term studies to allow the full evaluation of drugs administered in combination over long durations.

INTRODUCTION

The treatment of drug-susceptible tuberculosis (TB) requires the administration of multiple drugs for at least 6 months, and the programmatic management of TB requires daily supervision of treatment to ensure regular drug intake (1, 2). In addition to the incredible burden that 6 months of directly observed therapy places on patients, the substantial public health resources and infrastructure required to ensure successful treatment are an enormous burden on TB control programs (3, 4). Thus, although the first-line regimen is highly efficacious, its duration is a significant barrier to implementation and adherence, and inadequate treatment can lead to the selection and spread of drug-resistant Mycobacterium tuberculosis. With approximately 10 million new cases of TB occurring annually (>95% of which are drug-susceptible TB) (5), a shorter first-line treatment regimen could significantly improve TB control efforts around the world.

When incorporated into regimens for multidrug-resistant TB (MDR-TB), clofazimine, a key drug used in leprosy therapy, has consistently been associated with treatment-shortening activity in patients (6–11), and we have demonstrated that clofazimine specifically contributes significant bactericidal and sterilizing activity in a mouse model of MDR-TB treatment (12). This work led to the hypothesis that clofazimine could contribute treatment-shortening activity when incorporated into the first-line regimen for drug-susceptible TB, which subsequently was directly demonstrated in a mouse model (13).

While observational data suggest that the 100-mg dose of clofazimine is active against MDR-TB (7, 9–11), the choice of this dose is not evidence based, and the optimal dosing strategy for this drug remains unknown. Although considered safe for long-term use at the usual dose of 100 mg/day in patients with MDR-TB (6, 8, 11, 14, 15), the administration of clofazimine, a lipophilic red phenazine dye, is associated with reversible dose- and time-dependent tissue accumulation and skin discoloration (16–18). It is therefore prudent to define the minimal dosing scheme for clofazimine that preserves its treatment-shortening activity within TB treatment regimens. The objective of the present research was to systematically evaluate the effects of various clofazimine durations, doses, and companion drugs on the overall bactericidal and sterilizing (i.e., treatment-shortening) activity of the first-line regimen, including assessment of relapse-free cure 6 months after the completion of treatment, in a well-established mouse model of TB chemotherapy.

RESULTS

Study 1.

The objective of the first study was to determine whether clofazimine administered (i) only during the initial phase of treatment or (ii) alone during the continuation phase of treatment contributed treatment-shortening activity similar to that when it was given together with the standard first-line regimen throughout both phases of treatment. We compared the bactericidal and sterilizing activities of two test regimens, 2RHZC₂₅/2RH (where the slash indicates that the regimen has two phases [initial and continuation] of treatment, the number before the regimen phase indicates the duration of administration in months, R is rifampin at 10 mg/kg of body mass, H is isoniazid at 10 mg/kg, Z is pyrazinamide at 150 mg/kg, and C₂₅ is clofazimine at 25 mg/kg) and 2RHZC₂₅/2C₂₅ (Table 1), to a control clofazimine-containing regimen, 2RHZC₂₅/2RHC₂₅, shown previously to cure TB in mice after 3 months of administration (13) (see the study scheme in Fig. S1 and infection details in Tables S1 to S3 in the supplemental material). At the start of treatment (day 0), the mean bacterial burden was 6.67 log₁₀ CFU per lung (standard deviation [SD], 0.34 log₁₀ CFU per lung) (Fig. 1 and Table S4). Six of ten untreated negative-control mice became moribund and were euthanized prior to the month 2 time point. The remaining four mice were sacrificed at month 2 and yielded a mean of 8.37 log₁₀ CFU/lung (SD, 0.65 log₁₀ CFU/lung) (Table S4); these results were expected based on the CFU implanted and the lung CFU counts on day 0.

TABLE 1.

Regimens evaluated in three experimental chemotherapy studies

Control or test	Regimen(s)a
	Study 1	Study 2	Study 3
Negative control	Untreated	Untreated	Untreated
Positive control	2RHZE/2RH, 2RHZC25/2RHC25	2RHZE/4RH, 2RHZEC25/2RHC25	2RHZE/2RH, 2RHZEC12.5/2RHC12.5, 2RHZEC6.25/2RHC6.25, 2RHZEC3.1/2RHC3.1
Test	2RHZC25/2RH, 2RHZC25/2C25	2RHZEC12.5/2RHC12.5, 2RHZEC6.25/2RHC6.25, 2RHZEC3.1/2RHC3.1, 2RHZEC1.5/2RHC1.5	4RHZE, 4RHZEC12.5, 4RHZEC6.25, 4RHZEC3.1

^aA slash indicates that the regimen has two phases (initial and continuation) of treatment. The number before each regimen or regimen phase indicates the duration of administration in months, where 1 month equals 4 weeks. Drug abbreviations and doses are as follows: R, rifampin at 10 mg/kg; H, isoniazid at 10 mg/kg; Z, pyrazinamide at 150 mg/kg; E, ethambutol at 100 mg/kg; C, clofazimine at the dose in milligrams per kilogram indicated in subscript.

FIG 1.

Study 1 lung CFU counts before and during treatment. Data points represent the mean values, and error bars represent the SD. At month 2, 4 of the 10 untreated negative-control mice were still alive and were sacrificed to determine lung CFU counts. Raw CFU count data are presented in Table S3 (day after infection/month −0.5), Table S4 (day 0 and month 2, untreated), and Tables S5 to S7 (treated mice at months 2 to 4, respectively) in the supplemental material.

Study 1: assessment of bactericidal activity.

At the end of the initial phase (i.e., after the first 2 months) of treatment, the mean lung log₁₀ CFU counts declined to 2.77 log₁₀ CFU (SD, 0.63 log₁₀ CFU) in mice receiving RHZE (where E indicates ethambutol at 100 mg/kg) (Fig. 1; see also Table S5 in the supplemental material). As expected, the replacement of ethambutol with clofazimine at 25 mg/kg significantly increased the bactericidal activity, resulting in a mean of 1.72 log₁₀ CFU/lung (SD, 0.61 log₁₀ CFU/lung) (P = 0.0002). In the continuation phase of treatment, the mean bacterial burden was <1 log₁₀ CFU/lung in all groups after 3 months of treatment (Table S6). After 4 months of treatment, the lungs of all mice that received the standard regimen and the control clofazimine-containing regimen, 2RHZE/2RH and 2RHZC₂₅/2RHC₂₅, respectively, were culture negative, while the lungs of mice that received either test regimen remained culture positive at the end of treatment (Table S7), again with no statistically significant differences in lung CFU counts between any of the treatment groups.

Study 1: assessment of sterilizing activity.

The raw CFU/plate reads at relapse assessment for lung homogenates of mice treated for 2, 3, and 4 months are presented in Tables S8, S9, and S10, respectively, in the supplemental material. For mice that received 4 months of the standard regimen (2RHZE/2RH), approximately half (7/15; 46.7%) relapsed with culture-positive lungs 6 months after the completion of treatment (Table 2 and Table S10). Similar proportions of relapse were observed for mice treated for 3 months with 2RHZC₂₅/RH and 2RHZC₂₅/C₂₅, with 5/15 and 7/15 relapses, respectively (Table S9). After 4 months of treatment, mice that received 2RHZC₂₅/2RH had statistically significantly fewer relapses (1/14 mice) than did mice that received the standard regimen (P = 0.035), indicating that this regimen shortened treatment by 1 month compared to the control regimen. As observed previously (13), relapse-free cure resulted in all mice that received 2RHZC₂₅/RHC₂₅ for either 3 or 4 months (Table 2), while the mice that received 2 months of treatment with 2RHZC₂₅ experienced an 80% rate of relapse (12/15 mice) (Table S8), which was not statistically significantly different from the relapse proportion among mice treated for 4 months with the standard regimen (P = 0.13). Thus, compared to the standard regimen, 2RHZC₂₅/RHC₂₅ shortened the duration of treatment by up to 2 months. Based on the data from up to 4 months of treatment, the overall rank order of the sterilizing activity of each regimen was 2RHZC₂₅/RHC₂₅ > 2RHZC₂₅/RH > 2RHZC₂₅/C₂₅ ≈ 2RHZE/RH. The approximate bacterial burdens in the lungs of relapsing mice ranged from 2 CFU to >5 log₁₀ CFU, and differences in lung bacterial burdens were not observed between treatment groups or durations of treatment (Tables S8 to S10).

TABLE 2.

Study 1 relapse results^a

Regimen administered	No. of mice with culture-positive lungs/total no. of mice 6 mo after completion of treatment of duration of:
	2 mo	3 mo	4 mo
2RHZE/2RH	—	—	7/15
2RHZC25/2RHC25	12/15	0/14	0/13
2RHZC25/2RH		5/15	1/14
2RHZC25/2C25		7/15	3/15

^aThe relapse outcome for mice that received 2 months of treatment represents that for mice administered all clofazimine-containing regimens, as the regimens did not differ until the continuation phase of treatment (i.e., all mice in these groups received RHZC₂₅ for the first 2 months). —, not determined.

The results from this study indicated that the contribution of clofazimine, administered at 25 mg/kg, to the anti-TB activity of the first-line regimen was maximized when it was coadministered with the regimen for the entire duration of treatment. In a previous study, we found that the administration of clofazimine monotherapy at 25, 12.5, and 6.25 mg/kg to M. tuberculosis-infected BALB/c mice resulted in dose-independent bactericidal activity, with no difference in bacterial killing between the doses throughout 12 weeks of administration (18). Thus, we hypothesized that clofazimine at doses of <25 mg/kg could contribute similar anti-TB activity when coadministered with the first-line regimen in mice, and the next study was designed to test this hypothesis.

Study 2.

The objective of the second study was to determine the lowest dose of clofazimine that contributes treatment-shortening activity when added to the first-line regimen. We compared the bactericidal and sterilizing activities of four test regimens in which clofazimine was added to the first-line regimen at decreasing doses of 12.5, 6.25, 3.1, and 1.5 mg/kg (Table 1) to those of a control clofazimine-containing regimen, 2RHZEC₂₅/2RHC₂₅, as well as the 6-month standard first-line regimen, 2RHZE/4RH (see the study scheme in Fig. S1 and infection details in Tables S1 to S3 in the supplemental material). In this experiment, clofazimine was added to the complete first-line regimen and was not used in place of ethambutol during the initial phase to mimic more closely clinical conditions of treatment. At the start of treatment (day 0), the mean bacterial burden was 7.11 log₁₀ CFU/lung (SD, 0.10 log₁₀ CFU/lung) (Fig. 2 and Table S11). All 12 untreated negative-control mice were moribund and euthanized 3 weeks after infection; these results were expected based on the CFU implanted and the lung CFU counts on day 0.

FIG 2.

Study 2 lung CFU counts before and during treatment. Data points represent the mean values, and error bars represent the SD. All untreated mice were euthanized prior to the month 1 time point. Raw CFU count data are presented in Table S11 (day 0) and Tables S12, S13, S14, and S15 (months 1, 2, 3, and 4 to 6 of treatment, respectively) in the supplemental material.

Study 2: assessment of bactericidal activity.

After 1 month of treatment with the standard regimen, the mean lung CFU count fell to 5.25 log₁₀ CFU (SD, 0.19 log₁₀ CFU) (Fig. 2; see also Table S12 in the supplemental material). As expected, the addition of clofazimine at 25 mg/kg significantly increased the bactericidal activity of the standard regimen, killing an additional 0.80 log₁₀ CFU/lung (P < 0.0001). When clofazimine was added to the standard regimen at doses ranging from 1.5 to 12.5 mg/kg, an increase in the bactericidal activity was associated with each increase in the dose (Fig. 2). The mean CFU counts in the lungs of mice that received clofazimine at the lowest two doses, 1.5 and 3.1 mg/kg, were not significantly different from the mean counts in mice that received the standard regimen. Although the difference in the mean lung CFU counts was not significantly different from one dose to the next across any of the doses, the trend of the linear dose response, from no clofazimine (standard regimen) to clofazimine at up to 25 mg/kg, was highly statistically significant (P < 0.0001).

The same pattern of bactericidal activity was observed after 2 months of treatment. The mean lung CFU count in mice that received the standard regimen declined to 3.23 log₁₀ CFU (SD, 0.11 log₁₀ CFU), and the addition of clofazimine at 1.5, 3.1, 6.25, 12.5, and 25 mg/kg further decreased the bacterial burden by 0.17, 0.38, 0.75, 1.00, and 1.53 log₁₀ CFU/lung, respectively (Fig. 2 and Table S13). Again, the mean lung CFU counts in mice that received the lowest two doses of clofazimine were not significantly different from the counts in mice receiving RHZE, and the trend for a linear dose response was highly significant (P < 0.0001). The mean lung CFU counts were not significantly different from one dose of clofazimine to the next, except for clofazimine at 25 mg/kg, which yielded a mean lung CFU count that was statistically significantly lower than that associated with any other regimen (P ≤ 0.0001 at most for each regimen). Thus, during the initial phase of treatment, clofazimine administered at any dose added bactericidal activity to the standard regimen in a dose-dependent manner (P < 0.001 for linear dose response).

During the continuation phase of treatment, when the administration of pyrazinamide and ethambutol was discontinued, the lung CFU counts in mice receiving 2RHZE/RH decreased to 1.02 log₁₀ CFU (SD, 0.44 log₁₀ CFU) and 0.07 log₁₀ CFU (SD, 0.16 log₁₀ CFU) after 3 and 4 months of treatment, respectively, and all mice in this treatment group had culture-negative lungs after 5 months of treatment (Fig. 2 and Tables S14 and S15). As expected, all mice that received the standard regimen plus clofazimine at 25 mg/kg had culture-negative lungs after 3 months of treatment, as did mice that received clofazimine at 12.5 mg/kg. Although treatment with 2RHZEC_6.25/RHC_6.25 yielded lung CFU counts that were significantly lower than those with the standard regimen during the initial phase of treatment, the mean lung CFU counts were similar to those in the 2RHZE/RH control group after 3 and 4 months of treatment (Tables S14 and S15). Surprisingly, in mice that received regimens with the lowest two doses of clofazimine, 1.5 and 3.1 mg/kg, the mean lung CFU counts shifted to being statistically significantly higher than the mean count for the 2RHZE/RH control group after 3 months of treatment (P < 0.0001 for both groups). After 4 months of treatment, mice that received the lowest dose of clofazimine (2RHZEC_1.5/2RHC_1.5) had statistically significantly higher mean lung CFU counts than all other groups, including the 2RHZE/2RH control group (P < 0.05 at most for all regimens). Overall, during the continuation phase of treatment, the addition of clofazimine to the standard regimen continued to demonstrate significant dose-dependent bactericidal activity (P < 0.0001 for trend), with mice receiving clofazimine at the lowest two doses, 1.5 and 3.1 mg/kg, exhibiting dose-dependent antagonism of clofazimine with the control regimen.

Study 2: assessment of sterilizing activity.

The raw CFU/plate reads at relapse assessment for the lung homogenates of mice treated for 2, 3, 4, and 5 to 6 months are presented in Tables S16, S17, S18, and S19, respectively, in the supplemental material. Six months after the completion of 5 or 6 months of treatment with the standard regimen (2RHZE/RH), culture-positive relapse was detected in the lungs of 1/15 and 1/14 mice, respectively (Table 3 and Table S19). As expected, mice that received the standard regimen plus clofazimine at 25 mg/kg achieved a similar proportion of relapse-free cure after 3 months of treatment, and a comparable proportion of relapse was also observed for the mice that received 3 months of 2RHZEC_12.5/RHC_12.5 (Table S17). Of mice that received 2RHZEC_6.25/RHC_6.25 for 3 months, 6 out of 15 mice had culture-positive lungs at relapse assessment, a proportion which was higher than, but not statistically significantly different from, the relapse proportion observed for mice that received either the control regimen for 5 or 6 months (P = 0.08) or 2RHZEC_12.5/RHC_12.5 (P = 0.08) or 2RHZEC₂₅/RHC₂₅ (P = 0.2) for 3 months. No relapses were observed for mice that completed 4 months of treatment with the standard regimen plus clofazimine at 6.25, 12.5, or 25 mg/kg (Table S18). Thus, in this experiment, the addition of clofazimine to the standard regimen at either 12.5 or 25 mg/kg decreased the duration of treatment necessary to achieve relapse-free cure by up to 2 months, and the addition of clofazimine at 6.25 mg/kg decreased the duration by 1 month, compared to the standard regimen.

TABLE 3.

Study 2 relapse results

Regimen administered	No. of mice with culture-positive lungs/total no. of mice 6 mo after completion of treatment of duration ofa:
	2 mo	3 mo	4 mo	5 mo	6 mo
2RHZE/4RH	—	—	—	1/15	1/14
2RHZEC1.5/2RHC1.5	14/14	14/14	13/14	NA	NA
2RHZEC3.1/2RHC3.1	14/14	13/15	5/15	NA	NA
2RHZEC6.25/2RHC6.25	15/15	6/15	0/14	NA	NA
2RHZEC12.5/2RHC12.5	15/15	1/15	0/14	NA	NA
2RHZEC25/2RHC25	13/15	2/15	0/15	NA	NA

^a—, not determined; NA, not applicable.

Except for two mice in the 2RHZEC₂₅ group, all mice that received any clofazimine-containing regimen for a duration of 2 months had culture-positive lungs at relapse assessment (Table 3). For mice that received 2RHZEC_1.5/RHC_1.5, the proportion of relapse did not decrease with increasing durations of treatment, with culture-positive lungs in 14/14 and 13/14 mice that completed 3 and 4 months of treatment, respectively, although the bacterial burden in the lungs of positive mice tended to decrease with increasing treatment durations (Tables S16 to S18). For mice that received 2RHZEC_3.1/RHC_3.1 for 3 months, the relapse proportion remained high (13/15 mice were culture positive), but after 4 months of treatment, the relapse proportion decreased to 5/15 mice, a proportion which was not statistically significantly different from that observed with the standard regimen administered for 5 or 6 months (P = 0.17). Thus, the antagonism of the bactericidal activity observed during the continuation phase of treatment with the addition of clofazimine at 3.1 mg/kg to the control regimen was not evident in the assessment of the sterilizing activity of this regimen.

These results indicated that the addition of clofazimine to the standard regimen at 12.5 mg/kg contributed treatment-shortening activity equivalent to that with the addition of clofazimine at 25 mg/kg. Although the addition of clofazimine at 6.25 mg/kg was beneficial, the bactericidal activities of all regimens that included clofazimine at doses of <12.5 mg/kg decreased during the continuation phase of treatment, with regimens containing clofazimine at 1.5 or 3.1 mg/kg performing worse than the control regimen. During the initial phase of treatment, when pyrazinamide and ethambutol were coadministered, the addition of clofazimine at any dose increased bactericidal activity compared to the standard regimen alone. Thus, we hypothesized that if pyrazinamide and ethambutol were not removed from the regimen, the clofazimine dose-dependent increase in bactericidal activity would continue after the first 2 months of administration, which could increase the sterilizing activity of regimens that contained clofazimine at doses of <12.5 mg/kg. The next study was designed to test this hypothesis.

Study 3.

The objective of the third study was to determine whether continuing the administration of pyrazinamide and ethambutol during the continuation phase would potentiate the activity of clofazimine and abrogate the antagonism observed in study 2. The bactericidal and sterilizing activities of four test regimens, 4RHZE either alone and supplemented with clofazimine at doses of 3.1, 6.25, and 12.5 mg/kg, with pyrazinamide and ethambutol coadministered throughout the duration of treatment, were compared to those of similar regimens in which pyrazinamide and ethambutol were withdrawn after the 2-month initial phase of treatment (Table 1). All regimens were administered for up to 4 months (see the study scheme in Fig. S1 and infection details in Tables S1 to S3 in the supplemental material). At the start of treatment (day 0), the mean lung CFU count was 7.73 log₁₀ CFU (SD, 0.18 log₁₀ CFU) (Fig. 3 and Table S20). All untreated negative-control mice were moribund and euthanized in the third week after infection; these results were expected based on the CFU implanted and the lung CFU counts on day 0.

FIG 3.

Study 3 lung CFU counts before and during treatment. Data points represent the mean values, and error bars represent the SD. All untreated mice were euthanized prior to the month 1 time point. Raw CFU count data are presented in Table S20 (day 0) and Tables S21 to S24 (months 1 to 4 of treatment, respectively) in the supplemental material.

Study 3: assessment of bactericidal activity.

During the initial phase of treatment, the mean lung CFU counts in mice treated with the standard regimen decreased to 5.36 log₁₀ CFU (SD, 0.12 log₁₀ CFU) and 3.23 log₁₀ CFU (SD, 0.14 log₁₀ CFU) after 1 and 2 months of treatment, respectively (Fig. 3; see also Tables S21 and S22 in the supplemental material). As observed in study 2 (Fig. 2), the addition of clofazimine at any dose increased the bactericidal activity of the standard regimen in a dose-dependent manner throughout the initial phase of treatment. After 1 month of treatment, the mean bacterial counts in the lungs were significantly lower for mice that received any of the clofazimine-containing regimens than for those that received the control regimen. After 2 months of treatment, the mean lung bacterial counts were 2.87 log₁₀ CFU (SD, 0.18 log₁₀ CFU), 2.56 log₁₀ CFU (SD, 0.25 log₁₀ CFU), and 2.27 log₁₀ CFU (SD, 0.24 log₁₀ CFU) in mice that received RHZEC_3.1, RHZEC_6.25, and RHZEC_12.5, respectively (Table S22). At month 2, treatment with RHZEC_6.25 or RHZEC_12.5 was associated with statistically significantly lower lung CFU counts in mice than treatment with RHZE (P = 0.0044 and P < 0.0001, respectively), and the mean lung CFU count in mice treated with RHZEC_12.5 was significantly lower than that in mice treated with RHZEC_3.1 (P = 0.02). As in study 2, the dose-response trend of the clofazimine supplement was highly statistically significant (P < 0.0001).

During the continuation phase of treatment, the mean lung CFU counts in mice that received the standard regimen declined, as expected, to 1.85 log₁₀ CFU (SD, 0.20 log₁₀ CFU) and 0.13 log₁₀ CFU (SD, 0.28 log₁₀ CFU) after 3 and 4 months of treatment, respectively (Fig. 3 and Tables S23 and S24). No difference in lung CFU counts was observed for mice that received the standard regimen with the continued administration of pyrazinamide and ethambutol in the third and fourth months of treatment. All mice that received 2RHZEC_12.5/RHC_12.5 were culture negative after 3 months, except for one mouse with a single detectable CFU in the lungs, and of mice that received 3 months of continuous administration of RHZEC_12.5, three mice were culture positive, each with a single CFU per lung (Table S23). All mice that received either regimen containing clofazimine at 12.5 mg/kg were culture negative after 4 months of treatment.

As was observed in study 2 (Fig. 2), mice that received 2RHZEC_3.1/RHC_3.1 shifted from having lower CFU counts than those in mice receiving the control regimen during the initial phase of treatment to having higher CFU counts during the continuation phase of treatment (Fig. 3 and Tables S23 and S24). In mice that received 2RHZEC_6.25/RHC_6.25, the mean lung CFU count, which was significantly lower than that in control mice at month 2, was equivalent to that in the lungs of control mice at months 3 and 4. After 3 and 4 months of treatment, no differences in lung CFU counts were observed for mice that received the corresponding regimens in which the administration of pyrazinamide and ethambutol was maintained throughout the continuation phase. Thus, in this experiment, the continued coadministration of pyrazinamide and ethambutol did not impact the bactericidal activity of any regimen tested.

Study 3: assessment of sterilizing activity.

The raw CFU/plate reads at relapse assessment for the lung homogenates of mice treated for 2, 3, and 4 months are presented in Tables S25, S26, and S27, respectively, in the supplemental material. At relapse assessment, culture-positive lungs were observed for 5/14 mice treated with the standard regimen (2RHZE/2RH) for 4 months; a similar proportion of relapse (6/14 mice) occurred for mice that received the 4RHZE regimen (Table 4 and Table S27). For mice treated with 2RHZEC_12.5/RHC_12.5 for 3 months, the proportion of relapse was significantly lower than that for mice treated with the standard regimen for 4 months (P = 0.04), and 3 months of treatment with 2RHZEC_6.25/RHC_6.25 resulted in a relapse proportion that was not significantly different from that for mice treated with 2RHZE/2RH (P = 0.16). No culture-positive relapses occurred in mice treated for 4 months with any regimen containing clofazimine at 6.25 or 12.5 mg/kg. Thus, just as observed in study 2 (Table 3), the addition of clofazimine at 12.5 mg/kg to the standard regimen shortened the duration of treatment necessary to achieve relapse-free cure by more than 1 month, and the addition of clofazimine at 6.25 mg/kg shortened treatment by up to 1 month. The continued administration of pyrazinamide and ethambutol during the continuation phase of treatment did not affect the relapse assessment outcomes for these groups.

TABLE 4.

Study 3 relapse results^a

Regimen administered	No. of mice with culture-positive lungs/total no. of mice 6 mo after completion of treatment of duration ofa:
	2 mo	3 mo	4 mo
2RHZE/2RH	—	—	5/14
4RHZE	—	—	6/14
2RHZEC3.1/2RHC3.1	15/15	14/15	11/14
4RHZEC3.1		14/15	2/11
2RHZEC6.25/2RHC6.25	15/15	9/15	0/15
4RHZEC6.25		7/15	0/15
2RHZEC12.5/2RHC12.5	14/15	0/14	0/15
4RHZEC12.5		1/14	0/15

^aThe relapse outcome for mice that received 2 months of treatment represents that for mice from each pair of regimens receiving the same dose of clofazimine (or no clofazimine), as the regimens did not differ until after the first 2 months of treatment. —, not determined.

As observed in study 2, nearly all mice treated with the standard regimen plus clofazimine at 3.1 mg/kg for 3 months had culture-positive lungs at relapse assessment, and the same was observed for mice that received RHZEC_3.1, with relapse occurring in 14/15 mice from each treatment group (Table 4 and Table S26). For mice that completed 4 months of treatment, the relapse proportion was significantly higher for mice treated with 2RHZEC_3.1/2RHC_3.1 than for mice treated with 4RHZEC_3.1 (P = 0.005), indicating that the continuation of pyrazinamide and ethambutol may prevent the antagonistic effect of clofazimine at this dose. The approximate lung bacterial burdens in relapsing mice were similar for both treatment groups (Table S27).

DISCUSSION

Using a BALB/c mouse model of TB chemotherapy, we demonstrated that the treatment-shortening activity of clofazimine, when administered in combination with the first-line regimen, is (i) greatest when clofazimine is coadministered throughout the duration of treatment, (ii) equivalent when dosed at 12.5 mg/kg as when dosed at 25 mg/kg, and (iii) largely unaffected when the administration of pyrazinamide and ethambutol is continued throughout the second phase of treatment. These three key findings contribute to the evolving awareness that clofazimine may have a legitimate role to play in shortening TB treatment. However, these experiments also revealed several unexpected and intriguing findings associated with clofazimine that may have implications for its clinical usage.

In a previous study, we found that when administered for 12 weeks as monotherapy in six different dosing schemes (including doses ranging from 6.25 to 25 mg/kg) to M. tuberculosis-infected BALB/c mice, clofazimine exhibited dose-dependent serum and tissue concentrations but dose-independent bactericidal activity (18). This previous work was the foundation for the hypothesis tested here in study 2, namely, that lower doses of clofazimine might shorten treatment with the first-line regimen as effectively as the 25-mg/kg dose that was originally tested (13). However, in both study 2 and study 3, where clofazimine was added to the standard regimen in a dose-ranging manner, we observed very clear and statistically significant dose-dependent activity (Fig. 2 and 3). Part of this discrepancy could be due to experimental differences. BALB/c mice used in the monotherapy study were obtained via in-house breeding and were aged 6 to 8 weeks at the time of aerosol infection (18), while studies 2 and 3 presented here used BALB/c mice obtained from a commercial vendor, all aged 5 weeks at infection. There was indeed slightly more variability in the lung CFU counts in the former study, which perhaps hindered our ability to observe differences, although the clear dose-dependent distinctions observed in clofazimine serum and tissue concentrations in the monotherapy study suggest that interexperiment differences in mice may not fully explain the observed discrepancy in clofazimine's antimicrobial activity. Instead, these data allude to fundamental differences in drug activity observed in the context of single-drug versus multidrug therapy.

The in vivo activity of clofazimine alone against M. tuberculosis has been evaluated by our group and others using several different types of mouse models. When administered as monotherapy at doses of 12.5 to 25 mg/kg/day, clofazimine exhibited modest bactericidal (18–20), bacteriostatic (21–24), or no observable (25) activity in the lungs of tuberculous mice, indicating that clofazimine alone may have limited anti-TB activity. However, we have repeatedly observed that when combined with the first-line drugs rifampin, isoniazid, and pyrazinamide (with or without ethambutol), clofazimine at 12.5 or 25 mg/kg specifically contributes significant bactericidal and sterilizing activity to the combination regimen (13, 26) (studies 1 to 3 in this report). If clofazimine exerts limited antimicrobial activity when administered as monotherapy, the effects of dose ranging on this already limited activity may be difficult to observe. Likewise, the potent activity of clofazimine evident in combination therapy indicates that this context may be more appropriate for evaluating dose-ranging effects of the drug. Importantly, as monotherapy with any drug is never used for the treatment of active TB (1, 2, 27), preclinical assessment of clofazimine's activity within a multidrug regimen may provide data more relevant and informative for clinical translation.

That the activity of clofazimine appears enhanced in the presence of companion drugs is not an unusual phenomenon in TB treatment. In mouse models and in patients with TB, pyrazinamide, a key sterilizing drug, is known to be synergistic with other sterilizing drugs, namely, rifampin (27–30) and bedaquiline (31–34). Clofazimine, administered at 12.5 to 25 mg/kg, has been shown to specifically contribute significant treatment-shortening activity in mouse models when added to a number of different pyrazinamide-containing regimens (12, 13, 35–37) (studies 1 to 3 in this report). Although the mechanism of action is not fully understood for either clofazimine or pyrazinamide, data from numerous biochemical studies indicated that these drugs perturb different components of the mycobacterial oxidative phosphorylation pathway, interfering directly with oxidative ATP synthesis and indirectly with proton gradient-dependent membrane transport systems (38–46). Zhang and colleagues reported the in vitro synergism of clofazimine and pyrazinamide against M. tuberculosis in low-pH culture medium (47), and several researchers have proposed models in which targeting multiple components of the oxidative phosphorylation pathway results in the synergistic killing of M. tuberculosis (39, 42–44, 48, 49). Considering these data and models, it is reasonable to hypothesize that clofazimine and pyrazinamide have synergistic activity in vivo against M. tuberculosis. In this case, one would expect that the contribution of clofazimine to the first-line regimen would be diminished by the removal of pyrazinamide.

Indeed, in study 2, we observed that during the continuation phase of treatment, when pyrazinamide and ethambutol were no longer administered, the bactericidal activity of clofazimine was reduced for doses of 6.25 mg/kg and lower, while the activity of clofazimine at doses of 12.5 and 25 mg/kg appeared unaffected by the removal of pyrazinamide and ethambutol (Fig. 2). This apparent difference between the higher and loser doses may indicate that the higher doses of clofazimine were potent enough that the removal of pyrazinamide and ethambutol had little impact on mice that were fast approaching lung culture conversion, and/or shorter (such as weekly) intervals between time points may have been necessary to observe any such impact. However, as one of our main objectives was to evaluate if lower doses of clofazimine could contribute treatment-shortening activity to the standard TB regimen, it was logical to evaluate the activity of low-dose clofazimine regimens without the removal of pyrazinamide and ethambutol after the first 2 months of treatment. In study 3, where clofazimine was added to the standard regimen at doses of 3.1, 6.25, and 12.5 mg/kg, we observed that, in contrast to our hypothesis, the continued administration of pyrazinamide and ethambutol had no discernible impact on the bactericidal activity of the clofazimine-containing regimens (Fig. 3) or the sterilizing activity of clofazimine at 6.25 mg/kg (Table 4). For the regimen that contained clofazimine at 3.1 mg/kg, there was significantly less relapse after 4 months of treatment in mice that received 4RHZEC_3.1 (18% relapse) than in mice that received 2RHZEC_3.1/2RHC_3.1 (70% relapse), suggesting that the continued administration of pyrazinamide and ethambutol was beneficial when combined with this relatively low dose of clofazimine. However, in study 2, 4 months of 2RHZEC_3.1/2RHC_3.1 resulted in a 33% relapse rate (Table 3), which is not different from the relapse proportion in mice that received 4RHZEC_3.1 in study 3. Therefore, it seems that, overall, the coadministration of pyrazinamide and ethambutol during the continuation phase did not affect the activity of clofazimine, although further studies are needed to more clearly evaluate the role of pyrazinamide and ethambutol with low-dose clofazimine (i.e., doses of ≤3.1 mg/kg) during the continuation phase.

In TB treatment, pyrazinamide is administered for only 2 months because a number of clinical trials demonstrated that continued administration beyond 2 months did not improve patient outcomes (50–53), a finding that has also been demonstrated in BALB/c mouse models of TB treatment (30) (study 3 in this report). Thus, the potent, synergistic, bactericidal, and sterilizing activity of the rifampin-pyrazinamide combination, which permitted the shortening of the TB treatment regimen from 9 to 6 months (27, 54), not only occurs but also is maximized when pyrazinamide is coadministered only for the first 2 months. In the mouse model, a similar finding was observed when pyrazinamide was added to a bedaquiline-containing regimen (55). Therefore, it still seems reasonable to hypothesize that clofazimine and pyrazinamide may have synergistic activity in vivo against M. tuberculosis, which occurs, and is perhaps maximized, when pyrazinamide is coadministered for the first 2 months of treatment. Additional studies are necessary to better understand the relationship between clofazimine and pyrazinamide in TB therapy.

Although the above-stated hypothesis may explain why a dose-ranging effect of clofazimine was so evident when it was coadministered with the first-line regimen and why the continued administration of pyrazinamide and ethambutol beyond 2 months did not impact treatment outcomes, we are still left with the question of why, during the continuation phase of treatment, clofazimine at doses of 1.5 and 3.1 mg/kg was antagonistic with the control regimen (Fig. 2 and 3). The finding that the lung CFU counts in mice receiving these doses of clofazimine continued to decline in a dose-dependent manner after 3 and 4 months of treatment indicates that selection of drug-resistant M. tuberculosis was unlikely to be the cause. It is possible that there is some specific antagonism, which may perhaps be concentration dependent, between clofazimine and rifampin, isoniazid, and/or the rifampin-isoniazid combination. It would be interesting to know if this antagonism would be evident earlier in treatment, and also with higher doses of clofazimine, if pyrazinamide was given for only 1 month or omitted from the treatment regimen altogether.

Another consideration is the role that pharmacokinetics (PK) may have played in the observed antagonism between low-dose clofazimine and the standard regimen during the latter phase of treatment. In humans and mice, rifampin is known to induce its own clearance by upregulating the activity of drug-metabolizing enzymes and drug transporter proteins, including P-glycoprotein (56–61). Therefore, once rifampin induces its own clearance, drug exposure levels will decrease gradually until steady state is eventually reached, which in humans has been estimated to take from 1 week to up to 6 weeks (56, 60, 61). Considering the results observed in studies 2 and 3, if rifampin exposures were reduced over time, one would expect this to affect the overall activity of all treatment groups and not just the low-dose clofazimine groups. Thus, rifampin exposures alone most likely do not account for the observed antagonism.

In contrast to rifampin, PK properties, including drug-drug interactions, of clofazimine have not been well studied. There are conflicting data from biochemical and cell culture assays regarding clofazimine's ability to induce (62) or inhibit (63) human cytochrome P450 enzymes. Similarly, there are conflicting reports regarding clofazimine's ability to block (64) or be exported by (62) P-glycoprotein, and short-term clinical PK studies have found that clofazimine reduces (65) or has no impact on (66, 67) rifampin exposures. These discrepant results may be due, at least in part, to the different methods by which free-drug concentrations were estimated and applied in different analyses. In the BALB/c mouse model, we have observed that when clofazimine is dosed at 12.5 or 25 mg/kg either alone or in combination with the standard regimen, the serum clofazimine concentrations slightly decline after 2 months of administration (13, 18), which could indicate an autoinduction of liver enzymes and/or transporters, although this has yet to be specifically evaluated. If a similar effect occurs with lower doses of clofazimine, then it is possible that the serum concentrations shift to subinhibitory concentrations. If clofazimine-induced enzymes and/or transporters enhance and/or are different from those induced by rifampin, then clofazimine-driven drug clearance could further reduce the amount of rifampin to levels below those in mice receiving the standard regimen only. Additional experiments, designed to specifically capture rifampin and clofazimine levels throughout the course of treatment, are needed to test this hypothesis.

Finally, a critical issue remains to be addressed, namely, how do these preclinical findings translate to patients? Unfortunately, and despite decades of clofazimine administration to millions of leprosy patients, in which clofazimine is typically taken daily for 1 or more years (68, 69), no long-term clinical PK-focused study of clofazimine has been conducted in humans. Translation of our preclinical data requires an understanding of the PK correlations between mice and humans for the drugs and regimens involved. For the first-line anti-TB drugs, mouse dosing is based on achieving drug exposures that closely recapitulate those obtained with the recommended dosing in humans (29, 70). Although there is some debate about whether these officially recommended doses are indeed the optimal doses, especially for rifampin (71, 72), there are clinical PK and pharmacodynamic (PD) data and target PK/PD parameter values known to be associated with safe and effective treatment, supporting the recommended doses for each of the first-line drugs (1, 2). For clofazimine, an evidence base for dosing in patients with TB does not exist, and basic PK/PD parameters, including what exposure levels are required for antimicrobial activity, are unknown. There is no officially endorsed dose of clofazimine for the treatment of drug-resistant TB, even though clofazimine is part of the shorter MDR-TB treatment regimen conditionally recommended by the World Health Organization (6). A daily dose of 100 mg is usually used when clofazimine is included in second-line regimens, and long-term administration of this dose has been generally well tolerated and not associated with severe adverse effects (7–11, 14, 15). It follows that we should establish what clofazimine dose in mice results in exposures similar to those with the 100-mg dose in humans. However, the human data, which are limited in number and variable in outcome, combined with the peculiar PK characteristics of clofazimine, preclude a straightforward analysis.

Clofazimine is a highly lipophilic drug with extensive protein binding and tissue accumulation properties and an extremely long half-life (days to months, depending on the dose and the duration of administration) (73). Therefore, although several single-dose clofazimine PK studies have been conducted with both mice and humans (23, 67, 74–76), such studies are of limited value for predicting long-term, steady-state PK parameters. In 1974, Banerjee and colleagues reported that in patients with leprosy who received clofazimine at 100 mg/day for “periods averaging about 4 years,” the average peak serum concentration was 0.7 μg/ml (77). The international package leaflet for Lamprene, the trade name for clofazimine produced by Novartis, states that after 42 consecutive days of dosing at 100 mg in leprosy patients, the mean trough plasma concentration was about 0.4 μg/ml, although it was noted that the steady state was not yet reached (17). In a previous experiment in which mice were administered clofazimine at 20 mg/kg for 4 weeks, we found mean peak and trough serum levels of 0.78 and 0.47 μg/ml, respectively, indicating that the 20-mg/kg dose in mice was likely equivalent to the 100-mg dose in humans (35). At the time when we began to specifically investigate the treatment-shortening potential of clofazimine, we had interest in comparing the activity of the drug directly, on a milligram-per-kilogram basis, with that of bedaquiline, which is administered at 25 mg/kg in preclinical mouse models of TB (31, 78). As the single-dose PK profiles of clofazimine at 20 and 25 mg/kg were overlapping (18, 35), we considered that either of these doses in mice could reflect the 100-mg daily dose in humans. As we obtained more PK data from preclinical studies, we found that a dose of 25 mg/kg in mice resulted in steady-state trough serum levels of around 1 to 1.2 μg/ml and that a dose of 12.5 mg/kg may more accurately reflect the above-stated human PK parameters associated with the 100-mg dose (13, 18, 21, 26). Because we observed that clofazimine at 12.5 mg/kg is as sterilizing as clofazimine at 25 mg/kg when added to the standard regimen (Tables 3 and 4), this shift in our understanding of the human-mouse PK relationship seemingly should have little bearing on our interpretation of the results with the first-line regimen.

However, just as our understanding of the PK of clofazimine in mice shifted during the several years that it took to complete studies 1 to 3, so too has new information regarding human PK of clofazimine become available. Recently, researchers from Novartis reported results from a physiologically based PK model based on data from leprosy patients; the model predicted that the mean steady-state maximum plasma concentration of clofazimine is 0.797 μg/ml and is achieved after 126 consecutive doses or 18 weeks of treatment (63). Prideaux and colleagues reported data on steady-state clofazimine levels for three patients with mycobacterial lung disease, with one patient taking clofazimine at 100 mg/day (the doses given to the other two patients were not stated); trough plasma concentrations were 0.450, 0.295, and 0.238 μg/ml (79). Data from those studies support the use of 12.5 mg/kg in mice as the bioequivalent dose for 100 mg in humans. However, Diacon and colleagues, who conducted a 14-day early-bactericidal-activity study in which patients with TB were administered clofazimine at 300 mg/day for 3 days followed by 100 mg/day for the next 11 days, reported a median maximum observed plasma concentration of about 0.24 μg/ml (80). Although this was only a 2-week study, the reported clofazimine levels were lower than would be expected, especially with a loading dose, based on the human PK data reported previously (17, 77) and were also lower than levels that we observed with the 12.5-mg/kg dose in mice (18, 21). The difference in mouse and human exposures after 2 weeks may reflect only differences in the time to achieve steady state, which is apparently shorter in mice, but if the time to reach steady state in humans takes 16 to 18 weeks, as reported by Novartis (17, 63), the actual steady-state concentration may not be relevant when the objective is to shorten the total treatment duration to 12 or 16 weeks. In this case, understanding the target exposure levels of clofazimine necessary for its activity is essential.

The MIC of clofazimine for M. tuberculosis strain H37Rv used in our studies is 0.25 μg/ml (22). The median clofazimine MIC value for the isolates obtained from TB patients in the above-described early-bactericidal-activity study was 0.125 μg/ml, with a range of 0.03 to 0.25 μg/ml (80), indicating that PK/PD determined with the H37Rv strain may be relevant to clinical strains. We observed that, as monotherapy, the antimicrobial activity of clofazimine in the lungs of M. tuberculosis H37Rv-infected BALB/c mice is associated with total serum clofazimine levels; after the initial 1- to 2-week delay in the onset of any observable activity, clofazimine exhibits bacteriostatic or bactericidal activity when total serum clofazimine levels are at/below or above the MIC, respectively (21, 26). Any potential synergism with other drugs may further decrease the threshold serum level needed to observe activity when clofazimine is administered as part of a multidrug regimen.

Overall, this work adds significantly to the growing body of evidence that clofazimine has potent treatment-shortening potential when added to TB treatment regimens. Importantly, this work also suggests potential negative impacts of clofazimine at low exposure levels, achieved either through low daily doses or by discontinuing clofazimine administration before TB treatment is completed. To maximize the translational utility of the data from these and other preclinical studies, it is essential to better understand the long-term PK of clofazimine in humans. Specifically, more human PK data, especially for long-term administration and with the coadministration of other drugs, are needed to provide a high level of confidence in the population PK analyses of clofazimine. Addressing this critical knowledge gap will allow additional relevant preclinical studies, for example, to specifically dissect relationships between dose, duration of administration, and drug-drug interactions, that would not be possible to systematically study in patients.

MATERIALS AND METHODS

Three experimental chemotherapy studies were conducted; the overall scheme for each study is presented in Fig. S1 in the supplemental material. In each study, the primary endpoint was relapse-free cure (a measure of sterilizing activity), and the secondary endpoint was the decline in lung CFU counts during treatment (a measure of bactericidal activity). Study 1 was conducted at the KwaZulu-Natal Research Institute for TB-HIV, now the Africa Health Research Institute, in facilities located at the University of KwaZulu-Natal Biomedical Resources Unit at the Westville campus and the Nelson R. Mandela School of Medicine campus in Durban, South Africa. All animal procedures were approved by the Animal Ethics Subcommittee (now the Animal Ethics Committee) of the University of KwaZulu-Natal (permit numbers 050/13/Animal and 025/14/Animal). Studies 2 and 3 were conducted at the Johns Hopkins University School of Medicine Center for Tuberculosis Research in Baltimore, MD. All animal procedures were approved by the Johns Hopkins University Animal Care and Use Committee (permit numbers MO13M23 and MO15M479). All work with live M. tuberculosis was performed in biosafety level 3 laboratories, and all work with infected animals was performed in biosafety level 3 vivaria.

Animals.

Female BALB/c mice (n = 200), aged 6 to 8 weeks and obtained through in-house breeding, were used for study 1, and female BALB/c mice, aged 5 weeks and obtained from Charles River Laboratories, were used for study 2 (n = 421) and study 3 (n = 411). In all studies, mice were housed in individually ventilated cages (five mice per cage) with sterile wood shavings for bedding and with access to food and water ad libitum. Room temperature was maintained at 22°C to 24°C with a 12-h light/dark cycle. Mice were sacrificed by cervical dislocation under isoflurane-induced anesthesia (drop method).

Aerosol infections.

The same stock of mouse-passaged M. tuberculosis strain H37Rv (American Type Culture Collection strain ATCC 27294) was used for all experiments. This stock is susceptible to all drugs used in these experiments (22, 81). Bacterial stocks were cultured for infections as described previously (21). For all studies, mice were infected by aerosol with nebulized bacterial suspensions using a Glas-Col full-size inhalation exposure system according to the manufacturer's instructions; details of the aerosol infections for each study are presented in Table S1 in the supplemental material. The bacterial concentrations of the suspensions used for infection were determined as described previously (21) and as outlined in Table S2. To determine lung implantation, three to five mice from each infection run were sacrificed the day after aerosol infection, and lung CFU counts were determined as described previously (21) and as presented in Table S3.

Treatment.

Mice were randomized into treatment groups and timed sacrifice cohorts a few days before the start of treatment, with mice from each infection run being kept aside (i.e., not randomized) for sacrifice on the day of treatment initiation (day 0) for measurement of the pretreatment lung CFU counts: 5 per infection run (10 mice total) for study 1 and 3 per infection run (12 mice total) each for studies 2 and 3. Lung CFU counts for day 0 were determined as described previously (21). Treatment was initiated 14, 13, and 18 days after infection for studies 1, 2, and 3, respectively. The regimens evaluated in each study are presented in Table 1. In all regimens, rifampin (R), isoniazid (H), pyrazinamide (Z), and ethambutol (E) were administered at 10, 10, 150, and 100 mg/kg of body mass, respectively; these doses are well established for approximating the area under the plasma concentration-time curve produced by standard doses administered to patients with TB (29, 70). The dose of clofazimine, which varied within and between experiments, is denoted by subscript; e.g., C₂₅ indicates the administration of clofazimine at 25 mg/kg. Consistent with our previous clofazimine-related studies of regimens administered for >2 weeks (12, 13, 18, 26), treatment regimens were administered daily (5 days per week, Monday to Friday) by gavage, as detailed in Table S28 in the supplemental material. Rifampin was administered at least 1 h before HZ or HZE combinations to avoid adverse PK interactions (82, 83). In all studies, untreated mice comprised the negative-control group to validate the virulence of the M. tuberculosis strain, and the standard first-line regimen was a positive control, either 2RHZE/4RH administered for the full 6 months (study 2) or a 2RHZE/2RH regimen truncated to 4 months (studies 1 and 3). All drugs were purchased from Sigma-Aldrich/Millipore Sigma. Rifampin, isoniazid, pyrazinamide, and ethambutol were prepared as solutions in distilled water; clofazimine was prepared as a suspension in 0.05% (wt/vol) agarose. All drug solutions/suspensions were prepared to deliver the drug(s) and dose(s) in a total volume of 0.2 ml per gavage, based on an average mouse body mass of 20 g. Drug stocks were prepared in single or multidrug formulations for administration, as outlined in Table S28; drug solutions and suspensions were prepared weekly and stored at 4°C.

Assessment of bactericidal activity.

Bactericidal activity was measured by the decline in lung CFU counts during treatment. For each study, the time points and the number of mice sacrificed at each time point are presented in Fig. S1 in the supplemental material. Mice were sacrificed 72 h after the last treatment dose was administered. Lungs were dissected from the mice and stored in 2.5 ml phosphate-buffered saline (pH 7.4) at 4°C for at least 48 h before homogenization. Lungs were homogenized in TenBroeck glass tissue grinders, and undiluted lung homogenate as well as 10-fold serial dilutions of the homogenate were prepared and cultured as described previously (21), with a volume of 0.5 ml per agar plate. Plates were incubated at 37°C, with (study 1) or without (studies 2 and 3) 5% CO₂, in sealed plastic bags for at least 4 weeks before the final reading.

Assessment of sterilizing activity.

Sterilizing activity was measured by determining the culture status of the lungs 6 months after treatment completion. For each study, the number of mice sacrificed at each relapse time point is presented in Fig. S1 in the supplemental material. Lungs were removed and homogenized as described above for the assessment of bactericidal activity. The entire volume of the undiluted lung homogenate was cultured by using 0.5 ml per agar plate. The plates were incubated at 37°C in sealed plastic bags for at least 6 weeks before the final reading. Relapse was defined as having culture-positive lungs (i.e., the growth of ≥1 CFU from the entire lung homogenate).

Media.

M. tuberculosis was grown in 7H9 broth supplemented with a 10% (vol/vol) oleic acid-albumin-dextrose-catalase (OADC) enrichment, 0.5% (vol/vol) glycerol, and 0.1% (vol/vol) Tween 80. All plating was done on 7H11 agar supplemented with 10% (vol/vol) OADC and 0.5% (vol/vol) glycerol. Lung homogenates (and their cognate 10-fold dilutions) were plated onto selective 7H11 agar, i.e., 7H11 agar further supplemented with 50 μg/ml carbenicillin, 10 μg/ml polymyxin B, 20 μg/ml trimethoprim, and either 10 μg/ml amphotericin B (study 1) or 50 μg/ml cycloheximide (studies 2 and 3), to selectively cultivate mycobacteria while inhibiting the growth of any contaminating bacteria or fungi (84). To detect and limit drug carryover (known to be an issue with clofazimine [12, 37]), lung homogenates from at least all mice that received clofazimine were also plated onto selective 7H11 agar further supplemented with an adsorbent agent, 0.4% activated charcoal, as described previously (37). In study 2, when charcoal-containing agar was used, the concentrations of selective drugs were doubled (2 times the above-stated concentrations) at the month 3, 4, 5, and 6 time points. Difco Middlebrook 7H9 broth powder, Difco Mycobacteria 7H11 agar powder, and BBL Middlebrook OADC enrichment were obtained from Becton, Dickinson and Company. Glycerol and Tween 80 were obtained from Fisher Scientific, and activated charcoal was obtained from J. T. Baker. All selective drugs were obtained from Sigma-Aldrich/Millipore Sigma. Stock solutions of trimethoprim and amphotericin B were prepared in dimethyl sulfoxide, and all other selective drugs were dissolved in distilled water. All drug stocks were filter sterilized, as was the OADC enrichment, prior to use.

Statistical analyses.

All CFU/lung estimates (x) were log transformed as log₁₀(x + 1). For any lung sample that was plated in parallel onto both plain (i.e., charcoal-free) and charcoal-containing selective 7H11 agar, the log₁₀ CFU/lung value determined from the agar type that yielded the higher CFU/lung estimate was used when calculating the mean values and SD. The lower limit of detection was calculated based on the proportion of undiluted sample plated. Comparisons of CFU data between treatment groups and time points were analyzed by using two-way analysis of variance (ANOVA) corrected with Tukey's test for multiple comparisons. Dose-response relationships (dose of clofazimine on lung CFU counts) were analyzed by using one-way ANOVA with a posttest for linear trends. The proportions of mice with culture-positive relapse were compared by using Fisher's exact test. All statistical analyses were performed by using GraphPad Prism 7.02.