Respirable Clofazimine Particles Produced by Air Jet Milling Technique Are Efficacious in Treatment of BALB/c Mice with Chronic Mycobacterium tuberculosis Infection

ABSTRACT

Tuberculosis (TB) remains a major cause of morbidity and mortality, particularly in low- and middle-income countries where access to health care workers, cold-chain storage, and sterile water sources may be limited. Inhaled drug delivery is a promising alternative to systemic delivery of antimycobacterial drugs, as it enables rapid achievement of high infection-site drug concentrations. The off-patent drug clofazimine (CFZ) may be particularly suitable for this route, given its known systemic toxicities. In this study, micronized CFZ particles produced by air jet milling were assessed for shelf-stability, pharmacokinetics, and anti-TB efficacy by the oral and pulmonary routes in BALB/c mice. Intratracheal instillation of micronized CFZ particles produced several-fold higher lung concentrations after a single 30 mg/kg dose compared to delivery via oral gavage, and faster onset of bactericidal activity was observed in lungs of mice with chronic Mycobacterium tuberculosis infection compared to the oral route. Both infection status and administration route affected the multidose pharmacokinetics (PK) of micronized CFZ. Increased lung and spleen accumulation of the drug after pulmonary administration was noted in infected mice compared to naive mice, while the opposite trend was noted in the oral dosing groups. The infection-dependent PK of inhaled micronized CFZ may point to a role of macrophage trafficking in drug distribution, given the intracellular-targeting nature of the formulation. Lastly, air jet milled CFZ exhibited robustness to storage-induced chemical degradation and changes in aerosol performance, thereby indicating the suitability of the formulation for treatment of TB in regions with limited cold chain supply.

INTRODUCTION

Tuberculosis (TB) remains a major cause of global morbidity and mortality (1). International efforts by the World Health Organization (WHO) and other agencies are under way to reduce the global incidence of TB to one case per million by year 2050; however, to achieve this goal, novel therapies and more efficient administration of existing TB therapies are desperately needed. Perhaps the greatest threat to global TB control is the emergence of multidrug-resistant (MDR) and extensively drug resistant-TB (XDR-TB) strains. Conventional treatment of these patients consists of high doses of 4–6 drug combinations delivered orally and parenterally for 18–24 months (1, 2) and often requires the use of second-line drugs that are associated with severe side effects. The continued rise in the number of MDR-TB and XDR-TB cases may be attributed to the transmission of already resistant organisms, poor disease management, drug shortages, use of low-quality drugs, or inappropriate drug combinations, and a lack of patient adherence, particularly due to the lengthy treatment regimens required for drug-susceptible TB (3–5). While new TB drugs are in the development pipeline, clinical development can take more than 2 decades, cost billions of dollars, and end in failure (3, 6). The high cost of developing new chemical entities in this space is particularly concerning given that TB tends to predominate in lower- to middle-income countries (1). A less risky and more cost- and time-efficient approach is to repurpose, improve, and rethink older drugs with known anti-TB potential. This is especially true for alternatively delivering existing TB drugs via direct pulmonary inhalation, which carries the potential to achieve shorter treatment regimens, overcome drug resistance, and rapidly reduce transmission while being cost-effective due to the repurposed nature of the product development.

Clofazimine (CFZ) is an existing, off-patent, red phenazine dye approved in the United States (US) as an oral capsule formulation (Lamprene; Novartis) for the treatment of lepromatous leprosy (7). While CFZ was initially developed by Barry et al. for the treatment of TB and exhibited greater in vivo activity compared to isoniazid (8), it quickly fell out of favor upon the development of other antibiotics like rifampicin, pyrazinamide, and ethambutol due to its unusual pharmacokinetics and adverse effect profile (9, 10). CFZ is a highly lipophilic compound (logP 7.48 [11]) with generally poor aqueous solubility (10 mg/L at neutral pH) (11, 12). Two amine functional groups are present that exhibit protonation under acidic conditions (predicted pKa of 9.29 and 2.31 [13]). Upon prolonged oral administration, CFZ has been observed to massively accumulate in the liver, spleen, intestines, lungs, and skin in a crystalline form (8, 14). Its systemic adverse effect profile includes a notable discoloration of the skin to a reddish-purple color (15, 16), abdominal pain (16, 17), and in rare cases, splenic infarctions (17) and gastrointestinal bleeding (18). The extremely lengthy half-life of CFZ, which is estimated to be 70 days in humans with prolonged administration (19), further increases the potential severity of these side effects, given that even after discontinuation the drug will remain in the body for a substantial period of time. Oral administration is also limited by the slow onset of action. In mice, Swanson et al. found that at least 30 days of administration was necessary to reach steady-state concentrations, and a delay in bactericidal activity occurs for up to 2 weeks after oral dosing, regardless of the dose administered (20). Clinically, this has resulted in the use of large loading doses (21).

Despite the adverse effect profile of oral CFZ, the growing prevalence of MDR- and XRD-TB in the past decade has prompted its clinical reevaluation as an anti-TB drug (22). The antimycobacterial activity of CFZ is postulated to derive from the its membrane-disruption effects that result from its cationic amphiphilic properties at neutral-to-acidic pH, as well as the redox potential of the molecule, which contributes to the generation of antimicrobial reactive oxygen species (ROS) (8, 10, 23). The mean inhibitory concentration (MIC) of CFZ ranges from 0.06 to 2 μg/mL in susceptible Mtb (24), and its nonspecific mechanisms of action against mycobacteria may reduce the risk of drug resistance developing. In a pivotal study performed in Bangladesh with a large cohort of 427 MDR-TB patients, orally administered CFZ in combination with gatifloxacin, ethambutol, and pyrazinamide for 9 months followed by a 4-month regimen of kanamycin, prothionamide, and high dose isoniazid resulted in a relapse-free cure in 87.9% of patients (25). Oral CFZ was also utilized as an add-on therapy in a 21-month regiment containing five other WHO-recommended agents in a multicenter study in China, which resulted in significantly higher rates of cavity closure in patients receiving CFZ versus controls (26). In the most recently published 2019 WHO Consolidated Guidelines on Drug-Resistant Tuberculosis Treatment, oral CFZ is listed as one of two recommended options for treatment of MDR-TB in combination with a fluoroquinolone, bedaquiline, and linezolid (2).

The propensity of CFZ to accumulate in certain organs is hypothesized to be due to its sequestration within the macrophage cells in a crystal-like structure and subsequent transport through the reticuloendothelial system (14, 27). Interestingly, CFZ intracellular accumulation has been observed in vitro (28) and in vivo (14, 29) to promote anti-inflammatory activity by macrophages. Given the intracellular nature of Mtb infection, this phenomenon of macrophage sequestration may be intimately linked to the antimycobacterial activity of CFZ, though this has not been fully elucidated. Since the adverse effects of CFZ are dose-related and frequently GI related, administration of CFZ by an alternative route to oral delivery could alleviate or at least limit its side effects. More specifically, administration by the inhalation route as an alternative to systemic administration could be used to target CFZ to Mtb-infected alveolar macrophages, thereby more quickly achieving therapeutic concentrations and reducing GI exposure. In vivo, CFZ has been previously demonstrated to be efficacious via pulmonary delivery as a spray-dried powder in Mtb-infected mice (30) and as a nebulized suspension in Mycobacterium avium infected mice (31). Brunaugh et al. (32) reported that micronization of CFZ via air jet milling produced a respirable crystalline dry powder that demonstrated inhalation flow-rate independent aerosol performance without the need for excipients, and which was rapidly phagocytosed by macrophages and transformed to intracellular drug inclusions similar to those previously reported (27, 33, 34). Air jet milling is a scalable technique for particle size reduction that requires no organic solvents. The simple and economical nature of the manufacturing method and formulation approach would be ideal in the development of a low-cost treatment for TB in lower- to middle-income countries. Additionally, the utilization of a dry powder inhaler (DPI) would have advantages over liquid nebulizer delivery, as these systems are portable and require no external power source and no sterile water source for drug delivery.

Building upon this previously published data (32), we hypothesized that compared to oral delivery, pulmonary administration of air jet milled CFZ particles would result in faster onset of antimycobacterial activity in diseased tissues by improving drug targeting to the infection site (lung) and enhancing drug transport to secondary lymphoid tissues (spleen) by a phagocyte-targeted delivery approach, similar to the manner in which Mtb itself establishes infection and is potentially disseminated (35, 36). To test this hypothesis, a single-dose and multidose pharmacokinetic (PK) study was conducted in healthy BALB/c mice dosed with 30 mg/kg of micronized CFZ via intratracheal instillation (mCFZ-IT) or oral gavage (mCFZ-OG), followed by an efficacy study at equivalent doses in BALB/c mice chronically infected with Mtb. To additionally demonstrate the feasibility of inhaled air jet milled CFZ for the treatment of TB in lower- to middle-income countries, the effects of storage at accelerated and intermediate International Council for Harmonization (ICH) stability conditions on CFZ physicochemical properties and aerosol performance was determined. This study provides additional insights into the PK and pharmacodynamics (PD) of CFZ while additionally providing important justification for continued development of inhaled drug products for the treatment of TB, particularly using economical manufacturing techniques like air jet milling.

RESULTS

Delivery of micronized CFZ by the pulmonary route is safe and produces higher lung concentrations and lower systemic exposure than oral delivery in BALB/c mice.

BALB/c mice were administered both a single 30 mg/kg dose and 4 weeks of twice-weekly doses of an aqueous suspension of micronized CFZ delivered by the pulmonary route (mCFZ-IT) and oral routes (mCFZ-OG) to compare basic PK parameters and determine the effect of the administration route on tissue accumulation. Based upon the known prolonged half-life of CFZ and reported lung and spleen tissue accumulation in mice (20), a sparse sampling approach was utilized in which terminal time points were taken between 1–96 h post administration in the case of the single dose PK study, and 96 h after the previous dose in the multidose study. Lung tissue, spleen tissue, the pelleted cellular component of the bronchoalveolar lavage (BAL) fluid and plasma samples were collected from mice at each terminal time point. The concentrations obtained for each animal at each terminal time point were averaged to calculate the composite PK profiles using noncompartmental analysis. CFZ was rapidly cleared from the BAL fluid cell pellet after intratracheal (IT) administration, with a half-life of 4.43 h (Table 1, Fig. 1A). Interestingly, mCFZ-OG treated mice did not show quantifiable levels of CFZ in the BAL fluid cell pellet until 96 h postadministration (Table 1, Fig. 1B). Mice treated with a single dose of mCFZ-IT exhibited 6-fold higher peak lung tissue concentrations and 4-fold higher lung tissue area under the curve (AUC) than mCFZ-OG with a faster Tmax (1-h post dose versus 6-h post dose) (Table 1, Fig. 1C and D). The half-life of CFZ in the lungs of mCFZ-OG treated mice was longer than that of mCFZ-IT treated mice. Plasma levels between the two treatment groups were similar, though the plasma elimination phase of the mCFZ-OG treated mice could not be determined due to rise in CFZ plasma levels occurring between 72–96 h postadministration (Table 1, Fig. 1E and F), which was possibly due to enterohepatic recycling known to occur with oral administration of the drug (37). CFZ spleen levels following mCFZ-IT administration were low, with only one mouse exhibiting a spleen concentration above the lower limit of quantitation (LLOQ), which occurred 6 h postadministration. In contrast, quantifiable drug levels were observed in the spleens of mCFZ-OG treated mice, and absorbance into and elimination from the spleen was like that observed in the lungs (Table 1, Fig. 1G and H).

TABLE 1.

Noncompartmental PK analysis of micronized CFZ administered by two different extravascular routes (pulmonary; IT and oral; OG)

	mCFZ–IT				mCFZ–OG
Parameter	Plasma	Lung tissue	Spleen tissue	BAL fluid cell pellet	Plasma	Lung tissue	Spleen tissue	BAL fluid cell pellet
Cmax (μg/unita)	0.43	17.09	0.92	10.09	0.32	2.54	4.77	2.78
Tmax (hr)	1	1	6	1	6	6	6	96
AUC0-last (μgahr/unita)	10.02	210.43	9.23	106.28	7.00	45.43	112.51	33.47
AUC0-inf, predicted (μgahr/unita)	11.39	211.91	NAb	106.29	NA	46.07	121.70	NA
Half-life (hr)	32.54	11.77	NA	4.43	NA	17.81	24.03	NA
R2	0.64	0.88	NA	0.90	NA	0.70	1.00	NA
Adjusted R2	0.52	0.84	NA	0.80	NA	0.55	0.99	NA

^aUnit = g for lung and spleen tissue and BAL fluid cell pellet; mL for plasma.

^bNA indicates model could not be fit.

FIG 1.

Single-dose PK of 30 mg/kg mCFZ in BALB/c mice differs depending on whether the pulmonary (mCFZ-IT) or oral (mCFZ-OG) routes are utilized. At each terminal time point, lungs of mice treated with mCFZ-IT or mCFZ-OG were washed several times with PBS, the collected fluid centrifuged, and the pellet quantified for CFZ content. Mice treated with mCFZ-IT exhibited initially high CFZ levels in the BAL fluid pellet, which was rapidly cleared (A); in contrast, no detectable levels of CFZ were noted in the BAL fluid pellet until 96 h post administration in the mCFZ-OG treatment group (B). Administration of mCFZ-IT resulted in an earlier Cmax and a 10-fold higher peak concentration in the lung tissue (C) compared to mCFZ-OG (D) and maintained a mean concentration above the MIC until 24 h post administration. Both routes exhibited low plasma concentrations that dropped below the MIC by 24 h post dose, with the Cmax occurring earlier for mCFZ-IT (E) than mCFZ-OG (F). The spleen drug concentration for mice dosed with mCFZ-IT (G) was lower than in the lung tissue, with mice exhibiting a spleen concentration above the MIC. The spleen PK profile for mCFZ-OG (H) was like that observed in the lung tissue.

Toxicity and accumulation of CFZ was assessed by treating naive BALB/c mice with 30 mg/kg mCFZ-IT or mCFZ-OG twice weekly for 4 weeks. Mice treated with mCFZ-IT exhibited rising trough concentrations in the plasma, lung, and spleen tissue (Fig. 2A, C, and E, light shaded curves), though this did not reach the level of accumulation observed in naive mice treated with mCFZ-OG (Fig. 2B, D, and F). Minimal accumulation was noted in the BAL cell pellet (Table S1 in the supplemental material). At the conclusion of the study, representative cross sections of haired skin, spleen, lungs, salivary glands, mandibular lymph nodes, liver, mesenteric lymph nodes, and small intestine were examined by a blinded veterinarian pathologist. No lesions were observed, and there was no microscopic evidence of crystal deposition in the tissue sections examined.

FIG 2.

CFZ tissue accumulation after intratracheal (IT) or oral gavage (OG) dosing of micronized CFZ particles is significantly affected by the infection status of BALB/c mice. A significantly higher level of CFZ was noted in the lungs of infected mice treated with mCFZ-IT at day 21 (P = 0.0068) and day 28 (P <0.0001) of dosing compared to lungs of naive mice at the same time points (A), while no significant differences were noted between naive or infected mice treated with mCFZ-OG (B). Similarly, increased CFZ spleen accumulation was noted with infected mice treated with mCFZ-IT at 3 weeks (P = 0.00017) and 4 weeks (P <0.0001) compared to naive mice at the same time points (E). While nonsignificant differences were noted in plasma levels of mice dosed with mCFZ-IT (C), mice infected with Mtb and treated with mCFZ-OG exhibited significantly lower plasma levels after day 28 of dosing compared to naive mice (P = 0.0039) (D). In contrast to the mCFZ-IT treatment group (E), lower levels of spleen accumulation were noted in the mCFZ-OG treatment mice infected with Mtb at day 14 (P = <0.0001), day 21 (P = <0.0001) and day 28 (P = <0.0001) compared to naive treated mice (F). Results are presented as mean (n = 3) ± SEM; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Micronized CFZ delivered by the pulmonary route exhibits faster treatment of Mtb lung infection compared to oral delivery in BALB/c mice.

To confirm whether local delivery to the site of infection may incur advantages over systemic delivery for the treatment of TB, BALB/c mice were treated twice weekly for 4 weeks with a suspension of micronized CFZ via the oral or intratracheal routes 42 days after aerosol infection with Mtb (Fig. 3A). A subset of mice (n = 3) was sacrificed at varying time points during the treatment period to determine the time required for bactericidal activity and to examine the relationship between CFZ tissue concentrations and antibacterial activity. Mice treated with mCFZ-IT exhibited a significant decrease in lung CFU compared to the baseline control on day 1 posttreatment (P = 0.0021), day 14 (P = 0.0138), day 21 (P <0.0001), and day 28 (P <0.0001) (Fig. 3B). The mCFZ-OG group exhibited a statistically significant decrease in lung CFU starting at day 14 posttreatment (P < 0.0001), which was sustained throughout the treatment period (Fig. 3B). A statistically significant reduction in bacterial burden in the spleen versus the baseline control was achieved by day 28 for mCFZ-IT treated mice (P <0.0001), while the mCFZ-OG treatment group exhibited a significant decrease by day 14 of treatment (P = 0.0196) (Fig. 3C). A separate cohort of mice (n = 10) was utilized to determine endpoint efficacy of the micronized CFZ formulation after 4 weeks of dosing via each route compared to untreated controls. In this cohort, there were no statistically significant differences in the lung CFU count of mice treated with mCFZ-IT versus mCFZ-OG, but the mCFZ-OG treatment group exhibited a significantly lower CFU in the spleen (P = 0.0018) compared to the mCFZ-IT treatment group (Fig. 3C). Both groups exhibited a statistically significant reduction in lung and spleen CFU compared to the untreated control (Fig. 3C). Overall, the CFU decreases observed with 4 weeks of twice-weekly dosing of 30 mg/kg CFZ were comparable to those achieved in a previous study using 20 mg/kg CFZ dosed 5 days/week in BALB/c mice (20). Similar CFU read-outs were obtained in treated groups using both activated charcoal-containing plates and non-charcoal-containing plates, indicating that drug carryover was likely not observed in the homogenate tissue samples. Lack of drug carryover may be the result of the serial dilutions required for accurate bacterial enumeration, which placed the drug concentration well below the MIC of CFZ (0.06 to 2 μg/mL in susceptible Mtb [24]).

FIG 3.

(A) Forty-two days p.i. with Mtb, BALB/c mice were treated twice per week with 30 mg/kg intratracheal micronized CFZ (mCFZ-IT) or oral micronized CFZ (mCFZ-OG). A subset of mice (n = 3) was sacrificed weekly to determine CFZ levels in the lung and spleen tissue and bacterial burden relative to baseline (n = 6) while remaining mice (n = 9) were sacrificed day 70 p.i. Samples from both untreated controls and treatment groups at all time points were plated on 7H11/OADC agar containing 0.4% activated charcoal to prevent drug carryover effect. (B) A significant decrease in bacterial burden in lung tissue was noted 24 h after a single dose of mCFZ-IT; in comparison, a significant decrease in lung CFU in the mCFZ-OG group compared was not achieved until day 14 of treatment. In the spleen tissue, a significant decrease in CFU relative to baseline was achieved by day 14 of treatment in the mCFZ-OG group, while in the mCFZ-IT group a significant decrease was not achieved until day 28 of treatment. (C) At the conclusion of the study, mice treated with mCFZ-IT exhibited a significant decrease in bacterial burden in the lungs compared to untreated mice, with no significant difference versus mice treated via the oral route. Both treatment groups exhibited significantly lower spleen bacterial burden than untreated mice at the conclusion of the study, though mice treated with mCFZ-OG exhibited significantly lower bacterial burden than mCFZ-IT treated mice. (D) Cross sections of lungs from infected BALB/c mice at the start of treatment (untreated D42 p.i.) exhibit inflammatory cells composed of primarily macrophages (black arrow) and lymphocytes (blue arrow) with a small number of neutrophils (red arrow). At the conclusion of the study, untreated mice (untreated D70 p.i.) exhibited increased predominance of macrophages arranged in small collections surrounding necrosis (black arrows). In contrast, inflammation in the lungs of mice from the mCFZ-IT (mCFZ-IT D70 p.i., D28 treatment) and mCFZ-OG (mCFZ-OG D70 p.i., D28 treatment) groups was predominantly lymphoplasmacytic (black arrows) with fewer aggregates of macrophages admixed with neutrophils, necrotic debris (blue arrow). and cholesterol clefts (red arrow). Data presented as mean ± SEM; *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

In addition to bacterial enumeration, several cross sections of lungs of mice from the baseline control group (42 days postinfection [p.i.]), untreated control group (70 days p.i.), mCFZ-IT treatment group, and mCFZ-OG treatment group were stained and examined for inflammatory damage by a veterinarian pathologist (blinded to treatment groups) (Fig. 3D). All groups were observed to exhibit multiple foci of peribronchiolar and perivascular inflammation scattered throughout the pulmonary parenchyma. In lung section samples from mice of the baseline control group, multiple foci of perivascular, peribronchiolar, and alveolar inflammation were noted affecting approximately 30% of the pulmonary parenchyma. Inflammatory cells were composed of large numbers of lymphocytes, foamy macrophages, and lesser numbers of neutrophils and were found to flank the tunica adventitia of multiple small- to medium-caliber blood vessels and were noted to fill the lumen of multiple alveoli. In mice from the untreated control group, the type of inflammation was similar, with multiple foci of peribronchiolar and perivascular inflammation affecting approximately 60% of the pulmonary parenchyma. Inflammatory cells were composed of moderate to large numbers of foamy macrophages, lymphocytes, and neutrophils, with early granuloma formation characterized by a concentric collection of foamy macrophages flanked by moderate numbers of lymphocytes noted in multiple areas. Small amounts of necrotic cell debris were observed among foamy macrophages, and small numbers of macrophages and neutrophils infiltrated the lumen of multiple bronchioles. Lung sections obtained from mice in the mCFZ-IT treatment group exhibited multiple foci of peribronchiolar and perivascular inflammation that affects 20–40% of the pulmonary parenchyma. Inflammatory cells were composed of moderate to large numbers of foamy macrophages, lymphocytes, and neutrophils. In multiple areas, early granuloma formation was noted, characterized by a concentric collection of foamy macrophages flanked by moderate numbers of lymphocytes. In addition, foamy macrophages were often obscured by moderate to abundant amounts of necrotic cell debris. Similar findings were noted in the CFZ-OG treatment group mice, with the exception being multiple foci of peribronchiolar and perivascular inflammation that affected approximately 50% of the pulmonary parenchyma. No crystal deposition was associated with foci of inflammation in samples from either treatment group. From the standpoint of resolution of inflammation and lung damage associated with Mtb infection, micronized CFZ delivered via the pulmonary route appears as effective as CFZ delivered by the oral route.

Prescence of Mtb infection increases spleen and lung accumulation of CFZ after pulmonary delivery in BALB/c mice.

CFZ is known to accumulate in the tissues of the lung and spleen when dosed over a prolonged period via the oral route (14). To test whether this accumulation was affected by the presence of Mtb infection and to determine the relationship between drug tissue levels and lung and spleen CFU, a subset of mice (n = 3) was sacrificed on treatment day 1 (43 days p.i), treatment day 7, treatment day 14, treatment day 21, and treatment day 28. Lung and spleen tissues were homogenized, and a portion was serially diluted and plated for colony enumeration, while a separate portion was sterilized in methanol and then quantified for CFZ content using HPLC. BALB/c mice with chronic Mtb infection treated with mCFZ-IT exhibited significantly higher levels of drug in the lung tissue at treatment day 21 (P = 0.007) and day 28 (P <0.000001) (Fig. 2A), while no significant differences were noted in the lung tissue concentrations of naive versus Mtb-infected mice treated with mCFZ-OG at the time points tested (Fig. 2B). While there were no significant differences in plasma CFZ accumulation between naive and infected mice in the mCFZ-IT treatment group (Fig. 2C), a significant decrease was observed on day 28 of treatment (P = 0.004) between naive and infected mice in the mCFZ-OG group (Fig. 2D). Lastly, accumulation of CFZ in the spleen was significantly increased in infected mice treated with mCFZ-IT compared to naive mice at treatment day 21 (P = 0.0002) and day 28 (P = 0.00002) (Fig. 2E), while the opposite trend was noted for the mCFZ-OG treated mice. In this case, significantly less CFZ was noted in the spleen tissue of infected mice at treatment day 14 (P = 0.00004), treatment day 21 (P = 0.00002), and treatment day 28 (P <0.00001) (Fig. 2F). CFZ was below the LOD in the sterilized BAL fluid cell pellet samples obtained from infected BALB/c mice in both treatment groups.

Linear regression analysis was performed to examine the relationship between CFZ tissue levels noted at each time point versus the tissue CFU (Fig. 4A to H). A linear regression model with good fit (r² = 0.917) and predictive value (P = 0.0104) was generated for lung CFU versus the CFZ concentration in the lungs as well as for lung CFU versus the plasma CFZ concentration (r² = 0.849 and P = 0.0262). In contrast, linear regression models built for the spleen CFU using either spleen or plasma CFZ concentration from the mCFZ-IT treatment group exhibited poor fit and poor predictive value. In general, linear regression models built using data obtained from the mCFZ-OG dosing groups exhibited poor fit, with the exception being the plasma concentration versus lung CFU model. This may be a result of the greater variability in CFZ tissue concentrations observed in these mice as well as the differences in the time required to reach steady state for the oral versus the inhalation route.

FIG 4.

Linear regression analysis was performed on Mtb tissue CFU (lung and spleen) versus CFZ concentration in lung, spleen, and plasma at various time points in the treatment regimen to develop predictive models. The treatment day in which a significant decrease in CFU versus baseline was first noted is bolded. CFZ lung concentration after IT dosing appears predictive for lung CFU (A); in contrast, a linear regression model between CFZ lung concentration after OG dosing and lung CFU did not exhibit good fit (B). Plasma CFZ concentrations in the IT group were predictive of lung CFU (C) as were plasma CFZ concentrations in the OG group (D). Neither CFZ spleen concentration (E–F) nor CFZ plasma concentration (G–H) appear predicative of spleen CFU for either treatment route using linear modeling. Data presented as mean (n = 3) ± SEM.

Micronized CFZ exhibits stability in hot humid/tropical zone ICH conditions.

Stability of milled CFZ stored in hydroxypropyl methylcellulose (HPMC) capsules was assessed under ICH Zone IV conditions (hot humid/tropical zone) for 6 months to assess suitability of use in regions with limited access to cold-chain storage. Both intermediate (30°C ± 2°C and 65%RH ± 5%RH) and accelerated (40°C ± 2°C and 75%RH ± 5%RH) storage conditions were tested. Visual inspection of the encapsulated CFZ powder after 3 and 6 months of storage revealed no changes in powder color, agglomeration, or capsule appearance. Chemical assay of three encapsulated samples from each condition at each time point was performed using HPLC. Compared to an injected CFZ standard, no impurity peaks were noted in the chromatograms for either condition (Fig. S2), and CFZ assay remained uniform and well within typical limits of 90.0%–110.0% (Table S2).

The aerosol performance of milled CFZ at baseline and after 3 and 6 months of storage was also assessed. Statistical analysis was performed using analysis of variance (ANOVA) at an alpha level of 0.05. The aerosol performance of encapsulated CFZ was found to be stable. No statistically significant changes in the aerosol performance metrics of emitted fraction (EF), respirable fraction (RF), fine particle fraction (FPF), or mass median aerodynamic diameter (MMAD) were noted for the intermediate conditions (Table 2) or the accelerated conditions (Table 3).

TABLE 2.

Aerosol performance of 20 mg of encapsulated micronized CFZ powder through 6 months of storage in ICH Zone IV intermediate stability conditions (30°C/65%RH) at 1 kPa pressure drop through high resistance RS01 inhaler (n = 3; mean ± SD)

Performance metric	Baseline	3 mo	6 mo	P value
EFa (%)	48.7 ± 1.8	50.1 ± 4.8	54.6 ± 10.1	0.55
RFb < 5 μm (%)	34.0 ± 0.7	33.9 ± 4.2	33.7 ± 6.7	0.99
RF < 3 μm (%)	20.0 ± 0.8	17.1 ± 3.2	31.4 ± 19.9	0.32
FPFc < 5 μm (%)	69.8 ± 1.2	67.8 ± 7.2	61.9 ± 8.9	0.38
FPF < 3 μm (%)	44.9 ± 4.8	34.8 ± 10.1	38.7 ± 5.0	0.29
MMADd (μm)	4.4 ± 0.1	4.9 ± 0.7	4.8 ± 0.3	0.41

^aEF; emitted fraction (percentage of the loaded dose emitted from the device).

^bRF, respirable fraction (percentage of the loaded dose with an aerodynamic particle size less than 5 μm or 3 μm).

^cFPF, fine particle fraction (percentage of the dose emitted from the device with an aerodynamic particle size less than 5 μm or 3 μm).

^dMMAD, mass median aerodynamic diameter (the aerodynamic particle size in which 50% of the particles of the aerosol by mass were larger and 50% were smaller).

TABLE 3.

Aerosol performance of encapsulated micronized CFZ powder through 6 months of storage in ICH Zone IV accelerated stability conditions (40°C/75%RH) at 1 kPa pressure drop through high resistance RS01 inhaler (n = 3; mean ± SD)

Performance metric	Baseline	3 mo	6 mo	P value
EFa (%)	48.7 ± 1.8	44.8 ± 10.1	40.6 ± 17.3	0.55
RFb < 5 μm (%)	34.0 ± 0.7	28.3 ± 5.9	25.8 ± 14.4	0.99
RF < 3 μm (%)	20.0 ± 0.8	14.8 ± 1.4	15.1 ± 8.9	0.32
FPFc < 5 μm (%)	69.8 ± 1.2	63.4 ± 4.5	61.3 ± 8.7	0.38
FPF < 3 μm (%)	44.9 ± 4.8	34.2 ± 8.4	35.4 ± 8.3	0.29
MMADd (μm)	4.4 ± 0.1	4.9 ± 0.3	5.0 ± 0.4	0.41

^aEF, emitted fraction (percentage of the loaded dose emitted from the device).

^bRF, respirable fraction (percentage of the loaded dose with an aerodynamic particle size less than 5 μm or 3 μm).

^cFPF, fine particle fraction (percentage of the dose emitted from the device with an aerodynamic particle size less than 5 μm or 3 μm).

^dMMAD, mass median aerodynamic diameter (the aerodynamic particle size in which 50% of the particles of the aerosol by mass were larger and 50% were smaller).

Micronized CFZ particles exhibit limited dissolution in neutral pH environments.

A study examining the pH-dependent dissolution of aerosolized CFZ particles was undertaken to support the intracellular-targeted nature of the inhaled powder formulation. CFZ particles aerosolized from a high resistance RS01 DPI were collected on filter membranes and transferred to a Transwell system, and drug concentration in the receptor compartment was monitored over time using two different pH systems: a pH 7.4 aqueous medium with 0.2% polysorbate 80, representative of the neutral oral cavity and lung lining fluid, and a pH 4.5 aqueous medium with 0.2% polysorbate 80, representative of the acidic lysosomal compartment that the particles are anticipated to encounter upon phagocytosis. Quantifiable levels of CFZ in pH 7.4 dissolution conditions were not observed until 24 h, while quantifiable levels of drug in pH 4.5 conditions were observed starting at 45 min postdeposition, and 83.9 ± 2.2% of the deposited dose was dissolved by 24 h (Fig. S3).

DISCUSSION

CFZ is a highly active antimycobacterial drug known to exhibit unusual macrophage-based PK when administered over a prolonged period by the oral route. Though it holds great promise as an alternative therapy for treatment of MDR and XDR-TB, enthusiasm and widespread use of the existing oral formulation has been dampened by its systemic side effects. Administration of CFZ via inhalation could avoid many of the drawbacks associated with oral administration of the drug. Though efficacy of CFZ delivered via the pulmonary route has been assessed in mycobacterial infections with positive results (30, 31), these CFZ formulations have used costly particle engineering technologies (spray drying) or nebulizer-based delivery systems, which are unsuitable for the resource-poor regions of the world where TB predominates. In contrast, air jet milling has been demonstrated as a cost-effective technology for producing crystalline CFZ particles with high aerosolization efficiency (32). We hypothesized that formulation of CFZ in a poorly water-soluble particulate form could provide preferential targeting to alveolar macrophages on a faster timescale than oral administration while avoiding dissolution in the lung environment and subsequently systemic exposure. Our goal was to provide justification for further commercial development of lung macrophage-targeted therapies like micronized CFZ for the treatment of pulmonary mycobacterial infections.

In this study, we examined single-dose PK parameters of air jet milled CFZ particles administered by the oral and pulmonary routes in naive mice, tissue accumulation after repeated doses of CFZ in naive and Mtb-infected mice, and efficacy against chronic Mtb infection in mice. Linear regression analysis was performed to determine the relationship between tissue and plasma levels of CFZ and antimycobacterial activity. Additionally, physicochemical stability of the formulation at environmental conditions relevant for TB treatment was assessed. Treatment of Mtb-infected mice with mCFZ-IT resulted in a significant decrease in lung bacterial burden after only a single dose, and continual decreases were noted throughout the course of the 4-week treatment period. Though CFZ has been historically considered a slow acting drug against Mtb (38), this previously observed delay in bactericidal activity in the BALB/c TB mouse model may have been due to time required to achieve sufficient intracellular drug concentrations when CFZ is administered by the oral route. The rapid bactericidal activity observed after a single day of dosing with mCFZ-IT confirms previous in vitro results by Verma et al. in which 99% killing of Mtb was achieved after 2 days of dosing of respirable CFZ particles to infected THP-1 macrophages (30); combined, these results indicate that the pharmacological activity of CFZ may be enhanced through direct host-cell targeting. A nonstatistically significant increase in CFU was observed between treatment day 1 and day 7 in the mCFZ-IT dosing group, and there was no statistically significant difference between the baseline control and the treatment day 7 subset in the mCFZ-IT dosing group; this finding may indicate that increased frequency of CFZ administration may be required to achieve sustained decreases in bacterial load, as these treatment day 7 mice were sacrificed at the CFZ trough concentration. Lastly, it is important to point out that the number of mice utilized for these PK subset time points was lower than that used in the final day 28 efficacy time point (n = 3 versus n = 10). Utilization of a larger number of mice at each time point may have reduced the variability among the different groups and enabled a clearer identification of CFU decrease over time.

Though less effective than mCFZ-OG treatment, a significant decrease in spleen bacterial burden relative to untreated controls was also achieved at the conclusion of the study. Mtb infection resulted in an increase in CFZ accumulation in the lungs and spleen of mCFZ-IT treated mice, while opposing trends were noted in the mCFZ-OG treated mice. Our findings with oral CFZ align with previously published work. Similar to findings by Swanson et al. (20), we found that 2 weeks of treatment with oral CFZ was required to achieve significant decreases in lung and spleen bacterial burden. Likewise, as previously reported by Swanson et al., significant reductions in lung and spleen bacterial burden with oral dosing appear to occur once the plasma concentration of CFZ is above the MIC; indeed, we found a stronger linear relationship between plasma drug concentration and lung CFU than between lung tissue drug concentration and lung CFU for the oral dosing group.

Changes in CFZ spleen sequestration in infected BALB/c mice have been previously reported (39), and we experimentally confirm herein that infected mice treated with mCFZ-OG exhibited significantly reduced spleen accumulation and a decreased (though not significant) reduction in lung accumulation compared to their uninfected counterparts. Lower plasma concentrations throughout the course of treatment were also observed in naive versus infected mCFZ-OG treated mice. Dunne et al. (39) hypothesized that these infection-dependent changes in tissue sequestration of CFZ with prolonged oral dosing may be a result of a delay in the precipitation of CFZ in the lysosomes of infected macrophages relative to uninfected macrophages due to changes in lysosomal pH. It is also possible that bioavailability of oral CFZ was reduced in this disease model, which subsequently affected plasma and tissue drug levels. CFZ exhibits pH-dependent solubility, with increased solubility in acidic conditions. Though not reported in literature, the presence of Mtb infection in BALB/c mice could result in changes in stomach acidity, as TB infection progression has been associated with gastric anacidity in humans (38). Overall, these changes in the PK of oral CFZ according to infection status may explain the delay in bactericidal activity of the drug against Mtb when administered via the oral route.

The time-dependent and infection-dependent tissue accumulation of CFZ administered by the pulmonary route, as well as the time-dependent pharmacodynamics, has not been previously reported, to our knowledge. A single dose of mCFZ-IT produced severalfold higher Cmax in the lung tissue and BAL fluid pellet of mice compared to an equivalent dose of mCFZ-OG, and a more rapid onset of bactericidal activity in the lungs was observed in the mCFZ-IT group compared to the mCFZ-OG group in a murine model of chronic Mtb infection. These findings point to the utility of directly targeting alveolar macrophages via particle inhalation versus reliance upon the natural macrophage sequestration mechanisms noted when CFZ is systemically administered. The differences in lung and spleen accumulation of CFZ after pulmonary administration in naive versus infected mice provide insights into its mechanisms of clearance from the lung and transport through the body. Progression of TB in murine models has been marked by increased infiltration of dendritic, macrophage, and monocyte populations in the lungs (40), which would thereby increase the number of phagocytes available for CFZ particle uptake and sequestration within the lung tissue. Though lung tissue accumulation of CFZ was observed in both naive and infected mice, BAL fluid cell pellet accumulation in naive mice was not observed until week 4 of treatment, while in infected mice, levels were consistently below the LOD. A lack CFZ accumulation in the BAL cell pellet appears in contrast to previous studies that observed macrophage sequestration with prolonged CFZ dosing via systemic routes (14, 27, 41). However, though these studies did not utilize BAL techniques to collect macrophages on the lung epithelial surface, it is possible that alveolar macrophages do not sequester CFZ in this manner. Our multidose PK results indicate that the lung phagocytes are effectively clearing the CFZ particles prior to administration of the next dose. Clearance may be achieved through movement of macrophages up the mucociliary escalator (followed by subsequent swallowing or removal by cough), or by migration of particle-loaded phagocytes to the draining lymph nodes (42). Macrophage penetration through the alveolar epithelium into the lung interstitium has been reported to be induced by increased IL-1 production during Mtb infection (36). This latter mechanism may account for the increased CFZ lung tissue accumulation that was observed in infected mice versus healthy mice receiving CFZ particles via IT instillation. Regardless, the effective clearance of CFZ particles from the epithelial lining fluid has important implications for safety as particle overload and impairment of alveolar macrophage clearance can result in chronic inflammation (43). Elucidation of the mechanism of CFZ transport from the macrophages on the surface of the lung epithelium into the lung tissue and to distal tissues, and how this mechanism may be influenced by macrophage infection status, requires further investigation and will be evaluated in future studies.

Simple linear regression was performed as an initial analysis to determine if CFZ tissue and plasma concentration relates to bactericidal activity at a constant rate. The variability of CFZ concentration across mice, particularly in the mCFZ-OG treatment group, as well as observed differences in accumulation and time to steady state for each route, must be considered when interpreting the fit of these models. However, the strong correlation between CFZ lung concentration and antimycobacterial activity in mCFZ-IT treated mice can provide a preliminary basis for the selection of a human dose of inhaled CFZ powder. Assuming an average human lung weight of 1000 g (44), and given the lowest effective dose noted in the mCFZ-IT treated mice (3.1 μg/g lung tissue, treatment day 7), we estimate that a delivered dose of 3.1 mg to the lungs of humans could demonstrate efficacy. This delivered dose is within the lowest respirable fraction noted in aerosol performance study of encapsulated micronized CFZ after storage at a 1-kPa pressure drop using the high resistance RS01 device (20 mg loaded dose; 17% respirable fraction less than 3 μm aerodynamic diameter, estimated 3.4 mg delivered to the lungs). Additionally, the strong linear correlation between CFZ plasma levels and lung CFU for the mCFZ-IT treatment group provides justification for the use of plasma drug levels as a marker for determining tolerable and effective doses of inhaled CFZ in Phase I clinical trials, rather than more invasive methods like BAL; however, this must be further validated in other preclinical disease models.

Dramatically increased accumulation of CFZ in the spleen of infected mice treated with mCFZ-IT was noted, which did not correspond with increased plasma levels when compared to healthy mice. Spleen accumulation of CFZ was dramatically increased in the infected mice treated with mCFZ-IT compared to their healthy counterparts. However, this did not correspond to a significant increase in plasma concentration, which supports our hypothesis that increased spleen concentration may be achieved via inhalation of CFZ particles based upon a phagocyte-trafficking drug distribution mechanism that occurs in a similar manner to Mtb dissemination: namely, by phagocyte transport through the lymphatic system by lung draining lymph nodes (45, 46). An increase in the phagocytes in the lung with Mtb infection would also correspond to increase in clearance to the spleen. However, the lack of a linear relationship between spleen CFZ concentration and spleen CFU for both the mCFZ-IT and mCFZ-OG treated mice may indicate that not all the accumulated drug in the spleen is pharmacologically active; for example, it may be in a precipitate form, bound to proteins in the tissue or sequestered in an inaccessible intracellular compartment. If CFZ is indeed transported to the spleen via phagocyte trafficking mechanisms after inhalation, there may not be a mechanism for the drug to be released from the cell to interact with other infected cells or extracellular bacilli. Antibacterial activity observed in the spleen may instead be derived from the limited systemic absorption and plasma-based drug transport that was observed with the inhalation route. This may also explain the faster onset action in the spleen observed with oral CFZ dosing, which resulted in greater systemic exposure on a faster timescale. Further investigation into the proportion of drug in the spleen that is protein bound versus unbound, solubilized versus precipitated, and extracellular versus intracellular would enable the generation of a physiologically relevant nonlinear or multivariate model that better predicts CFZ dose response within the spleen.

BALB/c mice were selected for this study based upon their previous utilization in a PK/PD study on oral clofazimine (20), which enabled direct comparison and confirmation of findings. This particular mouse model exhibits an intracellular infection pathology (47), which provided an important proof-of-concept for our formulation design. However, penetration into granulomas remains a major unknown for inhaled TB therapies. Indeed, Irwin et al. (48) report that oral CFZ monotherapy is ineffective in the treatment of TB in the C3HeB/FeJ mouse model, which exhibits extensive, heterogenous formation of lung lesions and a predominantly extracellular infection pattern. We are currently assessing the efficacy of micronized CFZ via the oral and pulmonary routes using the C3HeB/FeJ TB model, the results of which will be presented in a future publication. A limitation in this study is the utilization of suspension of micronized CFZ particles for the in vivo mouse studies, rather than the dry powder itself. Reproducible dispersion of powder into the lungs of mice remains a challenge, particularly for excipient-free, high-dose inhaled powders. To ensure accurate delivery of the desired dose to mice, a method of suspending micronized particles in an aqueous, isotonic solution containing a small amount of surfactant was developed. The particle size distribution of the CFZ suspension was like that of the dry powder (Fig. S4), and less than 0.013% of the suspended CFZ particles dissolve in this media over 72 h (Table S3). PK and toxicity studies of the inhaled micronized CFZ formulation presented in this study are currently ongoing in rats and dogs, and utilize passive, nose-only inhalation chambers. These studies will form the basis for selecting the starting dose for a future Phase I clinical trial.

A major barrier to the clinical use of CFZ is its systemic side effects, including skin discoloration. Inhaled drug delivery of CFZ is a promising alternative to oral CFZ delivery, and the feasibility of this delivery approach is further supported by the data presented herein. Though systemic exposure is lessened with direct delivery to the lungs, one potential off-target area of exposure is the oropharyngeal area due to the impaction-based deposition of undispersed aggregates of CFZ particles upon inhalation. Using a Transwell-based dissolution method for aerosolized drugs, we have demonstrated that micronized CFZ particles exhibit limited dissolution in neutral aqueous environments (i.e., pH 7.4). Given that normal saliva has a pH range of 6.0–7.5 (49), we anticipate staining and discoloration of the mouth to be minimal if patients are appropriately counseled to rinse their mouths with water immediately after actuation to remove any deposited particles that may dissolve over time. A similar approach is counseled for poorly water-soluble inhaled corticosteroids (50). Due to the coughing associated with TB lung infections, it is also possible that patients may observe red sputum following inhaled CFZ use. The prevalence and severity of this side effect would be assessed as part of clinical trials, and if necessary, counseling information regarding the normal occurrence of sputum discoloration could be provided in the package insert of the approved drug product.

In conclusion, respirable CFZ particles produced by air jet milling represent a potentially more efficacious and tolerable alternative to oral CFZ administration for the treatment of TB. Additional studies are ongoing related to the multidose tolerability of inhaled CFZ powder, its efficacy in combination with other anti-TB drugs, and its potential utility as a treatment for nontuberculous mycobacterium infections. Our results provide several avenues of future investigation. Firstly, the cellular mechanisms underlying the differing infection-dependence of accumulation patterns of CFZ when dosed systemically versus targeted directly to lung macrophages remain to be elucidated. This finding may be related to underlying differences between tissue-resident macrophages and their response to both CFZ and Mtb. Additionally, the availability of tissue/macrophage sequestered CFZ and its availability to interact with mycobacteria requires further investigation, as this would provide further guidance to decisions regarding therapeutic drug monitoring as well as drug molecule and formulation design. Lastly, an investigation on the effect of CFZ particle size, shape, rugosity, and solubility and how these factors impact the host-cell response may further refine development of this inhaled drug product.

MATERIALS AND METHODS

Preparation of clofazimine formulations.

CFZ (Sangrose Laboratories Pvt. Ltd., Mavelikara, IN) was micronized to a respirable size using a Model 00 Air Jet Mill (Fluid Energy Group, Calgary, CA) at a grind pressure of 100 lb/in², a feed pressure of 95 lb/in², and a feed rate of 1 g/min. The particle size distribution (PSD) of the micronized material was determined using laser diffraction with a dry powder disperser attachment (Sympatec GmbH, Clausthal-Zellerfeld, DE). The powder sample was dispersed at 3-bar pressure with 20% rotation of the feeder. Time slices of the dispersed plume with an optical concentration between 5 and 25% were averaged to determine the final PSD measurement. The micronized CFZ was hand filled into size 3 transparent HPMC capsules (Qualicaps, Whitsett, USA) at a target weight of 20 ± 1 mg, placed in HDPE bottles and heat sealed in foil pouches to assess stability under ICH Zone IV (hot/humid) conditions. Stability at both intermediate (30°C ± 2°C/65% RH ± 5% RH) and accelerated (40°C ± 2°C/75% RH ± 5% RH) was assessed through storage for 6 months in a monitored environmental chamber.

Though the intended end product for the CFZ formulation under investigation is dry powder for inhalation, intratracheal powder administration through insufflation is variable, and it is difficult to administer large powder boluses in mice due to the limited volume of the lungs and the amount of air required for dispersion (51). As an alternative, a suspension formulation was developed that exhibited similar particle size distribution to the dried micronized CFZ powder and limited dissolution in neutral aqueous buffer. Micronized CFZ powder (30 mg) was weighed into a 15-mL conical vial. The powder was wetted with 100 μL of a 10 mg/mL polysorbate 80 (Millipore Sigma, Burlington, USA) in low-endotoxin 0.9% wt/vol sodium chloride (NS; Teknova S5818, Hollister, USA). Additional low-endotoxin NS was added in 0.5 mL aliquots until a volume of 3 mL was reached. Upon each aliquot addition, the suspension was vortexed and/or sonicated to coat the particles with the surfactant. The suspension was then mixed using a rotor stator homogenizer (Fisherbrand Homogenizer 150; Thermo Fisher, Waltham, USA) at 35,000 RPM for 1 min. The homogenizer blades were then rinsed with 500 μL aliquots of NS until a total volume of 5 mL was reached. The suspension was centrifuged for 5 min at 100 × g to reduce foaming, and then sonicated for 30 s to redisperse particles. The CFZ suspension was stored under ambient conditions and made fresh every 48 h.

Analytical methods for clofazimine.

All reagents utilized for analysis were ACS or HPLC grade and obtained from Thermo Fisher (Waltham, USA), unless otherwise stated. For quantitation of CFZ in plasma, lung, and spleen tissues, an HPLC method was adapted from Queiroz et al. (52) that utilized a 125 × 4 μm Lichrospher 100 RP-8, 5-mm column, a 20 μL injection volume, flow rate of 0.9 mL/min, and a mobile phase consisting of a 26% aqueous phase containing 0.3 M sodium acetate water (triple charcoal filter, in-house) adjusted to pH 3.0 using glacial acetic acid and a 74% organic phase consisting of methanol. Detection was performed at 285 nm, and the retention time was approximately 10 min. Drug was extracted from plasma by precipitating the proteins with exposure to chilled acetonitrile, followed by centrifugation at 12,000 × g for 10 min to pellet plasma proteins, transfer of supernatant to a new vial, and evaporation and reconstitution of the remaining film in the mobile phase. Drug was extracted from lung and spleen tissue and BAL fluid cell pellet by homogenizing the samples in methanol at a volume (mL) that was 30× the tissue weight (g) using a rotor stator homogenizer, followed by centrifugation at 12000 × g for 10 min. The analytical method produced standard curves with suitably linearity, precision, and accuracy and a lower limit of quantitation of 50 ng/mL. Diclofenac was utilized as an internal standard at a spike concentration of 10 μg/mL. CFZ content in lung and spleen tissues was normalized to the tissue weight (μg/g), while CFZ content in the BAL fluid cell pellet was normalized to the protein content in the BAL fluid cell pellet. Protein content was determined using a Pierce Coomassie Plus (Bradford) assay kit (Thermo Fisher, Waltham, USA).

The United States Pharmacopeia (USP) HPLC method for quantitation of CFZ (53) was utilized for assay and related substance analysis of CFZ before and after storage, quantitation of CFZ deposition in cascade impaction studies, and the dissolution assay. This method utilizes a mobile phase consisting of a 65:35 ratio of acetonitrile to a buffer consisting of 4.5 g/L sodium dodecyl sulfate, 1.7 g/L tetrabutylammonium hydrogen sulfate, and 21.8 g/L disodium phosphate in water, adjusted to pH 3.0 with o-phosphoric acid. The column used was a LiChromspher RP-8 100 Å, 250 mm × 4.6 mm, 5 μm pore size (Millipore Sigma, Burlington, USA). Flow rate was 1.0 mL/min, injection volume was 20 μL, and the column temperature was ambient. Analysis was conducted using an Agilent 1100 HPLC with a diode array detector. Prior to analysis, 20 mg of CFZ in transparent size 3 HPMC capsules (Qualicaps, Whitsett, USA) was dissolved in 100 mL of the mobile phase to achieve a concentration of 0.20 mg/mL. The relative area of the drug peak and related substances for samples at the baseline and stability time points was compared to an equivalent concentration of a USP reference standard for clofazimine (United States Pharmacopeia, Rockville, USA).

In vitro evaluation of aerosol performance.

The aerosol performance of the air jet milled CFZ powder was assessed at baseline and after storage at intermediate and accelerated conditions assessed by utilizing a “worst-case scenario” approach for aerosol performance in which a high resistance RS01 device (Plastiape S.p.A, Osango, Italy) actuated at 28.3 L/min, which is equivalent to an approximately 1 kPa pressure drop for this device and is reflective of a patient with severely reduced inspiratory ability (54). Aerosol performance evaluation was conducted using an Andersen cascade impactor (ACI) (Copley Scientific, Nottingham, UK) using methods previously described (32) and in accordance with USP guidance (55). Briefly, four metrics of aerosol performance were assessed: the emitted fraction (EF), which is the percentage of the loaded dose emitted from the device; the respirable fraction (RF_<5 μm and RF_<3 μm), which is the percentage of the loaded dose exhibiting an aerodynamic particle size less than 5 μm and 3 μm, respectively, based upon the stage cutoffs of the ACI at 28.3 L/min; and the fine particle fraction (FPF_<5 μm and FPF_<3 μm), which is the percentage of the emitted dose exhibiting a particle size distribution less than 5 μm and 3 μm, respectively; and, the mass median aerodynamic diameter (MMAD) which was determined by plotting the cumulative percentage of mass less than the stated aerodynamic size cut (expressed as probits) against the aerodynamic diameter (log scale). Statistical analysis of aerosol performance changes before and after storage was determined using ANOVA with Tukey post hoc analysis (alpha level = 0.05) in Prism 9 (Dotmatics, Boston, USA).

In vitro dissolution assay.

An adapted Transwell-based dissolution method from Arora et al. (56) was used to assess the dissolution rate of micronized CFZ particles in a neutral aqueous (phosphate buffer saline, pH 7.4) and acidic media (potassium phosphate buffer, pH 4.5). Surfactant (0.2% w/v) was added to each medium to aid wetting of the poorly water-soluble milled drug; this level was based upon polysorbate 80 levels used in a previously published aerosol dissolution study of poorly water-soluble drugs (57). Two mg of encapsulated milled CFZ was dispersed from a high resistance RS01 Monodose DPI (Plastiape) into an Andersen Cascade Impactor (ACI). The ACI was modified to include six 24-mm glass microfiber filters in stage 4 of the impactor. A 0.45 μm 24 mm insert Transwell system was used to mimic a donor and receptor compartment for the dissolution study. A 1.4-mL aliquot of the dissolution medium (pH 7.4 or pH 4.5) was added to the receptor compartment of the Transwell system. The filter containing deposited drug was placed drug-side down onto the membrane of the Transwell system. An aliquot of 0.4 mL of dissolution medium was added over the filter on the donor side to initiate the dissolution process. The Transwell was placed in a chamber maintained at 37°C and 80% relative humidity. Three replicates were analyzed for each dissolution condition. At each time point of the study, 0.5 mL was removed from the receptor, and 0.5 mL of fresh, prewarmed medium was added. Time points were collected at 10 min, 20 min, 30 min, 45 min, 60 min, 90 min, 2 h, 3 h, 4 h, 5 h, 6 h, and 24 h following the initiation of the dissolution process. At the conclusion of the study, the filters were collected and washed with solvent to determine the amount of undissolved drug remaining. All samples were analyzed for CFZ content using HPLC. Addition of the amount of CFZ concentration remaining on the filter, plus the CFZ concentration at each time point, was used to calculate the total loading concentration on the filter and the percentage dissolved at each time point. The amount of CFZ removed from each time point for sample analysis was accounted for in this calculation.

In vivo evaluation of clofazimine pharmacokinetics and toxicity via oral and pulmonary routes.

The PK, toxicity, and efficacy of micronized CFZ delivered by the oral and pulmonary routes were assessed in naive 6–8-week-old female BALB/c mice obtained from Jackson Laboratories (Bar Harbor, USA). The average weight of the mice was 0.020 kg at the time of treatment. Micronized CFZ suspensions were prepared as described in the previous sections. For pulmonary administration, mice were anesthetized with isoflurane and two sequential administrations of 50 μL each of 6 mg/mL CFZ suspension were administered dropwise into the trachea, resulting in a dose of 0.6 mg CFZ or ~30 mg/kg total. Dropwise intratracheal instillation, rather than utilization of a microsprayer device, was selected as the method of pulmonary administration based upon a previous publication in which more uniform distribution across all lung lobes was achieved using this method (58). For oral administration, 200 μL of the 6 mg/mL suspension was combined with 200 μL of NS, and 200 μL of this dilution was administered via oral gavage, resulting in a total dose 0.6 mg CFZ or ~30 mg/kg total.

To determine the absorption and clearance of drug from the lung and spleen tissues after a single dose of CFZ, a sparse sampling study design was utilized in which n = 3 mice were sacrificed at each time point (1, 6, 24, 48, 72, and 96 h postadministration). These prolonged postdosing time points were selected to capture the elimination phase of the drug based upon the previously reported long half-life of oral CFZ in mice (20) and the unknown efficiency of clearance of the poorly water-soluble CFZ particulates from the lung. At each time point, mice were bled first and immediately euthanized using carbon dioxide. Blood was collected via the submandibular route into a tube with lithium heparin (BD 365965), and plasma was separated by centrifugation at 10,000 × g for 5 min at 4°C. Thereafter, the plasma was collected and stored at −80°C until analysis. Immediately following euthanasia, the tracheas of mice were exposed and the lungs were washed three times with 0.5 mL of PBS. The collected lavage fluid was then centrifuged at 300 × g for 5 min to pellet the cellular component. Spleen and lungs were removed, and the lungs were separated into individual lobes (right upper, right lower, middle, accessory, and left lobe). All samples were flash frozen on dry ice and stored at −80°C until analysis.

Noncompartmental PK analysis of lung, spleen, plasma, and the BAL fluid pellet samples collected from mCFZ-IT and mCFZ-OG mice was performed in the opensource software R (59) using the ubiquity package (60). The goodness of fit of the models was assessed using the adjusted R squared value, which accounts for nonpredictive independent variables and was calculated in the ubiquity package using the equation below, where n is the number of independent variables in the model.

$$\mathit{adj}.r.\mathit{squared}=1\hspace{0.17em}-\hspace{0.17em}\left(1\hspace{0.17em}-\hspace{0.17em}r.\mathit{sq}\right)\text{\hspace{0.17em}*\hspace{0.17em}}\frac{n\hspace{0.17em}-\hspace{0.17em}1}{n\hspace{0.17em}-\hspace{0.17em}2}$$

For the purposes of the PK analysis, all samples that were below the LLOQ (50 ng/mL; equivalent to 1,500 ng/mL when accounting for the 30-fold spleen and lung tissue homogenate dilution) but above the LOD (10 ng/mL) were inputted at a concentration equivalent to the LLOQ/2 (25 ng/mL for plasma samples, 750 ng/g for lung and spleen tissue samples, and 12.5 ng/g for BAL fluid cell pellet samples). Samples below the LOD were inputted as LOD/10 (1 ng/mL for plasma, 30 ng/g for lung and spleen tissue samples, 0.51 ng/g for BAL fluid cell pellet samples). For the single-dose PK study, each lung lobe was analyzed individually as well as in combination (i.e., total lung concentration) by multiplying CFZ concentration (ng/g) of each lobe by the lobe weight to get the mass of drug in the lobe, adding all lobe drug masses together, and then dividing by the total lung weight. For the multidose PK studies, only the left lobe of the lung was analyzed, as this lobe was also utilized for bacterial enumeration to account for potential uneven drug distribution and accumulation across lung lobes.

The effect of administration route on CFZ accumulation in BALB/c mice was determined by administering 30 mg/kg micronized CFZ by the oral and pulmonary routes as previously described twice weekly (Monday and Thursday) for a period of 4 weeks. Mice (n = 3) were sacrificed on days 1 (24-h post dose), 7, 14, 21, and 28 of the study, with the 7- through 28- day sacrifices occurring 96 h after the previously administered dose (i.e., sacrifice was conducted on Mondays, with the last dose taking place the previous Thursday). Plasma and tissue samples were collected and analyzed as previously described.

In vivo evaluation of clofazimine efficacy against Mycobacterium tuberculosis.

Female BALB/c mice (6–8 weeks old) were utilized for the evaluation of efficacy of micronized CFZ delivered by the oral or pulmonary routes for the treatment of chronic TB infection. BALB/c mice were selected based upon their previous utilization in TB efficacy studies, and the ability to generate a sustained infection in the model (32, 38). Infection was performed using the Mtb Erdman strain (TMC107; ATCC 35801). Prior to infection, the bacteria were originally grown as a pellicle to generate low-passage-number seed lots. Working stocks were generated by growing to mid-log phase in Proskauer-Beck medium containing 0.05% Tween 80 (Sigma-Aldrich, St. Louis, USA) in three passages, divided into 1.5-mL aliquots and stored at −80°C until use. Aliquots of the stocks before and after freezing were enumerated by serial dilution on 7H11 agar plates. The mice were infected with a low-dose aerosol infection using the Glass-Col System to deliver ~50–100 bacilli per mouse. Mice were monitored daily; their weights were recorded, and they were euthanized upon displaying symptoms associated with disease or if they had lost more than 20% of their original weight. After the inoculation, mice were rested for 6 weeks to enable chronic infection to develop. At this point, n = 3 mice were sacrificed to determine the baseline bacterial burden, while the remaining mice were randomly divided into study groups (n = 10) and used for efficacy testing of CFZ administered via intratracheal instillation or oral gavage, dosed twice weekly with 30 mg/kg CFZ for 4 weeks total. An additional subset of animals (n = 3) from each treatment group was sacrificed at days 1, 7, 14, 21, and 28 posttreatment initiation to establish pharmacokinetic parameters in the context of an Mtb infection and to determine the timeline of infection resolution.

Following euthanasia at the defined time points, mouse tissues (left lung lobe and a portion of the spleen) were homogenized using the Next Advance Bullet Blender (Averill Park, USA). Briefly, the left lobe of the lung or spleen was placed in a 1.5-mL sterile, safe lock Eppendorf tube containing 0.5 mL of sterile saline and 3 × 3.2 mm, sterile stainless-steel beads; thereafter, the tubes were placed in the Bullet Blender and homogenized during 4 min and 8,000 rpm. After homogenization, an additional 0.5 mL of PBS was added to each tube and samples were used for bacterial enumeration as follows. For bacterial load enumeration, five serial dilutions of homogenized organs from both untreated controls and treatment groups at all time points were prepared and plated in duplicate onto Middlebrook 7H11 agar plates supplemented with oleic acid-albumin-dextrose-catalase (OADC) (Gibco BRL, Gaithersburg, USA), 0.03 mg/mL cycloheximide, and 0.05 mg/mL carbenicillin, and 0.4% activated charcoal to prevent a potential false lowering of the CFU count due to drug carryover effect (12). Homogenates were also plated on charcoal-free 7H11/OADC agar plates. The use of charcoal plates also enabled a direct comparison of CFU to previously published studies on oral CFZ efficacy in BALB/c mice (20, 61, 62), all of which utilized charcoal plates for bacterial enumeration. The number of colonies in charcoal-containing plates was counted after at least 42 days of incubation at 37°C to account for the slower growth of bacteria on this medium. The data were expressed as the mean log₁₀ CFU ± the standard error of the mean for each group. Statistical differences between untreated controls and treatment groups, as well as between the treatment groups, were determined using ANOVA with Sidak’s multiple-comparison test in Prism 9 (Dotmatics, Boston, USA). The alpha level was set at 0.05.

Histopathology analysis was performed on lung tissue samples collected from mice at the infection baseline (prior to treatment), from mice treated for 4 weeks, and from the untreated control group to assess the efficacy of CFZ in reducing the tissue inflammation associated with chronic TB infection. The right middle lobe of the lungs of each mouse was placed into a histology cassette and fixed in 4% paraformaldehyde. Samples were inactivated in 4% paraformaldehyde solution for 48 h and then processed using standard histological protocols for sectioning and staining with Hematoxylin-Eosin (H&E). A blinded assessment of the slides was then performed by a veterinarian pathologist.

In a subset of infected mice (n = 3 per time point), euthanasia and necropsy were performed on days 1, 7, 14, 21, 28 of the treatment period to collect samples for PK analysis in order to determine the relationship, if any, between CFZ tissue concentration and bacterial burden. The left lung lobe and a portion of the spleen were plated for bacterial enumeration as previous described. A portion of these samples was sterilized prior to PK analysis using a previously reported method (63). Briefly, sterilization was achieved by dilution of 50 μL of the homogenate in 200 μL of methanol. Samples were stored at 4°C for 24 h, after which a portion of the sample was centrifuged, supernatant removed, and pellet resuspended in sterile PBS prior to plating for bacterial enumeration. Drug quantitation was performed with HPLC as previously described using a standard curve that matched that used for the sterilization procedure. Multiple unpaired t tests were performed in Prism 9 (Dotmatics, Boston, USA) to compare drug levels in naive and Mtb-infected mice. The samples were assumed to exhibit Gaussian distribution, and the Holm-Sidak method was used to correct for multiple comparisons. Alpha level was set at 0.05.

A simple linear regression analysis was performed in Prism 9 (Dotmatics, Boston, USA) to examine the relationship between the CFZ concentration in plasma, lung, and spleen tissues at varying time points (independent variable) and the bacterial burden within lung and spleen tissue (response variable). The mean values of each point were used to generate the models, and no constraints were placed upon the models (i.e., the line was not forced through the origin). The coefficient of determination (R²) was used to measure the goodness of fit of the model. An F test was performed to test if the independent variable in the linear regression model was significant, and a P value threshold of <0.05 was set a priori. Residual plots were generated for each model (Fig. S5).