Efficacy of Bedaquiline, Alone or in Combination with Imipenem, against Mycobacterium abscessus in C3HeB/FeJ Mice

Mycobacterium abscessus lung infections remain difficult to treat. Recent studies have recognized the power of new combinations of antibiotics, such as bedaquiline and imipenem, although in vitro data have questioned this combination. We report that the efficacy of bedaquiline-imipenem combination treatment relies essentially on the activity of bedaquiline in a C3HeB/FeJ mice model of infection with a rough variant of M. abscessus. The addition of imipenem contributed to clearing the infection in the spleen.

ABSTRACT

Mycobacterium abscessus lung infections remain difficult to treat. Recent studies have recognized the power of new combinations of antibiotics, such as bedaquiline and imipenem, although in vitro data have questioned this combination. We report that the efficacy of bedaquiline-imipenem combination treatment relies essentially on the activity of bedaquiline in a C3HeB/FeJ mice model of infection with a rough variant of M. abscessus. The addition of imipenem contributed to clearing the infection in the spleen.

INTRODUCTION

Mycobacterium abscessus is a rapidly growing mycobacterial species whose infections remain very difficult to treat due to the limited panel of available antibiotics (1). Among them, the β-lactams imipenem (IPM) and cefoxitin (FOX) are part of M. abscessus multidrug therapy, along with amikacin (AMK) and clarithromycin (CLR) (2–5). In addition, the development of specific β-lactamase inhibitors, enhancing the efficacy of IPM in vitro and in vivo, broadens the use of IPM in M. abscessus drug therapy (6–8). Other studies highlighted the potential of testing new drug combinations that include IPM and are associated with increased efficacy against M. abscessus infection (6, 9, 10); however, the relevance of the bedaquiline (BDQ) plus IPM combination has been questioned (11). BDQ targets ATP synthase and exhibits activity against a wide panel of M. abscessus clinical isolates in vitro and in infected zebrafish, although its effect is bacteriostatic only (12). A recent study suggested that by reducing the intracellular pool of ATP in M. abscessus isolates, BDQ suppresses the effect of IPM and FOX, although the effect of the BDQ-IPM combination was considered additive (11). This led the investigators to conclude that addition of BDQ to a β-lactam-containing regimen may negatively affect the treatment outcome (11). In comparison, data from the hollow-fiber model highlight that β-lactam is the most active and important part of the M. abscessus treatment regimen (13). Because these studies focused exclusively on the interaction of β-lactams and BDQ in vitro, confirmatory results in a preclinical animal model are warranted.

Herein, we explored the therapeutic efficacy of BDQ and IPM, alone and in combination, using the immunocompetent C3HeB/FeJ mouse model of M. abscessus infection. C3HeB/FeJ mice are highly susceptible to mycobacterial infections, particularly to Mycobacterium tuberculosis due to a deletion of the Ipr1 (intracellular pathogen resistance 1) gene located within the locus sst1 (14, 15). All animal experiments were performed according to ethical guidelines and with ethical committee (no. 047 with agreement A783223) agreement APAFIS 11465.

First, we evaluated the in vitro interaction between BDQ and several β-lactams or CLR against M. abscessus strain CIP104536 in cation-adjusted Mueller-Hinton broth (CaMHB) (Becton, Dickinson, Le Pont-de-Claix, France) using a 2-dimensional microdilution checkerboard method, as previously described (16–19). Our results confirm that the β-lactam plus BDQ combinations are indifferent, as is the case for the CLR-BDQ combination (Table 1).

TABLE 1.

Interaction between bedaquiline and other drugs against M. abscessus strain CIP104536^T

Compound	MICa (mg/liter)	Interaction with BDQ
		FICIb score (mean ± SD)	Outcomec
BDQ	0.125
IPM	16	0.55 ± 0.06	Indifferent
FOX	32	0.52 ± 0.03	Indifferent
CLR	2	0.61 ± 0.09	Indifferent
AMP	>512	0.57 ± 0.02	Indifferent

^aMICs were evaluated by REMA (resazurin microtiter assay) checkerboard assay in CaMHB (Becton, Dickinson, Le Pont-de-Claix, France). 10⁵ bacteria were diluted in Mueller-Hinton media (Sigma-Aldrich). Plates were incubated for 3 days at 30°C, after which 20 μl (10% vol/vol) of resazurin 0.025% was added to the wells; the plates were incubated overnight at 30°C.

^bThe fractional inhibitory concentration index (FICI) was calculated as follows: FICI = (MIC drug A in combination/MIC drug A alone) + (MIC drug B in combination/MIC drug B alone), where drug A was bedaquiline (BDQ) and drug B was clarithromycin (CLR) (Sigma-Aldrich, France), imipenem (IPM, Mylan S.A.S., France), cefoxitin (FOX, Panpharma, France), or ampicillin (AMP, Euromedex, France). Values are means of 4 independent experiments.

^cInteraction between the 2 compounds was defined as synergistic when FICI was ≤0.5, indifferent when 0.5 < FICI ≤ 4, and antagonistic when FICI was >4.

Next, the performances of pulmonary and intravenous (i.v.) infection routes were compared in C3HeB/FeJ mice. Mice were infected intratracheally using agar bead-embedded bacteria to maintain a persistent infection, as reported previously for Pseudomonas aeruginosa (20). A significant increase in mortality was noted when mice were infected intratracheally with a solution of agar beads containing 2.10⁵ CFU/mouse in 50 μl, leading to only 40% mouse survival at 14 days postinfection (dpi) (see Fig. S1A in the supplemental material), which correlated with an important increase in the CFU at 14 dpi, suggesting accelerated bacterial growth in the lungs (see Fig. S1B). In contrast, persistence occurred for up to 25 days after i.v. infection with 10⁶ CFU/mouse, as evidenced by CFU counting after plating of the organ homogenates (Fig. 1; see also Fig. S2A and B in the supplemental material), although as soon as the injected dose was <10⁶ CFU, persistence in the organs was reduced (Fig. S2B). This represents an important improvement over results found in previously described murine models, characterized by a more rapid bacterial clearance (21–23).

FIG 1.

Bacterial persistence of M. abscessus strain CIP104536^T (rough variant) in lungs, spleens, and livers of C3HeB/FeJ mice after i.v. infection in the tail vein with 10⁶ CFU/mouse in a total volume of 200 μl of water containing 0.9% sodium chloride. The following day, 3 mice were euthanized, and whole organs were harvested to determine baseline bacterial burden. Mouse lungs, spleens, and livers were homogenized, serially diluted, and plated onto VCAT (vancomycin, colistin sulfate, amphotericin B, and trimethoprim) chocolate agar plates (bioMérieux, France) and incubated for 5 to 6 days at 37°C before CFU counting. Results are expressed as log₁₀ CFU at 1, 12, and 25 dpi.

The i.v. route of infection was subsequently used to evaluate and compare the activities of BDQ and AMK. Because AMK is bactericidal against M. abscessus isolates and BDQ is bacteriostatic in vitro, we wondered whether BDQ would be more effective than AMK in an in vivo infection model. CFU were significantly reduced in mice receiving BDQ 30 mg/kg (orally) compared with mice receiving AMK 150 mg/kg (subcutaneously) in lungs and spleen at 12 and 25 dpi (Fig. 2A and B). No significant differences were observed between the BDQ- and AMK-treated animals in the spleen at 12 dpi, but bacterial loads in these two groups were significantly lower than those in the control group (oral administration of dimethyl sulfoxide [DMSO]) (Fig. 2C).

FIG 2.

M. abscessus R-infected C3HeB/FeJ mice (9.2 × 10⁵ CFU/mouse) treated with bedaquiline (BDQ) or amikacin (AMK). Bacterial counts in the lungs (A), liver (B), and spleen (C) of C3HeB/FeJ mice infected i.v., as described in Fig. 1. Antibiotic treatment began at 2 dpi. Mice were treated starting on day 2 for 7 days (D12) or 17 days (D25) by daily subcutaneous injections of AMK 150 mg/kg (Mylan Laboratories) in saline solution or daily oral gavage of BDQ 30 mg/kg in a total volume of 200 μl (BDQ solution in DMSO was diluted in 20% 2-hydroxypropyl-β-cyclodextrin). A control group received a daily subcutaneous injection of saline and oral gavage of DMSO containing 20% 2-hydroxypropyl-β-cyclodextrin. All solutions were administered 5 times weekly for later time point. Mice were euthanized 3 days after antibiotic cessation to allow for antibiotic clearance. Furthermore, given the long half-life and high protein binding capacity of BDQ, spleens, livers, and lungs from drug-treated and control mice were homogenized in water supplemented with 10% bovine serum albumin before dilution (30). Experimental groups of mice were evaluated for bacterial burden on day 1 (before treatment started), 12, and 25 as described in Fig. 1. Five mice were used per group. Bacterial load in each group is expressed as log₁₀ CFU ± SD. Differences between means were analyzed by two-way analysis of variance (ANOVA) and the Tukey posttest, allowing for multiple comparisons. n.s., nonsignificant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. Experiment was performed once.

The efficacy of BDQ in this infection model prompted us to compare it with subcutaneous IPM, alone or as a companion drug, for 15 days of treatment. No significant differences were noticed between the animals treated with BDQ alone and the animals treated with BDQ-IPM at 12 and 20 dpi, with the exception of the liver at 12 dpi (Fig. 3A to C), indicating that the overall activity of the BDQ-IPM combination was mainly due to the intrinsic activity of BDQ. In general, BDQ alone or in combination with IPM exhibited increased activity compared with that of IPM alone in the liver and spleen but not in the lungs (Fig. 3). The spleens of treated and untreated mice were weighed as an additional marker of the effectiveness of the various treatments. These measures indicated that only treatments with BDQ plus IPM or IPM alone were associated with lower spleen weights than those of the untreated or BDQ-treated mice (Fig. 3D). Collectively, the reduced bacterial burden and the lower spleen weights represent a marker for improved outcome of the infection.

FIG 3.

M. abscessus R-infected C3HeB/FeJ mice treated (2.7 × 10⁵ CFU/mouse) with bedaquiline (BDQ), imipenem (IPM), or BDQ plus IPM. Bacterial loads in the lungs (A), liver (B), and spleen (C) were determined as reported in Fig. 1. Antibiotic treatment began 2 days after infection. Mice were treated starting on day 2 for 7 days (D12) or 13 days (D20) with twice-daily (i.e., every 12 h) subcutaneous injection of IPM (MSD Laboratories, France) in saline solution at 100 mg/kg or daily oral gavage of BDQ as described in Fig. 2 or IPM-BDQ. Experimental groups of mice were evaluated for bacterial burden on day 1 (before treatment started), 12, and 20 as described in Fig. 1. (D) Relative weights of spleen to mouse. Mouse spleens were weighed at 1, 12, and 20 dpi. Values represent the relative weight of each spleen relative to the weight of the mouse from which it was collected. Five mice were used per experiment. Bacterial load in each group is expressed as log₁₀ CFU ± SD cells. Differences between means were analyzed by two-way ANOVA and the Tukey posttest, allowing for multiple comparisons. n.s., nonsignificant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. Experiment was performed once.

BDQ is a diarylquinoline approved by the Food and Drug Administration and the European Medicines Agency for the treatment of multidrug-resistant tuberculosis. It is bacteriostatic against M. abscessus isolates in vitro, displaying an MIC₅₀ of 0.125 μg/ml and an MIC₉₀ of >16 μg/ml; and epidemiological cutoff values showed that BDQ exhibits moderate activity (16, 24). Discordant results regarding the efficacy of BDQ were generated in various immunocompromised mouse models, raising the question of the influence of immunosuppression on antibiotic efficacy (25, 26). However, efficient responses to BDQ were observed in other animal models, such as zebrafish (12). Two studies reported poor or negative results for BDQ administration against nontuberculous mycobacteria-infected patients (27, 28). However, recent studies showed that the activity of BDQ can be potentiated with adjunctive therapy, thus improving BDQ-based treatments (16, 29). This study provides evidence that treatment with the BDQ-IPM combination remains superior to treatment with IPM alone and equivalent to that with BDQ alone, as judged by the comparable bacterial clearance in the spleens of the mice treated with BDQ-IPM versus BDQ alone.

In summary, the BDQ-IPM combination enhances clearance of the infection. This also supports the importance of evaluating antibiotic activity in combination rather than separately against this highly drug-resistant mycobacterium.