M A J O R A R T I C L E

In Vivo Evaluation of Antibiotic Activity AgainstMycobacterium abscessus

Isabelle Lerat,1 Emmanuelle Cambau,3,4 Romain Roth dit Bettoni,1 Jean-Louis Gaillard,5,6 Vincent Jarlier,1,2

Chantal Truffot,1 and Nicolas Veziris1,2

1Laboratoire de Bactériologie-Hygiène, Faculté de Médecine, Université Pierre et Marie Curie, 2Laboratoire de Bactériologie-Hygiène, Hôpital Pitié-Salpêtrière, Centre National de Référence des Mycobactéries et de la Résistance des Mycobactéries aux Antituberculeux, 3Université Paris Diderot, and4Service Bactériologie, Hôpital Lariboisière, Paris, 5Université de Versailles – Saint-Quentin-en-Yvelines (UVSQ), France and 6 Laboratoire deMicrobiologie-Hygiène, Hôpital Ambroise Paré, Boulogne, France

Background. The prognosis of Mycobacterium abscessus infections is poor due to the lack of effective drug treat-ment. The objective of this study was to set up an animal model suitable to test antibiotic activity againstM. abscessus.

Methods. The following mouse strains were evaluated: Swiss, BALB/c, C57BL/6, nude, beige, A/J, and GKO. Anti-biotic activity was tested for clarithromycin, amikacin, cefoxitin, tigecycline, and bedaquiline (TMC207). Finally, weevaluated the 3-drug combination clarithromycin, cefoxitin, and amikacin.

Results. Nude and GKO mice fulfilled criteria for the model but only nude mice offered sufficient availability forlarge therapeutic experiments. Among the 3 drugs usually combined for treatment ofM. abscessus infection, cefoxitinwas the most active because it improved survival and reduced bacillary loads in spleen whereas clarithromycin andamikacin prevented death but had little impact on bacillary loads. The triple-drug combination was not more activethan cefoxitin alone. Tigecycline displayed bactericidal activity whereas bedaquiline was almost inactive.

Conclusions. Nude mice are an adequate model for in vivo chemotherapy studies. Among tested drugs, cefoxitinand tigecycline showed promising in vivo activity againstM. abscessus. The best drug combination remains to be deter-mined.

Keywords. Mycobacterium abscessus; clarithromycin; amikacin; cefoxitin; tigecycline; bedaquiline; TMC207;cystic fibrosis; murine model; nude mouse.

Mycobacterium abscessus [1, 2] is a rapidly growing my-cobacterium that is responsible for environmentally ac-quired infections ranging from localized abscess torespiratory [3–5] and disseminated diseases in immuno-compromised patients [6–8]. Localized skin and soft-tissue infections are acquired after accidental trauma orfrom medical equipment [9, 10]. Respiratory infectionsinvolve particularly young patients with cystic fibrosis orelderly female patients with bronchiectasis [6, 11].

Infections due to M. abscessus are difficult to treat[12] because of natural resistance to the classic

antituberculosis drugs and also to most of the antibiot-ics that are currently available [13]. Mycobacteriumabscessus has been presented recently as a complex in-cluding subspecies, such as M. abscessus sensu stricto,M. abscessus massiliense, and M. abscessus bolletii. Inthe rare studies in which the subspecies are distin-guished, it was reported that Mycobacterium massi-liense is more susceptible to antibiotics and infectionsdue to this subspecies are more often cured [14, 15].However, M. abscessus sensu stricto ([M. abscessus]) isthe most often isolated subspecies and is intrinsicallyresistant to most antimicrobial agents [16].

The treatment of M. abscessus infections is currentlybased on in vitro results and on few reported clinical casesbecause, to our knowledge, no controlled studies havebeen conducted. Because antibiotic susceptibility testingis not fully standardized, clinical response to drugs doesnot correlate well with in vitro susceptibility test results.Moreover, failure occurs frequently despite administra-tion of 2 or 3 antibiotics for several months [12, 17].

Received 17 June 2013; accepted 3 October 2013; electronically published 18November 2013.

Correspondence: Nicolas Veziris, MD, PhD, Laboratoire de Bactériologie, Facultéde Médecine, Université Pierre et Marie Curie, 91 Blvd de l’Hôpital, 75634 ParisCedex 13, France (nicolas.veziris@upmc.fr).

The Journal of Infectious Diseases 2014;209:905–12© The Author 2013. Published by Oxford University Press on behalf of the InfectiousDiseases Society of America. All rights reserved. For Permissions, please e-mail:journals.permissions@oup.com.DOI: 10.1093/infdis/jit614

Mycobacterium abscessus in vivo Therapy • JID 2014:209 (15 March) • 905

at Universite L

aval on July 10, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

The study of antibiotic activity in experimental models hasbeen very useful for designing the treatment of tuberculosis[18], leprosy, and some nontuberculous mycobacterial infec-tions, such as Mycobacterium avium and Buruli ulcer [19–21].Few studies have been carried out for rapidly growing mycobac-teria infections, because there is no validated experimentalmodel to test treatment regimens.

We up an experimental model for evaluating antibiotic activ-ity against M. abscessus infection, focusing on disseminated in-fections, which are more difficult to cure. Because M. abscessusis less virulent than Mycobacterium tuberculosis, the methodused for the tuberculosis model was adapted to obtain a modelwith high bacterial loads in organs persisting for a long time,essential for the study of both the elimination of bacteria andthe selection of resistant mutants under treatment. We usedthis model to study the in vivo activity of the 3 drugs usuallyused in the treatment of M. abscessus infection, amikacin, ce-foxitin and clarithromycin alone and combined. We also evalu-ated the activity of 2 new antibiotics, tigecycline andbedaquiline (TMC207).

MATERIALS ANDMETHODS

Mycobacterial StrainFor all the experiments we used the reference strain M. absces-sus ATCC 19977, purchased from the Institute Pasteur Collec-tion (CIP 104536T).

Development of Mouse ModelWe followed the experimental guidelines provided by themedical faculty of Université Pierre et Marie Curie. Severalmice strains were tested for the model ofM. abscessus infection:BALB/c, C57BL/6J, nude (athymic mice with depletion of Tcells), GKO mice (gamma interferon knot-out), A/J mice (defi-cient macrophagic functions), and beige mice (deficient innatural killers). BALB/c, Swiss, C57BL/6J, and nude mice werepurchased from Janvier Breeding Center, A/J and beige micefrom the Jackson Laboratory, and GKO mice from Centre Na-tional de la Recherche Scientifique. All mice were 4–6 weeksold, and all were female, except GKO mice owing to low avail-ability.

Each mouse was inoculated intravenously in the tail veinwith 0.5 mL of a bacterial suspension containing 106–108

colony-forming units (CFU) freshly prepared from a 4-dayculture on trypticase soy agar (TSA). Seven experiments wereconducted successively, because it was not possible to handleall the mice in parallel in a single experiment. The first 5 experi-ments were allocated to evaluate the development of M. absces-sus infection in the different mouse strains. All included BALB/c mice as control. The 2 others were conducted to confirmresults obtained with the nude strain which appeared as themore promising. The numbers of mice used in each experiment

are presented in Supplementary Table 1. Mice were euthanizedon days 1, 7, 14, 21, and 28 after inoculation and, when possi-ble, on days 45, 60, and/or 75. The CFU counts were deter-mined in the spleen, liver, lungs, and kidneys from 2 mice forexperiments 1–5 and from 5 mice for experiments 6 and 7.

Experimental Chemotherapy TrialsIn vitro antibiotic susceptibility was assessed before inoculationand after treatment experiments in mice by determining theminimum inhibitory concentration (MIC) in liquid medium.The microdilution method with standardized microtiter plates(Sensititre; Trek Diagnostic System) was used for clarithromy-cin, amikacin, cefoxitin, and tigecycline. The results were as-sessed after 3–5 days of incubation, as recommended by theClinical and Laboratory Standards Institute [21], but for clari-thromycin results were assessed after 14 days of extended incu-bation to detect inducible resistance [19]. For bedaquiline, amacrodilution method in brain-heart infusion broth was usedwith 3-day incubation. Minimum bactericidal concentration(MBC) was measured after 14 days of incubation for bedaqui-line. For that purpose, liquid cultures not showing any macro-scopic growth were numerated onto TSA as well as the initialinoculum before incubation. The MBC was defined as thelowest drug concentration that killed ≥99.99% of the initialpopulation.

Three trials were conducted in nude mice. Trial 8 evaluatedthe activity of clarithromycin, cefoxitin, and amikacin aloneand combined. Trial 9 evaluated the activity of tigecycline. Trial10 evaluated the activity of bedaquiline. Drug dosing waschosen to mimic human pharmacokinetics at the usual dosing[22–26]. The details of experimental scheme are presented inSupplementary Table 2. Clarithromycin, cefoxitin, amikacin,and tigecycline were purchased in forms for human use (Zeclar[Abbott], Mefoxin [Merck Sharp & Dohme], Amiklin [Bristol-Myers Squibb], and Tygacil [Pfizer], respectively); bedaquilinewas generously provided by Johnson and Johnson laboratories.

Assessment of ResultsMortality, macroscopic lesions in organs, and CFU counts inorgans were used as parameters for assessing the in vivo multi-plication of the bacilli and the severity of infection. Kidneylesions were scored according to the number of abscesses: 0 in-dicated no abscess; 1, 1–10 abscesses; 2, 11–20 abscesses; and 3,>20 abscesses.

The CFU counts were determined in the spleen and lungs inall experiments and also in the kidneys and liver in the develop-ment of the model. The organs were aseptically removed andhomogenized without decontamination. Suspensions weremade up to 2 mL for each organ. At least 4 serial 10-fold dilu-tions of the suspension were plated onto TSA (0.05 mL each)for enumeration of colonies. Because no antibiotic was addedto TSA medium, each colony that was not recognized as a

906 • JID 2014:209 (15 March) • Lerat et al

at Universite L

aval on July 10, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

mycobacteria colony was checked with Ziehl-Neelsen staining.For assessment of bedaquiline results, to avoid carryover [27],Lowenstein-Jensen medium was used in place of TSA. Theresults of the culture were recorded after incubation at 30°C for7 days. The organs of animals that died the day before sacrificewere cultivated (that is, if the animals were not found in late de-composition).

Statistical AnalysisStatistical analysis was performed using the Fisher test for sur-vival and the Mann–Whitney test for mean bacterial loads andkidney lesions.

RESULTS

Mouse Model DevelopmentThe results of the systematic evaluation of M. abscessus infec-tion of each strain of mice (experiments 1–5) during the first

month of infection are presented in Table 1. A few animalsdied among Swiss (0%), C57BL/6 (0%), BALB/c (3%), beige(0%), and nude mice (4%). Mortality rates were higher amongGKO (10%) and AJ (17%) mice.

No gross lesions were observed in the liver or lungs of mice.However, enlargement of spleen and kidney gross lesions(white spots) were observed in all mice, with no clear differenc-es between mouse strains. Based on this observation, a scorewas established that was subsequently used in the chemothera-py experiments (see Materials and Methods).

Among BALB/c, Swiss and C57BL/6 mice, CFU counts de-creased regularly in the spleen, liver and lungs from day 0 today 45 after infection. In the kidneys, CFU counts increasedinitially in BALB/c and Swiss mice, but there was no increase inC57BL/6 mice.

The CFU counts also decreased in the spleen, liver, and lungsof nude mice from day 1 to day 14, but to a lower extent than inthe 3 mouse strains listed above. After day 14, the bacterial load

Table 1. Mortality, Gross Lesions and Mean Colony-Forming Unit (CFU) Counts During the First Month After Inoculation of DifferentMouse Strains WithMycobacterium abscessusa

Outcome

BALB/c Miceb

(Experiments1–5)

NudeMice

(Experiment 2)

GKOMice

(Experiment 3)

A/JMice

(Experiment 4)

BeigeMice

(Experiment 5)

SwissMice

(Experiment 2)

C57BL/6 Micec

(Experiments3 and 5)

Mortality on day30

2/60 0/14 1/10 2/12 0/12 0/13 0/27

Gross lesions

Kidneys Present Present Present Present Present Present Present

Spleen, liver,lungs, tail

Absent Absent Absent Absent Absent Absent Absent

CFU counts, log10SpleenDay 0 7.0 ± 0.3 7.4 7.1 6.7 6.5 7.2 6.7 ± 0.7

Day 28 4.2 ± 0.3 6.6 5.6d 5.4e 3.3 4.5 3.1 ± 0.1

Changef −2.8 −1.1 −1.5 −1.3e −3.2 −2.7 −3.7Liver

Day 0 7.7 ± 0.3 7.9 7.8 7.8 7.3 7.8 7.3 ± 0.4

Day 28 4.2 ± 0.6 6.1 7.4d 5.1e 3.4 3.9 3.8 ± 0.3Changef −3.4 −1.8 −0.4 −2.7e −3.8 −3.9 −3.5

Lungs

Day 0 5.6 ± 0.3 6.0 5.1 5.6 5.4 5.7 5.3 ± 0.0Day 28 3.3 ± 0.4 5.0 3.5d 3.1d 3.0 3.2 3.0 ± 0.6

Changef −2.3 −1.0 − 1.6 −2.5 −2.4 −2.6 −2.3KidneysDay 0 4.8 ± 0.6 5.0 5.0 4.7 4.3 4.7 4.3 ± 0.2

Day 28 6.2 ± 0.4 7.3 10.1d 7.4 6.1 5.3 4.1 ± 0.7

Changef +1.4 +2.3 +5.2 +2.7 +1.8 +0.6 −0.2a The inoculum was log10 8.2, 8.4, 8.2, 8.0, and 7.4 for experiments 1, 2, 3, 4, and 5, respectively. The strain was ATCC 19977.b Mean results of experiments 1–5.c Mean results of experiments 3 and 5.d Only 1 mouse.e Because the culture for day 30 was contaminated, the results for day 21 are reported.f Change from day 0 to day 28.

Mycobacterium abscessus in vivo Therapy • JID 2014:209 (15 March) • 907

at Universite L

aval on July 10, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

reached a plateau at 6 log10 in spleen and liver and at 4–5 log10 inlungs until day 75, with no evidence of a decline trend. In con-trast, CFU increased in kidneys regularly from day 1 to day 28.The increase was about 2 times higher than among BALB/c mice.The bacterial load reached a plateau at 7 log10 at days 28–75.

The CFU counts decreased from day 1 to day 21 in thespleen and liver of GKO mice and, as in nude mice, reached atday 28 a plateau at 5 log10 in the spleen and at 6–7 log10 in theliver. In the lungs, CFU counts decreased from day 1 to day 14(by approximately 1 log10). In kidneys, as for nude mice, CFUincreased regularly from day 1 to day 28. The increase wasabout 3 times higher in GKO than in BALB/c mice. The limitedavailability of GKO mice at the time of the study made impossi-ble to assess CFU counts after day 28. The CFU counts de-creased regularly in the spleen, liver, and lungs of A/J and beigemice from day 1 to day 28. The decrease was approximately thesame as for BALB/c mice (2–4 log10). The bacterial load was <6log10 at day 28. The CFU counts increased markedly in kidneysfrom day 1 to day 14, but this increase was followed by a decrease(starting at day 14 for AJ and at day 28 for beige mice).

Confirmation of the Results Obtained With the Nude MouseModelResults of experiments 6 and 7 were compared with those of ex-periment 2. No deaths occurred, and lesions were observedonly in the kidneys. The evolution of CFU counts was similarin the 4 organs, ending in a plateau in the spleen, liver (approx-imately 6 log10), and lungs (approximately 5 log10) and reach-ing 7–8 log10 in the kidneys. The variations in the results fromone experiment to another were small for the spleen, liver, and

lungs (standard deviation, <0.85 log10). Variations were moreimportant for the kidneys, especially at days 28 and 45, with thestandard deviations reaching 1.5 and 2.4 log10 on days 28 and45, respectively (Figure 1).

In Vivo Efficacy of the Recommended Antibiotic Therapy(Clarithromycin, Amikacin, and Cefoxitin)Mice were inoculated with 6.3 log10 CFU. Pretreatment MICswere 32 µg/mL for cefoxitin and 16 µg/mL for amikacin. Forclarithromycin, the MIC increased from 1 µg/mL after 5 daysof incubation to >16 µg/mL after 14 days.

In the control group, 8 of 21 mice (38%) died. Among thetreated animals, a single mouse died in the group receiving thetriple-drug combination and none in the groups receivingmonotherapy. Compared with untreated animals, the differencewas statistically significant for the monotherapy groups (0/10;P = .03) but not for those receiving the triple-drug combination(1/10; P = .2), although the absolute difference between the 2groups was only 1 mouse (Supplementary Table 3). The kidneylesions were less statistically important in all treated groupsthan in the untreated control group after 2 months, but after 3months the difference remained significant only for cefoxitinand the triple-drug combination.

The bacterial load remained roughly stable in the spleen andlungs of control mice between day 0 and month 2 (Figure 2).This load decreased at month 3 but was assessed in the 4 sur-viving mice. In spleen, compared with untreated mice, alltreated groups showed a significant decrease in CFU counts atmonth 2, whereas at month 3 the difference remained sig-nificant only for cefoxitin and the triple-drug combination

Figure 1. Colony-forming unit (CFU) counts in the organs of nude mice infected with Mycobacterium abscessus; the counts represent mean values forexperiments 2, 6, and 7.

908 • JID 2014:209 (15 March) • Lerat et al

at Universite L

aval on July 10, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

(P = .03). In the lungs, none of the bacillary loads in the treatedgroups differed significantly from those in the control group.

The differences between the untreated control group and the4 treatment groups can be ranked as follows: cefoxitin ≥ triple-drug combination > clarithromycin = amikacin (SupplementaryTable 3). There was no modification of antibiotic MICs aftertreatment. In particular, the strain displayed the same clarithro-mycin-inducible resistance.

Activity of the New Antibiotics Tigecycline and BedaquilineThe MIC of tigecycline for the pretreatment isolate was 0.5 µg/mL.Mice were inoculated with 6.3 log10 CFU. In the control group, 6of 11 mice (55%) died. All the euthanized mice (4 at month 1) andall the dead mice except 1 (1 mouse died at day 56) had numerousgross kidney lesions and the bacterial load remained stable fromday 1 to month 2 in their spleen and lungs. Tigecycline preventeddeath (mortality rate, 0/12; P= .05 vs control) as well as kidneylesions (P = .01 at month 1 and P= .1 at month 2 vs untreatedcontrol mice; Supplementary Table 3). Compared with untreatedmice, CFU counts were significantly smaller in tigecycline-treatedmice in the spleen at months 1 and 2 (P = .02 and .004) and inlungs at month 2 (P = .008) whereas the difference was almost

significant in lungs at month 1 (P = .07; Figure 3). There was no in-crease in the tigecycline MIC for the posttreatment isolate.

The MIC of bedaquiline for the pretreatment isolate was 0.5µg/mL in agar and 0.06 µg/mL in brain-heart infusion broth.The MBC was >2 µg/mL.

Mice were inoculated with 6 log10 CFU. In the control group,2 of 10 mice (20%) died, on day 55. All the mice (sacrificed anddead) had numerous gross kidney lesions. The bacterial load re-mained stable from day 1 to month 2 in spleen but decreasedby about 1.5 log10 in lungs. Bedaquiline did not prevent death(2/10; 20%) or kidney lesions and did not modify the decreaseof bacillary load except in the spleen at month 1 (P = .02) (Sup-plementary Table 3 and Figure 4). There was no increase in thebedaquiline MIC for the posttreatment isolate.

DISCUSSION

The results of the experiments comparing the evolution of M.abscessus infection in 7 strains of mice showed that nude andGKO mice are the most suitable for establishing a model of M.abscessus infection for chemotherapy studies, that is, allowinghigh, stable, and reproducible bacterial loads in organs. Similar

Figure 2. Lung (A) and spleen (B) colony-forming unit (CFU) counts innude mice treated with cefoxitin (300 mg/kg, 4 times daily) clarithromycin(100 mg/kg/d), and amikacin (150 mg/kg/d), alone and in combination(trial 8).

Figure 3. Lung (A) and spleen (B) colony-forming unit (CFU) counts innude mice treated with tigecycline (2.5 mg/kg, twice daily; trial 9).

Mycobacterium abscessus in vivo Therapy • JID 2014:209 (15 March) • 909

at Universite L

aval on July 10, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

results have already been published for GKO mice [28].However, only nude mice fulfilled all the desirable criteria,because GKO mice are expensive and have limited availability,and experimental chemotherapy requires high numbers ofanimals. Moreover, nude mice are largely used for other myco-bacterial infections [29–31] and have been recently chosen as amodel for Mycobacterium xenopi infection [32]. However,because CFU counts initially decrease in nude mice beforereaching a plateau, antibiotic activity should be evaluated bycomparing treated and untreated mice at simultaneous timepoints. The reproducibility of the model, as assessed by stan-dard deviations of mean CFU counts observed in 3 indepen-dent experiments was good for liver and lungs but not forkidneys. Although the CFU count increase in kidneys seems in-teresting, the high variability in the CFU counts, probably dueto the presence of abscesses, prevented us from choosing bacte-rial count in this organ for assessing drug treatment efficacy.We did, however, use the gross lesions seen on kidneys to estab-lish a score allowing a rough estimate of drug treatment efficacy(see Material and Methods).

The second objective of this work was to evaluate the in vivoactivity of antibiotics against M. abscessus. In the 3 therapeutictrials (experiments 8, 9, and 10), lung and spleen CFU and

kidney lesions of untreated control mice evolved as in the 3nude mice model experiments (experiments 2, 6, and 7). Com-pared with model experiments, mortality in untreated controlmice was equivalent in therapeutic trials 8 and 10 (2/10 vs 2/61after 2 months; P = .09) but was higher in therapeutic trial 9 (3/11 vs 1/54 [P = .02] during the first month and 3/6 vs 1/17[P = .03] during the second). We do not think that this diffe-rence limits the interpretation of the results, because tigecyclineclearly prevented death. However, this high difference in mor-tality rates between treated and control groups limits the signifi-cance of CFU count comparison, especially at late time points.Conversely, the mortality observed in bedaquiline-treated micehelps show differences compared with control animals, becausethis drug was not able to prevent death.

The reduction in bacterial loads after clarithromycin treat-ment was limited and comparable to that obtained with bacter-iostatic antituberculous drugs, such as ethambutol. For the M.abscessus complex, as for M. avium [33], the MBC of clarithro-mycin is high, even for strains with low MICs (data notshown). The reference strain used in our study, displays in vitroinducible clarithromycin resistance, as illustrated by the in-creased MIC after 14 days of incubation [34]. Human andmouse data ,recently published, have demonstrated that thelack of activity of clarithromycin againstM. abscessus correlateswith a functional erm gene [14, 35]. However, it should be em-phasized that clarithromycin had a favorable impact on mortalityand consequently should not be considered as completely inac-tive in vivo. These results obtained in nude mice are reminiscentof the retrospective analysis by Jeon et al [12] who found that amultidrug regimen including clarithromycin had favorable effecton symptoms in 75% of M. abscessus infection cases, on high-resolution computed tomography lesions in 42% of cases but re-sulted in bacteriological sputum conversion in only 25%.

Amikacin showed little activity in the nude mouse model, aresult that is consistent with in vitro data showing high MBCcontrasting with low MIC [36]. Recent clinical data showed thatthe combination of clarithromycin plus moxifloxacin was moreactive than that of clarithromycin plus amikacin [37]. Taken to-gether with its long-term toxicity, these results do not supportthe long-term use of amikacin againstM. abscessus infections.

Cefoxitin was the most active drug among those tested intrial 8. It has been shown recently that the peptidoglycan of M.abscessus contains predominantly 3-3 cross-links generated byL,D-transpeptidases [38]. Thus, it is not surprising that cefoxi-tin showed activity in vivo, given that cephems have in vitro ac-tivity against L-D transpeptidases [39]. Because the dosing usedfor mice in our trial is equipotent to that used in humans [22],we think that cefoxitin should be further evaluated for thetreatment of M. abscessus infection in humans. Imipenem, a β-lactam that has been shown to be a potent inhibitor of L,D-transpeptidases, may also have in vivo activity and may displaysynergistic activity with other drugs [39–41].

Figure 4. Lung (A) and spleen (B) colony-forming unit (CFU) counts innude mice treated with bedaquiline (25 mg/kg/d; trial 10).

910 • JID 2014:209 (15 March) • Lerat et al

at Universite L

aval on July 10, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

The triple-drug combination cefoxitin, amikacin, and clari-thromycin was not more active than cefoxitin alone and mayeven be antagonistic, as suggested by CFU counts after 3months. Similar antagonism has been shown recently betweenmacrolides and moxifloxacin [42]. These results do not supportthe use of this combination with the aim to increase antibacte-rial activity. However, because the main goal of drug combina-tions in mycobacterial infections is to prevent drug resistance,it seems reasonable to use at least a 2-drug combination. In-cluding >2 drugs could increase toxicity without increasingefficacy.

The in vitro activity of tigecycline against M. abscessus isalready established [43], but our current results are the firstshowing the in vivo activity of this drug. Tigecycline could alsobe useful in drug combinations because it displays synergisticactivity with clarithromycin and amikacin [44]. However, thereare concerns about its safety of tigecycline, which may limit itsuse against M. abscessus infections that require long-termtherapy [45].

Bedaquiline did not show any activity in the nude mousemodel; it was the only tested drug that did not prevent death.This lack of activity is probably explained by high MBC, asalready described forM. avium [46].

A caveat must be emphasized in interpreting the results ofthese studies. We did not select drug-resistant mutants duringmonotherapy and thus could not assess the capacity of com-bined therapy to prevent drug resistance. We believe that this isdue to insufficient initial bacillary load (5.5–6 log10). However,it must be recalled that all the published work on experimentalchemotherapy for M. abscessus showed initial bacillary loads<7 log10 [35, 42].

Apart from the drugs we tested, some other seem promisingand may improve the treatment of M. abscessus infections.Among these, moxifloxacin has been shown to be active in amurine model and also in clinical setting [37, 42]. Clofaziminehas been shown to act synergistically in vitro with many of thedrugs used against M. abscessus [47, 48]. These 2 drugs, givenorally and having little toxicity, could be interesting choices forthe long-term treatment required for M. abscessus infections.However, because we and others have described antagonismbetween antibiotics in vivo, these drug combinations should betested in animal models before clinical use.

In conclusion, we showed that nude mice can be used as amodel for in vivo evaluation of antibiotic activity againstM. ab-scessus. Cefoxitin and tigecycline show promising in vivo activi-ty and should be further evaluated.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseasesonline (http://jid.oxfordjournals.org/). Supplementary materials consist ofdata provided by the author that are published to benefit the reader. The

posted materials are not copyedited. The contents of all supplementary dataare the sole responsibility of the authors. Questions or messages regardingerrors should be addressed to the author.

Notes

Acknowledgments. We thank Aurélie Chauffour and Maureen Baudoinfor technical assistance and Florence Doucet-Populaire for helpful discussion.Financial support. This work was supported by a grant MODEXA-

MEDICEN pôle de compétitivité Santé Paris Région and by grants fromVaincre La Mucoviscidose and Comité d’Assistance Respiratoire à Domiciled’Ile-de-France.Potential conflict of interest. E. C., V. J., C. T., and N. V. have conduct-

ed research experiments on bedaquiline (TMC207) activity against M. tu-berculosis that have been supported by Janssen Laboratory. All otherauthors report no potential conflicts.All authors have submitted the ICMJE Form for Disclosure of Potential

Conflicts of Interest. Conflicts that the editors consider relevant to thecontent of the manuscript have been disclosed.

References

1. Adékambi T, Berger P, Raoult D, Drancourt M. rpoB gene sequence-based characterization of emerging non-tuberculous mycobacteria withdescriptions of Mycobacterium bolletii sp. nov., Mycobacterium phocai-cum sp. nov. and Mycobacterium aubagnense sp. nov. Int J Syst EvolMicrobiol 2006; 56:133–43.

2. Adékambi T, Reynaud-Gaubert M, Greub G, et al. Amoebal cocultureof ‘Mycobacterium massiliense’ sp. nov. from the sputum of a patientwith hemoptoic pneumonia. J Clin Microbiol 2004; 42:5493–501.

3. Esther CR Jr, Esserman DA, Gilligan P, Kerr A, Noone PG. ChronicMycobacterium abscessus infection and lung function decline in cysticfibrosis. J Cyst Fibros 2010; 9:117–23.

4. Sermet-Gaudelus I, Le Bourgeois M, Pierre-Audigier C, et al. Mycobac-terium abscessus and children with cystic fibrosis. Emerging Infect Dis2003; 9:1587–91.

5. Griffith DE, Girard WM, Wallace RJ Jr. Clinical features of pulmonarydisease caused by rapidly growing mycobacteria: an analysis of 154 pa-tients. Am Rev Respir Dis 1993; 147:1271–8.

6. Griffith DE, Aksamit T, Brown-Elliott BA, et al. An official ATS/IDSAstatement: diagnosis, treatment, and prevention of nontuberculous my-cobacterial diseases. Am J Respir Crit Care Med 2007; 175:367–416.

7. Garrison AP, Morris MI, Doblecki Lewis S, et al. Mycobacterium ab-scessus infection in solid organ transplant recipients: report of threecases and review of the literature. Transpl Infect Dis 2009; 11:541–8.

8. Doucette K, Fishman JA. Nontuberculous mycobacterial infection inhematopoietic stem cell and solid organ transplant recipients. ClinInfect Dis 2004; 38:1428–39.

9. Leao SC, Tortoli E, Viana-Niero C, et al. Characterization of mycobac-teria from a major Brazilian outbreak suggests that revision of the taxo-nomic status of members of the Mycobacterium chelonae-M. abscessusgroup is needed. J Clin Microbiol 2009; 47:2691–8.

10. Viana-Niero C, Lima KVB, Lopes ML, et al. Molecular characterizationof Mycobacterium massiliense and Mycobacterium bolletii in isolatescollected from outbreaks of infections after laparoscopic surgeries andcosmetic procedures. J Clin Microbiol 2008; 46:850–5.

11. Van Ingen J. Strategies to improve outcome of drug treatment for My-cobacterium abscessus pulmonary disease. Clin Infect Dis 2011;52:1281–2.

12. Jeon K, Kwon OJ, Lee NY, et al. Antibiotic treatment ofMycobacteriumabscessus lung disease: a retrospective analysis of 65 patients. Am JRespir Crit Care Med 2009; 180:896–902.

13. Brown-Elliott BA, Nash KA, Wallace RJ. Antimicrobial susceptibilitytesting, drug resistance mechanisms, and therapy of infections withnontuberculous mycobacteria. Clin Microbiol Rev 2012; 25:545–82.

Mycobacterium abscessus in vivo Therapy • JID 2014:209 (15 March) • 911

at Universite L

aval on July 10, 2014http://jid.oxfordjournals.org/

Dow

nloaded from

14. Koh WJ, Jeon K, Lee NY, et al. Clinical significance of differentiation ofMycobacterium massiliense from Mycobacterium abscessus. Am JRespir Crit Care Med 2011; 183:405–10.

15. Kim HS, Lee KS, Koh WJ, et al. Serial CT findings of Mycobacteriummassiliense pulmonary disease compared with Mycobacterium absces-sus disease after treatment with antibiotic therapy. Radiology 2012;263:260–70.

16. Nessar R, Cambau E, Reyrat JM, Murray A, Gicquel B. Mycobacteriumabscessus: a new antibiotic nightmare. J Antimicrob Chemother 2012;67:810–8.

17. Adékambi T, Drancourt M. Mycobacterium bolletii respiratory infec-tions. Emerging Infect Dis 2009; 15:302–5.

18. Veziris N, Ibrahim M, Lounis N, Andries K, Jarlier V. Sterilizing activi-ty of second-line regimens containing TMC207 in a murine model oftuberculosis. PLoS ONE 2011; 6:e17556.

19. Ji B, Chauffour A, Robert J, Lefrançois S, Jarlier V. Orally administeredcombined regimens for treatment of Mycobacterium ulcerans infectionin mice. Antimicrob Agents Chemother 2007; 51:3737–9.

20. Bermudez LE, Kolonoski P, Young LS, Inderlied CB. Activity of KRM1648 alone or in combination with ethambutol or clarithromycinagainst Mycobacterium avium in beige mouse model of disseminatedinfection. Antimicrob Agents Chemother 1994; 38:1844–8.

21. Bermudez LE, Nash KA, Petrofsky M, Young LS, Inderlied CB. Effectof ethambutol on emergence of clarithromycin-resistant Mycobacteri-um avium complex in the beige mouse model. J Infect Dis 1996; 174:1218–22.

22. Lepeule R, Ruppé E, Le P, et al. Cefoxitin as an alternative to carbape-nems in a murine model of urinary tract infection due to Escherichiacoli Harboring CTX-M-15-type extended-spectrum β-lactamase. Anti-microb Agents Chemother 2012; 56:1376–81.

23. Lounis N, Veziris N, Chauffour A, Truffot-Pernot C, Andries K, JarlierV. Combinations of R207910 with drugs used to treat multidrug-resis-tant tuberculosis have the potential to shorten treatment duration.Antimicrob Agents Chemother 2006; 50:3543–7.

24. Andries K, Verhasselt P, Guillemont J, et al. A diarylquinoline drugactive on the ATP synthase of Mycobacterium tuberculosis. Science2005; 307:223–7.

25. Koomanachai P, Crandon JL, Banevicius MA, Peng L, Nicolau DP.Pharmacodynamic profile of tigecycline against methicillin-resistantStaphylococcus aureus in an experimental pneumonia model. Antimi-crob Agents Chemother 2009; 53:5060–3.

26. Meagher AK, Ambrose PG, Grasela TH, Ellis-Grosse EJ. The pharma-cokinetic and pharmacodynamic profile of tigecycline. Clin Infect Dis2005; 41:S333–40.

27. Lounis N, Gevers T, Van Den Berg J, Verhaeghe T, van Heeswijk R,Andries K. Prevention of drug carryover effects in studies assessingantimycobacterial efficacy of TMC207. J Clin Microbiol 2008; 46:2212–5.

28. Ordway D, Henao-Tamayo M, Smith E, et al. Animal model of Myco-bacterium abscessus lung infection. J Leukoc Biol 2008; 83:1502–11.

29. Lounis N, Ji B, Truffot-Pernot C, Grosset J. Comparative activities ofamikacin against Mycobacterium avium complex in nude and beigemice. Antimicrob Agents Chemother 1997; 41:1168–9.

30. Atkinson BA, Bocanegra R, Graybill JR. Treatment of Mycobacteriumhaemophilum infection in a murine model with clarithromycin, rifabu-tin, and ciprofloxacin. Antimicrob Agents Chemother 1995; 39:2316–9.

31. Ji B, Perani EG, Petinon C, Grosset JH. Bactericidal activities of singleor multiple doses of various combinations of new antileprosy drugsand/or rifampin against M. leprae in mice. Int J Lepr Other MycobactDis 1992; 60:556–61.

32. Andréjak C, Almeida DV, Tyagi S, Converse PJ, Ammerman NC, GrossetJH. Improving existing tools for Mycobacterium xenopi treatment: assess-ment of drug combinations and characterization of mouse models of in-fection and chemotherapy. J Antimicrob Chemother 2013; 68:659–65.

33. Heifets LB, Lindholm-Levy PJ, Comstock RD. Clarithromycin minimalinhibitory and bactericidal concentrations against Mycobacteriumavium. Am Rev Respir Dis 1992; 145:856–8.

34. Bastian S, Veziris N, Roux AL, et al. Assessment of clarithromycin sus-ceptibility in strains belonging to the Mycobacterium abscessus groupby erm(41) and rrl sequencing. Antimicrob Agents Chemother 2011;55:775–81.

35. Choi GE, Shin SJ, Won CJ, et al. Macrolide treatment for Mycobacteri-um abscessus andM. massiliense infection and inducible resistance. AmJ Respir Crit Care Med 2012; 186:917–25.

36. Greendyke R, Byrd TF. Differential antibiotic susceptibility ofMycobac-terium abscessus variants in biofilms and macrophages compared tothat of planktonic bacteria. Antimicrob Agents Chemother 2008; 52:2019–26.

37. Choi WS, Kim MJ, Park DW, et al. Clarithromycin and amikacin vs.clarithromycin and moxifloxacin for the treatment of post-acupuncturecutaneous infections due toMycobacterium abscessus: a prospective ob-servational study. Clin Microbiol Infect 2011; 17:1084–90.

38. Lavollay M, Fourgeaud M, Herrmann JL, et al. The peptidoglycan ofMycobacterium abscessus is predominantly cross-linked by L,D-trans-peptidases. J Bacteriol 2011; 193:778–82.

39. Mainardi JL, Hugonnet JE, Rusconi F, et al. Unexpected inhibition ofpeptidoglycan LD-transpeptidase from Enterococcus faecium by the β-lactam imipenem. J Biol Chem 2007; 282:30414–22.

40. Lavollay M, Arthur M, Fourgeaud M, et al. The peptidoglycan of sta-tionary-phase Mycobacterium tuberculosis predominantly containscross-links generated by L,D-transpeptidation. J Bacteriol 2008; 190:4360–6.

41. Miyasaka T, Kunishima H, Komatsu M, et al. In vitro efficacy of imipe-nem in combination with six antimicrobial agents against Mycobacteri-um abscessus. Int J Antimicrob Agents 2007; 30:255–8.

42. Choi GE, Min KN, Won CJ, Jeon K, Shin SJ, Koh WJ. Activities ofmoxifloxacin in combination with macrolides against clinical isolates ofMycobacterium abscessus and Mycobacterium massiliense. AntimicrobAgents Chemother 2012; 56:3549–55.

43. Wallace RJ Jr, Brown-Elliott BA, Crist CJ, Mann L, Wilson RW. Com-parison of the in vitro activity of the glycylcycline tigecycline (formerlyGAR-936) with those of tetracycline, minocycline, and doxycyclineagainst isolates of nontuberculous mycobacteria. Antimicrob AgentsChemother 2002; 46:3164–7.

44. Huang CW, Chen JH, Hu ST, et al. Synergistic activities of tigecyclinewith clarithromycin or amikacin against rapidly growing mycobacteriain Taiwan. Int J Antimicrob Agents 2013;41:218–23.

45. Vardakas KZ, Rafailidis PI, Falagas ME. Effectiveness and safety of tige-cycline: focus on use for approved indications. Clin Infect Dis 2012;54:1672–4.

46. Lounis N, Gevers T, Van Den Berg J, Vranckx L, Andries K. ATP syn-thase inhibition of Mycobacterium avium Is not bactericidal. Antimi-crob Agents Chemother 2009; 53:4927–9.

47. van Ingen J, Totten SE, Helstrom NK, Heifets LB, Boeree MJ, Daley CL.In vitro synergy between clofazimine and amikacin in treatment ofnontuberculous mycobacterial disease. Antimicrob Agents Chemother2012; 56:6324–7.

48. Shen GH, Wu BD, Hu ST, Lin CF, Wu KM, Chen JH. High efficacy ofclofazimine and its synergistic effect with amikacin against rapidlygrowing mycobacteria. Int J Antimicrob Agents 2010; 35:400–4.

912 • JID 2014:209 (15 March) • Lerat et al

at Universite L

aval on July 10, 2014http://jid.oxfordjournals.org/

Dow

nloaded from