Abstract

Objectives

This study aimed to provide basic pharmacodynamic information for key antibiotics used to treat Mycobacterium avium and Mycobacterium xenopi pulmonary disease.

Methods

M. avium subspecies hominissuis IWGMT49 and M. xenopi ATCC 19250 type strains were used; the MICs of clarithromycin, amikacin and moxifloxacin were determined by broth microdilution. Time–kill assays were performed, exposing bacteria to 2-fold concentrations from 0.062× to 32× the MIC at 37°C for 240 h for M. avium or 42 days for M. xenopi. The sigmoid maximum effect (Emax) model was fitted to the time–kill curve data.

Results

Maximum killing of M. avium by amikacin was obtained between 24 and 120 h (0.0180 h−1) and was faster and higher than with clarithromycin (0.0109 h−1); however, regrowth and amikacin-resistant mutants were observed. Killing rates for M. xenopi were higher, 0.1533 h−1 for clarithromycin and 0.1385 h−1 for moxifloxacin, yet required 42 days. There were no significant differences between the Hill's slopes determined for all of the antibiotics tested against M. avium or M. xenopi (P = 0.9663 and P = 0.0844, respectively).

Conclusions

The killing effect of amikacin and clarithromycin on M. avium subspecies hominissuis was low, although amikacin activity was higher than that of clarithromycin, supporting its role in a combined therapy. Clarithromycin and moxifloxacin may have similar activity within treatment regimens for M. xenopi disease. Future studies of in vitro and in vivo pharmacokinetic/pharmacodynamic interactions are needed to improve the current regimens to treat these two important slowly growing mycobacteria in pulmonary disease.

Introduction

Pulmonary disease is the most common clinical manifestation of non-tuberculous mycobacteria (NTM) infection.1 Several NTM can cause such disease, although their geographical distribution varies from region to region.2Mycobacterium avium complex (MAC) is the most frequently isolated slowly growing mycobacteria (SGM) involved in pulmonary disease,2,3 but Mycobacterium xenopi has also been frequently isolated in Northern Europe as an important infectious species.2,3

Treatment of pulmonary disease caused by M. avium and M. xenopi is a challenge and the outcome is uncertain. Only macrolides, like azithromycin or clarithromycin, are active both in vitro and in vivo against M. avium,1,4 hence treatment regimens include macrolides, but rifampicin and ethambutol are added in order to prevent acquired macrolide resistance. Amikacin is also used, during the first 2–3 months of therapy, despite clinical evidence; only the in vitro MIC supports its role.1,4 Likewise, the recommended regimen for M. xenopi includes isoniazid, rifampicin, ethambutol and clarithromycin, with aminoglycosides added during the first months. Moxifloxacin has also been suggested to play a role in the treatment of such mycobacterial disease.1,4 However, there is little information on the effectiveness of those antibiotics against M. avium and M. xenopi pulmonary disease and cure rates are low.4,5 Moreover, patient cohorts of M. xenopi pulmonary disease stand out for their high mortality rates.3

The limited information on the activity of the antibiotics currently used in these regimens emphasizes the need to better understand the fundamentals of SGM treatment, in particular the potency and activity of the antimicrobials that currently play a major role. In the present study we aimed to study the basic pharmacodynamics for key antibiotics used to treat M. avium and M. xenopi pulmonary disease by performing in vitro time–kill kinetics. Because of the slow growth of SGM and the long duration of treatment, some experiments were conducted over up to 42 days of exposure.

Materials and methods

Bacterial strains and antibiotics

We used M. avium subspecies hominissuis IWGMT49 (International Working Group on Mycobacterial Taxonomy) and M. xenopi ATCC 19250 (ATCC, Manassas, VA, USA) as test strains. Stock vials of each mycobacteria in the early logarithmic phase of growth were preserved at −80°C in trypticase soy broth with 40% glycerol and were thawed for each assay. Moxifloxacin, amikacin and clarithromycin were obtained from Sigma-Aldrich (Zwijndrecht, The Netherlands) and dissolved in water, except for clarithromycin, which was dissolved in methanol, following CLSI recommendations.6 Stock solutions, stored at −80°C, were thawed for each experiment to prepare the different concentrations to be tested.

Susceptibility testing

The MIC of each of the tested antibiotics was determined by broth microdilution in CAMHB (BD Bioscience, Erembodegem, Belgium) at 37°C, according to CLSI guidelines,7 using commercial panels (SLOWMYCO Sensititre®, Trek Diagnostics/ThermoFisher-Landsmeer, The Netherlands).

Time–kill assays

Individual bottles of 20 mL of Middlebrook 7H9 (BD Bioscience, Erembodegem, Belgium) plus OADC growth supplement (BD Bioscience) and 0.05% Tween 80 (Sigma-Aldrich), containing ten 2-fold increasing concentrations of each antibiotic (from 0.062× to 32× the MIC) were cultured with the inoculum (density ∼105–106 cfu/mL) at 37°C. All bottles were shaken (100 rpm) and ventilated with air through a bacterial filter (FP 30/0.2 Ca/S, Whatman GmbH, Germany). In order to perform cfu counting of the bacterial population, samples of 200 μL were taken from each bottle and serial 10-fold dilutions in 0.85% sterile saline solution were plated on Middlebrook 7H11 (BD Bioscience) plates, at different time intervals (12, 24, 48, 72, 96, 120, 144, 168 and 240 h for M. avium and 3, 6, 9, 12, 15 18, 21, 28 and 42 days for M. xenopi).

Mutation frequency

Mutants were counted at time 0 and after 240 h for M. avium exposed to both amikacin and clarithromycin, by plating the bacterial suspension and 10-fold dilutions on Middlebrook 7H11 plates containing 4× the MIC of each antibiotic. Owing to the higher mutation rate expected for amikacin, mutation frequency was determined throughout the experiment for bottles containing 1×, 4×, 8×, 16× and 32× MIC. No mutation frequency was determined for M. xenopi, since no resistant mutants were expected.

Curve fitting and analysis

Log cfu values were plotted against time for each antibiotic and analysed using Graphpad Prism 5.03 (Graphpad Inc., San Diego, CA, USA). The sigmoid maximum effect (Emax) model (four-parameter Hill's equation8) was fitted to the kill rate data, analysing each assay to determine the pharmacodynamic relationship between antibiotic concentration and bacterial growth or death. Emax, 50% effective concentration (EC50), Hill's slope (ɣ) with 95% CI and R2 were calculated for each assay.

Results

Susceptibility

Clarithromycin and amikacin MICs for M. avium were 4 and 8 mg/L. Clarithromycin and moxifloxacin MICs for M. xenopi were both 0.25 mg/L.

Time–kill assays

M. avium

Kill curves of M. avium exposed to clarithromycin and amikacin are shown in Figure 1. In the experiments with clarithromycin we observed a lag phase of 24 h; thereafter a slow decline in bacterial cfu/mL was noted for each concentration, except for the lowest of 0.062× MIC, which only produced inhibition of growth. Killing appeared to be maximized at relatively low multiples of the MIC and was hardly concentration dependent. Regrowth was observed after 168 h for concentrations in the lower range—0.062×, 0.25×, 0.5× and 2× MIC. In contrast, amikacin killing was faster and higher for concentrations above 2× MIC. However, regrowth was observed earlier than with clarithromycin, after 96–120 h, except for 4× MIC.

Figure 1.

Time–kill curves of M. avium IWGMT49 with (a) clarithromycin and (b) amikacin. Antibiotic concentrations are indicated by different symbols.

Figure 1.

Time–kill curves of M. avium IWGMT49 with (a) clarithromycin and (b) amikacin. Antibiotic concentrations are indicated by different symbols.

M. xenopi

Kill curves of M. xenopi exposed to clarithromycin and moxifloxacin are shown in Figure 2. When clarithromycin was used, the lag phase involved 3 days and thereafter a slow decline in cfu/mL was observed at almost all concentrations; killing was maximum after 42 days. Low clarithromycin concentrations showed killing, but similar to M. avium, regrowth was observed at the lower concentrations of 0.062×, 0.125×, 0.25× and 0.5× MIC. Concentrations of 1× MIC and higher produced constant killing with no regrowth; only concentrations 4× MIC and higher resulted in undetectable cfu/mL after 42 days of exposition. When moxifloxacin was tested, a lag phase of 3 days was also observed, but only for concentrations of 0.062 to 1× MIC; concentrations 2× MIC and higher already showed a killing effect by that time. After 3 days, low concentrations of 0.062× to 0.25× MIC initially showed some killing, but regrowth appeared after 9–12 days, depending on the concentration. Constant killing with no regrowth was observed with concentrations 0.5× MIC and higher, but only concentrations 1× MIC and higher resulted in undetectable cfu/mL after 42 days of exposition.

Figure 2.

Time–kill curves of M. xenopi ATCC 19250 with (a) clarithromycin and (b) moxifloxacin. Antibiotic concentrations are indicated by different symbols.

Figure 2.

Time–kill curves of M. xenopi ATCC 19250 with (a) clarithromycin and (b) moxifloxacin. Antibiotic concentrations are indicated by different symbols.

High mutation frequency in amikacin-exposed M. avium

No amikacin-resistant or clarithromycin-resistant mutants were detected at time 0 for the M. avium inoculum used in the experiments. After 240 h of incubation of the growth control, we observed a mutation frequency of 3.24 × 10−7 for amikacin and a mutation frequency of 4.05 × 10−8 for clarithromycin.

In amikacin-exposed cultures, the first M. avium resistant mutants appeared after 72–96 h of incubation. Mutation frequency after 240 h of exposure was on average 7.3 × 10−3, excluding 4× MIC where no mutants appeared (Table 1).

Table 1.

Mutation frequency of M. avium IWGMT49 exposed to amikacin after 240 h

Amikacin concentration cfu/mL Mutation frequency One mutant in cfu/mL 
Growth control 4.00 × 109   
1× MIC (8 mg/L) 2.80 × 107 7.00 × 10−3 143 
4× MIC (32 mg/L) — — 
8× MIC (64 mg/L) 3.23 × 107 8.03 × 10−3 124 
16× MIC (128 mg/L) 5.33×107 1.33 × 10−2 75 
32× MIC (256 mg/L) 3.33 × 106 8.33 × 10−4 1200 
Amikacin concentration cfu/mL Mutation frequency One mutant in cfu/mL 
Growth control 4.00 × 109   
1× MIC (8 mg/L) 2.80 × 107 7.00 × 10−3 143 
4× MIC (32 mg/L) — — 
8× MIC (64 mg/L) 3.23 × 107 8.03 × 10−3 124 
16× MIC (128 mg/L) 5.33×107 1.33 × 10−2 75 
32× MIC (256 mg/L) 3.33 × 106 8.33 × 10−4 1200 

Emax model

The Emax model fitted well; the highest kill rate for M. avium was observed after 24–120 h of exposure to amikacin, 0.0180 h−1. There was no significant difference in Hill's slope estimated for amikacin and clarithromycin (P = 0.9663), indicating that the differences in effect modality are primarily determined by the maximum kill rate of each antibiotic.

For M. xenopi, the maximum kill rate was observed in a different time interval, 3–42 days of exposure. The Emax values were higher than for M. avium, but very similar between clarithromycin and moxifloxacin. Again, no significant differences were found between the Hill's slope calculated for the two antibiotics tested (P = 0.0844).

Table 2 shows the parameter estimates obtained with the Emax model for each experiment and Figure 3 shows the best fitted sigmoid curves for the kill rate and concentration for the two SGM.

Table 2.

Parameter estimates, with 95% CI, derived from the inhibitory Emax

Strain Antibiotic Time Emax (h−195% CI EC50 95% CI Hill's slope (ɣ)a 95% CI R2 
M. avium IWGMT49 clarithromycin 24–120 h 0.0109 0.0089–0.0130 0.149 0.010–0.222 1.838 1.265–2.410 0.98 
 amikacin  0.0180 0.0154–0.0205 1.427 1.160–1.755   0.99 
M. xenopi ATCC 19250 clarithromycin 3–42 days 0.1533 0.1356–0.1711 0.191 0.149–0.245 2.543 1.484–3.602 0.98 
 moxifloxacin 0.1385 0.1232–0.1539 0.076 0.059–0.099  0.96 
Strain Antibiotic Time Emax (h−195% CI EC50 95% CI Hill's slope (ɣ)a 95% CI R2 
M. avium IWGMT49 clarithromycin 24–120 h 0.0109 0.0089–0.0130 0.149 0.010–0.222 1.838 1.265–2.410 0.98 
 amikacin  0.0180 0.0154–0.0205 1.427 1.160–1.755   0.99 
M. xenopi ATCC 19250 clarithromycin 3–42 days 0.1533 0.1356–0.1711 0.191 0.149–0.245 2.543 1.484–3.602 0.98 
 moxifloxacin 0.1385 0.1232–0.1539 0.076 0.059–0.099  0.96 

aPooled value for the two antibiotics evaluated by species.

Figure 3.

Best fitted sigmoid curves obtained from the Emax model of (a) M. avium IWGMT49 exposed to clarithromycin and amikacin, between 24 and 120 h, and (b) M. xenopi ATCC 19250 exposed to clarithromycin and moxifloxacin, between 3 and 42 days. A different y-axis scale is used for (a) and (b). Serum Cmax for MAC-infected patients when treated with amikacin (AMK) or clarithromycin (CLR) is included.13

Figure 3.

Best fitted sigmoid curves obtained from the Emax model of (a) M. avium IWGMT49 exposed to clarithromycin and amikacin, between 24 and 120 h, and (b) M. xenopi ATCC 19250 exposed to clarithromycin and moxifloxacin, between 3 and 42 days. A different y-axis scale is used for (a) and (b). Serum Cmax for MAC-infected patients when treated with amikacin (AMK) or clarithromycin (CLR) is included.13

Discussion

This study presents basic pharmacodynamic information for antibiotics used to treat pulmonary disease caused by M. avium and M. xenopi. Amikacin showed a higher killing effect on M. avium than clarithromycin, although the killing was relatively low. For M. xenopi, the killing obtained with clarithromycin or moxifloxacin was similar and very slow for both agents.

The amikacin killing rate was higher than that of clarithromycin for M. avium. Its effect was faster, especially at concentrations over 2× MIC; therefore, its role as a support antibiotic in combined regimens holds promise. Strong and rapid in vitro killing by amikacin has been noted for Mycobacterium tuberculosis;9 however, no early bactericidal activity was detected in a group of patients with untreated, smear-positive, pulmonary TB in South Africa.10 Superior activity of amikacin over clarithromycin was also observed in vitro against M. avium subspecies avium11 and M. abscessus.12

Only the higher concentrations of amikacin, 2× MIC and above (≥16 mg/L), were effective against M. avium in this study, some of which can be close to the achievable levels in serum.13 Whether clinically achievable levels are good enough to attain the maximum kill is a question that should be answered in the future. The use of inhaled amikacin, currently in trials (ClinicalTrials.gov Identifier NCT01315236), may overcome this pharmacodynamic barrier since higher concentrations can be obtained in lungs with potentially less nephrotoxicity and ototoxicity.14,15

Regrowth of M. avium was observed soon after the initial drop in cfu obtained with amikacin. A high mutation frequency was determined in this study after in vitro exposure of M. avium subspecies hominissuis to amikacin concentrations of 1×, 8×, 16× and 32× MIC (8, 64, 128 and 256 mg/L). The emergence of resistant mutants has been previously observed in M. tuberculosis, when exposure to an amikacin concentration window between 8 and 256 mg/L resulted in the selection of resistant bacteria, which harboured the A1401G mutation in the rrs gene.9 Previous work done with M. avium subspecies avium did not show the early and high mutation rate that we observed in this study.11

Macrolides are considered the cornerstone for the treatment of MAC disease, particularly because of the established in vitro–in vivo correlation of their activity.4,16 In this study, clarithromycin showed a slow killing for M. avium subspecies hominissuis, basically at all concentrations, which was maximum after 240 h; however, the regrowth observed after 168 h underlines that macrolide monotherapy should not be used. In addition, if there is unknown resistance to all of the companion antibiotics, clarithromycin monotherapy will favour the emergence of resistance.

When clarithromycin was tested against M. avium subspecies avium, it resulted in both concentration-dependent and time-dependent effects.11 However, concentration was not the most important factor for clarithromycin effect in our study, which may be evidence of the intrinsic differences in the interaction of the same antibiotic with different M. avium subspecies.

In the set of experiments with M. xenopi, killing took a lot of time; it was maximal after 42 days, but we found no significant difference among the effect obtained with clarithromycin or moxifloxacin. Our observation is concordant with the characterization of the activity of anti-mycobacterial drugs against M. xenopi in the mouse model performed by Andréjak et al.,5 who found no differences between clarithromycin- and moxifloxacin-containing regimens. Rifampicin/ethambutol synergy in vitro is absent in M. xenopi isolates,17 which points out the role of clarithromycin during therapy. That, along with the underlying diseases in M. xenopi patients, may explain the poor results with treatment.18 The absence of correlation between the in vitro activity of clarithromycin and moxifloxacin and the in vivo outcomes of treatment with regimens based on these antimicrobials, as a result of these underlying diseases, could be hampering our understanding of M. xenopi and how to deal with it.

Hill's slopes were not significantly different for amikacin and clarithromycin against M. avium, but the values of Emax were different. This is a confirmation of a previous observation made in a set of experiments that we conducted with rapidly growing mycobacteria,12 where the total effect observed was mainly determined by the Emax and not by the Hill's slope. Hill's slopes were also not significantly different for clarithromycin and moxifloxacin against M. xenopi, although neither were their Emax values. The Hill's slope helps to infer the concentration dependency of an antibiotic. In this study both antibiotics showed similar steeper slope values for both Mycobacterium species; therefore the difference in concentration between the maximum and minimal effect was small. The overall effect is thus more of a time-dependent nature than concentration dependent.

We acknowledge several limitations of this study. Although 42°C is the optimal growing temperature for M. xenopi, we performed our experiments at 37°C. We preferred the physiological temperature of the human host; however, we do not know how this could have affected the results. There is a discrepancy between the in vitro MIC and the concentration that leads to 2 log kill in the time–kill curves (99% kill) at the time for MIC readout. These discrepancies result in part from technical aspects, including the fact that MICs were determined in CAMHB, while time–kill assays were conducted in Middlebrook 7H9, which has a lower pH that affects clarithromycin activity.7 Bottles used in the time–kill experiments are shaken constantly and they have a big head space and filtered vents, which allows a stable oxygen supply.

Our experiments were designed to evaluate the role of each individual antibiotic in the therapy; however, it is necessary to evaluate combined regimens as they are used in clinical practice. Dynamic models for pharmacokinetic and pharmacodynamic determinations can provide more representative data for the in vivo situation; the next step is to continue with the evaluation of current regimens with those models to help in the design of new and hopefully more effective therapies.

In conclusion, the killing effect of amikacin and clarithromycin on M. avium subspecies hominissuis is low, although amikacin activity was higher than that of clarithromycin, supporting its role in a combined therapy. Time–kill assays with M. xenopi showed no differences between clarithromycin and moxifloxacin, which did have high killing rates yet required a lot of time. Clarithromycin and moxifloxacin may have identical activity within treatment regimens for M. xenopi disease. Future studies of in vitro and in vivo pharmacokinetic/pharmacodynamic interactions are needed to be able to improve the current regimens against these two important SGM in pulmonary disease.

Funding

This work was performed as part of the routine work of our organization. B. E. F. was supported by a Doctoral Fellowship from the Instituto Colombiano para el Desarrollo de la Ciencia y la Tecnología Francisco José de Caldas, COLCIENCIAS, Colombian government (no. 529-2012).

Transparency declarations

None to declare.

References

1
Griffith
DE
,
Aksamit
T
,
Brown-Elliott
BA
et al
.
An official ATS/IDSA statement: diagnosis, treatment, and prevention of nontuberculous mycobacterial diseases
.
Am J Respir Crit Care Med
 
2007
;
175
:
367
416
.
2
Hoefsloot
W
,
van Ingen
J
,
Andréjak
C
et al
.
The geographic diversity of non-tuberculous mycobacteria isolated from pulmonary samples: an NTM-NET collaborative study
.
Eur Resp J
 
2013
;
42
:
1604
13
.
3
Andréjak
C
,
Lescure
FX
,
Pukenyte
E
et al
.
Mycobacterium xenopi pulmonary infections: a multicentric retrospective study of 136 cases in north-east France
.
Thorax
 
2009
;
64
:
291
6
.
4
van Ingen
J
,
Ferro
BE
,
Hoefsloot
W
et al
.
Drug treatment of pulmonary antituberculous mycobacterial disease in HIV-negative patients: the evidence
.
Expert Rev Anti Infect Ther
 
2013
;
11
:
1065
77
.
5
Andréjak
C
,
Almeida
DV
,
Tyagi
S
et al
.
Improving existing tools for Mycobacterium xenopi treatment assessment of drug combinations and characterization of mouse models of infection and chemotherapy
.
J Antimicrob Chemother
 
2013
;
68
:
659
65
.
6
Clinical and Laboratory Standards Institute
.
Performance Standards for Antimicrobial Susceptibility Testing: Twenty-first Informational Supplement M100-S21
 .
CLSI
,
Wayne, PA, USA
,
2011
.
7
Clinical and Laboratory Standards Institute
.
Susceptibility Testing of Mycobacteria, Nocardia, and Other Aerobic Actynomycetes—Second Edition: Approved Standard M24-A2
 .
CLSI
,
Wayne, PA, USA
,
2011
.
8
Mouton
JW
,
Dudley
MN
,
Cars
O
et al
.
Standardization of pharmacokinetic/pharmacodynamic (PK/PD) terminology for anti-infective drugs
.
Int J Antimicrob Agents
 
2002
;
19
:
355
8
.
9
de Steenwinkel
JEM
,
de Knegt
GJ
,
ten Kate
MT
et al
.
Time–kill kinetics of anti-tuberculosis drugs, and emergence of resistance, in relation to metabolic activity of Mycobacterium tuberculosis
.
J Antimicrob Chemother
 
2010
;
65
:
2582
9
.
10
Donald
PR
,
Sirgel
FA
,
Venter
A
et al
.
The early bactericidal activity of amikacin in pulmonary tuberculosis
.
Int J Tuberc Lung Dis
 
2001
;
5
:
533
8
.
11
Bakker-Woudenberg
IAJM
,
van Vianen
W
,
van Soolingen
D
et al
.
Antimycobacterial agents differ with respect to their bacteriostatic versus bactericidal activities in relation to time of exposure, mycobacterial growth phase, and their use in combination
.
Antimicrob Agents Chemother
 
2005
;
49
:
2387
98
.
12
Ferro
BE
,
van Ingen
J
,
Wattenberg
M
et al
.
Time–kill kinetics of antibiotics active against rapidly growing mycobacteria
.
J Antimicrob Chemother
 
2015
;
70
:
811
7
.
13
van Ingen
J
,
Egelund
EF
,
Levin
A
et al
.
The pharmacokinetics and pharmacodynamics of pulmonary Mycobacterium avium complex disease treatment
.
Am J Respir Crit Care Med
 
2012
;
186
:
559
65
.
14
Davis
KK
,
Kao
PN
,
Jacobs
SS
et al
.
Aerosolized amikacin for treatment of pulmonary Mycobacterium avium infections: an observational case series
.
BMC Pulm Med
 
2007
;
23
:
2
.
15
Olivier
KN
,
Shaw
PA
,
Glaser
TS
et al
.
Inhaled amikacin for treatment of refractory pulmonary nontuberculous mycobacterial disease
.
Ann Am Thorac Soc
 
2014
;
11
:
30
5
.
16
van Ingen
J
,
Boeree
MJ
,
van Soolingen
D
et al
.
Resistance mechanisms and drug susceptibility testing of nontuberculosis mycobacteria
.
Drug Resist Updat
 
2012
;
15
:
149
61
.
17
van Ingen
J
,
Hoefsloot
W
,
Mouton
JW
et al
.
Synergistic activity of rifampicin and ethambutol against slow-growing nontuberculous mycobacteria is currently of questionable clinical significance
.
Int J Antimicrob Agents
 
2013
;
42
:
80
2
.
18
van Ingen
J
,
Boeree
MJ
,
de Lange
WC
et al
.
Mycobacterium xenopi clinical relevance and determinants, the Netherlands
.
Emerg Infec Dis
 
2008
;
14
:
385
9
.