|Home | About | Journals | Submit | Contact Us | Français|
Organisms of the Mycobacterium avium-intracellulare complex (MAC) have been demonstrated to be susceptible to moxifloxacin. However, clinical data on how to utilize moxifloxacin to treat disseminated MAC are scanty. In addition, there have been no moxifloxacin pharmacokinetic-pharmacodynamic (PK/PD) studies performed for MAC infection. We utilized an in vitro PK/PD model of intracellular MAC to study moxifloxacin PK/PD for disseminated disease. Moxifloxacin doses, based on a serum half-life of 12 h, were administered, and the 0- to 24-h area under the concentration-time curve (AUC0-24) to MIC ratios associated with 1.0 log10 CFU/ml per week kill and 90% of maximal kill (EC90) were identified. The AUC0-24/MIC ratio associated with 1.0 log10 CFU/ml kill was 17.12, and that with EC90 was 391.56 (r2 = 0.97). Next, the moxifloxacin MIC distribution in 102 clinical isolates of MAC was identified. The median MIC was 1 to 2 mg/liter. Monte Carlo simulations of 10,000 patients with disseminated MAC were performed to determine the probability that daily moxifloxacin doses of 400 and 800 mg/day would achieve or exceed 1.0 log10 CFU/ml per week kill or EC90. Doses of 400 and 800 mg/day achieved the AUC0-24/MIC ratio of 17.12 in 64% and 92% of patients, respectively. The critical concentration of moxifloxacin against MAC was identified as 0.25 mg/liter in Middlebrook media. The proposed susceptibility breakpoint means that a larger proportion of clinical isolates is resistant to moxifloxacin prior to therapy. For patients infected with susceptible isolates, however, 800 mg a day should be examined for safety and efficacy for disseminated M. avium disease.
Mycobacterium avium complex (MAC) organisms are an important cause of morbidity and mortality in patients immunocompromised by the human immunodeficiency virus (HIV) and posttransplant antirejection medications, among others (10, 12, 16, 25, 26, 31). In patients with AIDS, disseminated disease is particularly common when the CD4+ cell count declines below 50/μl and is associated with a 3-fold increased risk of death (8, 37). Current guidelines recommend antimicrobial therapy with at least two agents, one a macrolide and the other ethambutol with or without rifabutin (29). The treatment is lifelong unless immune restoration is achieved by antiretroviral therapy. However, even with the combination therapy of ethambutol and macrolide, a complete microbiologic response is achieved in only 21 to 69% of patients (4, 9, 19). The long duration of therapy and the response rates suggest that more-optimal therapies still need to be developed.
Moxifloxacin is an 8-methoxyquinolone compound shown to be active against M. avium in the beige mouse model of disseminated disease (5). Currently, moxifloxacin is one of the agents recommended for macrolide-resistant disease (20, 21). However, the efficacy of this drug is still unclear. In the past, when other quinolones, such as ciprofloxacin, had been examined in clinical trials for the treatment of disseminated MAC, their contribution had been minimal (28). We hypothesize that one of the reasons is that such studies tended to employ standard doses developed for nonmycobacterial infections, which may not have much relevance to intracellular MAC. In addition, it is unclear if the MAC isolates from patients in such studies were really susceptible to ciprofloxacin, given that fluoroquinolone susceptibility breakpoints for MAC have not been defined. Indeed, the true distribution of moxifloxacin MICs for clinical isolates is unknown.
A possible approach to optimizing the effectiveness of moxifloxacin is to examine the drug's antimicrobial pharmacokinetic-pharmacodynamic (PK/PD) properties. To our knowledge no moxifloxacin PK/PD studies have ever been performed for the treatment of MAC infection. We have developed an in vitro PK/PD model of intracellular M. avium infection (15) in which a bacillary burden similar to that encountered in patients with disseminated disease was exposed to the same moxifloxacin concentration-time profiles encountered in the serum of patients. This disease model was utilized to identify the relationship between moxifloxacin exposure and microbial kill. Exposures associated with optimal kill were then examined in clinical trial simulations, and optimal moxifloxacin doses were determined.
M. avium subspecies hominissuis (ATCC 700898), first isolated from an AIDS patient with disseminated MAC, was used for experiments throughout the present study. This subspecies causes the majority of disseminated MAC in patients (35, 49). Stock cultures were stored at −80°C in Middlebrook 7H9 broth and 10% glycerol. For each study, the bacterial stock was thawed and incubated in Middlebrook 7H9 broth with 10% oleic acid-albumin-dextrose-catalase (OADC) at 37°C in a water bath, under shaking conditions, for 4 days to achieve exponential-phase growth.
Hollow fiber cartridges were purchased from FiberCell. RPMI 1640 and bovine serum albumin (BSA) were purchased from Sigma, while fetal bovine serum (FBS) was purchased from SAFC Biosciences. FBS was heat inactivated and filtered prior to use. Moxifloxacin hydrochloride solution of 400 mg/250 ml of 0.8% saline was purchased from UT Southwestern Medical Center Aston Pharmacy. The drug was serially diluted using RPMI 1640 to the drug concentrations required for study.
M. avium cultures on day 4 of log-phase growth were adjusted to a McFarland standard of 0.5 and then diluted to a bacterial density of 1.5 ×103 CFU/ml, after which serial dilutions were plated onto Middlebrook 7H10 agar supplemented with moxifloxacin at concentrations of 0, 0.0625, 0.125, 0.25, 0.5, 1.0, and 2 mg/liter in duplicate. The cultures were incubated at 37°C under 5% CO2 for 14 days, after which colonies were counted, and the minimum concentration associated with 99% inhibition was identified. The experiment was performed twice on two different occasions.
Between 1 January 2006 and 31 October 2009, 671 MAC clinical isolates from the Cleveland Clinic Foundation were tested for clarithromycin susceptibility in the Microbiology Laboratory at the Cleveland Clinic. At least 97% of the isolates had a clarithromycin MIC of ≤4.0 mg/liter. Of these isolates, 102 also had moxifloxacin MICs performed. All susceptibility testing was performed in an MIC tray prepared by Trek Diagnostics, Cleveland, OH. The susceptibility testing was performed according to the Clinical and Laboratory Standards Institute methodology (11).
Human-derived THP-1 macrophages (ATCC TIB-202) were cultured in prewarmed RPMI 1640 medium and 10% FBS. A culture of 1.5 ×105 CFU/ml was used to infect 1.5 ×106 macrophages per ml by coincubating overnight at 37°C under 5% CO2, giving a bacillus-to-macrophage multiplicity of infection of 1:10. The infected macrophages were washed twice with warm RPMI 1640 and 10% FBS by centrifugation at 100 × g for 5 min and then examined in a hemocytometer for cell counts and viability after staining with trypan blue. This resulted in an intracellular infection with a bacillary burden of 3 to 4 log10 CFU/ml.
In order to study the activity of moxifloxacin against extracellular MAC, bacteria on day 4 of log-phase growth were adjusted to a bacterial density of 1.5 ×107 CFU/ml. Bacterial suspension in Middlebrook 7H9 broth with 10% OADC was treated with moxifloxacin at concentrations of 0, 0.0625, 0.125, 0.25, 0.5, 1.0, 2.0, 4.0, 8.0, 16.0, and 32.0 mg/liter in test tubes, in triplicate, and incubated at 37°C under 5% CO2. On day 7, the cultures were washed, serially diluted, and plated onto Middlebrook 7H10 agar. The cultures were then incubated at 37°C under 5% CO2 for 14 days, and the colonies were counted. The experiment was performed twice.
Infected macrophages were prepared as described above and then coincubated in 24-well plates in triplicate with the same moxifloxacin concentrations used for extracellular bacilli. On day 7 of incubation, 1 ml of cell suspension was centrifuged at 100 × g for 5 min, supernatant was discarded, and 1 ml of 0.5% Triton X-100 in phosphate-buffered saline was added. This concentration of Triton does not kill MAC. The suspension was then vortexed to completely lyse the macrophages and then serially diluted. The cultures were plated onto Middlebrook 7H10 agar with 10% OADC and incubated for 14 days as described above.
We wanted to assess the impact of protein binding on moxifloxacin concentrations achieved inside macrophages and on microbial kill of MAC. Two experiments were carried out. In the first, 4 mg/liter of moxifloxacin, the peak concentration achieved in serum of patients treated with 400 mg, was incubated with 1 × 105 THP-1 macrophages/ml in RPMI 1640 with and without 5.4 g/liter of BSA at 37°C, and samples were collected at 5, 15, 30, and 60 min. BSA binds moxifloxacin to the same extent as human serum albumin and to the same extent as whole human plasma (27, 36), while the concentration of 5.4 g/liter is the upper limit of normal in human serum. After sampling, macrophages were washed and ruptured as described above, and the moxifloxacin concentration was measured using fluorescence at excitation and emission wavelengths of 296 and 505 nm (FluoroMax-3; Jobin Yvon). The experiment was performed twice. In a second experiment, the relationship between protein binding in the extracellular milieu and intracellular microbial kill inside was examined. We were interested in determining if moxifloxacin potency (50% effective concentration [EC50]) against intracellular MAC decreased with increasing content of FBS in RPMI 1640. If it were the free drug concentration that was best associated with intracellular kill, then the moxifloxacin potency would decrease with increasing FBS concentrations. If it were the total moxifloxacin concentration, then potency would not change. Macrophages were infected with MAC and then treated with moxifloxacin concentrations of 0, 0.5, 1.0, 2.0, 4.0, 6.0, 8.0, 12.0, and 16.0 mg/liter in RPMI 1640 with 0%, 2%, 10% and 20% FBS for 3 days and then lysed for bacterial burden determination as described above.
Hollow fiber systems have been used for antimicrobial PK/PD studies of many bacteria, including Mycobacterium tuberculosis (6, 23, 32, 43, 46). The full PK model has been described in detail for extracellular bacteria in the past (22, 24). We adapted the system by inoculating 20 ml of infected THP-1 macrophages into the external compartment to mimic the 3 to 4 log10 CFU/ml bacilli encountered in blood cultures of severely immunocompromised patients with MAC (9, 18, 19). The external compartment is separated by semipermeable hollow fibers from a central compartment in which RPMI 1640 with 10% FBS circulates. The medium crosses from the central compartment to the external compartment, but neither the macrophages nor MAC can cross into the central compartment because the hollow fibers have a molecular cutoff of 40 kDa.
Moxifloxacin was dosed into the central compartment. Since the molecular mass of moxifloxacin is 0.437 kDa, it freely crosses back and forth between the central and external compartments so that infected macrophages are exposed to the same drug concentrations as those in the central compartment. Each hollow fiber system was treated with once-daily moxifloxacin under the control of a computerized syringe pump. Human equivalent moxifloxacin doses administered were 0, 50, 100, 200, 400, 600, 800, and 1,600 mg/day. Computer-controlled peristaltic pumps pumped fresh RPMI 1640 and 10% FBS in and pumped used media out of the systems to achieve a drug dilution rate that resulted in the moxifloxacin half-life of 12 h as encountered in patients (45).
Drug concentrations achieved in each system were examined by sampling the central compartment at 1, 3, 5, 9, 12, 18, 23, 25, 27, 29, 33, 36, 42, and 47 h after the seventh infusion. The external compartment was sampled on days 2 and 7, and macrophage viability was assessed. The macrophages were then processed for bacterial density determination as described above. The moxifloxacin-resistant subpopulation was determined by cultures on Middlebrook agar that had been supplemented with moxifloxacin at three times the MIC.
Samples collected from the central compartment were analyzed for moxifloxacin concentration by using a liquid chromatographic technique with UV detection (292 nm). Levofloxacin was used as the internal standard (IS). Each sample (200 μl) was mixed with the IS (200 μl) and acetonitrile (500 μl), vortexed for 10 min, and centrifuged at 4°C at 1,500 rpm. The supernatant was dried at 37°C using a speed vacuum centrifuge. Each sample was reconstituted with 200 μl methanol for analysis. A sample volume (20 μl) was injected into a Shimadzu liquid chromatography (LC)/UV system, separated using a 10 mM sodium dodecyl sulfate, 10 mM tetrabutylammonium acetate and 25 mM citric acid, and acetonitrile (57:43 [vol/vol]) mobile phase at 1 ml/h and Grace Adsorbosphere HS C18 7μ (150- by 4.6-mm) column. The assay was linear between 0.2 and 100 mg/liter (r2 = 0.999), and the relative standard deviation was within 5%.
The drug concentrations from each hollow fiber system were modeled by using ADAPT II software and applying a one-compartment model with first-order input and elimination (14). Since microbial kill by moxifloxacin and other fluoroquinolones is closely dependent on the ratio of the 0- to 24-h area under the concentration-time curve (AUC0-24) to MIC (2, 40), the relationship between the AUC0-24/MIC ratio and bacterial burden was analyzed in an inhibitory sigmoid Emax model using GraphPad Prism 5 software. Based on this, the AUC0-24/MIC ratios associated with 80 and 90% of maximal effect (EC80 and EC90) were identified. In addition, the AUC0-24/MIC ratio associated with 1.0 log10 CFU/ml per week was also identified.
Monte Carlo simulations were performed to examine how likely moxifloxacin doses of 400, 800, and 1,600 mg a day would achieve the AUC0-24/MIC ratio associated with EC80, or EC90, or 1.0 log10 CFU/ml reduction per week, in 10,000 virtual patients with disseminated M. avium. The PK parameter estimates and dispersion measures by Simon et al. (41) were utilized as prior data in subroutine PRIOR of ADAPT II. Both normal and log-normal distributions were examined, and the final distribution was chosen based on the ability to closely reproduce the PK parameters published by Simon et al. The moxifloxacin MIC distribution of MAC clinical isolates was based on data from the Cleveland Clinic Foundation. The minimum dose that achieved or exceeded the EC80 in ≥90% of 10,000 patients was defined as the optimal dose. The breakpoint MIC was defined as the MIC for which the standard dose of 400 mg a day, or the high dose of 800 mg a day, achieved the EC90 in ≥90% of 10,000 patients (1, 13, 50).
The moxifloxacin MIC for our laboratory MAC strain was 1 mg/liter. For the 102 clinical MAC isolates, the median MIC was 1 to 2 mg/liter, with an MIC distribution as shown in Fig. Fig.1.1. Thus, the MIC of our laboratory isolate was similar to the median for clinical isolates.
In 7-day exposure effect studies against extracellular MAC, the maximum kill (Emax) could not be achieved because there was continued decline in bacterial burden with each successive concentration, so that the flat portion on the Emax portion of the curve was not reached. Therefore, the Emax was ≥7 log10 CFU/ml, and the EC50 was ≥32 mg/liter (r2 = 0.90). However, against intracellular bacilli, the Emax was only 3.17 log10 CFU/ml, and the EC50 was 8.24 mg/liter for the same duration of exposure (r2 = 0.99).
The effect of physiological concentrations of albumin on intramacrophage concentrations of moxifloxacin is shown in Fig. Fig.2,2, which demonstrates similar intracellular moxifloxacin concentrations with and without albumin. In terms of microbial kill, the moxifloxacin EC50 against intracellular MAC was 7.54 ± 2.44, 8.09 ± 2.25, 5.07 ± 0.42, and 5.21 ± 1.48 mg/liter in media with 0%, 2%, 10%, and 20% FBS, respectively. Since both results suggest a lack of correlation between extracellular protein binding and the intracellular kill of MAC by moxifloxacin, we subsequently utilized total serum concentrations of moxifloxacin for further PK/PD analysis.
We mimicked the moxifloxacin serum decline encountered in patients, resulting in the observed concentrations in the hollow fiber systems that are shown in Fig. Fig.3.3. Based on these concentrations, a one-compartmental model PK analysis revealed a ka of 3.76 ± 0.49 h−1, a systemic clearance of 14.55 ± 1.92 liters/h, a volume of 251.9 ± 32.14 liters, and a half-life of 12.05 ± 1.19 h. The parameters for each regimen were utilized to calculate the AUC0-24 and AUC0-24/MIC ratio achieved in each system. The relationship between the AUC0-24/MIC ratio and the day 7 MAC burden is shown in Fig. Fig.44 and was described by the following equation: E(log10 CFU/ml) = 4.04 − 3.03 × (AUC0-24/MIC)0.93/[36.660.93 + (AUC0-24/MIC)0.93], where 4.04 log10 CFU/ml is the bacterial burden with no therapy and the AUC0-24/MIC ratio of 36.66 is the EC50 (r2 = 0.97). The EC80 was an AUC0-24/MIC ratio of 163.37, and the EC90 was 391.56. The AUC0-24/MIC exposure associated with 1.0 log10 CFU/ml per week was 17.12. There was no emergence of resistance during the 7 days of study.
In 10,000-patient Monte Carlo simulations, a log-normal distribution was associated with a serum clearance of 14.30 ± 0.78 liters/h and a volume of 62.91 ± 5.42 liters. This distribution adequately recapitulated the original PK parameter estimates in patients. The probabilities of different doses achieving the EC80 are shown in Fig. Fig.5A.5A. The target attainment probabilities for 400, 800, and 1,600 mg per day were only 2%, 13%, and 36%, respectively. This means that optimal kill is not expected to be achieved in an acceptable proportion of patients by any of these doses. On the other hand, a microbial kill of 1 log10 CFU/ml/week would still represent excellent kill compared to current regimens, even though it is not the maximal kill by the drug. The probabilities of the three doses achieving exposures associated with such an effect are shown in Fig. Fig.5B.5B. The target attainment probability for 400, 800, and 1,600 mg is 64%, 92% and 98%, respectively. Thus, 800 mg a day would be the lowest dose to achieve 1.0 log10 CFU/ml per week in an acceptable proportion of patients. The probabilities that 400 and 800 mg a day would achieve the EC90 are shown in Fig. Fig.5C.5C. The highest MIC that would allow ≥90% of patients treated with 800 mg a day to achieve EC90 was 0.125 mg/liter. The minimum concentration at which these doses would fail to achieve EC90 in >10% of patients will be 1 tube dilution higher, which is a critical concentration of 0.25 mg/liter.
Environmental M. avium subsp. hominissuis infects humans via the gut and lung mucosal surfaces, multiplies locally, eventually enters the bloodstream, and disseminates to many visceral organs (48, 49). The highest bacterial burden is in reticuloendothelial organs, with a range of 4 to 7 log10 CFU/g in mesenteric lymph nodes and 2.8 to 6.8 log10 CFU/g in the spleen (47). Blood bacillary burden correlates with burden in these tissues and is between 1.5 and 3.9 log10 CFU/ml. With the standard combination of clarithromycin and ethambutol, the blood bacillary load decreased at about 0.5 log10 CFU/ml per week during the first month and between 3.0 and 3.5 log10 CFU/ml during the first 16 weeks (4). However, this combination results in a complete response in only half of patients. New treatment regimens that have kill rates of higher than 0.5 log10 CFU/ml per week would be expected to result in higher success rates in patients. To develop such regimens, however, a preclinical model in which bacillary burden and kill rates mirror those in blood of patients with disseminated MAC is needed. Our PK/PD model was designed to fulfill these criteria. We have demonstrated elsewhere (15) that, in this model, weekly microbial kill rates for 15 mg/kg/day ethambutol monotherapy mirror those in patients, so that the PK/PD relationships derived in this model likely have clinical meaning.
The relationship between protein binding and antibiotic PD effect is well known. However, the effect of extracellular protein binding on intramacrophage PD effect is less well known. This is a particularly fascinating problem given that moxifloxacin is 50% protein bound (44) but is also concentrated 600 to 740% inside infected macrophages (3, 38). Our data demonstrate that intracellular moxifloxacin concentrations were not altered by physiological concentrations of albumin. In addition, there was no difference in potency against MAC even as the concentration of FBS was increased. This suggests that the effect of serum protein binding is overwhelmed by intracellular accumulation processes that are driven by both passive diffusion and active uptake of the drug (38). On the other hand, moxifloxacin killed approximately 7 log10 CFU/ml of extracellular MAC in 1 week, based on assays in 24-well plates. However, the kill rates for intracellular MAC in 24-well plates were at least several hundred-fold lower by Emax than those for extracellular bacilli. Both experiments employed static drug concentrations. In order to determine optimal exposures that have better in vivo relevance, we employed a model with a PK system. In that model, the maximal kill was 3.03 log10 CFU/ml at 1 week, and even that could be achieved only at high moxifloxacin AUC0-24/MIC exposures. This effect of reduced moxifloxacin efficacy inside macrophages has also been demonstrated for Staphylococcus aureus (3, 39). This means that despite achievement of a high intracellular concentration, there is some aspect of intracellular macrophage physiology that retards moxifloxacin efficacy.
The ability of several doses to achieve the high exposures associated with optimal effect was examined in clinical trial simulations by use of the Monte Carlo method (7, 17, 34). No dose between 400 and 1,600 mg a day could achieve optimal kill in an acceptable proportion of patients. If 1.0 log10 CFU/ml per week is chosen as a therapeutic target, based on the premise that it would be at least two times better than the kill rates achieved by the current standard drugs, 400 mg would still be inadequate, but 800 mg would at least be able to achieve this in >90% of patients. Unfortunately, the safety of this dose is yet to be established.
An intriguing riddle in the treatment of disseminated MAC has been the finding that except for macrolides, susceptibilities to other drugs failed to predict clinical outcome (21, 42). However, no study has specifically examined the relationship between moxifloxacin MIC and microbial outcome in patients. Based on moxifloxacin PK/PD properties and PK variability, we propose a critical concentration of 0.25 mg/liter. Unfortunately, the breakpoint we propose would designate a large proportion of wild-type clinical MAC isolates as resistant to the moxifloxacin. Patients infected by these isolates will likely have a low likelihood of responding to therapy, while those deemed susceptible should be treated with 800 mg of moxifloxacin a day. Our proposed breakpoints will need to be validated in the clinical setting. On the other hand, as discussed above, even the combination of a macrolide and ethambutol is not able to kill >1.0 log10 CFU/ml/week, so that it could be conceded that use of a moxifloxacin EC90 exposure, associated with kill of 2.7 log10 CFU/ml/week, for susceptibility breakpoint determination is overly stringent in this case, limiting clinical applicability. In that case, if 1.0 log10 CFU/ml/week is accepted to represent fairly effective kill and is used for breakpoint determination, examination of Fig. Fig.5B5B leads to a critical concentration of 2.0 mg/liter. Even then, based on Fig. Fig.1,1, a large proportion of clinical isolates would still be considered resistant. Nevertheless, the EC90 is the most commonly used standard for susceptibility breakpoints in other bacteria and is the one recommended (50). The EC90 has the advantage of being indexed to maximal kill, which differs from drug to drug, and can be used for susceptibility breakpoint determination even for drugs with less than 1.0 log10 CFU/ml per week kill.
There are some limitations to our study. First, an examination of more isolates in the PK/PD studies could change the AUC0-24/MIC ratio associated with effect. It has been demonstrated that AUC0-24/MIC ratios associated with optimal effect are in fact a distribution (33). When such a distribution was taken into consideration in one S. aureus study, the overall clinical breakpoints decreased (33). This means that even higher moxifloxacin doses and even lower critical concentrations than those we determined would likely be determined if a larger number of isolates were examined. This would still support our central findings that the standard dose of 400 mg a day is too low for the treatment of disseminated MAC and that the susceptibility breakpoints of 1.0 mg/liter used for other bacteria are likely too high. A second limitation is that with examination of a larger number of clinical isolates than the 102 we examined, the MIC distribution would change. However, to our knowledge, there are no studies that have examined larger numbers. In a recent study by Kohno et al., the MIC90 in 72 clinical isolates from Japan was 2 mg/liter (30), which does not differ substantially from that of our own isolates. A third limitation is that there was no emergence of moxifloxacin resistance. A likely reason is that the subpopulation resistant to moxifloxacin was below the limits of assay detection and could have become apparent with much longer durations of therapy. Thus, exposures associated with resistance suppression remain unknown. Finally, our findings are restricted to disseminated MAC; studies of pulmonary MAC will need to be performed independently.
In summary, we identified moxifloxacin exposures associated with optimal kill of intracellular MAC. Based on these, we conclude that the standard dose of 400 mg a day is likely to be suboptimal. In addition, we propose a new susceptibility breakpoint for identifying isolates that would not respond to moxifloxacin therapy in disseminated disease.
Published ahead of print on 12 April 2010.