ABSTRACT
Invasive pulmonary aspergillosis (IPA) is a devastating disease of immunocompromised patients. Pharmacodynamic (PD) examination of antifungal drug therapy in IPA is one strategy that may improve outcomes. The current study explored the PD target of posaconazole in an immunocompromised murine model of IPA against 10 A. fumigatus isolates, including 4 Cyp51 wild-type isolates and 6 isolates carrying Cyp51 mutations conferring azole resistance. The posaconazole MIC range was 0.25 to 8 mg/liter. Following infection, mice were given 0.156 to 160 mg/kg of body weight of oral posaconazole daily for 7 days. Efficacy was assessed by quantitative PCR (qPCR) of lung homogenate and survival. At the start of therapy, mice had 5.59 ± 0.19 log10 Aspergillus conidial equivalents (CE)/ml of lung homogenate, which increased to 7.11 ± 0.29 log10 CE/ml of lung homogenate in untreated animals. The infection was uniformly lethal prior to the study endpoint in control mice. A Hill-type dose response function was used to model the relationship between posaconazole free drug area under the concentration-time curve (AUC)/MIC and qPCR lung burden. The static dose range was 1.09 to 51.9 mg/kg/24 h. The free drug AUC/MIC PD target was 1.09 ± 0.63 for the group of strains. The 1-log kill free drug AUC/MIC was 2.07 ± 1.02. The PD target was not significantly different for the wild-type and mutant organism groups. Mortality mirrored qPCR results, with the greatest improvement in survival noted at the same dosing regimens that produced static or cidal activity. Consideration of human pharmacokinetic data and the current static dose PD target would predict a clinical MIC threshold of 0.25 to 0.5 mg/liter.
INTRODUCTION
The incidence of invasive pulmonary aspergillosis (IPA) is increasing in parallel with a growing population of immunocompromised patients. Recent surveillance data of transplant patients identified this infection as the second most common fungal pathogen in solid organ transplant recipients and the most common pathogen in bone marrow transplantation (1, 2). Despite the development of new antifungal drugs with enhanced potency against these organisms, morbidity and mortality associated with IPA remain unacceptably high. Numerous factors have been shown to impact treatment efficacy. One clinical factor under the control of the clinician is the antifungal dosing regimen. Pharmacokinetic/pharmacodynamic (PK/PD) investigations have been critical for defining the optimal antimicrobial exposure relative to a measure of in vitro potency, the MIC (3–6).
Posaconazole is the most recently approved advanced-generation triazole with potent anti-Aspergillus activity (7–12). Clinical efficacy has been demonstrated in the prevention and treatment of IPA (13–15). Furthermore, analysis of posaconazole concentration monitoring in these scenarios has suggested a strong clinical concentration-efficacy relationship (15–18). However, definitive determination of the optimal dose and the impact of variation in in vitro potency (MIC) remain unclear. The recent emergence of Aspergillus fumigatus isolates exhibiting reduced triazole susceptibility underscores the potential impact of these explorations (19–22). The goals of the current study were to utilize animal model pharmacodynamic approaches to define the optimal posaconazole exposure, discern the impact of MIC variation associated with emerging resistant A. fumigatus strains, and provide a basis for design of dosing strategies to successfully treat infections due to these resistant mutants.
MATERIALS AND METHODS
Organisms.Ten Aspergillus fumigatus isolates were chosen, including 9 clinical isolates with and without Cyp51 mutations and one laboratory isolate with an Fks1 mutation. Organisms were grown and subcultured on potato dextrose agar (PDA) (Difco Laboratories, Detroit, MI). Only organisms with similar fitness, as determined by growth in lungs (see Table 1) and mortality (see Table 3) in untreated mice over 7 days, were chosen.
Drug.Posaconazole solution for in vivo studies was obtained from the University of Wisconsin Hospital and Clinics pharmacy. Drug solutions were prepared on the day of study using sterile saline as the diluent and vortexed vigorously prior to administration by oral-gastric gavage. Posaconazole powder for in vitro susceptibility was obtained from Merck.
In vitro susceptibility.MICs were determined by broth microdilution using the CLSI M38-A2 methods (23). MICs were performed in duplicate three times; the median value is reported in Table 1.
Animals.Six-week-old Swiss/ICR specific-pathogen-free female mice weighing 23 to 27 g were used for all studies (Harlan Sprague-Dawley, Indianapolis, IN). Animals were housed in groups of five and allowed access to food and water ad libitum. Animals were maintained in accordance with the American Association for Accreditation of Laboratory Care criteria. The Animal Research Committee of the William S. Middleton Memorial VA Hospital and University of Wisconsin—Madison approved the animal studies.
Infection model.Mice were rendered neutropenic (polymorphonuclear cells < 100/mm3) by injection of 150 mg/kg of body weight cyclophosphamide subcutaneously (s.c.) on days −4, −1, and +3. Prior studies have shown this to sustain neutropenia for the 7-day experiment (24–26). Additionally, cortisone acetate (250 mg/kg given s.c.) was administered on day −1 as previously described (25, 27, 28). Throughout the 7-day neutropenic period, mice were also given ceftazidime, 50 mg/kg/day s.c., to prevent opportunistic bacterial infection. Uninfected animals given the above immune suppression and antibiotic had 100% survival to study endpoint (data not shown).
Organisms were subcultured on PDA 5 days prior to infection and incubated at 37°C. On the day of infection, the inoculum was prepared by flooding the culture plate with 5 ml of normal saline with 0.05% Tween 20. Gentle agitation was applied to release the conidia into the fluid. The conidial suspension was collected and quantitated by using a hemacytometer (Bright-Line; Hausser Scientific, Horsham PA). The suspension was diluted to a final concentration of 1 × 107 to 2 × 107 conidia/ml. Viability was confirmed by plating the suspension on PDA and determining CFU.
An aspiration pneumonia model was utilized. Mice were anesthetized with a combination of ketamine and xylazine. Fifty microliters of a 1 × 107 to 2 × 107 conidial suspension was pipetted into the anterior nares and aspirated into the lungs. The procedure produced invasive aspergillosis in more than 90% of animals and 100% mortality in untreated infected mice by day 3 or 4.
Lung processing and organism quantitation.Processing and quantitation of the lung burden were performed as previously described (29, 30). Briefly, at the time of sacrifice for moribund animals or at the end of therapy (7 days), lungs were aseptically removed and placed in a 2-ounce sterile polyethylene Whirl-Pak bag (Nasco, Fort Atkinson, WI) containing 2 ml of sterile saline. The lungs were manually homogenized using direct pressure (31). One milliliter of the primary homogenate was placed in a sterile bead beating tube (Sarstedt, Newton, NC) with 700 μl of 425 to 600 μm acid-washed glass beads (Sigma-Aldrich, St. Louis, MO). The primary homogenate was bead beaten in a Bio-spec mini bead beater (Bartlesville, OK) for 90 s at 4,200 rpm to yield a secondary homogenate. One hundred microliters of the secondary homogenate was mixed with 100 μl of buffer ATL (Qiagen, Valencia, CA) and 20 μl of proteinase K (Qiagen, Valencia, CA) and incubated overnight at 56°C with gentle agitation. DNA was then isolated following the DNeasy Blood and Tissue protocol (Qiagen, Valencia, CA). A final DNA elution step was carried out with a 100-μl volume. The DNA was stored at −20°C until the day of quantitative PCR (qPCR).
qPCR plates were prepared on the day of the assay. Standard amounts of conidia were prepared by hemacytometer counts and were used for generating standard curves. The results of qPCR are therefore reported as conidial equivalents (CE) per ml of primary lung homogenate. Samples were assayed in triplicate using a Bio-Rad CFX96 real-time system (Hercules, CA). A single-copy gene, Fks1, was chosen for quantitation (32). Primer sequences included the following: forward primer (5′-GCCTGGTAGTGAAGCTGAGCGT-3′), reverse primer (5′-CGGTGAATGTAGGCATGTTGTCC-3′), and probe (6-carboxyfluorescein [FAM]-AGCCAGCGGCCCGCAAATG-MGB-3′) (Integrated DNA Technologies, Coralville, IA). The fks1 mutation (EMFR S678P) was not located in the primer-probe area of the genome and did not affect the quantitation reaction for that isolate (data not shown). The primer-probe set was validated for all isolates by determining the kinetics and quantitative results for known quantities of conidia over the dynamic range (102 to 108) (data not shown). Additionally, conidium-spiked uninfected lung homogenate was used to test for the presence of PCR inhibitors that may adversely affect qPCR results (data not shown).
Pharmacokinetics.Murine posaconazole pharmacokinetic data, including the area under the concentration-time curve (AUC) and protein binding, were derived from our previous study of this animal model (33).
Pharmacodynamic index and magnitude.The AUC/MIC ratio was used as the PD index for exploration of exposure-response relationships based upon previous PK/PD investigations (33–35). Both total and free (not protein-bound) concentrations were considered. Neutropenic mice were infected as described above. Treatment consisted of 0.156 to 160 mg/kg/24 h of posaconazole administered once daily for 7 days by oral gavage (OG). The doses were selected to vary the effect from maximal to no efficacy. Controls were utilized for each isolate and included a zero-hour and untreated control groups at the end of the study period. Four mice were used for each group (zero-hour control, no-treatment control, and each dosing regimen).
Data analysis.The qPCR data were modeled according to a Hill-type dose response equation: log10 D = log10 (E/Emax − E)/N + log10 ED50, where D is the drug dose, E is the growth (as measured by qPCR and represented as CE/ml of lung homogenate) in untreated control mice, Emax is the maximal effect, N is the slope of the dose-response relationship, and ED50 is the dose needed to achieve 50% of the maximal effect. The AUC/MIC for total and free (non-protein-bound) drug was determined for each organism and associated drug exposure. The coefficient of determination (R2) was used to estimate the percent variance in the change of log10 CE/ml of lung homogenate over the treatment period for the different dosing regimens that could be attributed to the PD index, AUC/MIC. The dose necessary for net stasis (static dose) and 1-log kill and the associated PD targets total and free drug AUC/MIC associated with these endpoints were determined. The PD targets were compared between wild-type and Cyp51 mutants by using the t test for normally distributed data and by using the Mann-Whitney rank sum test for non-normally distributed data.
Survival.Survival to the end of the study period (7 days) was also recorded for each group. A laboratory technician not aware of the study design or expected results was responsible for determining the time of sacrifice of moribund animals in accordance with accepted laboratory standards for the humane treatment of research animals (American Association for Accreditation of Laboratory Care criteria). Survival in different groups was compared by t test for normally distributed data and by Mann-Whitney rank sum test for non-normally distributed data. Logistic regression was also performed using survival as the outcome using the software program Sigma Stat (Systat Software, Inc., San Jose, CA).
RESULTS
Organism susceptibility and in vivo fitness.Posaconazole susceptibility testing, genetic mutations where applicable, and the relative fitness in the in vivo murine model of each isolate are shown in Table 1. Cyp51 wild-type MICs ranged from 0.25 to 0.5 mg/liter and from 1 to 8 mg/liter in Cyp51 mutants. The organisms exhibited similar in vivo fitnesses. At the start of therapy, mice had 5.59 ± 0.19 log10 CE/ml of lung homogenate, and the infectious burden increased to 7.11 ± 0.29 log10 CE/ml of lung homogenate in untreated animals. Each isolate produced 100% mortality prior to the end of the study in untreated animals (see Table 3).
In vitro susceptibilities and in vivo fitnesses of select A. fumigatus isolates
Pharmacokinetics.Data from our prior PK study of posaconazole in this mouse model were used for the current study (33). The AUC over the dose range was linear. Thus, for dose levels that were not directly measured, the AUC was estimated using linear extrapolation or interpolation. The total drug AUC range was 1.78 to 800 mg · h/liter over the dose range of 0.156 to 160 mg/kg/24 h. Protein binding was 99%.
Dose-response curves.A sigmoid dose-response relationship was observed for each isolate, and higher doses were necessary to achieve similar outcomes against organisms with elevated posaconazole MICs (Fig. 1). A net static outcome was observed with all 10 isolates, a 1-log kill was achieved against 9 isolates (3 of 4 wild-type isolates and 6 of 6 mutants), and for 6 strains (3 of 4 wild-type strains and 3 of 6 mutants), a 2-log kill was observed. The dose-response curves were steep, with a 4-fold change in drug exposure producing a 2 to 3 log10 change in antimicrobial effect.
Dose-response curves for each isolate are shown, with solid symbols representing wild-type Cyp51 organisms and open symbols Cyp51 mutants. Mice were given 0.156 mg/kg to 160 mg/kg of posaconazole once daily for 7 days. Each data point is the mean ± SD in log10 CE/ml of lung homogenate for four mice. The horizontal dashed line represents the net stasis of burden from the start of therapy. Points above the line represent an increase in burden (i.e., net growth), whereas those below the line represent a decrease in burden.
PD index and target.The dose and AUC/MIC needed to produce growth suppression compared to the start of therapy (static dose) and the regimens associated with a 1-log reduction in organism burden (1-log kill) for each isolate are reported in Table 2. The static dose and 1-log kill dose (when achieved) in Cyp51 wild-type organisms ranged from 1.09 to 2.16 mg/kg/24 h and 2.28 to 4.22 mg/kg/24 h, respectively. Comparative values for the Cyp51 mutant group were much higher, at 14.5 to 51.9 mg/kg/24 h and 22.4 to 150 mg/kg/24 h, respectively. The differences for both static dose and 1-log kill dose (mg/kg) between wild-type and mutant groups were statistically significant (P = 0.01 and 0.04, respectively). The total and free drug AUC/MIC PD targets, however, were comparable among this diverse organism group. While the posaconazole exposure associated with these endpoints based upon dose in mg/kg varied nearly 50-fold, expression of the exposure as AUC/MIC for the same endpoints varied only 4-fold. This supports the relevance of the AUC PD index and even more so the MIC.
Dose and total and free drug AUC/MIC needed to achieve net stasis and 1-log kill endpoints (when achieved) for each A. fumigatus isolatea
The mean free drug AUC/MIC associated with net stasis was 0.67 for the wild-type group and 1.36 for the Cyp51 mutant group. This difference was not statistically significant (P = 0.09). Similarly, the 1-log-kill free drug AUC/MIC target was numerically higher for Cyp51 mutants (mean of 1.52 for the wild type versus 2.35 for the mutant group) but was not statistically significant (P = 0.28). The PD target free drug AUC/MIC for a 1-log kill was roughly 2-fold larger than the stasis endpoint (2.07 versus 1.09). The free drug AUC/MIC for all organisms was fit to the Hill sigmoid-dose-response model, and the relationship is shown in Fig. 2. The free drug AUC/MIC was a robust predictor of the observed outcome based upon a high coefficient of determination (R2 = 0.79).
The free drug AUC/MIC and microbiological effect are plotted for each of the 10 A. fumigatus isolates tested. Solid symbols denote wild-type Cyp51 organisms, and open symbols denote Cyp51 mutants. The horizontal dashed line represents the net stasis of infectious burden from the start of therapy. Points above the line represent an increase in burden (i.e., net growth), whereas those below the line represent a decrease in burden. The horizontal dotted lines represent 1- and 2-log kills, respectively. The coefficient of determination (R2) based on the Hill equation is shown in the upper corner with associated PD parameters, including Emax, ED50, and slope (N).
Survival.Survival to the end study endpoint mirrored the qPCR results, with higher survival rates (50 to 100%) when higher doses of drug were administered and uniform fatality with very low concentrations (Table 3). In the wild-type group, the average survival rate was 89.6% in animals administered ≥2.5 mg/kg/24 h of posaconazole; however, animals that received less than this had an average survival rate of only 18.8%. As expected, the dosage breakpoint that correlated with significant differences in the survival rate for the Cyp51 mutants was shifted higher. In this group, the average survival rate was 70.6% in animals administered ≥10 mg/kg/24 h of posaconazole but an average of only 10.9% in animals that received <10 mg/kg/24 h. The differences in survival using the above dosing cutoffs were statistically significant by the Mann-Whitney rank sum test (P < 0.001). Survival was assessed for each dosage regimen, and differences in effects for wild-type versus Cyp51 mutant groups were explored. Large differences in survival between the two organism groups were seen at the 2.5-mg/kg/24 h dose. At this dose, survival was 100% in the wild-type group versus 20.8% in the mutant group (P = 0.01). The free drug AUC at this dose level is approximately 0.3, and the free drug AUC/MIC for wild-type organisms would range from 0.6 to 1.2. This free AUC/MIC value range is similar to that associated with a static effect based upon qPCR (free AUC/MIC, 0.44 to 0.99). The free AUC/MIC for the CYP51 mutant group would be only 0.04 to 0.3. These values did not produce appreciable efficacy using the qPCR endpoint. Finally, logistic regression was also performed using survival as the outcome. There was a statistically significant 17% increase in survival associated with every 1-log10 decrease in burden as measured by qPCR. The Hosmer-Lemeshow test showed a good model fit, and the difference in exposure associated with survival and death was statistically significant (P < 0.0001).
Animal survivala for each A. fumigatus isolate to end of study (7 days) for a given posaconazole daily dose
DISCUSSION
Invasive aspergillosis is a devastating infection for the immunocompromised host (1, 2, 36–40). The development of new-generation triazoles, such as posaconazole and voriconazole, has improved the ability to prevent and treat these infections. Despite this therapeutic advance, up to half of patients continue to succumb to progressive aspergillosis. Numerous factors account for treatment failure, including persistent host immunodeficiency, late diagnosis, and inadequate antifungal exposure. Insufficient drug exposure can be due to an inadequate dose level, pharmacokinetic variability, and more recently the development of Aspergillus species triazole resistance (19–22). Determining the optimal antifungal exposure is requisite for addressing these pharmacologic shortfalls. Animal model pharmacokinetic/pharmacodynamic investigation has proven useful for designing optimal dosing regimens and delineating resistance levels which can be overcome with dose adjustments (3–6). These approaches have been underutilized for filamentous fungal infections.
Results from the current studies demonstrated a strong relationship between dose and effect for a triazole in therapy of invasive pulmonary aspergillosis. Furthermore, across the relatively large group of organisms included in the present experiments, efficacy was also closely linked to the MIC. More posaconazole, on a mg/kg basis, was required for efficacy against organisms with reduced in vitro susceptibility. The AUC/MIC index was utilized as the PK/PD index for subsequent exposure-response analysis based upon the previous in vivo models and clinical results for invasive candidiasis and aspergillosis (15, 33–35, 41–55). This index provided a useful measure of exposure for modeling the present data using a sigmoid Emax model.
The primary treatment endpoint chosen for these studies was quantitative lung PCR. The rationale for this choice was based upon the relatively large dynamic range between effective and ineffective therapy and reproducibility among biologic replicates. Our experience with this measure was similar to that previously reported (56–59). The quantitative burden of Aspergillus fumigatus in mouse lungs over the treatment range was closely related to animal survival over the study period. However, we did find that survival was less sensitive at detecting changes in microbiological efficacy. When the survival data were fit to a sigmoid Hill model examining the relationship of the free drug AUC/MIC ratio (fAUC/MIC) to survival, the relationship was strong but less than with the qPCR endpoint (R2 of 0.63 with survival endpoint, compared to 0.79 for qPCR).
It is unclear which qPCR endpoint in this infection model might correlate with optimal treatment effect in patients. Treatment results from a similar immunocompromised model of invasive candidiasis using an ED50 endpoint have correlated well with patient survival and clinical success (44, 45, 47, 50, 52, 53, 55). We report the posaconazole AUC/MIC associated with a net stasis or inhibitory endpoint, as well as that necessary to produce a further 1-log10 CE/ml reduction in organ burden. While the dose associated with these endpoints varied nearly 50-fold across the group of Aspergillus isolates, the AUC/MIC varied only 4-fold. The unbound AUC/MIC values associated with the stasis and killing endpoints were 1.09 and 2.07, respectively. We were somewhat surprised at the relatively low values for these estimates compared to those demonstrated for multiple triazole antifungals in similar animal models and clinical trials of invasive candidiasis. The AUC/MIC PD targets identified in the present study are more than 10-fold lower than those for disseminated candidiasis for posaconazole in a similar immunocompromised murine model (33). The basis for these differences is not evident but clearly is a fertile area for future mechanistic investigations. Previous PK/PD investigation with posaconazole and other triazoles in experimental aspergillosis have also explored the question of PD index magnitude (34, 35). We are encouraged to observe congruence in the PK/PD exposure-response relationships across animal models and laboratories. The posaconazole AUC/MIC associated with 50% of maximal effect in a similar model with a single A. fumigatus isolate was a free drug ratio of 1.67, compared to the present study value of 1.76 (34). Similarly, the free drug AUC/MIC needed to protect 50% of mice from mortality in an acute disseminated aspergillosis model for three A. fumigatus strains was a value of 3.2 (35).
An additional question probed by the present study includes delineating the impact of MIC variation due to the majority of defined mutations in the gene conferring resistance in the triazole target. The observations were similar to that described for other antimicrobial agents, such as quinolones with pneumococci compared to Gram-negative bacilli (60, 61). Principally, MIC is a relatively robust predictor of efficacy for susceptible and resistant strains across resistance mechanisms.
Experimental PK/PD analyses have been shown to be useful for predicting clinical outcomes. If one considers posaconazole pharmacokinetics in patients and the MIC distribution for Aspergillus fumigatus isolates from surveillance studies, one would predict treatment success for the majority of patients. The kinetics of the current formulation using a regimen of 200 mg every 6 h given with a high-fat meal would be expected to produce an AUC exposure as high as 60 mg · h/liter (i.e., free AUC, 0.6 mg · h/liter) (62). If the stasis endpoint from the present study is relevant for clinical outcome, the highest MIC for which the current posaconazole formulation and regimen would be predicted to produce a favorable outcome is 0.5 μg/ml. For the 1-log-kill endpoint, this value would shift a single dilution lower (0.25 μg/ml). The single clinical posaconazole experience for which concentration monitoring was available identified the optimal plasma concentration associated with clinical efficacy as a value of ≥1.25 μg/ml (15). Unfortunately, organism MICs were not available for additional PK/PD analysis. Examination of the outcomes of future clinical investigations of the present and new formulations of posaconazole in treatment of invasive aspergillosis will be important to explore these experimental PK/PD predictions. In the absence of this clinical evidence, the present studies should be used to guide preliminary susceptibility breakpoint determination.
ACKNOWLEDGMENT
We thank David Perlin for providing Aspergillus isolates DPL EC S 1 and EMFR S678P.
FOOTNOTES
- Received 20 June 2012.
- Returned for modification 16 August 2012.
- Accepted 7 November 2012.
- Accepted manuscript posted online 12 November 2012.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
