Final Report Summary - IN VITRO PKPD SYSTEM (In vitro pharmacokinetic/pharmacodynamic system for antifungal combination therapy against filamentous fungi)
Project context and objectives
Invasive fungal infections caused by filamentous fungi have emerged as a leading cause of morbidity and mortality in severely immunocompromised patients, such as cancer patients, transplant recipients, patients with diabetes, AIDS or other immunodeficiencies (chronic granulomateous disease). Antifungal agents that are commonly used to treat these infections belong to three classes: polyenes, which target ergosterol in the fungal cell membrane, azoles, which target lanosterol C14a-demethylase inhibiting ergosterol biosynthesis and echinocandins, which target b-D-glucan synthase, an enzyme catalysing the cell wall biosynthesis of most fungi. Although these agents have shown some efficacy in treating infections caused by Aspergillus spp, the mortality remains high (60-80 %). In vitro studies can be used to assess the efficacy (pharmacodynamics) of combination therapy and to define the final regimen design for in vivo studies and clinical trials. A challenge for current in vitro methodologies for testing antifungal drug combinations constitutes the simulation of in vivo conditions in order to account for pharmacological, toxicological and immunological factors that may affect antifungal efficacy.
The objective of this proposal is the development of a new in vitro pharmacokinetic/ pharmacodynamic (PK-PD) system for filamentous fungi that will simulate the microenvironment at the site of infection and account for drug-host-fungus factors and their interactions. In this model, drug concentrations will fluctuate over time, simulating in vivo plasma pharmacokinetics; host defence cells will be incorporated to simulate the pathophysiology of fungal infections at different tissues and clinical isolates with diverse microbiological characteristics will be tested in order to reflect variability of virulence factors, growth rates and drug susceptibilities. Similar but simpler systems have been developed to study the effect of combination regimens against yeast and bacteria. However, their application for filamentous fungi was problematic due to the filamentous growth of these pathogens and difficulties in quantifying their growth over time.
Work performed
The in vitro model that was developed consisted of floating tubes (Float-A-lyzer) with walls made of semi-permeable cellulose ester membranes, which would allow free diffusion of small molecules (nutrients and drugs) but retain growing fungi and their macromolecular products (galactomannan). These tubes were placed inside a glass beaker whose content was diluted with a peristaltic pump. Thus, drug concentration in the beaker would be in equilibrium with drug concentration inside the Float-A-lyzer, while galactomannan could be used as a marker of fungal growth. Float-A-lyzers with semi-permeable membranes of different molecular weight cut off (MWCO) (10, 20, 50 kDa) and volume (1, 5 and 10 ml) were tested and pharmacokinetic (PK) and pharmacodynamic (PD) data were analysed. The best results were achieved with a semi-permeable membrane of a 20kDa MWCO and a volume of 10 ml. For PK studies, various bioassays were evaluated in order to measure voriconazole, amphotericin B and caspofungin levels in the in vitro system using different nutrient media (Sabouraud, RPMI1640) and susceptible yeast (Candida kefyr, Candida albicans) and mould (Paecilomyces variotii, Aspergillus fumigatus) strains. For PD studies, the galactomannan kinetics inside and outside the Float-A-Lyzer were studied for increasing inocula (102-106 CFU/ml) of A. fumigatus with a commercial ELISA ( Platelia, Biorad).
The system that was finally developed consists of an internal compartment (a 10 ml dialysis tube made out of semi-permeable cellulose membrane allowing the free diffusion of molecules with MW <20kDa) placed inside an external compartment (a 700 ml glass beaker) whose content is diluted by a peristaltic pump working at the same rate as the clearance of the antifungal drugs in human plasma. Aspergillus conidia were inoculated inside the internal compartment and antifungal drugs were added in the external and internal compartment. The model has now been adapted to include two drugs with different half-lives, i.e. different flow rates for each drug, thus enabling the study of drug combinations.
The human plasma drug concentration-time profile of voriconazole, amphotericin B, caspofungin and posaconazole were simulated in the new in vitro system reliably and reproducibly (intra- and inter-day assay variation <10 %). The in vitro PK parameters were similar to those observed in humans for all four drugs except amphotericin B for which the half-life in the in vitro system was 12 hours (h) as opposed to 24h, which is the mean residence time of amphotericin B in plasma. The half-life of 12h is similar to the half-life that characterises the second phase of elimination of amphotericin B observed 6-24h after administration The galactomannan-time data followed a sigmoidal curve reaching a plateau after 12h for the drug-free control, whereas in the presence of drugs the curve was shifted to the left, reaching a lower plateau depending on the drug and dose.
The pharmacodynamic characteristics of voriconazole, amphotericin B (AMB) and caspofungin were determined against three clinical isolates of A. fumigatus, A. flavus and A. terreus with identical minimum inhibitory concentrations (MICs) (1 mg/l for AMB, 0.5 mg/l for voriconazole) and minimum effective concentrations (MECs) (0.5 mg/l for caspofungin). Despite the same MICs, AMB completely inhibited (100 %) A. fumigatus but not A. flavus and A. terreus, whose growth was only delayed for 7.53h and 22.8h respectively (J. Meletiadis et al., 20th ECCMID, 2010). Voriconazole partially inhibited A. fumigatus (49.5 %) and ?. flavus (27.9 %) but not ?. terreus and delayed fungal growth by 3.99h (A. fumigatus) and 5.37h (?. terreus). The new in vitro system simulated human pharmacokinetics of antifungal drugs and revealed important pharmacodynamic differences in their activity that cannot be disclosed by conventional MIC testing (J. Meletiadis et al., 49th ICAAC 2009).
The PD of voriconazole activity against Aspergillus spp. were studied in the new in vitro PK system and the PK-PD data were bridged with data on human drug exposure to assess efficacy of standard dosing (J. Meletiadis et al., 50th ICAAC, 2010). For A. fumigatus infections, standard treatment with 4 mg/kg of voriconazole was not associated with maximal drug efficacy, while dosing with 5 mg/kg appeared to be more effective. For A. terreus, even 5 mg/kg dosing was not sufficient, whereas no voriconazole dose provided adequate drug exposure against A. Flavus.
Real time polymerase chain reaction (PCR) of Aspergillus DNA confirm that the total increase of PCR conidial equivalents (PCR CE) was associated with the total increase of galactomannan over time, as assessed with the area under the corresponding curves (r2 = 0.98). Thus, PCR CE could be used to assess the pharmacodynamics of caspofungin in the new in vitro PK-PD model (Meletiadis J. et al., AAC 2012).
Optimisation studies showed an excellent correlation between in vitro PK-PD analysis of multi-dose pharmacodynamics of voriconazole using the 103 CFU/ml inoculum, and in vivo survival of mice infected with four A. fumigatus strains with variable voriconazole susceptibility and treated with increasing doses of voriconazole for 15 days. The in vitro log10AUC0-24/MIC associated with 50 % of maximal activity was 11.53 (8-16), which is very close to the corresponding value found in the in vivo animal model that was 10.5 f?UC0-24/MIC (Mavridou E. et al., AAC 2010;54(11)) after adjusting for the 58 % protein binding (Siopi M., 52nd ICAAC 2012).
Human voriconazole pharmacokinetics were simulated in the in vitro PK-PD model with Cmax varying from 1.75-7 mg/l and t1/2 of 6h using the optimised in vitro conditions derived from the in vitro-in vivo correlation study. The pharmacodynamic target associated with near maximal activity of voriconazole was 56 fAUC/MIC in the in vitro model. Monte Carlo simulation analysis was employed in order to calculate the percentage of patients that will attain the pharmacodynamic target for different MICs after standard dosing of 4 mg/kg, which corresponded to a fAUC0-24 of 24.8 (13-46.7) mg*h/L. The susceptibility, intermediate and resistant MIC ranges were defined as the MICs that >80 %, 10-80 and <10 % of the patients will attain the PD target of maximal antifungal activity. Monte Carlo simulation analysis revealed that isolates with MICs =0.25 0.5 and =1 could be considered susceptible, intermediate and resistant, respectively (Siopi M., 52nd ICAAC 2012).
The in vitro PK-PD analysis of posaconazole using an inoculum of 103 CFU/ml resulted in AUC/MIC corresponding to 50 % of maximal activity of 355.1 which is very close to the 321.3 AUC/MIC observed in vivo (Mavridou E. et al., AAC 2010;54(2)). However, this finding is conflicting with the free-drug hypothesis based on which only the free fraction of the drug is pharmacodynamically active, i.e. that the in vitro AUC/MIC should be 1 % of the in vivo AUC/MIC since posaconazole in plasma is 99 % protein bound. In order to explain this difference, the effect of serum on the pharmacodynamics of posaconazole was studied against A. fumigatus. Posaconazole pharmacokinetics with Cmax 3.1 mg/l and t1/2 24h were simulated in the in vitro model and the pharmacodynamics were determined in the presence of serum by monitoring galactomannan levels. No significant differences were found in the galactomannan index time-curve with and without serum. Thus, posaconazole activity was not altered in the presence of serum, which explains the in vitro-in vivo correlation and challenging the free-drug hypothesis.
The in vitro model showed a very good simulation of the monophasic and biphasic plasma concentration profile of voriconazole and amphotericin B, respectively, for different dosing regimens. In vitro PK-PD analysis of voriconazole and amphotericin B against Aspergillus species showed a differential activity of both drugs, despite identical MICs of the tested isolates (0.5 mg/l for voriconazole and 1 mg/l for amphotericin B). A. fumigatus was the most susceptible in the in vitro PK-PD model, whereas higher concentrations were required for the inhibition of A. flavus and A. terreus (Al-Saigh R. et al., AAC 2012 accepted).
The pharmacodynamic characteristics of voriconazole + amphotericin B combination were studied in the in vitro PK-PD model simulating clinically administered doses of both drugs. The combination was antagonistic at higher doses of amphotericin B, since voriconazole reversed the complete inhibition of galactomannan production by amphotericin B alone when both drugs were combined. The combination was synergistic at lower doses of amphotericin B, since the combination reduced galactomannan production more than each drug did alone (Siopi M., 52nd ICAAC 2012).Taking into account the protein binding of the two drugs, the synergistic interaction occurred at clinically achievable drug concentrations, whereas the antagonistic interaction did not.
The presence of neutrophils within the dialysis tube did not alter the effect of voriconazole since galactomannan-time curves were similar. Further, experiments are required in order to optimise the in vitro conditions.
Main results
The results of the novel in vitro PK-PD system developed in this project were very well correlated with the results of complex, expensive and time-consuming experiments with animal models. One-to-three day experiments with the in vitro PK-PD model gave similar results obtained from animals infected with the same Aspergillus isolates and treated for 15 days with antifungal drugs. These findings render the novel system as one of the few in vitro models that provides results that correlate precisely with those obtained with animal experiments and, hence, an excellent alternative for replacing / decreasing animal testing.
Another major finding of the second period of the project was the determination of susceptibility breakpoints for voriconazole. Voriconazole is the drug of choice for aspergillosis and therefore the most commonly used drug to treat these infections. Nonetheless, no reliable susceptibility breakpoints have yet been established, thus prohibiting the precise characterisation of resistant isolates. Previously determined epidemiological cut-off values for voriconazole did not demonstrate clear clinical correlates, whereas animal experimentation was hampered by the different voriconazole pharmacokinetics in animals and in humans. These limitations were overcome with the in vitro PK-PD model developed in this project, since human voriconazole pharmacokinetics were reliably simulated for different dosages and the pharmacodynamics against clinical isolates causing invasive aspergillosis were determined. It appears that the breakpoints = 0.25 for susceptibility and = 1 for resistance can be used to detect resistant isolates and thus treat effectively.
Finally, another important finding is the dose dependent synergistic interaction of the amphotericin B + voriconazole combination. This combination is one of the most commonly used antifungal combination regimens, particularly for refractory aspergillosis. In vitro and in vivo experiments produced conflicting results with antagonism being the most commonly observed interaction, despite the broad clinical use of the combination and no evidence of clinical antagonism. The concentration-dependent nature of this interaction was previously hypothesised but the conventional in vitro susceptibility testing used constant drug concentrations (Meletiadis J et al., IJAC 2009). Those results could not be extrapolated to humans since the changes over time of the concentrations of the two drugs alone and in combination were not simulated. The in vitro PK-PD model confirmed the dose-dependent interaction of amphotericin B + voriconazole combination. Furthermore, the PK-PD model showed that the synergistic and not the antagonistic interaction was observed at clinically relevant dosing regimens of both drugs, thus explaining the discrepancy between clinical experience and in vitro/in vivo findings.
Invasive fungal infections caused by filamentous fungi have emerged as a leading cause of morbidity and mortality in severely immunocompromised patients, such as cancer patients, transplant recipients, patients with diabetes, AIDS or other immunodeficiencies (chronic granulomateous disease). Antifungal agents that are commonly used to treat these infections belong to three classes: polyenes, which target ergosterol in the fungal cell membrane, azoles, which target lanosterol C14a-demethylase inhibiting ergosterol biosynthesis and echinocandins, which target b-D-glucan synthase, an enzyme catalysing the cell wall biosynthesis of most fungi. Although these agents have shown some efficacy in treating infections caused by Aspergillus spp, the mortality remains high (60-80 %). In vitro studies can be used to assess the efficacy (pharmacodynamics) of combination therapy and to define the final regimen design for in vivo studies and clinical trials. A challenge for current in vitro methodologies for testing antifungal drug combinations constitutes the simulation of in vivo conditions in order to account for pharmacological, toxicological and immunological factors that may affect antifungal efficacy.
The objective of this proposal is the development of a new in vitro pharmacokinetic/ pharmacodynamic (PK-PD) system for filamentous fungi that will simulate the microenvironment at the site of infection and account for drug-host-fungus factors and their interactions. In this model, drug concentrations will fluctuate over time, simulating in vivo plasma pharmacokinetics; host defence cells will be incorporated to simulate the pathophysiology of fungal infections at different tissues and clinical isolates with diverse microbiological characteristics will be tested in order to reflect variability of virulence factors, growth rates and drug susceptibilities. Similar but simpler systems have been developed to study the effect of combination regimens against yeast and bacteria. However, their application for filamentous fungi was problematic due to the filamentous growth of these pathogens and difficulties in quantifying their growth over time.
Work performed
The in vitro model that was developed consisted of floating tubes (Float-A-lyzer) with walls made of semi-permeable cellulose ester membranes, which would allow free diffusion of small molecules (nutrients and drugs) but retain growing fungi and their macromolecular products (galactomannan). These tubes were placed inside a glass beaker whose content was diluted with a peristaltic pump. Thus, drug concentration in the beaker would be in equilibrium with drug concentration inside the Float-A-lyzer, while galactomannan could be used as a marker of fungal growth. Float-A-lyzers with semi-permeable membranes of different molecular weight cut off (MWCO) (10, 20, 50 kDa) and volume (1, 5 and 10 ml) were tested and pharmacokinetic (PK) and pharmacodynamic (PD) data were analysed. The best results were achieved with a semi-permeable membrane of a 20kDa MWCO and a volume of 10 ml. For PK studies, various bioassays were evaluated in order to measure voriconazole, amphotericin B and caspofungin levels in the in vitro system using different nutrient media (Sabouraud, RPMI1640) and susceptible yeast (Candida kefyr, Candida albicans) and mould (Paecilomyces variotii, Aspergillus fumigatus) strains. For PD studies, the galactomannan kinetics inside and outside the Float-A-Lyzer were studied for increasing inocula (102-106 CFU/ml) of A. fumigatus with a commercial ELISA ( Platelia, Biorad).
The system that was finally developed consists of an internal compartment (a 10 ml dialysis tube made out of semi-permeable cellulose membrane allowing the free diffusion of molecules with MW <20kDa) placed inside an external compartment (a 700 ml glass beaker) whose content is diluted by a peristaltic pump working at the same rate as the clearance of the antifungal drugs in human plasma. Aspergillus conidia were inoculated inside the internal compartment and antifungal drugs were added in the external and internal compartment. The model has now been adapted to include two drugs with different half-lives, i.e. different flow rates for each drug, thus enabling the study of drug combinations.
The human plasma drug concentration-time profile of voriconazole, amphotericin B, caspofungin and posaconazole were simulated in the new in vitro system reliably and reproducibly (intra- and inter-day assay variation <10 %). The in vitro PK parameters were similar to those observed in humans for all four drugs except amphotericin B for which the half-life in the in vitro system was 12 hours (h) as opposed to 24h, which is the mean residence time of amphotericin B in plasma. The half-life of 12h is similar to the half-life that characterises the second phase of elimination of amphotericin B observed 6-24h after administration The galactomannan-time data followed a sigmoidal curve reaching a plateau after 12h for the drug-free control, whereas in the presence of drugs the curve was shifted to the left, reaching a lower plateau depending on the drug and dose.
The pharmacodynamic characteristics of voriconazole, amphotericin B (AMB) and caspofungin were determined against three clinical isolates of A. fumigatus, A. flavus and A. terreus with identical minimum inhibitory concentrations (MICs) (1 mg/l for AMB, 0.5 mg/l for voriconazole) and minimum effective concentrations (MECs) (0.5 mg/l for caspofungin). Despite the same MICs, AMB completely inhibited (100 %) A. fumigatus but not A. flavus and A. terreus, whose growth was only delayed for 7.53h and 22.8h respectively (J. Meletiadis et al., 20th ECCMID, 2010). Voriconazole partially inhibited A. fumigatus (49.5 %) and ?. flavus (27.9 %) but not ?. terreus and delayed fungal growth by 3.99h (A. fumigatus) and 5.37h (?. terreus). The new in vitro system simulated human pharmacokinetics of antifungal drugs and revealed important pharmacodynamic differences in their activity that cannot be disclosed by conventional MIC testing (J. Meletiadis et al., 49th ICAAC 2009).
The PD of voriconazole activity against Aspergillus spp. were studied in the new in vitro PK system and the PK-PD data were bridged with data on human drug exposure to assess efficacy of standard dosing (J. Meletiadis et al., 50th ICAAC, 2010). For A. fumigatus infections, standard treatment with 4 mg/kg of voriconazole was not associated with maximal drug efficacy, while dosing with 5 mg/kg appeared to be more effective. For A. terreus, even 5 mg/kg dosing was not sufficient, whereas no voriconazole dose provided adequate drug exposure against A. Flavus.
Real time polymerase chain reaction (PCR) of Aspergillus DNA confirm that the total increase of PCR conidial equivalents (PCR CE) was associated with the total increase of galactomannan over time, as assessed with the area under the corresponding curves (r2 = 0.98). Thus, PCR CE could be used to assess the pharmacodynamics of caspofungin in the new in vitro PK-PD model (Meletiadis J. et al., AAC 2012).
Optimisation studies showed an excellent correlation between in vitro PK-PD analysis of multi-dose pharmacodynamics of voriconazole using the 103 CFU/ml inoculum, and in vivo survival of mice infected with four A. fumigatus strains with variable voriconazole susceptibility and treated with increasing doses of voriconazole for 15 days. The in vitro log10AUC0-24/MIC associated with 50 % of maximal activity was 11.53 (8-16), which is very close to the corresponding value found in the in vivo animal model that was 10.5 f?UC0-24/MIC (Mavridou E. et al., AAC 2010;54(11)) after adjusting for the 58 % protein binding (Siopi M., 52nd ICAAC 2012).
Human voriconazole pharmacokinetics were simulated in the in vitro PK-PD model with Cmax varying from 1.75-7 mg/l and t1/2 of 6h using the optimised in vitro conditions derived from the in vitro-in vivo correlation study. The pharmacodynamic target associated with near maximal activity of voriconazole was 56 fAUC/MIC in the in vitro model. Monte Carlo simulation analysis was employed in order to calculate the percentage of patients that will attain the pharmacodynamic target for different MICs after standard dosing of 4 mg/kg, which corresponded to a fAUC0-24 of 24.8 (13-46.7) mg*h/L. The susceptibility, intermediate and resistant MIC ranges were defined as the MICs that >80 %, 10-80 and <10 % of the patients will attain the PD target of maximal antifungal activity. Monte Carlo simulation analysis revealed that isolates with MICs =0.25 0.5 and =1 could be considered susceptible, intermediate and resistant, respectively (Siopi M., 52nd ICAAC 2012).
The in vitro PK-PD analysis of posaconazole using an inoculum of 103 CFU/ml resulted in AUC/MIC corresponding to 50 % of maximal activity of 355.1 which is very close to the 321.3 AUC/MIC observed in vivo (Mavridou E. et al., AAC 2010;54(2)). However, this finding is conflicting with the free-drug hypothesis based on which only the free fraction of the drug is pharmacodynamically active, i.e. that the in vitro AUC/MIC should be 1 % of the in vivo AUC/MIC since posaconazole in plasma is 99 % protein bound. In order to explain this difference, the effect of serum on the pharmacodynamics of posaconazole was studied against A. fumigatus. Posaconazole pharmacokinetics with Cmax 3.1 mg/l and t1/2 24h were simulated in the in vitro model and the pharmacodynamics were determined in the presence of serum by monitoring galactomannan levels. No significant differences were found in the galactomannan index time-curve with and without serum. Thus, posaconazole activity was not altered in the presence of serum, which explains the in vitro-in vivo correlation and challenging the free-drug hypothesis.
The in vitro model showed a very good simulation of the monophasic and biphasic plasma concentration profile of voriconazole and amphotericin B, respectively, for different dosing regimens. In vitro PK-PD analysis of voriconazole and amphotericin B against Aspergillus species showed a differential activity of both drugs, despite identical MICs of the tested isolates (0.5 mg/l for voriconazole and 1 mg/l for amphotericin B). A. fumigatus was the most susceptible in the in vitro PK-PD model, whereas higher concentrations were required for the inhibition of A. flavus and A. terreus (Al-Saigh R. et al., AAC 2012 accepted).
The pharmacodynamic characteristics of voriconazole + amphotericin B combination were studied in the in vitro PK-PD model simulating clinically administered doses of both drugs. The combination was antagonistic at higher doses of amphotericin B, since voriconazole reversed the complete inhibition of galactomannan production by amphotericin B alone when both drugs were combined. The combination was synergistic at lower doses of amphotericin B, since the combination reduced galactomannan production more than each drug did alone (Siopi M., 52nd ICAAC 2012).Taking into account the protein binding of the two drugs, the synergistic interaction occurred at clinically achievable drug concentrations, whereas the antagonistic interaction did not.
The presence of neutrophils within the dialysis tube did not alter the effect of voriconazole since galactomannan-time curves were similar. Further, experiments are required in order to optimise the in vitro conditions.
Main results
The results of the novel in vitro PK-PD system developed in this project were very well correlated with the results of complex, expensive and time-consuming experiments with animal models. One-to-three day experiments with the in vitro PK-PD model gave similar results obtained from animals infected with the same Aspergillus isolates and treated for 15 days with antifungal drugs. These findings render the novel system as one of the few in vitro models that provides results that correlate precisely with those obtained with animal experiments and, hence, an excellent alternative for replacing / decreasing animal testing.
Another major finding of the second period of the project was the determination of susceptibility breakpoints for voriconazole. Voriconazole is the drug of choice for aspergillosis and therefore the most commonly used drug to treat these infections. Nonetheless, no reliable susceptibility breakpoints have yet been established, thus prohibiting the precise characterisation of resistant isolates. Previously determined epidemiological cut-off values for voriconazole did not demonstrate clear clinical correlates, whereas animal experimentation was hampered by the different voriconazole pharmacokinetics in animals and in humans. These limitations were overcome with the in vitro PK-PD model developed in this project, since human voriconazole pharmacokinetics were reliably simulated for different dosages and the pharmacodynamics against clinical isolates causing invasive aspergillosis were determined. It appears that the breakpoints = 0.25 for susceptibility and = 1 for resistance can be used to detect resistant isolates and thus treat effectively.
Finally, another important finding is the dose dependent synergistic interaction of the amphotericin B + voriconazole combination. This combination is one of the most commonly used antifungal combination regimens, particularly for refractory aspergillosis. In vitro and in vivo experiments produced conflicting results with antagonism being the most commonly observed interaction, despite the broad clinical use of the combination and no evidence of clinical antagonism. The concentration-dependent nature of this interaction was previously hypothesised but the conventional in vitro susceptibility testing used constant drug concentrations (Meletiadis J et al., IJAC 2009). Those results could not be extrapolated to humans since the changes over time of the concentrations of the two drugs alone and in combination were not simulated. The in vitro PK-PD model confirmed the dose-dependent interaction of amphotericin B + voriconazole combination. Furthermore, the PK-PD model showed that the synergistic and not the antagonistic interaction was observed at clinically relevant dosing regimens of both drugs, thus explaining the discrepancy between clinical experience and in vitro/in vivo findings.