Abstract
Combination therapy of voriconazole with an echinocandin is often employed in order to increase the efficacy of voriconazole monotherapy.
Four clinical Aspergillus fumigatus isolates with different in vitro susceptibilities to voriconazole (MIC 0.125–2 mg/L) and anidulafungin (MEC 0.008–0.016 mg/L) were tested in an in vitro pharmacokinetic/pharmacodynamic model simulating human serum concentrations of standard dosages of voriconazole and anidulafungin. Fungal growth was assessed using galactomannan production and quantitative PCR. Drug concentrations were determined with bioassays. Pharmacodynamic interactions were assessed using Bliss independence analysis (BI) and Loewe additivity-based canonical mixture response-surface non-linear regression analysis (LA). Probability of target attainment (PTA) was estimated with Monte Carlo analysis for different doses of anidulafungin (25, 50 and 100 mg) and azole resistance rates (5%–25%).
Synergy [BI 51% (8%–80%), LA 0.63 (0.38–0.79)] was found at low anidulafungin (fCmax/MEC <10) and voriconazole (fAUC/MIC <10) exposures, whereas antagonism [BI 12% (5%–18%, LA 1.12 (1.04–4.6)] was found at higher drug exposures. The largest increase in PTA was found with 25 mg of anidulafungin and voriconazole MIC distributions with high (>10%) resistance rates. PTAs for isolates with voriconazole MICs of 1, 2 and 4 mg/L was 78%, 12% and 0% with voriconazole monotherapy and 96%–100%, 68%–82% and 9%–20% with combination therapy, respectively. Optimal activity was associated with a voriconazole tCmin/MIC ratio of 1.5 for monotherapy and 0.75 for combination therapy.
The present study indicated that the combination of voriconazole with low-dose anidulafungin may increase the efficacy and reduce the cost and potential toxicity of antifungal therapy, particularly against azole-resistant A. fumigatus isolates and in patients with subtherapeutic serum levels. This hypothesis warrants further in vivo verification.
Introduction
Voriconazole is currently recommended as first-line therapy of invasive aspergillosis.1 However, its efficacy is limited against azole-resistant Aspergillus fumigatus isolates and in patients with subtherapeutic serum levels.2,3 Combination therapy with an echinocandin is often employed in order to increase efficacy and overcome the limitation of monotherapy regimens. Currently, the coadministration of an echinocandin with an azole is recommended by international guidelines only in a setting of salvage therapy and particularly in difficult-to-treat infections.4 A large, prospective, randomized clinical trial comparing voriconazole/anidulafungin versus voriconazole for primary therapy of invasive aspergillosis demonstrated non-statistically significant increased efficacy of combination therapy in the primary endpoint analysis.5 Since standard dosages of both drugs were used and the in vitro susceptibility of isolates was unknown, conclusions about the efficacy of different doses of combined drugs against azole-resistant isolates cannot be inferred.
Antifungal drug combinations can demonstrate complex in vitro and in vivo pharmacodynamic (PD) interactions, ranging from synergy to antagonism at different concentrations and doses.6,7 We recently showed that the efficacy of voriconazole/amphotericin B combination can be maximized by combining standard doses of voriconazole with low doses of amphotericin B in an in vitro dialysis/diffusion closed pharmacokinetic (PK)/PD model.8 Furthermore, target values for serum concentrations were determined for isolates with reduced in vitro susceptibility. In the present study, we applied the latter model in order to investigate the PD effects of voriconazole/anidulafungin combination against A. fumigatus isolates including azole-resistant isolates, simulating human serum concentration–time profiles of standard and lower dosages of anidulafungin. Target serum levels of combination regimens were determined for isolates with increasing MICs and target attainment rates were calculated for centres with different resistance rates.
Materials and methods
Fungal isolates
Four clinical A. fumigatus isolates with defined azole resistance mechanisms and distinct susceptibility profiles to voriconazole and anidulafungin were studied. These included a WT isolate without substitutions in the cyp51A gene (AZN8196) and three non-WT isolates with the following confirmed by sequence-based analysis: cyp51A mutations G54W (V59-73), M220I (V28-77) and TR35/L98H (V52-35).9In vitro susceptibility testing was performed in accordance with the standard CLSI M38-A2 method evaluating results after 48 h of incubation.10 Voriconazole MICs were determined as the lowest drug concentration corresponding to complete growth inhibition and were 0.125, 0.125, 0.25 and 2 mg/L, respectively, while anidulafungin minimum effective concentrations (MECs) were defined as the lowest drug concentration at which short, branched hyphal clusters were observed compared with the growth control well of the panel using an inverted microscope. All isolates had identical MECs of 0.016 mg/L except the V59-73 isolate, which had an MEC of 0.008 mg/L. All isolates demonstrate paradoxical growth at high (≥0.5 mg/L) concentrations of anidulafungin.
The isolates were stored in normal sterile saline with 10% glycerol at −70°C until the study was performed and prior to testing they were revived by subculturing them twice on Sabouraud dextrose agar with gentamicin and chloramphenicol (SGC2; bioMérieux) at 30°C for 5–7 days. Inocula suspensions were prepared in sterile saline with 0.1% Tween 20, conidia were counted using a Neubauer chamber in order to obtain a final concentration of 103 cfu/mL in the model and the count of each inoculum was confirmed by subsequent culturing on SGC2 plates.
Antifungal drugs and medium
Laboratory-grade standard powders of voriconazole and anidulafungin (Pfizer, Groton, CT, USA) were dissolved in sterile DMSO (Carlo Erba Reactifs-SDS, Val de Reuil, France) and stock solutions of 10 mg/mL were stored in small portions at −70°C until the day of the experiment. The medium used throughout was RPMI 1640 medium (with l-glutamine, but without bicarbonate) (AppliChem, Darmstadt, Germany) buffered to pH 7.0 with 0.165 M MOPS (AppliChem, Darmstadt, Germany) and supplemented with 100 mg/L chloramphenicol (AppliChem, Darmstadt, Germany).
In vitro PK/PD model
A previously optimized two-compartment PK/PD dialysis/diffusion closed model was used.11 The model consists of an external compartment comprising a conical flask connected to a peristaltic pump and an internal compartment of a 10 mL semipermeable dialysis tube (Spectra/Por® Float-A-Lyzer® G2 with a molecular weight cut-off of 20 kDa; Spectrum Laboratories, Breda, The Netherlands) inoculated with the conidial suspension (103 cfu/mL). The in vitro model has been adapted to accommodate two drugs with different t1/2, thus enabling the study of drug combinations.8 Repeated sampling of 160 μL was made from the internal compartment in order to determine the drug concentration (60 μL) and galactomannan level (100 μL)–time profile. All samples were stored at −70°C until tested.
In vitro PK
Mean PK parameters and their variation after standard dosages of 4 mg/kg voriconazole and 100 mg of anidulafungin were simulated in the in vitro PK/PD model. In particular, free drug concentration–time profiles were simulated in order to encompass low, intermediate and high drug exposures with voriconazole fCmax 0.35, 1.5 and 3 mg/L and t1/2 6 h dosed every 12 h and anidulafungin fCmax 0.01, 0.08 and 0.16 mg/L and t1/2 24 h dosed every 24 h. These fCmax values corresponded to the upper 95% CI limit, the average and the lower 95% CI limit, respectively, of the free drug levels observed in human plasma of patients on the basis of voriconazole and anidulafungin protein binding of 58% and 99%, respectively.12,13 Thus, nine different combination regimens including monotherapies and drug-free control were investigated. After inoculating the internal compartment with Aspergillus conidia, drugs were injected into both compartments of the model alone and in combination and the external compartment was placed on a heated (37°C) magnetic stirrer for 72 h. During each experiment, temperature and flow rate were measured to ensure that they were at the expected values. Voriconazole and anidulafungin levels were determined by microbiological agar diffusion assays using a Candida parapsilosis isolate and the WT AZN8196, respectively.14,15 The diameter of the inhibition zone regressed linearly on the log2 anidulafungin concentration (R2 = 0.89). The intra- and interexperiment coefficients of variation ranged from 3% to 12%, with an average of 6%. Because the targeted free anidulafungin concentrations were below the detection limit of the bioassay, the concentration–time profile of anidulafungin was calculated in dosing regimens with 10× the above-mentioned fCmaxs. A concentration–time curve was generated for each simulated dose and analysed by non-linear regression analysis using a one-compartment model described by the equation Ct = C0 * e−k*t, where Ct (dependent variable) is the concentration of drug at a given time t (independent variable), C0 is the initial concentration of the drug at time t = 0 h, e is the physical constant 2.718 and k is the rate of drug removal. The t1/2 was calculated using the equation t1/2 = 0.693/k and compared with the respective average values observed in humans.
In vitro PD
Galactomannan index (GI)
To estimate fungal growth and the antifungal effect of each monotherapy and combination dosing regimen, 100 μL was sampled from inoculated dialysis tubes at regular intervals up to 72 h. GI levels (100 μL of sample + 200 μL of saline) were determined using a commercially available sandwich enzyme-linked immunoassay (Platelia Aspergillus EIA; Bio-Rad Laboratories). A GI–time profile was constructed for each dose and isolate. Moreover, the area under the GI curve (AUCGI) was determined as a surrogate marker of fungal growth as previously described.16 The percentage of growth inhibition at each dose was calculated as 1 − AUCGI,DR/AUCGI,GC, where AUCGI,DR is the AUCGI at a certain dose of drug monotherapies and their combination, whereas AUCGI,GC is the AUCGI of the drug-free control. All experiments were carried out in duplicate and were independently performed on two different days with individually prepared inocula.
Quantitative PCR
A real-time PCR was developed in order to quantify fungal growth and killing with a sensitivity of 10 PCR conidial equivalent (CE). Aspergillus DNA was extracted (Argene DNA extraction kit; bioMérieux, Verniolle, France) from 3 mL samples from the internal compartment of the in vitro PK/PD model at 0 and 72 h combining enzymatic (incubation with protease K at 56°C for 10 min) and mechanical (4 h vortexing with glass beads) pretreatment. Real-time PCR was performed with a previously described assay (2Asp assay) using Aspergillus-specific primers (ASF1 and ADR1) and probe (ASP28P).17 The threshold cycle (Ct) of each sample, which identifies the cycle number during PCR when fluorescence exceeds a threshold value determined by the software, was converted into CE by interpolation from a four-point standard curve of Ct values obtained with 101–107Aspergillus cfu/mL. The change of log10 PCR CEs after 72 h of incubation compared with 0 h was calculated for each isolate and dosing regimen. For illustrative purposes, data were analysed with GraphPad Prism, version 5.0, for Windows (GraphPad Software, San Diego, CA, USA) and sigmoid curves were constructed with the four-parameter sigmoid Emax model sharing the Emax and Emin among all datasets for each concentration of voriconazole.
Drug interaction analysis
In order to assess the nature of in vitro interactions between voriconazole and anidulafungin, data were analysed using the Bliss independence and Loewe additivity models.18,19
Bliss independence
Bliss independence is described by the equation EIND = EANI × EVRC for a certain combination of two drugs, where EANI and EVRC are the percentage fungal growth in monotherapy regimens of anidulafungin and voriconazole, respectively, and EIND is the expected percentage fungal growth of a non-interactive (independent) theoretical combination. The difference (ΔE = EIND − EOBS) between the expected and the experimentally observed percentage growth describes the interaction of each combination of the concentrations of two drugs. If ΔE is >0 (EIND > EOBS), Bliss synergy was concluded; if ΔE is <0 (EIND < EOBS), Bliss antagonism was concluded. In any other case, Bliss independence was claimed. The ΔE was calculated for each combination and its statistical significance was determined by Student's t-test (P < 0.05).20
Loewe additivity
Equation (5), together with Equations (2), (3) and (4), was used to fit the global model to the percentage fungal growth using the non-linear platform of JMP5.0.1 software and weighted with the inverse of the standard deviation of the replicates (SAS Institute, Cary, NC, USA). The goodness of fit was checked with a variety of diagnostic tests such as R2 values, analysis of variance, lack-of-fit test, residual and leverage plot analysis, correlation matrix and the standard error of parameters. Parameters with coefficients not statistically different from 0 were removed from the final model. Simpler models with fewer parameters were statistically compared with complicated models with more parameters using the F-test. A P value <0.05 indicates that the complicated model is significantly better that the simpler model. Interaction indices were calculated using Equation (3) for different voriconazole/anidulafungin exposures. The interaction surface was constructed by subtracting the experimental response surface described by Equation (5) from the additive response surface described by Equation (5), with the interaction coefficients β and γ of Equations (2–4) being 0.
Monte Carlo simulation
In order to bridge the in vitro data with human PKs, Monte Carlo simulation analysis was performed using the Normal random number-generating function of Excel (MS Office 2007) for 10 000 patients infected with A. fumigatus isolates with voriconazole MICs ranging from 0.125 to 8 mg/L and anidulafungin MECs of 0.008–0.06 mg/L (this is the MEC range for clinical A. fumigatus isolates)22,23 and treated with the standard intravenous dosage of 4 mg/kg voriconazole twice daily alone and together with 100 mg of anidulafungin as a combination therapy regimen. For anidulafungin, this dosage resulted in a steady-state mean ± SD total maximum concentration in human serum (tCmax) of 7.2 ± 1.7 mg/L,24 which corresponds to a free maximum concentration (fCmax) of 0.072 ± 0.017 mg/L based on the 99% protein-binding rate of anidulafungin.25 For voriconazole, this dosage resulted in a mean ± SD total tAUC0–12 of 51 ± 11 mg · h/L,26 while the fAUC0–12 was calculated on the basis of the 58% protein binding of voriconazole in human serum and was 21.4 ± 9.2 mg · h/L.27 In order to estimate the percentage fungal growth for each of the 10 000 simulated patients treated with either the monotherapy regimens or the combination therapy regimen, the above-described response surface model was used. The input parameters of each simulated patient were the fCmax/MEC of anidulafungin and fAUC/MIC of voriconazole and the output parameter was the percentage of fungal growth for this particular patient. The percentage of patients with <50% of the estimated fungal growth was calculated for each voriconazole MIC. The PK/PD target corresponding to 50% growth (EI50) was previously found to be associated with 6 week survival for voriconazole and amphotericin B using the same in vitro PK/PD model.11,28 In addition to probability of target attainment rate (PTA) for each MIC, the total PTA was estimated for a collection of A. fumigatus isolates with a previously published anidulafungin MEC distribution (68% with MEC 0.008 mg/L, 21% with MEC 0.016 mg/L, 10% with MEC 0.03 mg/L and 1% with MEC 0.06 mg/L)23 and voriconazole WT MIC distribution adjusted to accommodate different resistance rates from 5% to 25% as shown in Figure 1.29 The frequency of each voriconazole MIC distribution was adjusted in order to keep the same relative frequencies for WT (MIC <1 mg/L) and resistant (MIC >1 mg/L) isolates. In order to find the optimal dose of anidulafungin, the PTA was also estimated for lower dosing regimens of anidulafungin of 50 and 25 mg, corresponding to mean ± SD tCmax (fCmax) of 4.2 ± 0.94 (0.042 ± 0.009) and 2.1 ± 0.47 (0.021 ± 0.005) mg/L, respectively.24
MIC distributions used for Monte Carlo simulation analysis (top) and the target attainment rates for each distribution combination regimen (bottom). The original dataset (broken line)29 was adjusted in order to accommodate increasing resistance rates from 5% to 25% (isolates with MIC >1 mg/L) while keeping the relative frequencies among the MICs the same. VRC, voriconazole; ANI, anidulafungin.
Therapeutic drug monitoring
The human serum levels required to attain clinically relevant in vitro EI50 were calculated in relation to voriconazole MICs and anidulafungin MECs. For that purpose, anidulafungin fCmax values and the voriconazole fCmin/MIC in monotherapy and combination therapy were calculated using Equation (5) after determining the best-fit parameters that described the entire exposure–response surface. Total drug concentrations tCmax of anidulafungin and tCmin of voriconazole were calculated on the basis of 99% and 58% protein-binding rates in human serum, respectively.25,27
Results
PK analysis
The in vitro model simulated well the steady-state human PK of voriconazole and anidulafungin.13,24 In particular, the initial fCmax in the internal compartment was 0.26±0.03, 1.31±0.16 and 2.62±0.32 mg/L with an average t1/2 of 7.4 h for voriconazole and 0.01±0.001, 0.08±0.005 and 0.18±0.001 mg/L with an average t1/2 of 21 h based on the extrapolated concentrations for anidulafungin. Voriconazole fAUC0–12 was 1.99, 9.96 and 19.91 mg · h/L and anidulafungin fAUC0–24 0.18, 1.44 and 2.88 mg · h/L for the three exposures, respectively. The supra-MEC peak concentrations of the two high anidulafungin exposures with targeted fCmax 0.08 and 0.16 mg/L were also verified by the presence of aberrant hyphae in the in vitro model. Concentration–time profiles of both monotherapies are depicted in Figure 2.
In vitro PK (continuous lines) of anidulafungin and voriconazole simulating free human plasma concentrations (broken lines).
PD analysis
Galactomannan
After 72 h of incubation, anidulafungin alone resulted in minimal (<10%) suppression of galactomannan production for all isolates at all three simulated exposures. On the other hand, voriconazole alone suppressed galactomannan production of voriconazole-susceptible isolates AZN8196 and V59-73 by >83% at all concentrations, whereas for V28-77 and V52-35 galactomannan suppression was 13%, 45% and 93% and 8%, 32% and 38% at fCmax 0.35, 1.5 and 3 mg/L, respectively. The GI–time curves of selected combination dosing regimens of anidulafungin and voriconazole are shown in Figure 3. For combinations with low anidulafungin fCmax (0.01 mg/L), GI was lower than both monotherapy regimens (Figure 3a–d), whereas at higher anidulafungin fCmaxs (0.08 and 0.16 mg/L) GI was similar to (Figure 3g and h) or higher (Figure 3e and f) than monotherapy dosing regimens.
GI curves for four A. fumigatus isolates with increasing voriconazole MICs of the voriconazole/anidulafungin combination demonstrating various levels of Bliss synergistic (a–d), antagonistic (e and f) or independent (g and h) interactions. The fCmax of each dosing regimen and the Bliss interactions are shown for each combination regimen. Synergistic interactions were found at low anidulafungin drug exposures, whereas antagonistic interactions were observed at high drug exposures. Error bars represent standard deviations. VRC, voriconazole; ANI, anidulafungin.
PCR
Real-time PCR data are shown in Figure 4. PCR CEs of all isolates increased by 3–4 log10 after 72 h compared with the PCR CE at 0 h. In the presence of voriconazole, sigmoid dose–response relationships were found with a 1 log10 PCR CE reduction observed for isolates with MIC 0.125 mg/L and stasis for isolates with MICs 0.25–2 mg/L at high voriconazole exposures (black lines, Figure 4). When voriconazole was combined with anidulafungin, the sigmoid dose–response curves were shifted to the left or down, indicating enhanced killing by the combination compared with monotherapy regimens particularly for the isolates with MICs 0.25–2 mg/L (black arrows). This effect was more pronounced at the two lowest anidulafungin exposures, whereas at the higher anidulafungin dosing regimen (fCmax = 0.16 mg/L) the effect was reversed (grey arrow), resulting in reduced killing compared with voriconazole alone for the resistant V52-35 isolate (broken line of the right bottom graph, Figure 4). Reduced killing was also observed for the susceptible V59-73 isolate at high voriconazole exposures (grey arrow).
Real-time PCR data of the voriconazole/anidulafungin combination. Changes in log10 PCR CEs of voriconazole alone and in combination are presented for each combination and isolate. The horizontal broken line corresponds to no change compared with PCR CE at 0 h (stasis). Black and grey arrows respectively indicate enhanced and decreased killing by voriconazole when combined with anidulafungin. VRC, voriconazole; ANI, anidulafungin.
PD interactions
Bliss interactions
In vitro PD analysis based on GI showed that the combination was indifferent at most dosing regimens against the two voriconazole-susceptible isolates with CLSI MIC 0.125 mg/L: WT AZN8196 and non-WT V59-73 with the cyp51A mutation G54W. The in vitro activity of the combination was similar to the activity of voriconazole alone at all three simulated fCmaxs of voriconazole and anidulafungin, except for the lowest drug exposures with fCmax 0.35 mg/L voriconazole and 0.01 mg/L anidulafungin where weak (8%–16%) Bliss synergistic interactions were found (sum 8%–36%) (Figure 5). Stronger Bliss synergy of 23%–80% (sum 270%) and 20%–24% (sum 65%) was found against the non-WT isolate V28-77 with CLSI MIC 0.25 mg/L and the resistant isolate V52-35 harbouring the TR35/L98H cyp51A mutation with CLSI MIC 2 mg/L, respectively (Figure 5). Synergistic interactions were found at intermediate and low drug exposures for both drugs with fCmax ≤1.5 mg/L voriconazole and ≤0.08 mg/L anidulafungin, whereas at higher drug exposures Bliss independent interactions were found. At the highest anidulafungin exposure, Bliss antagonistic interactions were found for all isolates although statistically significant Bliss antagonism was found only for V59-73 and V28-77. The presence of Bliss synergistic, independent and antagonistic interactions is illustrated with the GI–time curves (Figure 3) and the log10 PCR CE-exposure response curves (Figure 4) with the greatest shift of combination therapy curves observed for the isolates with reduced susceptibility to voriconazole.
Sum of Bliss interactions for all tested concentrations of the voriconazole/anidulafungin combination and each A. fumigatus isolate. The horizontal broken line represents the level of statistically significant interactions.
Interestingly, antagonism for V28-77 with voriconazole MIC 0.25 mg/L appeared at higher voriconazole exposures (fCmax 3 mg/L, fAUC 20 mg · h/L, fAUC/MIC 80) than for V59-73 with MIC 0.125 mg/L (fCmax 0.35 mg/L, fAUC 4 mg · h/L, fAUC/MIC 32), indicating exposure-dependent interactions. Thus, fAUC/MIC ratios >32 were associated with antagonism. Similarly, synergy for AZN8196 with voriconazole MIC 0.125 mg/L and anidulafungin MEC 0.016 mg/L was found at anidulafungin exposures up to fCmax 0.08 mg/L (5 fCmax/MEC) but not at fCmax 0.16 mg/L (10 fCmax/MEC), whereas for the V59-73 isolate with the same voriconazole MIC but lower anidulafungin MEC of 0.008 mg/L, synergy was found at lower anidulafungin exposures with fCmax 0.01 mg/L (1 fCmax/MEC) but not at fCmax 0.08 mg/L (10 fCmax/MEC). This indicates that synergy was observed at anidulafungin fCmax/MEC ratios <10. Exposure- and MIC-dependent killing was also found with quantitative PCR (Figure 4). For isolates with reduced susceptibility to voriconazole (MIC 0.25 and 2 mg/L), voriconazole exposures required to reach stasis reduced from fCmax ≥3 mg/L in monotherapy to fCmax ≤0.35 mg/L in combination therapy (Figure 4).
Loewe interactions
Exposure–response relationships of voriconazole in the presence of increasing concentrations of anidulafungin for four A. fumigatus isolates with different MICs. Voriconazole EC50 (exposure associated with 50% growth) statistically significantly increased at higher concentrations of anidulafungin, indicating enhancement of voriconazole activity (rightwards arrow) (F-test with 3 and 51 degrees of freedom = 5.887, P = 0.015). Note that some growth was observed at high voriconazole exposures when combined with anidulafungin, indicating reduction of voriconazole activity (upwards arrow). Error bars represent standard deviations.
Response surface modelling of the voriconazole/anidulafungin combination for all A. fumigatus isolates using the canonical mixture response surface Emax-based global model. (a) Three-dimensional scattergram of experimental data. (b) Actual versus predicted growth plot with the linear regression line and its 95% CI limits (broken lines). (c and e) Three-dimensional surface of growth versus voriconazole and anidulafungin fAUC/MIC. The mesh surface corresponds to growth levels if the two drugs were acting additively (additive surface). The solid surface corresponds to growth levels as predicted by the model (response surface). The white and black circles show the main areas of synergistic and antagonistic interactions. (d and f) Three-dimensional interaction surface obtained by subtracting the additive from the response surface. Volumes above and below the 0 plane correspond to synergistic (less growth than expected) and antagonistic (more growth than expected) interactions.
The final model contained eight coefficients, namely αD12 = 1.12 ± 0.40, βD12 = 0.79 ± 0.09, c = 10 ± 3.1, γD12 = 18.7 ± 4.9, αm1 = −8.1 ± 3.8, βm2 = −1.6 ± 0.24, γm12 = 18.4 ± 8.1 and δm12 = 15.6 ± 7.7 (mean ± SEM) and described well the entire response surface (Figure 7a; R2 = 0.95). When the full model was compared with the reduced model, the P value of the F-test was 0.063 indicating that the full model is not significantly better that the simple model. The coefficients of variations of most parameters were <30% and most differences between predicted and observed growth were <20% (Figure 7b and d). Synergistic interactions were found at low anidulafungin (fCmax/MEC <10) and voriconazole (fAUC/MIC <10) exposures, whereas antagonistic interactions were found at higher drug exposures (Figure 7c). The median (range) interaction indices of synergistic and antagonistic combinations were 0.63 (0.38–0.79) and 1.12 (1.04–4.6), respectively.
Monte Carlo analysis
The PTAs for EI50 are shown in Figure 8 for A. fumigatus isolates with different voriconazole MICs and anidulafungin MECs. The PTAs for voriconazole monotherapy were ≥78%, 12% and 0% for isolates with voriconazole MICs ≤1, 2 and ≥4 mg/L, respectively, and increased for all combination therapy regimens against all isolates with different anidulafungin MECs. For isolates with anidulafungin MECs 0.008 and 0.016 mg/L, the largest PTAs were found when voriconazole was combined with 25 mg of anidulafungin, reaching 96%–100%, 68%–82% and 9%–20% for isolates with voriconazole MICs 1, 2 and 4 mg/L, respectively. For isolates with anidulafungin MECs 0.03 and 0.06 mg/L, the largest PTA was found when voriconazole was combined with 50 and 100 mg of anidulafungin, respectively, reaching 100%, 82% and 20% for isolates with voriconazole MICs 1, 2 and 4 mg/L, respectively. The differences in PTA for a collection of isolates with different resistance rates and MIC distributions are shown in Figure 1. The PTA increased up to 6% for the 5%–10% resistance rate and 8%–12% for 15%–25% resistance rate between monotherapy and combination therapy with the highest increase observed at the lowest dose of anidulafungin.
Target attainment rates of voriconazole monotherapy and voriconazole/anidulafungin combination therapy for A. fumigatus isolates with different voriconazole MICs and anidulafungin MECs using a PD target associated with 50% growth. VRC, voriconazole; ANI, anidulafungin.
Therapeutic drug monitoring
The voriconazole and anidulafungin concentrations associated with the EI50 are shown in Figure 9 where the voriconazole tCmin/MIC was plotted against the anidulafungin tCmax for isolates with MECs 0.008, 0.016, 0.03 and 0.06 mg/L. The EI50 can be reached by voriconazole monotherapy by targeting 1.5 tCmin/MIC. Thus, isolates with voriconazole MIC >4 mg/L will require >6 mg/L voriconazole trough levels, which are associated with higher toxicity.30 The EI50 can be attained with combination therapy targeting lower concentrations corresponding to 0.75 tCmin/MIC providing that the anidulafungin tCmax is <0.5, 1, 2 and 4 mg/L for isolates with MECs 0.008, 0.016, 0.03 and 0.06 mg/L, respectively. Thus, with combination therapy the PD target could be attained for isolates with voriconazole MICs up to 4 mg/L without reaching toxic voriconazole concentrations. The 25 mg dose of anidulafungin is sufficient to achieve the maximal reduction of voriconazole levels required for target attainment for the most commonly observed A. fumigatus isolates with anidulafungin MECs 0.008 and 0.016 mg/L. Higher doses of 50 and 100 mg will minimize or reverse this positive effect. The latter doses will provide optimal anidulafungin levels only for isolates with MEC 0.03 and 0.06 mg/L, respectively.
Voriconazole and anidulafungin serum concentrations required to attain the EI50 PD target in monotherapy (black circles) and combination therapy. The largest reduction in voriconazole levels was observed at <0.5, 1, 2 and 4 mg/L of anidulafungin tCmax (white circles) for isolates with MECs of 0.008, 0.016, 0.03 and 0.06 mg/L, respectively. The black, dark grey and light grey arrows indicate the serum levels of anidulafungin after 25, 50 and 100 mg dosing, respectively. The dose of 25 mg is sufficient to achieve the maximal reduction of voriconazole levels required for target attainment for the most commonly observed A. fumigatus isolates with MECs of 0.008 and 0.016 mg/L. Higher doses of 50 and 100 mg will minimize or reverse this positive effect.
Discussion
Exposure- and MIC-dependent interactions were found for the combination of voriconazole/anidulafungin against azole-susceptible and -resistant A. fumigatus isolates in an in vitro PK/PD model simulating human PK. Synergistic interactions were found at low anidulafungin (fCmax/MEC <10) and voriconazole (fAUC/MIC <10) exposures, whereas antagonistic interactions were found at higher drug exposures. By bridging in vitro PK/PD combination data with human PK, the combination therapy of anidulafungin with standard-dose voriconazole (4 mg/kg) resulted in higher PTA than voriconazole monotherapy. The largest increase was found with the lowest dose of anidulafungin (25 mg), for isolates with anidulafungin MECs 0.008–0.015 mg/L and voriconazole MIC distributions with high (>10%) resistance rates. Optimal activity of the combination regimens was found with a serum ratio of 0.75 tCmin/MIC for voriconazole and an anidulafungin tCmax between 0.5 and 4 mg/L depending on the anidulafungin MEC, whereas the equally effective monotherapy regimen required a voriconazole tCmin/MIC of 1.5. This 2-fold reduction in the voriconazole tCmin/MIC ratio allowed target attainment for isolates with voriconazole MICs up to 4 mg/L with lower concentrations than those required for monotherapy regimens to efficiently treat the same isolates. Consequently, therapeutic drug monitoring of antifungal combination therapy can be employed in order to increase its efficacy against resistant isolates with minimal toxicity.
In vitro interactions between azoles and echinocandins are not always predictable. Most combination studies with voriconazole and anidulafungin found indifferent interactions against A. fumigatus,31,32 whereas synergistic interactions have been reported against azole-susceptible and -resistant isolates.33–36 Meanwhile, antagonistic effects have also been observed.35 However, standard chequerboard broth microdilution studies do not account for the fluctuating drug concentrations, the changing relative concentration ratios and the dynamic nature of fungal growth, whereas extrapolation of in vitro combination data to humans is difficult. We therefore employed an optimized in vitro PK/PD closed dialysis/diffusion model to assess interactions between voriconazole and anidulafungin simulating human drug exposures.11,15 The synergistic interactions were observed at low suboptimal drug exposures while antagonism was found at higher drug exposures. A previous in vitro PK/PD study failed to find these interactions, probably because a high inoculum of 105 cfu/mL was used thus saturating the fungal growth detection assay using galactomannan production and an inflexible mathematical model for assessing PD drug interactions was used.31 Although hyphae are the invasive form, a lower inoculum of 103 cfu/mL is more clinically relevant, allows detection of changes in fungal growth and resulted in comparable results with in vivo data.11,28 The mathematical model used in the present study is very flexible to capture complex PD interactions compared with other full response surface models.8,18,19,31 Although anidulafungin monotherapy did not significantly suppress galactomannan production, a significant decrease in galactomannan levels was observed when it was combined with voriconazole compared with monotherapies in concordance with previous in vivo studies.7,36 These changes were verified by quantitative PCR.
The most likely explanation for the synergistic interactions is that anidulafungin increases the entry of voriconazole into fungal cells by disrupting the cell wall integrity rather than that anidulafungin decreases voriconazole efflux by directly interacting with an efflux pump. Alterations on the fungal cell wall may influence the access of antifungal drugs to the plasma membrane and ultimately to the cytoplasm.37 Thus, a minimal concentration of anidulafungin (i.e. the MEC) is required to disrupt cell wall integrity and interact synergistically with voriconazole. Higher concentrations of anidulafungin did not increase the synergistic interactions as would be expected if anidulafungin had directly interacted with voriconazole's efflux pump. This is in line with findings showing a lack of cross-resistance between these two classes of antifungals.38 However, an indirect effect of anidulafungin on voriconazole efflux by disrupting cell wall integrity and thereby plasma membrane stability and efflux pump function cannot be excluded. The antagonism observed at high exposures of anidulafungin could be explained with the paradoxical effect where increased growth is observed at high concentrations of echinocandins, which is caused by the activation of the cell wall salvage pathway(s) that promote chitin synthesis and cell survival.39 This compensatory increase in the chitin content of A. fumigatus is activated in response to concentrations near the MEC.39 The restoration of cell wall integrity by increased chitin production may decrease voriconazole intracellular levels and increase fungal growth, resulting in antagonistic interactions.40 This phenomenon could be enhanced at high voriconazole concentrations since chitin increased after prolonged incubation with azoles.41
In vivo studies of voriconazole/anidulafungin combination in experimental animal models of invasive aspergillosis usually showed increased survival of combination therapy compared with monotherapy regimens.36,42 Dose-dependent interactions were found in a persistently neutropenic rabbit model of invasive pulmonary aspergillosis with synergy observed only at lower anidulafungin dose (5 mg/kg) and antagonism observed at higher anidulafungin dose (10 mg/kg) when combined with voriconazole (10 mg/kg).7 The latter combination was associated with higher galactomannan levels compared with voriconazole monotherapy observed late at the infection as we also found in the present study. Investigating the efficacy of combination therapy in a non-neutropenic murine model of disseminated aspergillosis using a broad range of doses, dose-dependent interactions were found at low doses (<5 mg/kg), which corresponded to low EIs (<5 AUC/MIC) and antagonism at higher drug exposures.43 In the clinical setting, a randomized, placebo-controlled multicentre trial failed to show superiority of combination therapy with voriconazole and anidulafungin for the treatment of invasive aspergillosis in haematological patients (overall survival 72% in monotherapy versus 80% in combination therapy).5 This is in line with the present findings where combination therapy with 100 mg of anidulafungin did not result in significantly higher (1%–2%) PTA than voriconazole monotherapy for commonly observed A. fumigatus isolates with anidulafungin MECs up to 0.016 mg/L, although the difference in the clinical trial was higher (8%). However, as the authors of the clinical trial acknowledged, the 6 week survival was lower than expected in voriconazole monotherapy, which could lead to overestimation of the difference between the two groups.5 An important finding of the present study that merits further clinical investigation is that a lower dose of anidulafungin (25 mg) resulted in higher PTA (6%–12%) than voriconazole monotherapy. The low dose would result in free anidulafungin levels just above the MEC for these isolates, maximizing the synergistic interactions; higher doses would result in antagonistic interactions. This is also compatible with the mechanism of interaction of the two drugs, where concentrations close to the MEC are required to disrupt the cell wall and produce a synergistic interaction, whereas higher concentrations do not further increase the synergistic interactions but produce antagonistic effects. Another interesting finding of the present study is the fact that the largest difference (10%–12%) in PTA was observed with a collection of isolates with high resistance rates (>10%). In the randomized clinical trial of voriconazole/anidulafungin combination therapy, significant difference was found only for a subset of patients with positive galactomannan and radiographic findings (73% for monotherapy versus 84% for combination therapy).5 Since the isolates were not available, it is unknown whether this subset of patients were infected with resistant isolates since many participating clinical centres were from Europe where resistance rates up to 25% were reported.44 The azole resistance rate of >10% determined in the present study was recently proposed by an international expert opinion panel as the cut-off to prompt initial empirical therapy with combination therapy of voriconazole and an echinocandin or liposomal amphotericin B.45
The optimal activity of voriconazole alone was found at exposures corresponding to a tCmin/MIC ratio of 1.5, in agreement with previous clinical studies where a trough/MIC ratio of 2 was associated with near-maximal probability of response.46 The same efficacy can be obtained with combination therapy with a lower tCmin/MIC ratio of 0.75. This is particularly important for the first days of treatment when voriconazole is not at steady-state and therefore subtherapeutic levels may persist for a long time period.3 Combination therapy can minimize this period, since the tCmin/MIC ratio of 0.75 will be reached sooner than the tCmin/MIC of 1.5 required for voriconazole monotherapy. Furthermore, this reduction in the tCmin/MIC ratio is also important for the management of infections with isolates with reduced susceptibility to voriconazole. Isolates with MIC 4 mg/L would require high toxic voriconazole levels (6 mg/L) when voriconazole is given alone,30 whereas for combination therapy isolates with MICs up to 4 mg/L could be treated with non-toxic voriconazole trough levels of 3 mg/L. Therapeutic drug monitoring of combination antifungal therapy can be challenging because of large intraindividual variation in serum levels, serum/tissue penetration and protein binding.
In conclusion, the in vitro combination of simulated human anidulafungin and voriconazole plasma concentration–time profiles revealed exposure- and MIC-dependent synergistic and antagonistic interactions. Low-dose anidulafungin may enhance synergistic interactions for most A. fumigatus isolates and reduce costs of antifungal combination therapy. PTA of combination therapy was significantly higher than those of monotherapy, particularly for centres with azole resistance rates >10%. Finally, therapeutic drug monitoring may increase the efficacy of combination therapy for patients with subtherapeutic serum concentrations infected with azole-resistant A. fumigatus isolates.
Funding
This study was supported by an unrestricted Investigator-Initiated Research Grant from Pfizer Hellas, Athens, Greece.
Transparency declarations
None to declare.
