The Pharmacokinetics and Pharmacodynamics of Micafungin in Experimental Hematogenous Candida Meningoencephalitis

Abstract

Background

Hematogenous Candida meningoencephalitis (HCME) is a relatively frequent manifestation of disseminated candidiasis in neonates that is associated with significant mortality and neurodevelopmental abnormalities. The outcome following antifungal therapy is often suboptimal with few therapeutic options. A limited amount of clinical data suggests that echinocandins may have a role in treatment of HCME.

Methods

We therefore studied the pharmacokinetics and pharmacodynamics of micafungin in a rabbit model of neonatal HCME, and bridged the results to neonates using population pharmacokinetics and Monte Carlo simulation.

Results

Micafungin exhibited linear plasma pharmacokinetics in the range 0.25-16 mg/kg. Micafungin penetrated most compartments of the central nervous system (CNS), but only following dosages >2 mg/kg. Micafungin was not reliably found in cerebrospinal fluid. With few exceptions, drug penetration into the various CNS subcompartments was not statistically different in infected versus non-infected rabbits. A dose-microbiological response relationship was apparent within the brain and near-maximal effect was apparent following dosages of approximately 8 mg/kg. Monte Carlo simulations revealed near-maximal antifungal effect was attained at neonatal dosages of 12-15 mg/kg.

Conclusions

These results provide a foundation for clinical trials of micafungin in neonates with HCME, and a model for antimicrobial bridging studies from bench-to-bedside in pediatric patients.

Introduction

Neonatal hematogenous Candida meningoencephalitis (HCME) represents a relatively common and serious manifestation of disseminated candidiasis in premature infants [1, 2]. This syndrome has been associated with a high rate of short-term mortality and neurodevelopmental abnormalities [1, 3]. Involvement of the central nervous system (CNS) by Candida spp. is likely to occur more frequently than suggested by the rate of positive blood and/or cerebrospinal fluid (CSF) cultures, which has implications for an accurate and timely diagnosis [4]. Consequently, neonatal HCME is a syndrome of significant public health importance.

Currently, there are relatively few therapeutic options for neonatal HCME, and the outcome of antifungal therapy is frequently suboptimal [1]. The echinocandins represent a novel class of antifungal agents, which exhibit broad candidacidal activity [5]. Limited clinical data suggest that echinocandins may have a role in the treatment of neonatal HCME [6, 7]. Micafungin is an echinocandin which been shown to be safe and well tolerated in neonates [8]. Unfortunately, however, there is no a priori information to guide the selection of clinical dosages which are likely to result in a favorable clinical outcome in a high proportion of neonates with HCME. Moreover, there is a paucity of pharmacokinetic and pharmacodynamic studies which directly bridge laboratory animal data to human infections, and to the best of our knowledge there are no such studies for pediatric patients.

Herein, using micafungin as an example, we consider a unique approach by which to identify candidate antifungal regimens which may be suitable for further study for the treatment of neonatal HCME. We defined the pharmacokinetics and pharmacodynamics of micafungin in a rabbit model of neonatal HCME, and bridged these experimental results to neonates using population pharmacokinetic techniques and Monte Carlo simulation. To our knowledge, this is the first pharmacodynamic study which has utilized such an approach to guide the treatment of a life-threatening fungal infection in pediatric patients.

Methods

Organism and model of neonatal HCME

A non-neutropenic rabbit model of neonatal HCME, described elsewhere [9], was used for all experiments. Briefly, female New Zealand White rabbits (Hazleton, Research Products, Inc., Denver, PA), weighing 2.4-3.7 kg were housed individually, and cared for in accordance to the National Institutes of Health guidelines for laboratory animal care and in fulfillment of the American Association for Accreditation of Laboratory Animal Care [10]. Prior to all experiments, a silastic tunneled central catheter was placed under general anesthesia [11], to enable repeated atraumatic venous access.

A well characterized clinical isolate of Candida albicans, NIH-8621 (ATCC MYA-1237), with a micafungin minimum inhibitory concentration (MIC) of 0.125 mg/L [12] was prepared as previously described [9]. The final inoculum was 1×10⁶ organisms per rabbit, which was administered i.v. and designed to establish a non-lethal model of HCME. Micafungin (Astellas Pharma US, Inc., Deerfield IL) therapy was initiated 48 hours post inoculation; seven dosages of drug were administered. All animals were sacrificed 0.5 hours after the seventh dose of micafungin.

Sample collection and preparation for extraction

Blood was collected in heparinized syringes. Tissue samples were obtained at autopsy. Plasma and all tissue samples were stored at -80 °C prior to analysis. Plasma, CSF, vitreous humour and aqueous humour were not processed prior to extraction. Representative samples of choroid and meninges were weighed, and twice and 5 times the volume of rabbit plasma, respectively, were added (to provide enough volume to enable homogenization). Samples of choroid were vortexed for 30 minutes; whereas, the meninges were sonicated for 10 seconds, and vortexed for a further 10 minutes. Both matrices were then centrifuged at 2300 g for 5 minutes and the supernatant submitted to extraction.

Samples of cerebrum, cerebellum, and spinal cord were sectioned in the same manner at each postmortem examination. Resected tissue samples were weighed and four times the volume of ice-cold 50 mM ammonium acetate, pH 4.0 (J.T. Baker, Phillipsburg NJ) was added. Samples were homogenized (PRO-200; PRO Scientific Inc, Oxford, Ct), and then centrifuged at 2300 × g for 5 min. Seven-hundred μL of the resultant supernatant was added to 700 μL of acetonitrile (J.T. Baker, Phillipsburg NJ), vortexed and submitted to extraction. Standards and quality control samples were prepared by adding known amounts of micafungin to the respective matrices: i.e. rabbit plasma (for plasma, choroid, and meninges), untreated rabbit tissue/fluids (for cerebrum, cerebellum, spinal cord, CSF, vitreous humour) and 0.9% saline (for aqueous humor).

Extraction

All samples were extracted using liquid-solid phase extraction using C8 bonded phase extraction cartridges (Varian, Palo Alto, CA) and a vacuum manifold (Analytichem International, Harbor City, CA). For plasma, CSF, meninges, choroid, vitreous humour, and aqueous humour, 1 cc cartridges were used; these were conditioned by completely draining 500 μL of acetonitrile, and then half-draining 1 mL of acetonitrile-ammonium acetate (10:90 v:v). Two-hundred μL of CSF and aqueous humour, and 300 μL of the other matrices were applied to the half-filled cartridges, together with 25 μL of internal standard (anidulafungin 50 mg/L; Vicuron Pharmaceuticals Inc, King of Prussia, Pa). Acetonitrile: ammonium acetate (10:90, v:v) was then added to a final volume of approximately 1 mL, and the sample was drained slowly under vacuum. For cerebrum and cerebellum specimens, 10 cc cartridges were used and these were conditioned as described above. One mL of sample, 25 μL of internal standard and 4 mL of ammonium acetate were added and drained.

Following drainage, cartridges were washed with 1 mL of acetonitrile: ammonium acetate; 10:90, and dried under vacuum. Subsequently, 1 mL chloroform (Mallinckrodt, Phillipsburg, NJ) was added, drained and the cartridges thoroughly dried under vacuum. Micafungin was eluted with 1 mL of acetonitrile: ammonium acetate (70:30 v:v). The eluant was dried in an evaporator (Zymark Corp., Hopkinston, MA) under a stream of air warmed to 40°C, and then reconstituted in 150 μL of methanol (J.T. Baker, Phillipsburg NJ)-ammonium acetate 50:50 (v:v), prior to transfer to microvial inserts for analysis.

Analytical method

Concentrations of micafungin were determined using reversed-phase HPLC (Waters 2695 Separation module). Acetonitrile-ammonium acetate 50:50 (v:v) was used as the mobile phase with an isocratic flow rate of 0.5 mL/min. A 150 × 4.6 mm, 5-μm, C8 column (maintained at 50°C) was preceded by a 5-μm, C8, 7.5 × 4.6 mm column guard (Alltech Inertsil, Alltech Associates, Deerfield IL). The injection volume was 75 μL for all matrices with the exception of CSF, and aqueous and vitreous humour, for which the injection volume was 150 μL. Micafungin and the internal standard, anidulafungin, were detected using UV light (wavelength 271 nm) and eluted between 7.6-8.1 and 12.7-13.1 minutes, respectively.

Standard curves which encompassed the expected experimental range of micafungin concentrations were constructed in their respective matrix. The lower limit of quantification was ≤ 0.075 mg/L for all matrices. The intra- and inter-day coefficients of variation were both <14%.

Pharmacokinetics of micafungin in rabbits

The plasma pharmacokinetics of micafungin were defined in 38 infected rabbits over the course of multiple experiments. There were 4-6 rabbits per dosage group. Sampling began immediately prior to the sixth dose and continued until the time of sacrifice (i.e. between 120-144.5 h post initiation of drug therapy). The pharmacokinetics of 0.25, 0.5, 1.0, 2.0, 4.0, 8.0 and 16.0 mg/kg were studied. Blood was drawn at time 0 (pre-level), 0.25, 0.5, 2, 4, 6, 8 and 24 h following the administration of the sixth dose; a plasma sample was also obtained immediately prior to sacrifice, 0.5 h after the administration of the seventh dose of micafungin.

The penetration of micafungin into the sub-compartments of the CNS was studied in both infected (6 rabbits per group) and non-infected rabbits (3 rabbits per group). The concentrations of micafungin within the cerebrum, cerebellum, meninges, CSF, aqueous humour, vitreous humour and choroid were determined at a single time point, 30 min. following the seventh dose of micafungin (144.5 h post initiation of antifungal therapy).

Pharmacodynamics of micafungin in rabbits

The pharmacodynamics of micafungin were defined over multiple experiments following the administration of 0-16 mg/kg of micafungin to groups of three rabbits. The endpoint was the fungal burden in the brain, which was estimated using quantitative cultures. Samples of the cerebrum and cerebellum, which weighed at least 1 g were obtained from each rabbit and homogenized. The samples from the cerebrum and cerebellum from an individual rabbit were treated as two replicate biopsies, and used to calculate the mean fungal burden within the brain. This approach allowed for sampling of the bulk of the CNS from its two major arterial supplies (internal carotid and vertebral arteries) with the potential limitation of increasing overall variance.

Neonatal pharmacokinetic data

The neonatal study was a phase I, single dose, multi-center open-label, sequential dose escalation trial. The safety data and the results from non-compartmental pharmacokinetic analysis from this study have been reported elsewhere [8]. The neonatal population consisted of 22 premature infants ≤ 40 weeks post-conceptional age, with an average weight of 1.35 kg, who required systemic antifungal therapy. Micafungin was administered at dosages of 0.75, 1.5 and 3.0 mg/kg, and was infused over 30 min. Plasma samples were collected immediately at the end of infusion (30 min) and then 2, 8, 12 and 24 h post the initiation of infusion.

Pharmacokinetic and pharmacodynamic modeling

The pharmacokinetic data from individual rabbits and neonates were modeled using a population methodology with the program Non-Parametric Adaptive Grid (NPAG) with adaptive γ [13]. For both rabbits and neonates, an open two-compartment model, with zero-order time-delimited input and first-order elimination from the central compartment was used. The data were weighted by the inverse of the estimated variance of the respective micafungin assays.

The pharmacodynamic data were modeled using an inhibitory sigmoid Emax model using ADAPT II [14]; the model took the form:

$$\text{Effect}({\log }_{10}\mathrm{CFU}∕\mathrm{g})={\mathrm{E}}_{\mathrm{con}}-\left(\frac{E_{max}\times {\left(\mathrm{exposure}\right)}^{\mathrm{H}}}{{{\mathrm{EC}}_{50}}^{\mathrm{H}}+{\left(\mathrm{exposure}\right)}^{\mathrm{H}}}\right)$$

where E_con is the Candida cerebral density in the absence of therapy, E_max is the maximum effect induced by micafungin, H is the slope function (Hill constant), EC₅₀ is the dose of micafungin which produced 50% of the maximum effect, and exposure is the dose of micafungin or the area under the concentration-time curve:MIC (AUC:MIC) ratio. The data were weighted by the inverse of the observed variance. On the basis of a previous study with caspofungin, the AUC:MIC ratio was used as the pharmacodynamic variable which optimally links micafungin exposure to efficacy [15]. The pharmacokinetic model was used to transform the dose-response to the AUC_0-144.5:MIC-response relationship (i.e. the AUC from the time of drug administration to 144.5 hours post drug initialization).

Bridging the experimental data to neonates: Monte Carlo simulations

Monte Carlo simulations were performed using ADAPT II [14]. The mean parameter values and the full covariance matrix was inserted into subroutine PRIOR of ADAPT II. An additional subroutine (courtesy Dr David D’Argenio, University Southern California) enabled micafungin to be administered to each of the 9,999 simulated neonates on a weight basis (milligram/kg). For each simulated neonate, the weight-based dose of drug in milligrams/kg was converted internally to an absolute dose of micafungin (milligrams) by multiplying by the simulated weight. To mimic the clinical trial, micafungin was infused over 30 min to simulated neonates. The AUC_0-144.5 was determined by integration for each simulated neonate, converted to the AUC_0-144.5:MIC ratio and the value inserted into the inhibitory sigmoid E_max model fitted to the data from rabbits. The mean ± standard deviation of the predicted residual fungal burden for the simulated population receiving 3-15 mg/kg of micafungin was calculated.

Statistical analysis

The tissue concentrations in infected and non-infected rabbits were examined using two-way analysis of variance (ANOVA), with the dependant variable (concentration) classified according to dosage group and infection status.

Results

Model of hematogenous Candida meningoencephalitis

The distribution of C. albicans in the untreated rabbit model of HCME is depicted in figure 1. Infection varied in intensity, depending upon the CNS subcompartment. The mean ± SEM fungal burden in the brain of untreated rabbits, measured as the average of the cerebrum and cerebellar tissue densities obtained at the end of the experiment, was log₁₀CFU/g 3.31 ± 1.34. Notably, the CSF cultures were persistently beneath the limit of quantification, despite established infection within the cerebrum, cerebellum, spinal cord and meninges.

Figure 1.

The intensity of infection within the various subcompartments of central nervous system in untreated rabbits with hematogenous Candida albicans meningoencephalitis at the end of the experiment (8 days post inoculation). The fungal density in the cerebrospinal fluid and aqueous humour was beneath the limit of detection, despite established infection within the meninges and contiguous parenchyma.

Pharmacokinetics of micafungin in rabbits

Micafungin (total concentrations) exhibited linear pharmacokinetics over the dosage range used in this study. The fit of the pharmacokinetic model to the data was excellent with a coefficient of determination of 0.95 after the Bayesian step. The estimates for the mean, median and standard deviation of the model parameters are summarized in table 1. The calculated AUC_0-144.5:MIC values corresponding to each of the micafungin dosage groups are summarized in figure 2H.

TABLE 1.

The estimates for the means, medians, standard deviations and coefficient of variation for each parameter from the population pharmacokinetic models of micafungin in rabbits and neonates

	Vc (liters)	SCL (liters/h)	Kcp (h-1)	Kpc (h-1)	t1/2 (hours)
Rabbits
mean	0.572	0.124	0.685	0.391	10.010
median	0.543	0.130	0.422	0.348
standard deviation	0.141	0.050	1.110	0.440
coefficient of variation (%)	24.650	40.323	162.044	112.532
Neonates
mean	0.340	0.077	3.424	2.635	7.189
median	0.330	0.050	0.580	0.860
standard deviation	0.109	0.065	5.721	6.172
coefficient of variation (%)	32.059	84.416	167.085	234.231

NOTE. V_c: volume of the central compartment; SCL: clearance from the central compartment; K_cp and K_pc: the first-order rate constants connecting the central and peripheral compartments; t_1/2: the elimination half-life. The half-life for rabbits and neonates was calculated using the mean parameter estimates from the respective models.

Figure 2.

The concentrations of micafungin within the sub-compartments of the central nervous system in infected and non-infected rabbits (Panels A-G) and the relationship between AUC:MIC ratio and the dosage of micafungin (Panel H). Panels A-G, data are mean ± SEM. The highest drug concentrations were found within the meninges and choroid. Micafungin was not reliably detected within the CSF. There were no significant differences in the concentrations in infected-versus-non-infected rabbits in the cerebrum (p=0.07), meninges (p=0.09) and choroid (p=0.87). The concentration of drug was significantly higher in the cerebellum (p<0.01) of infected rabbits, although there was a statistically significant interaction between infection status and dosage. The concentration of drug was significantly higher in the aqueous humour (p<0.01) of infected rabbits; whereas, the concentration was higher in the vitreous humour of non-infected rabbits (p<0.01). H, the relationship between the dosage of micafungin and the AUC_0-144.5:MIC value in plasma; the latter represent simulated values obtained from the pharmacokinetic model summarized in Table 1.

Concentrations of micafungin within the various subcompartments of the CNS

Micafungin penetrated most CNS compartments, but only following dosages of >2 mg/kg (figure 2). The highest drug concentrations were found within the meninges and choroid. Micafungin was not reliably detected within the CSF. There were no significant differences in the concentrations in infected-versus-non-infected rabbits in the cerebrum (p=0.07), meninges (p=0.09) and choroid (p=0.87). The concentration of drug was significantly higher in the cerebellum (p<0.01) of infected rabbits; however, there was a statistically significant interaction between infection status and dosage, meaning that it is not possible to claim these differences are entirely due to infection. The concentration of drug was significantly higher in the aqueous humour (p<0.01) of infected rabbits; whereas, the concentration was higher in the vitreous humour of non-infected rabbits (p<0.01).

Pharmacodynamic relationships

An exposure-response relationship was apparent within the brain, with an asymptotic reduction in the fungal burden observed following progressively higher drug exposures (figure 3). The fit of the sigmoid E_max model was acceptable, with a coefficient of determination of 0.61 (p<0.01). Half-maximal effect was achieved at a dosage of 2.15 mg/kg, and near-maximal effect was observed following a dosage of approximately 8 mg/kg. As demonstrated in figure 3, dosages higher than 8 mg/kg produced only marginal incremental effect. The variation around the mean decreased considerably as the dosage increased from 8 mg/kg to 16 mg/kg, suggesting that a true Emax effect was achieved with the latter dose. The AUC_0-144.5:MIC ratio which produced half-maximal effect was 2499 (Figure 3B).

Figure 3.

The inhibitory sigmoid Emax models describing micafungin exposure and effect determined by the residual fungal burden. A, Dose-response relationship; B, AUC_0-144.5:MIC-response relationship. The open triangle in both panels represents a data point which was not fitted; the significantly smaller standard deviation associated with this point meant that the sigmoid Emax model would include this point at the expense of all others.

Neonatal Population Pharmacokinetics

The estimates for the mean and standard deviation for the neonatal population pharmacokinetic model parameters are shown in Table 1. The fit of the model to the data was excellent, with a coefficient of determination of 0.98 for the observed-predicted values after the Bayesian step. The estimates for the neonatal pharmacokinetic model closely approximated those observed in rabbits (table 1). The Bayesian estimates for clearance for each of the 22 neonates are shown in figure 4. Of note, two neonates exhibited rapid clearance of drug, and were responsible the relatively large coefficient of variation for clearance of 84.4% (table 1). The mean parameter values and their dispersions could be recapitulated using a 9,999 patient Monte Carlo simulation, in which a log-normal distribution was assumed for each parameter. Figure 5 shows that the AUC_0-24 at steady state which developed following the administration of 9 mg/kg to 9,999 simulated neonates was comparable to 2 mg/kg administered to children aged 2-17 and adults receiving 150 mg (data from [16] and Gumbo et al, submitted). The % CV was the same for simulated neonates receiving 3 mg/kg and 9 mg/kg.

Figure 4.

A, The Bayesian estimates for clearance for each of the 22 neonates in the population pharmacokinetic analysis. Of note, two neonates exhibited rapid rates of clearance.

Figure 5.

The predicted mean ± standard deviation AUC_0-24 at steady state for neonates receiving 3 and 9 mg/kg, compared with children aged 2-17 years receiving 2 mg/kg and adults receiving a daily dosage of 150 mg. A larger neonatal dosage is required to produce comparable drug-exposure to those predicted on the basis of weight in children and adults.

Bridging and Monte Carlo simulations

The predicted effect for each of the 9,999 simulated neonates receiving 3-15 mg/kg of micafungin is shown in figure 6. The fungal burden in the absence of treatment was log₁₀CFU/g 3.20. The predicted mean ± standard deviation of the residual fungal burden in neonates ranged from log₁₀CFU/g 1.79 ± 0.52 to 0.78 ± 0.35, in neonates receiving 3 and 15 mg/kg, respectively. The simulations showed that with increasing dose, a progressively larger proportion of neonates attained a near-maximal decline in fungal burden. Importantly, however, was evidence of a persistent “tail” of simulated neonates who failed to reduce their fungal burden to near-maximal levels. These neonates had rapid rates of drug clearance, and as a consequence developed less drug exposure and less antifungal effect. The “tail” in the simulations is a direct reflection of the relatively large CV associated with the estimate of neonatal clearance.

Figure 6.

The predicted residual fungal burden in 9,999 simulated neonates receiving micafungin 3-15 mg/kg (Panels A-E). Each bar represents the absolute number (left y-axis) and proportion of the total simulated population (right y-axis) with a given residual CNS fungal burden following antifungal therapy. With progressively higher dosages of micafungin, a greater proportion of the simulated neonates attain near-maximal effect, which is approximately (log₁₀CFU/g) 0.4.

Discussion

This study demonstrates proof-of-principle that an echinocandin is active in the treatment of HCME, and provides a foundation for the design of a definitive clinical trial in neonates. This study also provides a rational basis for the selection of an optimal pediatric dosage of micafungin based upon objective pharmacodynamic endpoints. The methodology employed in this study serves as a paradigm by which the experimental activity of antimicrobial agents for a range of infectious syndromes can be bridged to pediatric patients.

This rabbit model of HCME demonstrates that C. albicans invades most subcompartments of the CNS. Previous models of CNS candidiasis have utilized an intracisternal route of inoculation [17, 18]; the hematogenous route used in the current study has the advantage of mimicking the pathogenesis of human disease in which involvement of the CNS is preceded by (frequently undiagnosed or subclinical) candidemia. By virtue of the relative mass of tissue, the bulk of the infectious burden in this experimental model is confined to the cerebrum, cerebellum and spinal cord. In contrast, however, the fungal burden in the CSF is beneath the limit of quantification, despite established infection within the meninges and contiguous parenchyma. This observation is consistent with previous experimental and clinical data, and underscores the fact that negative CSF cultures do not exclude a diagnosis of HCME [4, 9]. Consequently, the possibility of CNS involvement should be considered in every neonate with candidemia and, if suspected, treated accordingly.

The extent of penetration of the CNS by pharmacological agents is dependent on their ability to traverse the blood-brain and the blood-CSF-barrier; this, in turn, is a function of the presence of specific drug transporters as well as certain physicochemical characteristics of the drug such as molecular weight, extent of ionization and degree of lipophilicity [19, 20]. While there has been a progressive understanding of the pharmacological determinants of penetration into the CNS [20], the demonstration of antimicrobial efficacy at this site ultimately rests with experimental models and clinical data. Here, we demonstrate that micafungin, a 1292.26 Da water soluble compound [21], penetrates most subcompartments of the CNS and achieves concentrations which are sufficient to produce a candidacidal effect. The influence of meningeal inflammation on the penetration of antimicrobial agents into the CNS is variable, and appears to be enhanced for some drugs (e.g. beta-lactams and vancomycin [19, 22, 23]), but unchanged for others (e.g. amphotericin B, itraconazole [24-26]). Importantly, however, the majority of these studies have focused on bacterial meningitis, in which there is diffuse involvement of the meninges rather than the focal lesions observed histopathologically in HCME. Our results do not suggest that infection significantly influences the overall penetration of micafungin into the CNS in rabbits with HCME; nevertheless, it remains possible that there are increased levels of drug within focally abnormal lesions, which are characteristic of HCME. Our study suggests that relatively high plasma micafungin concentrations are required to achieve therapeutic tissue levels, indicating that CNS drug penetration is dependent on the generation of a concentration gradient of sufficient magnitude to drive drug from the plasma into the brain. Given reports that describe the failure of echinocandins to eradicate CNS candidiasis [27, 28], the identification of optimal dosages using pharmacokinetic and pharmacodynamic techniques represents a critical step in assuring effective treatment of fungal infections at this site.

The human population pharmacokinetic data and Monte Carlo simulations allow the experimental findings to be placed in a clinical context. The principal assumption of this process is that the extent of tissue penetration and the pharmacodynamic relationships in neonates and rabbits are the same. The simulations demonstrate that higher neonatal dosages result in a progressively larger proportion of simulated patients achieving near-maximal reduction in fungal burden (figure 6). In the absence of a priori information regarding the appropriate dosage of micafungin for the treatment of neonatal HCME, the following are relevant considerations: (1) a neonatal dosage of approximately 9 mg/kg results in a similar mean AUC_0-24 at steady state to an adult dosage of 150 mg and children aged 2-17 receiving 2 mg/kg (Figure 5); (2) micafungin has a favorable safety profile, suggesting that the use of higher dosages may be possible; in this regard, the use of neonatal dosages as high as 8 mg/kg [29] have been reported, and more recently 12-15 mg/kg have been used (Daniel Benjamin, personal communication); and (3) the simulations suggest that near-maximal effect is observed with neonatal dosages of 12-15 mg/kg (figure 6). Collectively, therefore, these data suggest that an appropriate dosage of micafungin to induce near-maximal effect in a majority of infants with neonatal HCME lies somewhere between 9-15 mg/kg. Providing the safety and linear pharmacokinetics of micafungin in neonates are confirmed at these higher dosages, the most conservative initial approach for clinical studies would be to choose a dosage in the upper portion of this range. In addition, the two neonates with highly accelerated clearance suggest that monitoring plasma levels of micafungin may identify a subset of patients who require dosage adjustment to avoid concentration-dependent therapeutic failure.