Fluconazole Pharmacokinetics in Galleria mellonella Larvae and Performance Evaluation of a Bioassay Compared to Liquid Chromatography-Tandem Mass Spectrometry for Hemolymph Specimens

ABSTRACT

The invertebrate model organism Galleria mellonella can be used to assess the efficacy of treatment of fungal infection. The fluconazole dose best mimicking human exposure during licensed dosing is unknown. We validated a bioassay for fluconazole detection in hemolymph and determined the fluconazole pharmacokinetics and pharmacodynamics in larval hemolymph in order to estimate a humanized dose for future experiments. A bioassay using 4-mm agar wells, 20 μl hemolymph, and the hypersusceptible Candida albicans DSY2621 was established and compared to a validated liquid chromatography-tandem mass spectrometry (LC–MS-MS) method. G. mellonella larvae were injected with fluconazole (5, 10, and 20 mg/kg of larval weight), and hemolymph was harvested for 24 h for pharmacokinetics calculations. The exposure was compared to the human exposure during standard licensed dosing. The bioassay had a linear standard curve between 1 and 20 mg/liter. Accuracy and coefficients of variation (percent) values were below 10%. The Spearman coefficient between assays was 0.94. Fluconazole larval pharmacokinetics followed one-compartment linear kinetics, with the 24-h area under the hemolymph concentration-time curve (AUC_{24 h}) being 93, 173, and 406 mg · h/liter for the three doses compared to 400 mg · h/liter in humans under licensed treatment. In conclusion, a bioassay was validated for fluconazole determination in hemolymph. The pharmacokinetics was linear. An exposure comparable to the human exposure during standard licensed dosing was obtained with 20 mg/kg.

INTRODUCTION

The invertebrate Galleria mellonella model has been introduced as a promising tool for screening in vivo virulence and response to treatment of bacterial and fungal isolates (1, 2). In vivo testing in mammalian models remains the cornerstone for predicted treatment response in the clinical situation when clinical treatment data are lacking; however, mammalian experimentation is becoming increasingly challenging due to ethical and financial constraints. Invertebrate models have been introduced as inexpensive and generally accessible screening models. The greater wax moth G. mellonella has proved especially promising, as a precise inoculum and treatment dose can be injected and the infection models can be studied at a human physiological temperature (3). The larvae have an innate immune system with phagocytic cells pivotal to combating fungal infections and produce an ever-increasing recognized array of peptides as a response to infection, much like humans (4). Treatment experiments in the Galleria model have suggested that the model can be useful for invasive fungal infections, as treatment efficacy was correlated with the susceptibility of the infecting species (with intrinsic or acquired resistance mechanisms predicting failure) (5–12). Comparative efficacy studies of different antifungal compounds require a detailed knowledge of the pharmacokinetics (PK) of each antifungal in the larvae in order to ensure selection of appropriate humanized dosage schemes. However, such data are limited and so far are available only for selected antibacterial drugs (13, 14) and for the antifungals only for posaconazole and voriconazole (15). The majority of data on drug concentrations in hemolymph have been obtained using an agar diffusion bioassay, as this method has the advantage of being inexpensive and generally available. Fluconazole is inexpensive, well tolerated, and used extensively for human Candida infections.

Our aim was to describe the pharmacokinetics of fluconazole in the G. mellonella model and to determine a humanized dose for future in vivo experimentation. For this purpose, we first developed and compared a bioassay against a validated liquid chromatography-tandem mass spectrometry (LC–MS-MS) method for measuring fluconazole concentrations in hemolymph. We then described the 24-h time-concentration profiles of various dosages of fluconazole and utilized this to determine a dose resulting in an exposure comparable to the human exposure during standard licensed dosing with fluconazole.

RESULTS

Bioassay establishment and validation.

The bioassay gave a linear standard curve from 1 to 20 mg/liter of fluconazole with a linear-regression coefficient (R²) of >0.99 and a reproducible formula for the standard curve [Y = 9.544 ± 0.243 × ln(x) + 4.428 ± 0.474, where Y is the zone of inhibition in mm] (the mean and standard deviation [SD] from all seven plates was used for validation) using five calibration standards (2, 4, 8, 14, and 20 mg/liter). The lower limit of detection was 1 mg/liter. When testing higher concentrations (25 to 30 mg/liter), a loss of linearity was observed. The results of the inter- and intra-assay validations are summarized in Table 1. For each quality control sample, the intra- and interrun precision and accuracy (coefficient of variability [CV%]) were within the recommended range of ±15%. No matrix effect was observed (data not shown).

TABLE 1.

Intra- and interrun validation in larval hemolymph of the fluconazole agar well bioassay

Nominal FLZa concn (mg/liter)	Validation
	Intrarun (n = 5)		Interrun (n = 5)
	Accuracy (±SD)	Assay precision (CV%)	Accuracy (±SD)	Assay precision (CV%)
2 (LLOQ)	94.0 ± 6.3	6.7	96.2 + 7.1	7.4
5 (QC)	86.0 ± 3.8	4.5	89.6 + 2.5	2.7
9 (QC)	90.9 ± 6.4	7.0	90.1 + 1.4	1.5
15 (QC)	92.7 ± 1.2	1.2	94.3 + 6.3	6.7
All	90.9 ± 3.5	4.9 ± 2.7	92.6 + 3.3	4.6 ± 2.9

^aFLZ, fluconazole.

The bioassay values were 11.5% (confidence interval [CI], 5.3 to 18.1) lower than the LC–MS-MS values (paired t test [log-transformed values for normality], P < 0.001) with absolute bioassay values from +10.2 to −29.7% lower than the LC–MS-MS values (Fig. 1). The methods deviated by an average of 5.4% (range, 0.03 to 17.4%) from the mean of the values from both analysis methods. The Spearman correlation coefficient between the bioassay and the LC–MS-MS samples was 0.94 (P < 0.0001).

FIG 1.

Correlation of fluconazole (FLZ) concentrations on the 23 split samples analyzed by agar well bioassay or LC–MS-MS.

As we saw a loss of linearity in the fluconazole concentration range, including the expected maximum concentration of drug in serum (C_max) in humans, and as at least four larvae were needed to obtain sufficient hemolymph for both the bioassay and LC–MS-MS determination, the latter was chosen for the pharmacokinetic characterization study.

Pharmacokinetics in larval hemolymph.

The mean (±SD) C_max and area under the hemolymph concentration-time curve (AUC) for all larvae treated with 5, 10, and 20 mg/kg of larval weight of fluconazole were 7.55 ± 0.71, 15.52 ± 2.95, and 27.33 ± 2.13 mg/liter and 92.7 ± 20.0, 173.1 ± 34.7, and 406.0 ± 66.1 mg · h/liter, respectively. The doses were linearly correlated with both the observed C_max (R² = 0.97) and AUC from 0 to the last sample time at 24 h (AUC_0–24) (R² = 0.98), with slopes of 19.57 ± 0.81 and 1.41 ± 0.04, respectively (Fig. 2). The time-concentration profile of fluconazole in hemolymph followed a one-compartment model (R² > 0.91; residuals normality test, P > 0.38; runs test, P > 0.52) (Fig. 2). The mean ± standard error of the mean (SEM) of the elimination rate (K_e) and elimination half-life (t_1/2) were 0.066 ± 0.022 h⁻¹ and 11.38 ± 3.7 h, respectively. The derived volume of distribution (V) and clearance (CL) parameters were 0.198 ± 0.015 ml and 0.013 ± 0.003 ml/h for an average larva of 0.287 g, i.e., 0.69 ± 0.05 liter/kg and 0.045 ± 0.012 liter/kg · h⁻¹. Key Galleria and human pharmacokinetic data are compared in Table 2.

FIG 2.

Pharmacokinetics (left) and dose proportionality (right) of 20, 10, and 5 mg/kg of fluconazole in hemolymph of larvae. Data from two independent experiments were analyzed with a one-compartment model. The error bars represent SEM.

TABLE 2.

Fluconazole PK data in larvae in comparison to human data^a

Parameter	Value
	G. mellonella (single dose) [mean (SD)]			Human
				Single doseb	Steady statec
FLZ dose (mg/kg)	5	10	20	1.4	∼5.3d
t1/2 (h)		11.38 (3.7)e		30–35	31–37.2
Vd (liter/kg)		0.69 (0.05)e		0.68	0.7
Mean body CL (liter/kg · h−1)		0.045 (0.012)e		0.017–0.022
Cmax (mg/liter)	7.55 (0.71)	15.52 (2.95)	27.33 (2.13)	2.1–2.8	18.9–30.6
AUC0–24 (mg · h/liter)	92.7 (19.96)	173.11 (34.68)	405.96 (66.09)	31.6–35.9	350
Unbound fraction (%)		NDf			88–89

^aLarval data for V_d and CL were standardized to a weight ratio to allow easy comparison to available human data. All human data values are not exact values but are meant to give an overall impression of corresponding values in humans.

^bSingle-dose data from available sources on intravenous (i.v.) administration of 100 mg fluconazole to healthy volunteers (30, 31). As the bodyweight was given as 69 ± 12 kg for one of the articles, a weight of 70 kg overall has been used.

^cData were compiled from the EUCAST rationale document (32) from available sources on healthy adults (20, 33, 34) on a standard licensed systemic dose of 400 mg/day. Steady state is on average achieved on day 6 to day 10 of dosing, with a 2.5-fold increase in the C_max over this period (20).

^dConsidering a dose of 400 mg/day and a body weight of 75 kg.

^eOne value calculated across the three dosages administered using 1-compartment i.v. pharmacokinetic model for all animals at each dose.

^fND, not determined.

DISCUSSION

The G. mellonella model has the advantage of being inexpensive. Likely for this reason, most pharmacokinetic studies in the larvae have so far been performed by measuring antifungal concentrations with a (low-cost) bioassay (13, 14). The bioassay was associated with some technical challenges. Fluconazole standards prepared in hemolymph produced much smaller zones than same-concentration standards prepared in horse serum (data not shown). We speculate this could be due to differential protein binding in the two matrices (16–18). The standards and quality controls (QCs) for the bioassay were difficult to mix due to the small volumes and clotting at room temperature, requiring mixing on ice. Difficulties in obtaining precise and reproducible determination of zone inhibition diameters were often encountered for the bioassay due to fuzzy zones for larval samples. These were never observed around standards, suggesting the dimethyl sulfoxide (DMSO) solvent in the standards somehow prevented the phenomenon. Finally, the bioassay required larger quantities of hemolymph and had a narrower range of measurable concentrations than the LC–MS-MS method. Samples with an expected concentration above 20 mg/liter would therefore require dilution before measurement, and trough concentrations for lower doses may be under the level for quantification, both reasons for favoring LC–MS-MS for our pharmacokinetic studies. Fluconazole has indeed previously proven rather difficult to assess in human serum, also using bioassays, due to a high lower limit of detection, a finding we confirmed here for hemolymph, and the low volumes of hemolymph obtainable potentiate this issue (19). Despite these challenges, we validated the bioassay for a fluconazole range of 1 to 20 mg/liter and found results using the two analytical methods to be comparable, suggesting that active metabolites of fluconazole are not generated in the larvae. Thus, although LC–MS-MS (or high-performance liquid chromatography [HPLC]) is preferred for fluconazole concentration determination in hemolymph, the bioassay is a valid low-cost alternative.

The Galleria pharmacokinetics of fluconazole was adequately described following one-compartment kinetics and, as in humans, with a linear increase in C_max and AUC_0–24. (20). The distribution was fairly rapid and probably represents the time needed for the hemolymph to circulate through the hemocoel. V in humans is equal to total body water, and the value was almost identical in the larvae. In humans, the primary route of elimination is renal, with 60 to 80% of fluconazole being excreted unchanged in the urine, and the half-life is significantly prolonged in severe renal insufficiency (21, 22). However, despite the lack of kidneys, the larval t_1/2 is only about a third of the human value, showing much greater clearance. The larva has a fat body with some functions matching those of the liver, but whether fluconazole is metabolized or excreted in not known.

Published pharmacokinetic data for posaconazole and voriconazole are scarce but show similarity to those for fluconazole, with a linear dose-AUC_{0–24 h} and C_max relationship (15). However, for these azoles, the AUCs were comparable to human values on a milligram-per-kilogram basis. For antibacterial drugs, drug-dependent differential pharmacokinetics compared to that in humans has been reported. For weight-adjusted doses, the C_max and AUC in the larvae exceed those in humans for several beta-lactam-derived drugs (ceftazidime, cefotaxime, meropenem, imipenem, and piperacillin), but the t_1/2 values are comparable (13, 14). For the fluoroquinolones, there is a higher C_max but differential elimination, with ciprofloxacin (14) having a much longer and levofloxacin (13) a shorter t_1/2 in the Galleria model than in humans. Amikacin parameters were overall comparable between the species, whereas colistin was reported to have a higher C_max and shorter t_1/2 (13). Thus, despite variable PK for some antimicrobials, the fluconazole data corresponded extremely well to C_max and AUC levels in humans achieved after single-dose exposure, whereas the t_1/2 in the larvae was shorter.

Studies of treatment response to various doses of fluconazole have been conducted in the Galleria model and have shown dose-dependent increased survival in susceptible Candida isolates. Overall, the variety of endpoints (CFU per milliliter or mortality), inoculum sizes, virulence of infecting isolates, larval sizes and vendors, doses and dose timing (pre- or postinoculation), and follow-up time (5 to 13 days for mortality; from 16 h to daily for up to 5 days for CFU) employed make direct comparison between studies difficult. Variability in larva vendors, diet, handling, and preexperimental temperature are known to impact susceptibility and treatment response (23–26), and these factors could have impacted the divergent results for the two groups investigating treatment efficacy following infection with Candida albicans SC5314. Nevertheless, postinfection treatment dosages in a range from 11.1 to 16 mg/kg in contrast to ≤4 mg/kg were associated with >20 to 50% better response rates for infections with susceptible isolates of C. albicans, Candida parapsilosis, and Candida tropicalis (5–8), supporting a dosing strategy of 15 to 20 mg/kg fluconazole, in agreement with our findings that 20 mg/liter is needed to obtain a humanized exposure.

Given that we do not know the AUC of the unbound fraction of fluconazole and that the protein content of the larval hemolymph surpassed that of human serum (16, 17, 27), a conservative high larval dose of 20 mg/kg fluconazole is advisable in order to be certain to achieve a humanized exposure in the larvae.

This study has some limitations. The humanized dose calculation presented does not take differential protein binding into account. As the protein content is higher in larval hemolymph, this may lead to an underestimation of the dose needed. We have not assessed the pharmacokinetics in infected larvae, and we have not measured the fluconazole exposure beyond 24 h or after repeated dosing and thus cannot predict to what degree accumulation would occur. Compared to an LC–MS-MS method, the results obtained using the bioassay were on average 11% lower. Although fluconazole is indeed very stable, we cannot rule out the possibility that differences in potency in the fluconazole stock solutions used for the standards in the two assays could have played a role.

In conclusion, despite the limitations mentioned above, we have validated a fluconazole bioassay for small-volume samples in hemolymph and for the first time have provided data on fluconazole PK in the Galleria larva model that will allow selection of humanized dosing in future experiments. As no specialized equipment is needed, the larva model can be implemented in most laboratories interested in in vivo experimentation. Future standardization measures could be directed at increasing reproducibility. Larvae are often available from local bait shops; however, they vary in size and quality. A research grade, guaranteed antibiotic-free, and genome-sequenced inbred colony variety (Biosystems Technology) has recently become available, although at a significantly higher cost, and could potentially increase intra- and interlaboratory reproducibility.

MATERIALS AND METHODS

Antifungals. (i) Larva model.

Fluconazole (Fresenius; 2 mg/ml for intended intravenous use) was employed and diluted as needed in sterile isotonic NaCl (B. Braun Melsungen AG, Melsungen, Germany). For the test of a matrix effect in the bioassay, selected larvae were inoculated with isotonic NaCl alone, phosphate-buffered saline (PBS) (SSI Diagnostica, Hillerød, Denmark), or 20 mg/liter ampicillin (Pentrexyl; Bristol-Meyers Squibb, Virum, Denmark) in PBS (as these agents are often used in the model).

(ii) Bioassay.

Fluconazole stock solutions (1,000 mg/liter; HPLC grade; Sigma-Aldrich, St. Louis, MO, USA) were prepared in DMSO (Sigma-Aldrich, Brøndby, Denmark).

(iii) LC–MS-MS analysis.

Fluconazole stock solutions (5,000 mg/liter; 99% purity; Alfa Aesar, VWR, Leicestershire, United Kingdom) were prepared in LC-MS grade methanol (Fisher Scientific, Loughborough, United Kingdom).

Larva model.

G. mellonella sixth-instar larvae (Minizoo, Copenhagen, Denmark) stored in the dark at 5 to 10°C in wood shavings were used within a week of receipt. Larvae without black spots indicative of disease were allocated into groups, ensuring equal representation of larvae from the following weight classes 250 to 275 mg, 275 to 300 mg, and 300 to 325 mg (the mean larval weight in each group was 286 mg). After a brief wash in 70% ethanol, the larvae were dried on a clean piece of swab, injected in the last left proleg with 10 μl of fluconazole suspension, and incubated at 37°C. The hemolymph was harvested by an incision in the frontal part of the dorsal midline at different time points (0.5 to 24 h). Hemolymph was allowed to drip into 1.3 ml EDTA-coated Eppendorf vials kept on ice. The hemolymph from 4 larvae was pooled when analyzed by both LC–MS-MS and bioassay, and hemolymph from 2 larvae was pooled when analyzed by LC–MS-MS only when concentrations below the limit of detection for the bioassay were expected (10 and 24 h) and for a confirmatory pharmacokinetics study (see below). All samples were immediately frozen at −70°C in order to avoid clotting and discoloration of the hemolymph. The experiment was repeated using same-size larvae from a different vendor (HPReptiles.dk, Copenhagen, Denmark) and 3 samples/time point from each of two larvae (including additional samples after 3 h).

Bioassay.

The azole-hypersusceptible C. albicans DSY2621 mutant constructed by targeted deletions of genes encoding membrane efflux transporters (cdr1Δ::hisG/cdr1Δ::hisG, cdr2Δ::hisG/cdr2Δ::hisG, flu1Δ::hisG/flu1Δ::hisG, and mdr1Δ::hisG/mdr1Δ::hisG) and calcineurin subunit A (cnaΔ::hisG/cnaΔ::hisG-URA3-hisG) (28) (kindly provided by D. Sanglard) was used as the indicator organism; the median fluconazole EUCAST MIC for the strain was 0.03 mg/liter (range, 0.016 to 0.03 mg/liter; 5 repetitions). An inoculum of C. albicans (approximately 10⁶ CFU/ml) was prepared in sterile deionized water, and 40 ml was poured over the surfaces of 200-ml YNBCG agar plates (24.5 by 24.5 mm; SSI Diagnostica, Hillerød, Denmark). Excess inoculum was poured off after 30 min, and the plates were allowed to dry for 30 min. A sterile 4-mm-diameter cork borer was used to make 30 wells in the plates. Overall, nine calibration standards with known fluconazole pure substance concentrations (0.5, 1, 2, 4, 8, 14, 20, 25, and 30 mg/liter) were prepared from a fluconazole stock solution (1,000 mg/liter in DMSO) in naive hemolymph (collected from larvae of the same size used in the experiments). The stock (DMSO) contents of the five standards (2 to 20 mg/liter) included in the validation and final assays were 1.5 to 9%. Standards and test samples were applied in duplicate (20 μl/well). The plates were incubated at 37°C for 18 h prior to measuring the horizontal and vertical diameters of the zone of inhibition from each well using calipers. A standard curve (ln diameter of zone of inhibition versus antibiotic concentration) was made from which the concentrations of fluconazole from larval samples could be interpolated.

Bioassay validation and comparison to LC–MS-MS results.

The bioassay was internally validated using an abridged validation scheme based on the European Medicines Agency (EMA) guideline on bioanalytical method validation (29); in short, a lower limit of quantification (LLOQ) (2 mg/liter) and three QC samples with fluconazole concentrations of 5 (low QC), 9 (middle QC), and 15 mg/liter (high QC) were prepared like the calibration standards (but using an independently produced stock solution) and were each tested on different plates (5 samples each/plate) in order to estimate intraday accuracy (measured value/nominal value × 100) and precision (coefficient of variability [CV%], calculated as follows: standard deviation of measured value/mean measured value × 100) for the bioassay. Furthermore, five samples of each were tested on five different plates on three different days in order to test interday accuracy and CV. The specificity of the assay was determined by testing hemolymph collected from each of four larvae (two samples of naive hemolymph; one sample each from larvae that had received an injection of 10 μl of NaCl/PBS/PBS with 20 mg/liter ampicillin and one sample of naive hemolymph mixed with 10% DMSO). As the samples clotted at room temperature, they were kept in an ice bath when thawed and immediately analyzed. Stability at room temperature or storage at −80°C was not evaluated. Due to the small volume of the samples, freeze-thaw cycles were not tested.

The performance of the bioassay was compared to the validated LC–MS-MS using split samples. Samples from larvae administered 20 mg/kg of fluconazole were excluded (as they were above the range of fluconazole concentrations tested in the bioassay according to the LC–MS-MS results), and selected samples were used before/during the validation, leaving 23 samples for comparison. Each analytical run of the bioassay contained duplicates of the five standards, the three QC standards, and up to seven samples. The results were discarded if any of the QCs were more than 15% from the expected value. Comparison of the performance of the bioassay versus LC–MS-MS was performed using GraphPad Prism 6.04 (La Jolla, CA) and a paired t test of log-transformed values from the two assays. A P value below 0.05 was regarded as significant. The differences between values obtained using the two methods were compared to their mean value. The nonparametric Spearman correlation coefficient was calculated.

LC–MS-MS method.

Fluconazole concentrations were measured using a validated ultrahigh-performance liquid chromatography tandem mass spectrometry method comprising an Agilent 6420 Triple Quad Mass spectrometer and an Agilent 1290 infinity LC system (Agilent Technologies United Kingdom Ltd., Cheshire, United Kingdom). Briefly, fluconazole was extracted by protein precipitation; 300 μl of acetonitrile containing the internal standard (IS) 6,7-dimethyl-2,3-di(2-pyridyl)quinoxaline at 0.01 mg/liter (Sigma-Aldrich, Dorset, United Kingdom) was added to 20 μl of hemolymph. The samples were then vortex mixed for 5 s and centrifuged at 13,000 × g for 3 min. One volume of the supernatant was then mixed with 1 volume of LC-MS grade water (Fisher Scientific, Loughborough, United Kingdom), and 5 μl was injected on an Agilent A Zorbax SB C₁₈ column (2.1 by 50 mm; 1.8-m particle size) (Agilent Technologies United Kingdom Ltd., Cheshire, United Kingdom). Chromatographic separation was achieved using a gradient with the following starting conditions: 90:10 (0.1% formic acid in water as mobile phase A and 0.1% formic acid in acetonitrile as mobile phase B). Mobile phase B was increased to 100% over 1.5 min and then reduced to the starting conditions for 1.5 min of equilibration. The mass spectrometer was operated in multiple-reaction-monitoring (MRM) scan mode in positive polarity. The precursor ion for fluconazole was m/z 307.11, and it was m/z 313.15 for the IS. The product ion for fluconazole was m/z 220.1, and it was m/z 284.1 for the IS. The source parameters were set as 4,000 V for the capillary voltage, 350°C for the gas temperature, and 60 lb/in² for the nebulizer gas. The standard curve for fluconazole encompassed the concentration range of 0.1 to 60.0 mg/liter and was constructed using blank hemolymph. The data were acquired and processed using the Mass Hunter Quantitation B.6.0 software package. For fluconazole in hemolymph, the limit of detection was 0.05 mg/liter and the limit of quantitation was 0.1 mg/liter. The CV% was 10.2% over the concentration range 0.1 to 60.0 mg/liter, and the intra- and interday variation was <10.8%.

Pharmacokinetics.

The single-dose pharmacokinetics of fluconazole were determined following administration of 5, 10, and 20 mg fluconazole/kg of larval weight administered in 10-μl volumes. Samples were collected over 24 h at the following time points: 0.5, 1, 2, 6, and 24 h, with an additional harvest at 10 h for 10 mg/kg fluconazole and at 3 h for experiment 2. In total, 136 samples were included using LC–MS-MS (80 samples from experiment 1 and 56 from experiment 2). One sample (10 mg/kg fluconazole; 24 h) was excluded as it was below the lower limit of quantification (and below all samples after a dose of 5 mg/kg, indicating that the administration had most likely failed). No outliers were removed.

For the pharmacokinetic calculations, the maximum concentration of each dose (C_max) was determined, and the AUC_0–24 was obtained by the linear trapezoidal rule. Dose proportionality was checked with linear regression analysis of doses versus AUC_0–24 and C_max. The concentration-time curves of each dose (D) were analyzed by nonlinear regression analysis using a one-compartment model after intravascular administration as described by the following equation: C_t = C_max × e^−Ke×t, where C_t (the dependent variable) was the concentration of drug at a given time t (the independent variable), C_max was the maximum concentration, and K_e was the elimination rate. In order to calculate the global K_e for all doses and larvae, the parameter was shared among all data sets. The elimination half-life (t_1/2) was calculated using the following equation: t_1/2 = 0.693/K_e. The volume of distribution (V) and clearance (CL) were determined as D/C₀ and V × K_e, where C₀ is the concentration at time zero. Goodness of fit was assessed with visual inspection of curves, R², runs test, and analysis of residuals using the D'Agostino and Pearson omnibus K2 normality test (GraphPad [La Jolla, CA, USA] Prism 6.04).