Bottlenose dolphins (Tursiops truncatus) can suffer from fungal infections, which can be treated with voriconazole. In common practice, the voriconazole doses are extrapolated from human doses by adjusting for body weight only, because no dose regimen is available yet. Therefore, the aim of this study was to define a dose regimen for voriconazole in bottlenose dolphins. Dolphins treated with voriconazole between November 2005 and September 2015 at the Dolfinarium Harderwijk, the Netherlands, and TDM was performed were included. Voriconazole plasma concentrations were analyzed with a HPLC-UV method. A population pharmacokinetic model was developed using Mw/Pharm in which the elimination rate constant (ke) and apparent total distribution volume (Vd) was calculated. The loading dose was calculated using the central Vd (Vdc). The maintenance dose was simulated in Mw/Pharm using ke and total Vd.
Eleven dolphins were included. Ke was estimated to be 0.0026 ± 0.0007 hour−1 (mean ± SD), Vd 5.3 ± 3.2 l/kg (mean ± SD), and Vdc ranged between 1.8 and 2.8 l/kg (mean 2.2 l/kg). In order to obtain a median top plasma concentration of 5 mg/l, the loading dose was calculated to be 10 mg/kg (range 9.0–14.2 mg/kg) divided in three administrations (3.3 mg/kg every 24 hours). The maintenance dose was calculated to be 4 mg/kg once a week (range 1.7–6.0 mg/kg); 17% of the dolphins did not reach the therapeutic window of 1–5 mg/l without TDM. A TDM-guided dosing algorithm was developed in order to obtain therapeutic plasma concentrations in this population.
Bottlenose dolphins (Tursiops truncatus), in this article referred to as “dolphins,” in human care can suffer from fungal infections, for example, Candida and Aspergillus species. Voriconazole, an azole antifungal, is commonly used to treat these infections. Azoles exert their effect by inhibition of the fungal cytochrome P450 51p (CYP51p) enzyme resulting in an inhibition of the ergosterol synthesis, an important component of fungal cell membranes.1,–4
In humans, therapeutic drug monitoring (TDM) is required for monitoring of voriconazole therapy due to non-linear pharmacokinetics and relative large inter-patient variability in order to prevent underdosing or toxicity.1,5 Obtaining therapeutic concentrations increases with subsequent TDM measurements.6 Voriconazole adverse events, mainly visual disturbances and hepatic dysfunction, are directly related to the plasma concentration. In humans the risk of visual adverse events is 18% at a plasma concentration of 1.0 mg/l and every increase in plasma concentration by 1 mg/l, increases the risk by about 4.7%.5 Hepatotoxicity increases by 7–17% in case the plasma concentration increases with 1 mg/l.7
The efficacy of azoles is correlated with the ratio of the area under the curve and the minimal inhibition concentration of the fungus for the drug (AUC/MIC ratio).1,5,8 For a maximal effect of voriconazole, the AUC/MIC ratio should be between 20 and 25.1,8 The plasma trough concentration is supposed to reflect the AUC/MIC efficacy and should be at least 1 mg/l with an observed maximal response at 3–4 mg/l.1,5 The rate of treatment success in humans is about 70% at a trough concentration of 1 mg/l.9
Since most Candida and Aspergillus pathogens are (in in vitro assays) susceptible to a 0.5–1 mg/l voriconazole concentration, 1 mg/l can be chosen as minimal trough concentration.8,9 As a result of the plasma protein binding (in humans about 50%), though, a trough of 2 mg/l can be advised to obtain a 1 mg/l free plasma concentration.5,9 Furthermore, a trough concentration of 1 mg/l can compare intra- and interpatient differences.9 At last no sufficient evidence is available with other pharmacokinetic or pharmacodynamic parameters to better predict the efficacy and safety of voriconazole.9
Based on the efficacy/risk balance, the therapeutic window of voriconazole is 1–5 mg/l for treatment of fungal infections.1 Evidence-based target values for prophylactic use of voriconazole are not established since this has only been investigated in small studies.5,10,–12 Therefore, the same lower limit is used for prophylactic and therapeutic use.11
So far, there are no published specified voriconazole dose regimens for bottlenose dolphins. In common practice cetacean doses are often estimated by extrapolation of human doses, which are adjusted for body weight. However, for various drugs differences in pharmacokinetics, in terms of half-life and distribution volume, have been observed between humans and these species.13,14 Extrapolation of human doses without monitoring might result in underdosing or toxicity.13,14 The aim of this study is to define a dose regimen for voriconazole in bottlenose dolphins.
Bottlenose dolphins (Tursiops truncatus) at the Dolfinarium Harderwijk, the Netherlands, that received voriconazole for treatment of fungal infections and where TDM was applied between November 2005 and September 2015 were included. Dolphins were included for the pharmacokinetic analysis and determination of the loading and maintenance dose, if TDM was performed on at least two blood samples. In addition, dolphins were included for defining the loading dose if the loading dose was administered in three administrations and the first TDM occurred within seven days after treatment initiation. Dolphins with TDM performed on at least one blood sample were used for the internal validation. If possible, customary diet was maintained during the voriconazole treatment.
The data originated from November 2005 till September 2015, and the external validation occurred in November–January 2016. The data analysis was conducted from September 2015 until February 2016 at the University Medical Centre Utrecht (UMCU), the Netherlands. The animals received voriconazole tablets (Vfend from Pfizer, Capelle aan den IJssel, the Netherlands) orally, adjusted for cetacean body weight ranging from 0.85 to 7.7 mg/kg. The drugs were administered through fish. The first fish served as a food acceptance control, while the second one contained the voriconazole tablets, which is a general accepted method to administer drugs to cetaceans. All animals were trained to facilitate blood sampling.
Blood was drawn from the fluke by a qualified veterinarian. Blood samples were drawn for routine clinical chemical and hematological monitoring as well as for voriconazole monitoring. Not every blood sample was collected prior to a subsequent voriconazole administration.
Whole blood was collected in citrate, heparin, EDTA and serum tubes depending on the routine clinical chemical and hematological laboratorium assays there were required. The blood was taken to the Dolfinarium laboratory. The serum tubes were coagulated for an hour and afterwards they were centrifuged at 4500 rpm for 15 minutes. Thereafter, the serum was transferred to another tube and stored at −20°C if not immediately transported to an analyzing laboratory. At least 0.5 ml serum was shipped for analysis at ambient temperature.
The serum samples were analyzed using a validated assay for voriconazole at the Clinical Pharmaceutical and Toxicological Laboratory of the UMCU, using high performance liquid chromatography (HPLC) with ultraviolet detection. Since serum was used for the analysis the type of collection vial (citrate, heparin, and/or EDTA) did not influence the results of the assay. A voriconazole reference substance was obtained from Sigma-Aldrich (Zwijndrecht, the Netherlands). Standards were prepared in newborn calf serum; stock solutions were prepared in methanol. Zinc sulfate and acetonitrile were added to 0.25 ml serum for protein precipitation. Methaqualon, obtained from Duchefa (Haarlem, the Netherlands), was added as the internal standard. Samples were centrifuged at 10.000 rpm for 5 minutes and afterward 50 microliters were injected into the HPLC system. A Zorbax-Eclipse XDB-C18, 4.6×150 mm, 5 micrometer column coupled to a diode array detector (Agilent Technologies 1260 infinity G2412B) was used to analyze the blood samples at a wave length of 254 nm. The range of linearity of the assay was 0.5–12.0 mg/l.
The unbound fraction of voriconazole was analyzed using a validated ultrafiltration assay for unbound fractions. 0.5 ml serum was filtered with a Centrifree YM-30 filter (cat. 4104, Amicon). The samples were transferred to a modificated block and centrifuged for 30 minutes at 25°C and 2500 rpm. When finished, the samples were treated as serum samples.
Population pharmacokinetic analysis
The collected voriconazole doses and plasma concentration data were processed in MW/Pharm version 3.60 (MediWare, Groningen, the Netherlands). The drug was entered as an oral dose administration (D) with a bioavailability (F) of 100% (D/F) in MW/Pharm, because of the lack of information about the bioavailability since no voriconazole peak concentrations were obtained or literature values are available. If a cetacean got multiple treatment cycles with voriconazole, one data set per cycle was established.
A population pharmacokinetic model was developed. An iterative two-stage Bayesian (ITSB) procedure in KinPop was performed to estimate the population mean and standard error of the elimination rate constant (ke) and the apparent total volume of distribution (Vd/F). The pharmacokinetic parameters were performed as a one-compartment model since only trough levels were analyzed. In order to asses a two-compartmental model, samples during the distribution phase would have been required. We did not draw blood during the distribution phase because we considered this unethical since these values would not have attributed to dose adjustments for the particular dolphin. Pharmacokinetic parameters were expected to have a log-normal distribution. The stop criterion was 0.0001. Ke was used to determine voriconazole half-life (t1/2) with the pharmacokinetic equation below.15 Ln2 is the natural logarithm of 2 (i.e., 0.693).
Estimation of loading and maintenance doses
With the aid of a simulation in MW/Pharm, the loading dose peak concentration was determined and used to calculate the central distribution volume (Vdc) (l/kg). Vdc was calculated for dolphins with at least two measured plasma concentrations, from which one in the first seven days after treatment initiation, and a loading dose divided in three administrations. The expected ideal loading dose (mg/kg) was calculated with the aid of the mean central distribution volume multiplied by the maximal plasma concentration of 5 mg/l. The equation used for Vdc was:
The maintenance dose was estimated by a simulation in MW/Pharm for each dolphin with at least two measured plasma concentrations based upon the new population model. The voriconazole steady state trough concentration was aimed to reach 3 mg/l in the simulations. The mean of these maintenance doses was calculated and used in a next simulation in which dolphins with at least one plasma concentration were included to validate the loading and maintenance dose in a 2–4 mg/l range and the therapeutic window of 1–5 mg/l. The dosing regimen was externally validated in one dolphin. This dolphin received voriconazole based upon the new dose regimen in the period of November 2015 until February 2016.
Hepatic and visual adverse events
Determination of liver enzymes was not part of the study protocol. However, among other parameters, aspartate aminotransferase (AST) and alanine aminotransferase (ALT) (transaminases) were routinely determined to check the health status of the dolphins. In case liver enzymes were determined at the same time as voriconazole concentrations, these measurements were used in this study.
Visual adverse events were assessed by checking the responses of the dolphins on signs of their trainers. If the dolphins did not react well on visual signs, this could be due to visual adverse events. Another sign of visual adverse events is squeezing with the eyes.
In total, 11 dolphins (dolphin #1–11) were included. Dolphins #9.2, #10, and #11 were excluded for the pharmacokinetic model development and determination of the loading and maintenance dose due to the fact that only one TDM analysis was performed. No co-medication that clinically significantly altered the pharmacokinetic profile of voriconazole was administered. Dolphin #6 had six treatment cycles with voriconazole from which two cycles were included (ID 6.1 and 6.2) in the pharmacokinetic model. Other cycles lacked data about voriconazole plasma concentrations. Though included in the dose regimen validation, no Bayesian fitting was possible in dolphins #9.2, #10, and #11. Dolphin #12 served as an external validation of the model. Characteristics of the studied dolphins are shown in table 1.
Population pharmacokinetic analysis
Dolphins #1 to #9.1 were included in the population model obtained with MW/Pharm. These dolphins all had two or more plasma concentrations available. Population parameter estimates are shown in table 2. The population half-life was estimated to be 267 hours, around 11 days. Dolphins #2, #3, and #9.1 were used to estimate the central distribution volume since these dolphins had a blood withdrawal relatively short after administration of the loading dose. This value ranged from 1.8 to 2.8 l/kg (mean 2.2 l/kg).
The plasma protein binding (%) of voriconazole was determined in serum of dolphin #12. The total (free and bound) voriconazole concentration was 3.88 mg/l and free voriconazole concentration was 1.56 mg/l. Thus, 40% of the voriconazole was free, resulting in a plasma protein binding of 60%.
Loading and maintenance dose
A maximal desired plasma concentration of 5 mg/l and the central distribution volume of dolphins #2, #3, and #9.1 were used to establish the loading dose. The mean loading dose (11.1 mg/kg, range 9.0–14.2 mg/kg) was rounded down to 10 mg/kg divided in three administrations (3.3 mg/kg every 24 hours). Repetitive simulations performed in the population pharmacokinetic model with different doses established an optimal weekly maintenance dose of 4 mg/kg (mean 3.8 mg/kg/week, range 1.7–6.0 mg/kg/week), started one week after the loading dose, in order to obtain a median through concentration of 3 mg/l in steady-state.
Internal validation of dose regimen
The loading (10 mg/kg in three doses) and maintenance (4 mg/kg/week) dosing regimens, rounded to the nearest available 50 mg formulation, were internally validated in dolphin ID 1–11 (dolphin 6 and 9 got two simulations, one for each period) by a simulation in MW/Pharm. In the simulation, 38% of the dolphins did not reach the 2–4 mg/l range and 15% did not reach the therapeutic window. Validation results are shown in table 3. In which total loading dose in three days, maintenance dose and estimated day 17 (first day of the treatment counted as 1) and steady state trough concentrations are shown.
|Rounded* total loading||Rounded* maintenance||17 days through||Steady state trough|
|Dolphin number||dose (mg)||dose (mg/week)||concentration (mg/l)**||concentration (mg/l)|
|Rounded* total loading||Rounded* maintenance||17 days through||Steady state trough|
|Dolphin number||dose (mg)||dose (mg/week)||concentration (mg/l)**||concentration (mg/l)|
*10 mg/kg loading dose and 4 mg/kg maintenance dose rounded to the nearest available 50 mg formulation.
**First day of the treatment counted as 1.
External validation of dose regimen
The dose regimen was tested in dolphin #12, which had a relatively long treatment cycle with multiple through-level blood samples taken. Dolphin #12 received multiple other drugs simultaneously with voriconazole. However, in literature, no clinical relevant interactions were found for voriconazole with these drugs. The treatment started at November 12, 2015. The maintenance doses scheduled at November 28 and December 20, 2015, were shifted to November 29 and December 21, 2015, respectively, because the dolphin did not accept the food given. The trough concentration of November 28, 2015 (day 17) was 3.9 mg/l. Since December 6 the maintenance dose was adjusted to 3 mg/kg (650 mg) per week to obtain a lower (effective) drug exposure. Trough concentrations at December 21 and December 28, 2015, were 2.63 and 2.56 mg/l, respectively. In February 2016 no signs of a fungal infection were present anymore, and the treatment was discontinued.
Hepatic and visual adverse events
In 26 voriconazole samples transaminases were also analyzed. The relationship between voriconazole plasma concentration and transaminases level is shown in figure 1. In only 10% (1/10) of the samples with a plasma concentration of <5 mg/l, elevated transaminases (AST and/or ALT) were observed. In 69% (11/16) of the samples with a plasma concentration >5 mg/l, elevated transaminases were observed. The incidence of transaminases level elevation increases at higher voriconazole plasma concentrations, especially at concentrations >5 mg/l. Normal transaminases levels were observed at voriconazole plasma concentrations of 0–8 mg/l. A χ2 analysis was performed on both groups but lacked significance (P > .05).
No visual adverse events were observed in dolphins 1–10. During the treatment of dolphin #12, the dolphin started to squeeze with his eyes around day 34. At this time, voriconazole plasma concentration was simulated at 3.8 mg/l.
Therapeutic drug monitoring
Based on the pharmacokinetic model and therapeutic window a dosing algorithm was developed which incorporated TDM. In order to obtain therapeutic plasma concentrations of voriconazole, TDM is required. To avoid long-term underdosing or toxicity, a voriconazole trough has to be taken after the first maintenance dose of the treatment. Based on this through concentration different advices are given, as shown in figure 2.
The aim of this study was to define a dose regimen for voriconazole in bottlenose dolphins. The present study showed the need for a loading dose, because of the long voriconazole half-life in dolphins. The loading dose is given to reach a therapeutic concentration within several days, instead of weeks. This study demonstrated a 10 mg/kg loading dose, given in three administrations (3.3 mg/kg every 24 hours). A divided loading dose was based upon earlier experiences and the fact that it's unknown whether toxicity and absorption rate are related to each other. In humans a loading dose is administered in multiple administrations as well, according to the instructions in the package leaflet.16
The maintenance dose was set at 4 mg/kg/week in one administration. The maintenance dose is low compared to humans, horses and dogs (all three species receiving 6 mg/kg/day) and pigeons (20 mg/kg/day) and is probably due to the half-life in dolphins.8,17,–19 Therefore, to avoid toxicity by voriconazole accumulation the dosing interval is set to one week. Doses have to be rounded with 25 mg, because only 50 and 200 mg tablets are available.
For establishing the maintenance dose, a plasma concentration of 3 mg/l was aimed where concentrations within the range of 2–4 mg/l were accepted because of relative high inter-dolphin variability in the model. This same method was used by Mitsani et al.11 Observed plasma protein binding was 60%, more or less the same as in humans (58%).1,8 Due to the 60% protein binding a minimal plasma concentration of 2 mg/l is required in order to obtain a free concentration of 1 mg/l in the blood. It is important to mention that in the simulations, 38% of the dolphins did not reach the 2–4 mg/l range and 15% did not reach the therapeutic window with this maintenance dose. Therefore, in order to increase the number of plasma concentration within the therapeutic window, therapeutic drug monitoring is required.
Based upon observed hepatic and visual side events an upper limit of the therapeutic window of 5 mg/l seems applicable in dolphins too. However, because of the small study population, further research is necessary to confirm the upper limit.
In this study the transaminases were used to predict hepatotoxicity of voriconazole. Though ALT levels seem to have no significant relationship with voriconazole concentrations in humans, ALT levels seem to be predicable for hepatic damage of voriconazole in dolphins.5 Elevated AST, bilirubin, and alkaline phosphatase (ALP) do have a significant relationship with voriconazole concentrations in humans, but no voriconazole toxicity cut-off concentration has been determined.5 In dolphins AST seems to have a relationship with voriconazole concentrations as well. ALP, however, is probably not a specific marker for hepatic damage inflicted by voriconazole because this parameter can also be elevated in healthy dolphins without liver injury.
The visual side events of dolphin #12 might be caused by voriconazole through blocking of the ON-bipolar cells.20,21 The underlying mechanism might be a blockade of transient receptor potential melastatin 1 (TRPM1) in the metabotropic glutamate receptor 6 (mGluR6)-TRPM1 pathway.21 The visual side effects are reversible and disappear when voriconazole treatment is discontinued or plasma concentrations decrease.20,21
Differences in gastro-intestinal anatomy between humans and dolphins makes it difficult to predict or extrapolate voriconazole absorption in dolphins. For example, the absorption may be delayed because food will not be digested immediately in dolphins. Dolphins have three stomachs with a mean pH of <2.0.22,23 The forestomach serves as a food storage, the mean digestion takes place in the second stomach, and the third stomach, the pylorus, regulates the flow of food into the intestines.22 Another problem might be the simultaneously administration of voriconazole and food. In humans the voriconazole absorption will be decreased and delayed when taken with food.1,5,8
Half-life was around 11 days in dolphins and about six hours in humans.8 Sharma et al. describe cases in which relative large animals need lower doses compared to smaller animals.14,24 Large animals have a lower rate of metabolism, because they are not able to lose heat like smaller animals.24 It is recommended to adjust doses for size-independent factors too.24 Since dolphins are larger than humans, it can be expected drugs have a prolonged half-life in dolphins.
The prolonged voriconazole half-life might be explained by cetacean evolution as well, because cetaceans evolved from terrestrial mammals.25 These mammals evolved CYP-enzymes to detoxify phytochemicals.26 Modern cetaceans, however, do not need CYP-enzymes anymore as they do not have a vegetable diet.
A second evolutionary explanation for a prolonged voriconazole half-life might be a reduced binding capacity of cetacean CYP-enzymes. Camels, which share a common ancestor with cetaceans, have a low evolution rate and binding capacity in their CYP1A1, CYP2C, and CYP3A-enzymes compared to humans.25,27 This reduced binding capacity can lead to a possible toxicity, because of the lower rate of metabolism compared to humans.27 Because of the common ancestor, it is plausible cetaceans have a reduced binding capacity and evolution rate of their CYP-enzymes too compared to humans.
First of all, this study was limited by the small study population. This might have led to a relatively high standard error of the pharmacokinetic parameters. Second of all, no bioavailability or absorption data of voriconazole in dolphins was available. Third of all, the population pharmacokinetic model was externally validated in one dolphin. Last of all, TDM did not occur at fixed times. Therefore, not every dolphin could be included in the pharmacokinetic analysis and the determination of the central distribution volume. This could have led to larger standard errors on the pharmacokinetic parameter estimates.
The aim of this study was to define a dose regimen for voriconazole in bottlenose dolphins. This study shows a therapeutic concentration can be reached by a 10 mg/kg loading dose in three administrations (3.3 mg/kg every 24 hours) at day 1, 2, and 3 of the treatment and a 4 mg/kg/week maintenance dose from day 10 of the treatment. However, due to the relative large interindividual variability in dolphins, TDM is still mandatory in order to prevent underdosing or toxicity of voriconazole in dolphins. Adverse events have to be taken in account as well by monitoring visual adverse events and liver injury (ALT). Further research is necessary to confirm this dose regimen.
Declaration of interest
The authors report no conflicts of interest. The authors alone are responsible for the content and the writing of the paper.