Autoinhibitory properties of the parent but not of the N‐oxide metabolite contribute to infusion rate‐dependent voriconazole pharmacokinetics

Abstract

Aims

The pharmacokinetics of voriconazole show a nonlinear dose–exposure relationship caused by inhibition of its own CYP3A‐dependent metabolism. Because the magnitude of autoinhibition also depends on voriconazole concentrations, infusion rate might modulate voriconazole exposure. The impact of four different infusion rates on voriconazole pharmacokinetics was investigated.

Methods

Twelve healthy participants received 100 mg voriconazole intravenous over 4 h, 400 mg over 6 h, 4 h, and 2 h in a crossover design. Oral midazolam (3 μg) was given at the end of infusion. Blood and urine samples were collected up to 48 h. Voriconazole and its N‐oxide metabolite were quantified using high‐performance liquid chromatography coupled to tandem mass spectrometry. Midazolam estimated metabolic clearance (eCLmet) was calculated using a limited sampling strategy. Voriconazole‐N‐oxide inhibition of cytochrome P450 (CYP) isoforms 2C19 and 3A4 were assessed with the P450‐Glo luminescence assay.

Results

Area under the concentration–time curve for 400 mg intravenous voriconazole was 16% (90% confidence interval: 12–20%) lower when administered over 6 h compared to 2 h infusion. Dose‐corrected area under the concentration–time curve for 100 mg over 4 h was 34% lower compared to 400 mg over 4 h. Midazolam eCLmet was 516 ml min^–1 (420–640) following 100 mg 4 h^–1 voriconazole, 152 ml min^–1 (139–166) for 400 mg 6 h^–1, 192 ml min^–1 (167–220) for 400 mg 4 h⁻¹, and 202 ml min^–1 (189–217) for 400 mg 2 h^–1. Concentration giving 50% CYP inhibition of voriconazole N‐oxide was 146 ± 23 μmol l^–1 for CYP3A4, and 40.2 ± 4.2 μmol l^–1 for CYP2C19.

Conclusions

Voriconazole pharmacokinetics is modulated by infusion rate, an autoinhibitory contribution voriconazole metabolism by CYP3A and 2C19 and to a lesser extent its main N‐oxide metabolite for CYP2C19. To avoid reduced exposure, the infusion rate should be 2 h.

What is Already Known about this Subject

• Voriconazole target plasma trough concentrations range between 1.5 and 4.5 μg ml^–1 and therapeutic drug monitoring is useful to achieve target concentrations.

• Voriconazole is a substrate and inhibitor of CYP3A4 and CYP2C19 and thus an autoinhibitor of its own metabolism.

• Dose‐dependent autoinhibition and saturation of metabolism have been proposed as underlying mechanisms for voriconazole's nonlinear pharmacokinetics.

What this Study Adds

• In addition to dose dependence, voriconazole single dose pharmacokinetics is also infusion rate‐dependent.

• The voriconazole N‐oxide metabolite exhibits inhibitory properties for CYP2C19 and CYP3A4, albeit only at higher micromolar concentrations.

• Longer duration of infusion or continuous infusion is not recommended by our data.

Tables of Links

TARGETS
Enzymes 2
CYP3A
CYP2C19

LIGANDS
Midazolam

These Tables list key protein targets and ligands in this article that are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 1, and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 2.

Introduction

The azole antimycotic voriconazole improves clinical outcome and reduces mortality in patients with invasive fungal infections. Analyses of the exposure–response relationship indicate that optimal target plasma trough concentrations range between 1.5 and 4.5 μg ml^–1 3 and therapeutic drug monitoring is recommended to maximize therapeutic success 4. Therefore, detailed knowledge is necessary to determine the optimal dose administration schedule. Voriconazole clearance depends on the activities cytochrome P450 (CYP) isozymes 3A, 2C19, 2C9 and flavin monooxygenase 3 as identified in vitro. Up to 6% of voriconazole is eliminated by the kidneys. The main circulating metabolite is voriconazole N‐oxide, while in urine the N‐oxide as well as hydroxy‐voriconazole, dihydroxy‐voriconazole, 4‐hydroxy‐voriconazole and their glucuronidated phase II‐metabolites are found 5, 6. The bioavailability of therapeutic 400 mg voriconazole doses is high 7. Therefore, switching the route of administration is possible without dose adaptation 8. Voriconazole's pharmacokinetics are nonlinear with dose‐dependent bioavailability and dose‐dependent elimination half‐life 9, 10. The bioavailability of smaller doses is considerably lower, e.g. 39% for subtherapeutic 50 mg doses 9. Voriconazole exposure increases 40‐fold when increasing the oral dose 8‐fold (from 50 to 400 mg) 9, and 3.8‐fold, when doubling a 200 mg dose 10. This has been attributed to saturation of presystemic and systemic voriconazole metabolism at therapeutic doses 8. In our recently published study we presented evidence for dose‐dependent CYP‐mediated autoinhibition as mechanism for dose nonlinearity 9 as it is the case for autoinhibitors such as telithromycin 11. In vitro, voriconazole K_i for competitive inhibition of CYP3A4‐mediated midazolam metabolism is 0.66 μmol l^–1 and 2.97 μmol l^–1 for noncompetitive CYP3A4 inhibition 12. Furthermore voriconazole is a potent competitive inhibitor of CYP2B6 (K_i < 0.5 μmol l^–1), CYP2C9 (K_i = 2.79 μmol l^–1), and CYP2C19 (K_i = 5.1 μmol l^–1) 12. Concentration giving 50% inhibition (IC₅₀) reported for of CYP2C9‐mediated tolbutamide metabolism was 8.4 μmol l^–1, for inhibition of CYP2C19 mediated S‐mephenytoin metabolism 8.7 μmol l^–1 13. Others found a solely competitive inhibition type for CYP3A4 with a K_i of 0.15 μmol l^–1, and CYP3A5 with a K_i of 0.2 μmol l^–1 14. A study building a semiphysiological model determined a K_i of 0.34 μmol l^–1 15. Voriconazole does not influence CYP1A2, 2E1 or 2D6 activity 16 CYP3A4, 2C9 and 2C19 dependent metabolism has been demonstrated in human liver microsomes (HLMs), among recombinant CYP isoforms CYP2C19 and CYP3A4 had voriconazole N‐oxidation activities, but not CYP2C9. Apparent Michaelis constant (K_M) and maximum velocity (V_max) values of CYP2C19 and CYP3A4 for voriconazole N‐oxidation were 14 μmol l^–1 and 0.22 nmol min^–1 nmol^–1 CYP2C19 and 16 μmol l^–1 and 0.05 nmol min^–1 nmol^–1 CYP3A4, respectively 17.

In vivo, the extent of CYP3A inhibition depends on route of administration and dose. Coadministration of a 400‐mg oral voriconazole dose increases the exposure of the CYP3A marker substrate midazolam given orally by 595% to 885% 9, 18. When coadministered with 400 mg intravenous (IV) voriconazole, oral midazolam exposure increases only by 294% 9. Similarly, an oral 50 mg voriconazole dose leads to an 84% increase of oral midazolam exposure while 50 mg IV has no effect 9. Even when being a CYP3A inhibitor in itself, other strong inhibitors of CYP3A can further increase voriconazole exposure as demonstrated by the decrease in clearance following the administration of ritonavir 19. Voriconazole clearance is also dependent on CYP2C19 as demonstrated by the 41% increase of AUC at steady‐state when coadministering omeprazole 20 Because CYP2C19 is polymorphic, voriconazole clearance is CYP2C19 genotype dependent in vivo as shown in a 66% lower clearance in poor metabolizers (PMs; *2/*2 allele) compared to extensive metabolizers (EMs; *1/*1 allele) and a genotype‐dependent bioavailability (95% in CYP2C19 EM, and 75% in PM) 21. When CYP3A was inhibited in CYP2C19 PMs, clearance dropped from 158 ± 54 to 22 ± 11 ml min^–1 19.

Because voriconazole systemic clearance (CL) and CYP3A inhibition are dose‐dependent 10, the question arises whether its exposure also depends on rate of infusion. We hypothesized that lower infusion rates would lead to less autoinhibition and hence supraproportionally lower peak plasma concentrations and / or exposure. If correct, then from a clinical point of view the highest possible and safe infusion rate should be chosen or in cases of low plasma concentrations, an increase of infusion duration could be attempted. From a mechanistic point of view concentration dependence of CYP‐mediated autoinhibition is in line with a competitive inhibition mechanism. We therefore designed a trial to investigate the pharmacokinetics of voriconazole and its N‐oxide metabolite in healthy volunteers after 400 mg doses administered IV over 2 h, 4 h and 6 h and compared it to 100 mg voriconazole administered over 4 h.

Materials and methods

After approval by the responsible Ethics Committee of the Medical Faculty of Heidelberg University and the Federal Institute of Drugs and Medical Devices (BfArM, Bonn, Germany), we performed a monocentre phase I pharmacokinetic trial in healthy participants (EudraCT 2012–000970‐52) at the DIN EN ISO 9001:2008 certified Clinical Research Centre of the Department of Clinical Pharmacology and Pharmacoepidemiology, Heidelberg University. It was conducted in agreement with the standards of Good Clinical Practice (as defined in the International Conference on Harmonisation E6 Guideline for Good Clinical Practice), the Declaration of Helsinki, and the specific legal requirements in Germany.

Study population

Twelve healthy participants [1 female, mean age 32.2 years (range: 23–52), mean body weight 80.8 kg ± 11.8, mean BMI 25.1 kg m^–2 ± 3.2, all participants were Caucasian) were enrolled after obtaining written informed consent. They were mentally and physically healthy as ascertained by medical history, clinical examination, electrocardiogram and routine laboratory analyses including haematology, blood chemistry, urine screening for illicit drugs and a quantitative pregnancy test in women to exclude pregnancy. All participants were informed to use a double barrier method (e.g. combination if oral contraception or IUP and a preservative) for contraception when being heterosexually active. None of the participants had been on any regular drug treatment (except oral contraception or L‐thyroxin) in the past 2 weeks prior to the study nor were they taking any compounds known to induce or inhibit drug metabolizing enzymes or transporters (e.g. St. John's wort) within a period of <10 times the respective elimination half‐life or 2 weeks, whatever was longer. Participants with known intolerance against midazolam or voriconazole were excluded as were participants with any of the following conditions: regular smoking, excessive alcohol drinking (>20 g daily), blood donation within the last 4 weeks, participation in another study within the last 6 weeks before inclusion, suspected non‐adherence, or inability to give written informed consent or to communicate well with the investigator.

Study design

Participants were recruited into a fixed‐sequence study comparing the IV pharmacokinetics of single therapeutic (400 mg) and subtherapeutic (100 mg) voriconazole doses with different infusion rates and concurrent changes in midazolam pharmacokinetics. Voriconazole administrations were separated by a wash‐out phase of at least 7 days. Participants received 100 mg voriconazole IV over 4 h (A: 100 mg 4 h^–1), 400 mg voriconazole IV over 6 h (B: 400 mg 6 h^–1), 400 mg voriconazole IV over 4 h (C: 400 mg 4 h^–1), and finally 400 mg voriconazole IV over 2 h (D: 400 mg 2 h^–1; Figure 1). Oral microdosed midazolam as a CYP3A probe drug was administered at the end of each infusion.

Figure 1.

(A) Trial design and (B) administration schedule of midazolam (MDL) in relationship to the voriconazole (VRZ) infusion and lunch on the different study visits. IV, intravenous; p.o., orally

Study medication

Voriconazole (Vfend; Pfizer Pharma GmbH, Berlin, Germany) for IV administration was prepared according to the manufacturer's instructions and dissolved in normal saline to yield a total volume of 50 ml.

We used oral microgram doses of midazolam, which do not exhibit pharmacological effects for CYP3A phenotyping 22, 23. All midazolam doses were prepared from commercially available vials containing 5 mg midazolam in 5 ml (Dormicum V: Roche, Grenzach‐Wyhlen, Germany). To prepare the oral solution, 3 μg of midazolam were diluted in 100 ml of water.

Blood and urine sampling

On days of voriconazole IV administration, blood and urine was sampled predose, and over a period of 48 h after start of infusion. In group A, blood samples were taken predose, and 5, 10, 15, 20, 25, 30, 40, 60, 75, 90, 105, 120, 140, 170, 200, 210, 240 min (end of infusion), 245, 250, 255, 260, 270, 280, 300 min, 6, 6.5, 7, 8, 10, 24, 32 and 48 h after start of infusion. In group B, blood samples were taken predose, and 5, 10, 15, 20, 25, 30, 40, 60, 75, 90, 105, 120, 140, 170, 200, 210, 240, 270, 300, 330, 360 min (end of infusion), 365, 370, 380, 390, 400 min, 8, 8.5, 9, 10, 24, 32 and 48 h after start of infusion. In group C, blood samples were taken predose, and 5, 10, 15, 20, 25, 30, 40, 60, 75, 90, 105, 120, 140, 170, 200, 210, 240 min (end of infusion), 245, 250, 255, 260, 270, 280, 300 min, 6, 6.5, 7, 8, 10, 24, 32 and 48 h after start of infusion. In group D, blood samples were taken predose, and 5, 10, 15, 20, 25, 30, 40, 60, 75, 90, 105, 120 min (end of infusion), 125, 130, 135, 140, 165, 180 min, 4, 4.5, 5, 6, 8, 10, 24, 32 and 48 h. All urine was collected over the 48 h period.

CYP2C19 genotyping

For genotyping, genomic DNA was isolated from white blood cells using the NucleoSpin Blood kit (Macherey‐Nagel, Düren, Germany) according to the manufacturer's instructions. The presence of the alleles *2 and *3 was determined using the simple probe/hybridization probe format applied by the LightMix Kit CYP2C19*2*3 (TIBMolbiol, Berlin, Germany) on a LightCycler 480 (Roche Applied Science, Mannheim, Germany) according to the manufacturer's instructions. The presence of the alleles *4 and *17 was determined using the hybridization probe format on a LightCycler 480 according to a previously published method 24.

Assay for inhibition of CYP2C19 and CYP3A4 by voriconazole‐N‐oxide

Inhibition of CYP2C19 and CYP3A4 was assessed with the P450‐Glo CYP2C19 Screening System P450‐Glo and the P450‐Glo CYP3A4 Screening System (Promega Corporation, Madison, WI, USA) according to the manufacturer's instructions. The kits contain a luminogenic substrate (luciferin‐H EGE for CYP2C19 and luciferin‐IPA for CYP3A4), which is converted by the respective CYP into luciferin. When incubated with the luciferin detection reagent of the kit a light signal is generated which correlates to the activity of the respective enzyme. Voriconazole‐N‐oxide was tested for its ability to inhibit the production of the luminescent signal. Eight concentrations in triplicates ranging from 0.05 up to 100 μmol l^–1 were tested and each experiment was conducted thrice. IC₅₀ values were calculated using a four‐parameter fit using Prism version 6.07 (GraphPad Software, Inc., La Jolla, CA, USA).

Analytical assays

Voriconazole, midazolam and their metabolites were quantified as reported earlier 9, 25. The lower limit of quantitation (LLOQ) was 0.6 ng ml^–1 for voriconazole, and 3.0 ng ml^–1 for voriconazole N‐oxide in plasma. In urine LLOQ was 30 ng ml^–1 for voriconazole and voriconazole N‐oxide and 230 ng ml^–1 for hydroxy‐voriconazole and dihydroxy‐voriconazole. The within‐batch and batch‐to‐batch accuracies for voriconazole and voriconazole N‐oxide in plasma and urine were always within 100 ± 15%, and the corresponding precisions were <15%, with the exception of voriconazole N‐oxide at the lowest urine quality control sample (29%).

The LLOQ of midazolam and 1′‐hydroxy‐midazolam was 0.1 pg ml^–1 and 0.28 pg ml^–1 with corresponding precision below 20%. The calibrated concentration ranges were linear for midazolam (0.1–250 pg ml^–1) and its metabolite (0.28–125 pg ml^–1) with correlation coefficients >0.99. Within‐batch and batch‐to‐batch precision in the calibrated ranges for both analytes were <14% and <12%.

CYP3A phenotyping

A validated limited sampling strategy based on plasma samples taken 2–4 h after midazolam dose was used to determine CYP3A activity 26. To calculate estimated midazolam clearance, we constructed the AUC_2–4 using molar midazolam plasma concentrations at t = 2 h, 2.5 h, 3 h and 4 h with Prism 6.07. These were normalized to a midazolam dose of 1 mg and then eCLmet was calculated using the following formula: eCLmet =5668/AUC_2–4 (h.nmol ml^–1)/mg midazolam 26.

Pharmacokinetic analysis and statistical evaluation

All pharmacokinetic parameters of voriconazole and its metabolites were determined using Kinetica 5.0 (Thermo, Waltham, MA, USA). AUC_0–∞ was calculated with the log‐linear model implemented in Kinetica 5.0 27. C_max was the maximum concentration reached, T_max was determined as time to reach maximum concentration. Half‐life (t_1/2) was calculated as ln2/k. The apparent volume of distribution at steady state (V_ss) was calculated as (Dose*MRT)/(AUC_0–∞). Systemic clearance (CL_sys) was calculated as CL_sys = dose/AUC_0–∞ 27.

Metabolic ratio in plasma was calculated as molar AUC_0–24 (voriconazole N‐oxide) / molar AUC_0–24 (voriconazole) while the metabolic ratio in urine was based on the amount excreted in 48 h urine (Ae): molar Ae (voriconazole N‐oxide) / molar Ae (voriconazole). When looking at bioequivalence, the geometric mean ratio and 90% confidence interval of AUC or C_max between test and comparator were calculated. When the 90% confidence interval was within the margins of 80–125%, we considered both administration schedules to be bioequivalent.

The statistical analysis was conducted using Prism 6.07. We have run a nonparametric one‐way ANOVA (Friedman test) corrected for multiple comparisons (Dunn's posthoc test).

Results

Voriconazole plasma kinetics

Plasma concentration–time profiles of voriconazole and its N‐oxide metabolite are shown in Figure 2 and Figure S1. The pharmacokinetic parameters of voriconazole following different administration schedules are reported in Table 1. C_max decreased with lower infusion rates. C_max after a 4 h infusion, and after a 6 h infusion was 29% and 51% lower compared to a 400 mg 2 h^–1 infusion and outside of the bioequivalence margins (0.80–1.25; Table 2). Dose‐disproportionality of voriconazole was also observed for the 100 mg dose, were dose‐normalised C_max was 28% lower compared to the 400 mg 4 h^–1 infusion (Table 2). Voriconazole exposure (AUC) was dose dependent and infusion rate dependent (Table 1). AUC after a 400‐mg dose decreased with longer infusion times. Administered over 4 h, and 6 h, the relative AUC was 89% (90% confidence interval: 81–98%; p = 0.051 vs. 2‐h infusion), and 84% (80–88%; p < 0.0001 vs. 2‐h infusion), respectively (Table 2). Hence, the voriconazole exposure (AUC) of the 400 mg 6 h^–1 administration schedule is 16% lower but still within bioequivalence margins (0.80–1.25) of the 400 mg 2 h^–1 administration schedule (Table 2). Dose‐normalised AUC was 34% lower following a 100 mg 4 h^–1‐infusion compared to a 400 mg 4 h^–1 infusion (Table 2). Concurrently, the total systemic clearance (CL_sys) was uniformly lower after 400 mg compared to 100 mg. Also, CL_sys was infusion rate‐dependent (Figure 3).

Figure 2.

Mean (± standard deviation) plasma concentration‐time profiles of voriconazole (main) and voriconazole N‐oxide (insert) of 12 healthy volunteers who were administered a single intravenous dose of 100 mg of voriconazole over 4 h (black diamonds), 400 mg of voriconazole over 6 h (black triangles), 400 mg of voriconazole over 4 h (black circles), and 400 mg of voriconazole over 2 h (black squares) on four different occasions with a washout phase of ≥7 days between each dosing

Table 1.

Voriconazole pharmacokinetics after intravenous (IV) administration of 100 mg voriconazole over 4 h and 400 mg voriconazole over 2, 4, and 6 h

	Voriconazole dose and duration of infusion
	A	B	C	D
	100 mg IV in 4 h	400 mg IV in 6 h	400 mg IV in 4 h	400 mg IV in 2 h
Voriconazole
Cmax (ng ml–1)	482 [413–563]	1830 [1580–2130]	2670 [2350–3030]	3730 [3377–4120]
tmax (h) [median and range]	4.00 [3.33–4.00]	5.75 [4.50–6.00]	4.00 [4.00–4.25]	2.00 [1.75–2.00]
AUCobs (h*ng ml–1)	2630 [2280–3030]	17 600 [14 700–21 000]	18 800 [14 900–23 700]	21 100 [17 700–25 100]
AUC0–∞ (h*ng ml–1)	2640 [2290–3040]	17 700 [14 800–21 200]	18 900 [15 000–23 900]	21 200 [17 800–25 200]
CLsys (ml min–1)	631 [548–727]	376 [315–450]	352 [279–445]	315 [264–375]
t1/2 (h)	4.83 [4.25–5.49]	6.59 [5.80–7.49]	6.55 [5.91–7.25]	6.50 [5.82–7.26]
Vss (L)	230 [199–266]	233 [199–272]	190 [160–225]	150 [132–173]
Voriconazole N‐oxide
AUCobs (h*ng ml–1)	11 800 [9700–14 200]	80 600 [69 600 – 93 400]	81 500 [71 000 –93 400]	76 900 [64 000 –92 300]
Molar metabolic ratios
AUCplasma (N‐Oxide) / AUCplasmavoriconazole	4.48 [3.68–5.46]	4.64 [3.62–5.94]	4.33 [3.40–5.52]	3.65 [2.80–4.76]

AUC_obs, observed area under the concentration–time curve; AUC_0–∞, extrapolated area under the concentration–time curve; C_max, peak concentration; CL_sys, systemic clearance; t_max, time of peak concentration; t_1/2, terminal elimination half‐life, Vz, volume of distribution. Geometric mean and 95% confidence interval for all parameters except t_max, median and range for t_max.

Table 2.

Comparison of voriconazole pharmacokinetic parameters

		Voriconazole dose and duration of infusion
		A vs. C	B vs. D	C vs. D
		100 mg IV in 4 h (dose normalized) vs. 400 mg in 4 h	400 mg IV in 6 h vs. 400 mg IV in 2 h	400 mg IV in 4 h vs. 400 mg IV in 2 h
Cmax	GMR [90% CI]	0.72 [0.66–0.80]	0.49 [0.44–0.54]	0.71 [0.65–0.79]
	P	< 0.0001 *	< 0.0001 *	< 0.0001 *
AUC0–∞	GMR [90% CI]	0.56 [0.50–0.62]	0.84 [0.80–0.88]	0.89 [0.81–0.98]
	P	< 0.0001 *	< 0.0001 *	0.0511
t1/2	GMR [90% CI]	0.74 [0.71–0.77]	1.01 [0.95–1.08]	1.01 [0.98–1.04]
	P	< 0.0001 *	0.6838	0.6784
Vss	GMR [90% CI]	1.21 [1.12–1.30]	1.55 [1.48–1.63]	1.27 [1.19–1.35]
	P	0.1062	< 0.0001 *	0.2390
AUC0–obs (VRZ‐N‐Ox)	GMR	0.58 [0.54–0.62]	1.06 [0.98–1.14]	1.04 [0.93–1.15]
	P	0.0001	>0.9999	>0.9999
MR (AUCplasma (VRZ‐N‐Ox) / AUCplasma VRZ)	GMR [90% CI]	1.03 [0.92–1.17]	1.20 [1.09–1.32]	1.19 [1.08–1.30]
	P	0.6335	0.0066 *	0.0066 *
MR (Aeurine (VRZ‐N‐Ox) / Aeurine VRZ)	GMR [90% CI]	0.92 [0.77–1.10]	1.35 [1.24–1.46]	1.00 [0.81–1.23]
	P	0.4283	< 0.0001 *	0.9950

Ae, Amount excreted; AUC_0–∞, Area under the plasma concentration–time curve extrapolated to infinity; CI, confidence interval; C_max, peak plasma concentration; GMR, geometric mean ratio; MR, metabolic ratio; t_1/2, terminal elimination half‐life; VRZ, voriconazole; VRZ‐N‐Ox, voriconazole N‐oxide.

^* and bolded, statistically significant.

Figure 3.

Individual systemic clearance of voriconazole of 12 healthy volunteers who were administered a single intravenous dose of 100 mg of voriconazole over 4 h (A), 400 mg of voriconazole over 6 h (B), 400 mg of voriconazole over 4 h (C), and 400 mg of voriconazole over 2 h (D) on 4 different occasions with a washout phase of ≥7 days between each dosing. A Friedman's test followed by Dunn's multiple comparisons test was run, an adjusted P‐value <0.05 was considered statistically significant

Voriconazole metabolites

In contrast to the parent compound, exposure (AUC) with voriconazole N‐oxide after 400 mg IV voriconazole was independent of infusion rate (Table 1). Corresponding to the changes of the parent drug, the plasma metabolic ratios were approximately 20% higher when voriconazole was infused over 4 h and 6 h compared to the 2 h infusion, but was independent of dose (Table 2). Dose‐adjusted AUC of voriconazole N‐oxide was 44% lower after a 100 mg dose (Table 2, p < 0.0001). Interestingly, urinary recovery was about 10% lower following a 100 mg dose compared to the 400 mg doses (group A vs. D P = 0.0016; Table 3), which was mainly due to approximately 10% reduced urinary recovery of the N‐oxide metabolite, but the recovery of the hydroxy‐metabolite and dihydroxy‐metabolite was significantly reduced also (Table 3). The molar metabolic ratio in urine was significantly increased by 24% when the 400‐mg dose was administered over 6 h compared to the 4 h and 2 h infusions (Table 3).

Table 3.

Geometric means [95% confidence interval] of amount excreted in urine and urinary recovery of voriconazole and metabolites of 12 healthy volunteers receiving 100 mg voriconazole over 4 h and 400 mg voriconazole over 2, 4, and 6 h

		Voriconazole dose and route of administration
		A		B	C	D
		100 mg IV in 4 h	Dose normalized to 400 mg	400 mg IV in 6 h	400 mg IV in 4 h	400 mg IV in 2 h
Voriconazole	Ae (μmol)	2.47 [1.73–3.51]	9.86 [6.93–14.0]	12.2 [9.50–15.7]	17.6 [12.6–24.5]	16.9 [13.2–21.6]
	P (vs. D)	< 0. 0001 *	0.0451	0.1417	>0.9999	n/a
	% of dose	0.86 [0.60–1.23]		1.07 [0.83–1.37]	1.54 [1.10–2.14]	1.47 [1.15–1.88]
	P (vs. D)	0.003 *		0.016 *	>0.9999	n/a
Voriconazole N‐oxide	Ae (μmol)	28.9 [22.5–37.0]	115 [90.1–157]	209 [170–257]	224 [189–265]	215 [183–253]
	P (vs. D)	< 0.0001 *	0.0298 *	>0.9999	>0.9999	n/a
	% of dose	10.1 [7.85–12.9]		18.2 [14.8–22.4]	19.5 [16.5–23.1]	18.8 [16.0–22.1]
	P	0.0017 *		0.9228	>0.9999	n/a
Hydroxy‐voriconazole	Ae (μmol)	12.2 [8.69–17.3]	49.0 [34.8–69.1]	55.3 [37.9–80.8]	67.2 [46.8–96.5]	76.8 [55.1–107]
	P	<0.0001	0.1417	>0.9999	>0.9999	n/a
	% of dose	4.30 [3.03–6.04]		4.83 [3.31–7.05]	5.86 [4.08–8.42]	6.70 [4.81–9.34]
	P (vs. D)	0.016 *		0.3467	>0.9999	n/a
Dihydroxy‐voriconazole	Ae (μmol)	13.3 [8.82–20.1]	53.2 [35.3–80.2]	51.8 [36.1–74.3]	57.6 [40.1–82.8]	61.8 [41.7–91.5]
	P (vs. D)	0.0006 *	>0.9999	>0.9999	>0.9999	n/a
	% of dose	4.64 [3.08–7.01]		4.52 [3.15–6.48]	5.03 [3.50–7.23]	5.40 [3.64–7.98]
	P (vs. D)	>0.9999		>0.9999	>0.9999	n/a
Parent + all metabolites	UR (%)	20.5 [15.8–26.7]		29.6 [24.0–36.6]	33.3 [28.3–39.2]	33.3 [27.2–40.7]
	P (vs. D)	0.0160 *		>0.9999	>0.9999	n/a
Aeurine (N‐Oxide) / Aeurine voriconazole		11.7 [7.56–18.1]		17.1 [13.0–22.6]	12.7 [9.09–17.8]	12.7 [10.1–16.1]

Ae, amount excreted into urine; clearance; U_R, urinary recovery; n/a, not applicable.

^* and bolded, statistically significant at an α = 5% level using Friedman's test followed by Dunn's multiple comparison's test.

In vitro inhibition of CYP2C19 and CYP3A4 by voriconazole N‐oxide

Mean IC₅₀ of voriconazole N‐oxide was 146 ± 23 μmol l^–1 for CYP3A4 and 40.2 ± 4.2 μmol l^–1 for CYP2C19 (Figure S2).

CYP3A phenotyping

Geometric mean (95% confidence interval) of eCLmet was 1070 ml min^–1 (800–1420). eCLmet of all 12 participants was 516 ml min^–1 (420–640) following the 100 mg 4 h^–1 voriconazole administration, 152 ml min^–1 (139–166) for the 400 mg 6 h^–1 schedule, 192 ml min^–1 (167–220) for the 400 mg 4 h^–1 infusion, and 202 ml min^–1 (189–217) for the 400 mg 2 h^–1 infusion (Figure 4). As a comparator, we added recent (<1 year) values for estimated metabolic midazolam clearance (eCL_met) as a marker for CYP3A activity for the eight participants for whom they were available from baseline measurements of a prior trial participation (Figure 4).

Figure 4.

Estimated metabolic midazolam clearance (CL_met) after administration of 100 mg of intravenous voriconazole over 4 h, or 400 mg over 2 h, 4 h or 6 h. p < 0.05 as determined by the Friedmann's test followed by Dunn's multiple comparisons test was seen as statistically significant. * Midazolam estimated metabolic clearance without comedication of eight participants from whom recent CYP3A data (< 1 year) was available from earlier trial participations 7.

CYP2C19 genotype data

No CYP2C19 PMs (expressing two defective alleles *2, *3 or *4) were among the 12 analysed participants. Genotype data on an individual level are reported in Table S1.

Safety

Only mild adverse events occurred during the trial. Two cases of CTCAE (Common Terminology Criteria for Adverse Events v.4.03) grade 1 nausea and grade 1 cephalgia were related to voriconazole. One case of grade 1 blurred vision, one case of grade 1 hyperkalaemia and two cases of grade 1 hyperbilirubinaemia were not related to voriconazole. All events abated spontaneously without treatment.

Discussion

In this clinical trial, we demonstrate that peak plasma concentration, exposure and volume of distribution of intravenous voriconazole are infusion rate‐dependent. The most important change that occurred in quantitative terms was the strong decrease in C_max when decreasing the infusion rate. The reduction of C_max by 50% when administering voriconazole over 6 h is suggestive for a faster elimination rate from plasma during the infusion. From a clinical point of view this finding is quite significant as the fungicide effect is concentration dependent and patients infected with species with a high minimum inhibitory concentration will benefit from quickly achieving high plasma concentrations. The reduction of voriconazole exposure was less pronounced than changes in C_max but statistically significant. The observed differences suggest that differences in administration regimens and thus exposure differences during the administration phase probably modified voriconazole clearance in a concentration‐dependent manner. Still the AUC after 400 mg were bioequivalent independent of the duration of the infusion. A comparable effect of infusion rate on parent drug and metabolite pharmacokinetics has recently been reported for CYP1A2‐dependent bendamustine biotransformation in rats, where longer infusion times correlated with lower AUC 28. Interestingly, V_ss increased 1.5‐fold when comparing the 400 mg 6 h^–1 group to the 400 mg 2 h^–1 group, while no difference in terminal elimination half‐life (t_1/2) was observed. The V_ss of the 400 mg 2 h^–1 group was equal to previously reported values after multiple 3 mg kg^–1 IV doses while a V_ss of 220 l has been reported after single 3 mg kg^–1 IV doses 29. In consequence, steady‐state plasma concentration will be achieved faster in patients when choosing higher infusion rates and higher doses. The extent of AUC reduction in our trial was only small, still within the bioequivalence range, and hence is not deemed clinically significant. However, duration of infusion should only be increased with caution in clinical practice and prolonging the administration or even administration by continuous infusion will yield the lowest exposures and thus the most expensive treatment regimen.

Concerning metabolite kinetics, no alteration of the N‐oxide metabolite exposure was observed when varying infusion rates. With a ~ 20% lower voriconazole AUC in the 400 mg 6 h^–1 group, the relative metabolite formation rate was 20% higher. This is suggestive for accelerated voriconazole biotransformation at lower infusion rate, reflecting rapid and concentration‐dependent hepatic autoinhibition of the metabolism. The trial was of an explorative nature using four different infusion rates, it was not powered to detect a difference of 20% of voriconazole‐N‐oxide formation.

Overall, comparing the 100 mg 4 h^–1 group with the 400 mg 4 h^–1 group, a 4‐fold increase in dose has the greater impact on pharmacokinetics than a reduction of infusion rate by 66%. While overall exposure with voriconazole N‐oxide was 8‐fold lower (Table 1), plasma metabolic ratio did not differ between the 100 mg 4 h^–1 schedule and the 400 mg 4 h^–1 schedule (Table 1). Interestingly though, urinary recovery of total dose was 10% lower with the 100 mg dose. This was mostly due to reduced N‐oxide recovery. IV doses of 100 mg voriconazole also inhibited CYP3A activity, but to a lesser extent than the 400 mg doses. A possible explanation is that at lower doses, metabolism is shunted towards a more CYP3A dependent elimination pathways. N‐oxide is both formed by CYP3A4 and CYP2C19, while formation of the 4‐hydroxy‐voriconazole metabolite is solely CYP3A4 dependent 5, 17.

Estimated midazolam metabolic clearance (eCL_met) was lower in the 400 mg 6 h^–1 infusion group than in the 400 mg 2 h^–1 infusion group (Figure 4). This finding seemed contradictory to the concept of increased CYP3A‐ and CYP2C19‐mediated voriconazole biotransformation at lower infusion rates. Exposure with voriconazole N‐oxide during the time interval from start of infusion until 2 h post end‐of‐infusion markedly differs between groups B, C and D (Figure 2) Considering the different time points of midazolam administration relative to the start of infusion, exposure over time with voriconazole N‐oxide was different and highest when midazolam was administered 6 h after start of infusion. We therefore postulated that voriconazole metabolites might inhibit CYP isozymes similar to itraconazole metabolites 30 and subsequently analysed the perpetrator potential of voriconazole N‐oxide. Peak voriconazole N‐oxide plasma concentrations were approximately 11 μmol l^–1 following the 400 mg IV single dose in our trial (Figure 2). The plasma protein binding of voriconazole is 58% and is not determined for its N‐oxide metabolite 3. Regarding the total and estimating the free plasma concentrations, N‐oxide mediated CYP3A4‐inhibition is unlikely (IC₅₀ = 146 μmol l^–1) but mild inhibition of CYP2C19 following a single dose is conceivable (IC₅₀ = 40.2 μmol l^–1). For voriconazole, end‐of‐infusion concentrations were 5.24 μmol l^–1 for the 6 h‐infusion, 7.64 μmol l^–1 for the 4 h‐infusion and 10.7 μmol l^–1 for the 2 h‐infusion, which is above the K_i for competitive (K_i = 0.66 μmol l^–1) CYP3A4‐inhibition, and CYP2C19 inhibition (Ki = 5.1 μmol l^–1) 12. We only observed the plasma compartment after the first dose, where drug distribution into the tissue still occurs. Also the higher V_ss for the 400 mg 6 h^–1 group is may be indicative of better tissue penetration and hence higher local concentrations in the hepatocyte where the drug interaction occurs. Hence, a later administration of midazolam with respect to the start of voriconazole infusion could explain lower midazolam clearance. Stronger CYP3A and CYP2C19 inhibition occur at steady state concentrations of both voriconazole and voriconazole N‐oxide. With voriconazole‐N‐oxide steady‐state trough plasma concentrations of 1700 ng ml^–1 in the target patient population 31, the inhibitory effect will mostly come from the parent drug and to a much lesser extent for CYP2C19 from the N‐oxide metabolite. eCL_met was reduced by about 50% compared to baseline in the 100 mg 4 h^–1 group, while we recently observed that 50 mg IV voriconazole does not alter oral midazolam clearance 9. Still with a 100 mg dose, midazolam clearance was 2–3 times higher than following a 400 mg dose (Figure 4).

Limitations

There are some limitations for the results of this trial. Only a single IV dose was investigated and the situation can be different when steady‐state is reached, and both metabolite and parent levels are higher, thus exerting a stronger inhibition of the CYP3A system 9, 18, 26. Concerning the midazolam drug interaction data, baseline values were not assessed during the study but recent values from other study participations were taken, when available (8 of 12). This might bias baseline as there is a slight seasonal variability in midazolam disposition 32. No CYP2C19 PMs were included in this trial. Smaller absolute effects but higher relative effects because of lack of CYP2C19 activity are expected in this population. Also, there are differences in the timing of food administration and intake of the midazolam oral solution that may affect drug disposition.

Conclusion

In conclusion, we have demonstrated a minor effect of rate of infusion on voriconazole exposure for 400‐mg doses. Lowering infusion rate by 67% leads to a 50% reduction in C_max and 16% reduction of exposure. These and previous findings confirm that voriconazole pharmacokinetics are modulated by autoinhibition. We expand this information to duration of parenteral voriconazole administration, and suggest an inhibitory contribution of the voriconazole N‐oxide metabolite to CYP2C19 dependent metabolism. In the light of the data presented here and from our previous work 9, dose is more relevant than infusion rate. Findings concerning CYP3A4 inhibition are consistent. Only participants with functional CYP2C19 protein were enrolled, hence the mechanism of CYP autoinhibition also extends to CYP2C19 mediated voriconazole metabolism in EMs, as voriconazole and its metabolite both inhibit 2C19.

In clinical practice, infusion rate should not deliberately be chosen or prolonged because this can trigger reduced exposure. Also, faster infusion rates could be explored, to reach higher plasma levels of voriconazole. While there is a positive association between C_max and visual adverse events for single dose voriconazole 33, oral doses were tested up to 1600 mg single dose for QT‐study, leading to a C_max of 10 000 ng ml^–1 34. These concentrations still were safe and only had marginally more (70 vs. 60) treatment‐related adverse events. Overall, voriconazole is well tolerated compared to other azole antimycotics and infusion reactions are rare 35.

Competing Interests

All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.

The authors acknowledge Kristina Lohmann for monitoring the study, Marlies Stützle‐Schnetz (study nurse) for the excellent assistance provided, Magdalena Longo, Andrea Deschlmayr, and Monika Wittnebel for technical support during the analytical procedures, and Corina Mueller and Jutta Kocher for CYP2C19 genotyping and conducting the CYP inhibition assays.

Contributors

N.H., R.K., A.B., J.W., J.B., W.E.H. and G.M. analysed and interpreted the data, and wrote the manuscript; G.M., W.E.H., R.K. and N.H. designed the study; N.H., A.B. and R.K. performed the trial; J.B. developed the analytical methods and analysed the samples; J.W. performed and analysed the genotype data and the CYP inhibition assays.

Supporting information

Figure S1 Log transformed mean (± standard deviation) plasma concentration‐time profiles of voriconazole (main) and voriconazole N‐oxide (insert) of 12 healthy volunteers who were administered a single intravenous dose of 100 mg of voriconazole over 4 h (black diamonds), 400 mg of voriconazole over 6 h (black triangles), 400 mg of voriconazole over 4 h (black circles), and 400 mg of voriconazole over 2 h (black squares) on four different occasions with a washout phase of ≥7 days between each dosing

Figure S2 Concentration‐dependent effect of voriconazole‐N‐oxide on CYP3A4 and CYP2C19 activity measured by using the P450‐Glo CYP1A2 Screening System and the P450‐Glo CYP2C19 Screening System P450‐Glo. Depicted is one experiment of three with each concentration tested in triplicate (mean ± standard deviation)

Table S1 Participant characteristics

Hohmann, N. , Kreuter, R. , Blank, A. , Weiss, J. , Burhenne, J. , Haefeli, W. E. , and Mikus, G. (2017) Autoinhibitory properties of the parent but not of the N‐oxide metabolite contribute to infusion rate‐dependent voriconazole pharmacokinetics. Br J Clin Pharmacol, 83: 1954–1965. doi: 10.1111/bcp.13297.