Therapeutic drug monitoring of voriconazole for treatment and prophylaxis of invasive fungal infection in children

Abstract

Voriconazole therapeutic drug monitoring is not consistently recommended due to its high interpatient and intrapatient variability. Here, we aimed to describe our experience with voriconazole for treatment and prophylaxis of invasive fungal infections in paediatric patients. A fully validated high‐performance liquid chromatography–mass spectrometry method was used to quantify voriconazole concentration in plasma, at the end of dosing interval. A high interindividual variability was shown. We enrolled 237 children, 83 receiving intravenous and 154 oral voriconazole. A positive correlation between drug dose and drug plasma exposure was observed. Considering intravenous route, patients with higher serum creatinine had higher voriconazole concentrations; a positive correlation was found among drug exposure and age. Sex significantly influenced drug levels: males had higher median drug concentrations than females (P < 0.001). Close voriconazole pharmacokinetics monitoring should help individualize antifungal therapy for children.

What is Already Known about this Subject

• Large interindividual and intraindividual variability in drug exposure has been observed.

• Therapeutic drug monitoring may play an important role in voriconazole treatment and prophylaxis management.

What this Study Adds

• In patients receiving oral voriconazole, a positive correlation was found between trough concentration and drug dose.

• In patients receiving intravenous voriconazole a positive correlation was found between pharmacokinetics and age, and pharmacokinetics and creatinine.

• Males receiving intravenous voriconazole had higher voriconazole concentrations.

Introduction

In the last 2 decades an increase in the prevalence and severity of invasive fungal infections, associated with significant morbidity and mortality, has been registered. The second‐generation antifungal voriconazole (VFend; Pfizer Pharmaceuticals, New York, NY, USA) is an antifungal agent with a broad spectrum of activity and commonly used for prophylaxis and treatment of invasive fungal infections 1. It is recommended to treat patients with invasive aspergillosis, candidaemia and disseminated infections such as those caused by Candida spp. and oesophageal candidiasis 2.

Voriconazole inhibits the fungal cytochrome P450 (CYP) enzyme lanosterol 14‐α‐demethylase, decreasing the synthesis of ergosterol, an important component of fungal cell membranes 3. It is metabolized in liver through CYP2C19 and to a lesser extent by CYP3A4 and CYP2C9. It is mainly excreted in urine (80%) 4. Even though voriconazole had been approved by the Food and Drug Administration for treatment of invasive fungal infections in patients older than 12 years, it is routinely used in younger children 3, 5. Voriconazole pharmacokinetics are complex and nonlinear: a saturable clearance (due to saturable hepatic metabolism) has been observed in children receiving drug dose higher than recommended. Large interindividual and intraindividual variability in drug exposure have been reported 6. Furthermore, the drug plasma exposure is considerably influenced by different factors: age, sex, body mass index (BMI), CYP2C19 polymorphism, liver disease, inflammation and drug–drug interactions 7, 8, 9. Voriconazole oral bioavailability in paediatric patients is lower than in adults, with further reduction when coadministered with food 10. Paediatric patients need higher loading and maintenance dosing of voriconazole [both intravenous (IV) or orally (PO) administered], due to enhanced hepatic clearance and first‐pass effect 11.

In patients aged 2–11 years, the package leaflet recommends a weight‐based dose of 9 mg kg^–1 IV or PO twice daily and a maximum dose of 350 mg. In patients older than 12 years the loading dose is 6 mg kg^–1 IV twice daily 12, 13. Numerous prospective studies reported that voriconazole trough concentration (C_trough) ≥ 1.5–2 mg l^–1 are associated with near maximal clinical response treatment of invasive fungal infections 9, 14, 15, 16, 17, 18, 19. Whereas, targets have not been validated for paediatric patients. These data suggest that therapeutic drug monitoring (TDM) may play an important role in voriconazole treatment and prophylaxis management. Despite this, TDM is not routinely used in many institutions due to costs and resource availability 20.

In this paper, we aimed to describe the obtained results from TDM in paediatrics and to investigate the relationships among voriconazole trough concentration (C_trough) and sex, age, BMI, ethnicity, serum creatinine and drug dose, thus as to guide the voriconazole individual dosage adjustment in children.

Methods

Patients and inclusion criteria

Plasma samples were collected at the Laboratory of Clinical Pharmacology and Pharmacogenetics (Department of Medical Sciences, University of Turin, ASL “Città di Torino”, Turin) and Clinical Pharmacology Service “Franco Ghezzo” (Department of Biological and Clinical Sciences, University of Turin, S. Luigi Gonzaga Hospital) from different hospitals in Piedmont (Italy). Inclusion criteria were: age < 18 years, a diagnosis of invasive fungal infection and treatment with voriconazole, for prophylaxis or therapy purposes, with an adherence of 90% (measured by pill counts for PO route). Patients with potential interacting drugs (such as HIV‐protease inhibitors and non‐nucleoside reverse transcriptase inhibitors, immunosuppressant, rifampin, rifabutin, carbamazepine and long‐acting barbiturates), allergy or intolerance to voriconazole, HIV infection, severe malnutrition, liver cirrhosis, chronic renal failure (with estimated creatinine clearance, eCRCl < 60 ml min^–1) or sepsis diagnosis were excluded. Study protocol (PkPG_J02AC Studio retrospettivo per la valutazione e farmacocinetica e farmaco‐genetica della terapia antimicotica con farmaci triazolici) was approved by the local ethics committee. For each patient, the approved protocol provides for the recording of the data to enable traceability, accuracy, completeness and timeliness of the drug administration and blood sample collection timings. A written informed consent for the study was obtained from each subject, signed by natural/biological father or mother of a child with full parental legal rights. For all the patients, following data were available: sex, age, BMI, ethnicity, serum creatinine and voriconazole dose.

Determinations of voriconazole plasma concentration

For each patient, blood samples were taken immediately before drug intake (C_trough), under steady‐state conditions (5 days after both IV and PO administration). Plasma samples were obtained by centrifugation at 2000 × g for 10 min at 4°C. 6,7‐dimethyl‐2,3‐di(2‐pyridyl)quinoxaline (QX), used as the internal standard, was purchased from Sigma–Aldrich Corporation (Milan, Italy), and voriconazole was purchased from Sigma–Aldrich. Acetonitrile (HPLC grade) and methanol (HPLC grade) were purchased from VWR (Milan, Italy). Formic acid was from Sigma–Aldrich. HPLC‐grade water was produced by a Milli‐DI system coupled with a Synergy 185 system by Millipore (Milan, Italy).

Voriconazole concentrations were determined using an HPLC‐mass spectrometry system (HLPC‐MS), according a fully validated method 21, 22.

Chromatographic separation was performed at 35°C, using a column oven, on a C18 Atlantis T‐3 5‐μm (150 × 4.6 mm, inner diameter) column (Waters, Milford, MA, USA), protected by a Security Guard with a C18 (4.0 × 3.0 mm, inner diameter) precolumn (Phenomenex, Torrance, CA, USA). The mobile phase composed initially of 50:50 water with formic acid (0.05%)/acetonitrile with formic acid (0.05%) was then ramped to 20:80 within 6.5 min. The flow rate was set at 1 ml min^–1.

Detector settings were as follows: electrospray ionization; capillary voltage, 3.5 kV; source temperature, 110°C; desolvation temperature, 350°C; nitrogen desolvation flow, 400 l h^–1; nitrogen cone flow, 50 l h^–1. The ion m/z values monitored were: 350.3 for voriconazole and 313.4 for QX, cone voltage was 25 V and 50 V, respectively. The lower limit of quantification was considered the lowest standard on the calibration curve. Therefore, the lower limit of quantification for voriconazole was 0.039 μg ml^–1. The considered limit of determination was 0.019 μg ml^–1.

Accuracy and precision were assayed using eight determinations for each quality control concentration on different days. Accuracy was calculated by determining the deviation in percentage of the mean from the true value. Intra‐ and interday precision were calculated by determining the relative standard deviation at each quality control concentration. Results of validation are shown in Table 1. In accordance with FDA. Food and Drug Administration guidelines, accuracy (expressed as percent deviation from the nominal value) and precision (expressed as percent relative standard deviation) did not exceed 15%.

Table 1.

Intra‐ and interday accuracy and precision of the used method. Nominal value refers to plasmatic concentration of voriconazole quality controls

Nominal value (μg ml –1 )	Accuracy (% deviation)	Precision (% RSD)
		Intraday	Interday
0.12	1.50	5.81	13.30
1.50	1.70	3.60	9.10
4.00	5.78	4.09	8.09
6.00	9.55	5.90	6.09

RSD, relative standard deviation

This work was carried out in a UNI EN ISO 9001:2008 and 13 485:2012 (CE‐IVD) certified laboratory.

Statistical analysis

For descriptive statistics, continuous and non‐normally distributed variables were summarized as average, standard deviation and median values; interquartile range (IQR), 25th to 75th percentiles, was calculated to measure the statistical dispersion of the data. Categorical variables were described as frequency and percentage. All the variables were tested for normality with the Shapiro–Wilk test. The correspondence of each parameter was evaluated with a normal or non‐normal distribution, through the Kolmogorov–Smirnov test.

The independent‐samples t test was used to compare the means of two independent groups, considering the level of statistical significance (P < 0.05). Pearson linear correlation coefficient (r) was used to investigate the strength of the association between two quantitative variables considering the level of statistical significance (P < 0.05). Mann–Whitney U test was used to probe the influence of categorical variables on continuous variables, considering the level of statistical significance (P < 0.05). Any predictive power of the considered variables was finally evaluated through univariate and multivariate linear (for pharmacokinetic parameters) regression analysis. Factors (β, β coefficient; CI, confidence interval at 95%) with P < 0.2 in univariate analysis were considered in multivariate analysis (P < 0.05).

All statistical analyses were performed with IBM SPSS Statistics 22.0 for Windows (Chicago, Illinois, USA).

Results

This study enrolled 237 paediatric patients (147 male, 62%) treated with voriconazole; the 63.3% (n = 150) of them were Caucasians. Of these, 150 (63.3%) received voriconazole antifungal prophylaxis; routes of administration were PO (n = 154; 65%) or IV (n = 83; 35%).

Mean, standard deviation, median and IQR values for age, BMI, creatinine serum levels, voriconazole plasma concentration, drug dose (mg kg^–1) and body weight, and number and percentage of patients for ethnicity and invasive fungal infection treatment and prophylaxis, considering all the cases and those receiving PO and IV administration, were compared in Table 2.

Table 2.

Upper panel: Mean, standard deviation, median and interquartile range for age, body mass index, body weight, drug dose, serum creatinine levels and voriconazole plasma concentration, considering all the cases and those receiving oral and intravenous route of administration

	All patients (n = 237)				Intravenous route (n = 83)				Oral route (n = 154)
Variable	Mean	SD	Median	IQR	Mean	SD	Median	IQR	Mean	SD	Median	IQR
Age (years)	10.19	4.443	10.00	7.00–14.00	8.47	4.97	12	7.00–15.00	8.47	4.97	8.00	3.00–12.00
BMI (kg m –2 )	18.20	3.87	17.37	15.38–20.65	18.41	4.56	17.50	15.39–20.42	18.41	4.56	17.14	15.29–20.98
Body weight (kg)	37.21	19.364	38.00	24.00–49.25	33.60	17.863	29.75	17.00–50.00	39.15	16.828	39	25.38–48.13
Drug dose (mg kg –1 )	11.159	4.039	10.256	8.00–14.00	11.5222	4.772	11.111	8.00–14.00	10.964	3.585	10.128	7.984–13.841
Serum creatinine (mg dl –1 )	0.94	1.45	0.45	0.31–0.69	0.46	0.36	0.52	0.36–0.80	0.46	0.36	0.37	0.29–0.48
Voriconazole C trough (μg ml –1 )	2.504	2.726	1.793	0.688–3.180	2.564	3.223	1.311	0.512–2.904	2.564	3.223	1.910	0.895–3.247

Ethnicity	n		%		n		%		n		%
Caucasian	150		63.6		49		59.0		101		65.6
Other	87		36.7		34		41.0		53		34.4

Voriconazole indication:	n		%		n		%		n		%
Treatment	150		63.6		35		42.2		52		33.8
Prophylaxis	87		36.7		48		57.8		102		66.2

SD, standard deviation; IQR, interquartile range; BMI, body mass index; C_trough, concentration at the end of dosing interval

In all 237 enrolled patients, a high interindividual variability was found between voriconazole C_trough concentrations: the median value was 1.793 μg ml^–1 and the IQR was 688.50 μg ml^–1 and 3180.00 μg ml^–1.

Evaluating the IV route (n = 83; 40 males, 48.2%), 49 patients (59.0%) were Caucasians and 48 (57.8%) received antifungal prophylaxis. The median voriconazole trough value was 1.311 μg ml^–1 and IQR 512.00–2904.00 μg ml^–1 (Table 2). Mann–Whitney U test showed a significantly influence of sex on drug exposure (P < 0.001): males (n = 43) had 2.258 μg ml^–1 (IQR: 1.009–5.618 μg ml^–1) median concentrations, while females (n = 40) had 0.669 μg ml^–1 (IQR: 0.220–1.789 μg ml^–1; Figure 2). Furthermore, Pearson test showed positive and significant correlations between C_trough and age (r = 0.230; P = 0.036; Figure 1A) and C_trough and serum creatinine levels (r = 0.521; P < 0.001; Figure 1B).

Figure 2.

Plots of sex influence on voriconazole trough concentration, considering the 83 patients with intravenous route of administration (p < 0.001). Boxes and black lines in boxes represent respectively interquartile ranges and median values; open dots and stars represent outlier values. Median values (horizontal line), interquartile range (bars), patient values (black square), highest and lowest value (whiskers) are shown. Males (n = 43) had 2.258 μg ml^–1 (IQR: 1.010–5.618 μg ml^–1) median concentrations; females (n = 40) had 0.669 μg ml^–1 (IQR: 0.220–1.789 μg ml^–1) median concentrations

Figure 1.

(A) Scatter plot and linear correlation between voriconazole trough concentration (μg ml^–1; x) and age (years; y), considering the 83 patients with intravenous route of administration. The Pearson correlation coefficient (r) was 0.230; P = 0.036. Trend line and interval of confidence at 95% were shown. (B) Scatter plot and linear correlation between voriconazole trough concentration (μg ml^–1; x) and creatinine serum levels (mg dl^–1; y), considering the 83 patients with intravenous route of administration. r = 0.521, P < 0.001. Trend line and interval of confidence at 95% were shown. (C) Scatter plot and linear correlation between voriconazole trough concentration (μg ml^–1; x) and drug dose (mg kg^–1; y), considering the 154 patients with oral route of administration. r = 0.184, P = 0.022. Trend line and interval of confidence at 95% were shown

Considering PO administration (n = 154; 104 males, 67.5%), 101 patients (65.6%) were Caucasians and 102 (66.2%) received antifungal prophylaxis. The median voriconazole trough value was 1.910 μg ml^–1 (IQR: 0.895–3.247 μg ml^–1, Table 2). Pearson test showed positive and significant correlation between C_trough and drug dose (r = 0.195; P = 0.016; Figure 1C).

Univariate linear regression analysis was performed to evaluate the effect of ethnicity, sex, age, BMI, serum creatinine levels and drug dose on voriconazole C_trough. Stepwise forward regression analysis was used to identify the minimum set of independent predictive variables of voriconazole exposure and estimate the contribution of each factor to pharmacokinetic variability. Considering all the enrolled patients (β: 0.191 and CI: 0.064–0.313) and the PO route (β: 0.162 and CI: 0.004–0.358), only drug dose remained in the final model with P = 0.003 and 0.045, respectively. In those receiving IV route, serum creatinine levels resulted as predictor of voriconazole C_trough (P < 0.001, β: 0.521 and CI: 2.319–6.912).

Discussion

We have reported our data on voriconazole TDM in paediatrics and analysed the effect of age, sex and creatinine serum levels on drug exposure. Highly variable C_trough was observed, as already reported 6. Pearson test showed positive and significant correlation between voriconazole C_trough and drug dose in evaluating PO (P = 0.016; Figure 1C) routes. In linear regression analysis, drug dose was retained as predictor of increased voriconazole concentrations, in all the enrolled patients (P = 0.003) and in those receiving PO drug (P = 0.045). There are no guidelines for dose adjustment of voriconazole in children, so it is often at the discretion of the clinicians. Bartelink et al. observed that, in paediatrics, a linear dose adjustment can improve the attainment of the target concentrations 23. Moreover, a strategy not still clinically validated suggests dose adjustments by 1 mg kg^–1 lead to increase in voriconazole C_trough by 0.5 mg l^–1 24, 25. Considering the sex effect on drug exposure, we observed that, in those receiving IV treatment, males had higher C_trough than females (p < 0.001; Figure 2). For the IV route, Pearson test showed positive and significant correlation between drug exposure and age (P = 0.036; Figure 1A), probably due to differences in drug metabolism and elimination 4. It is already known that age and sex may also influence voriconazole levels 6, 26 and a study on healthy volunteers observed that maximum voriconazole concentration and area under the curve were higher in elderly male subjects and in women compared with younger men 3. Evaluating children receiving IV drug, we observed higher C_trough in patients with higher serum creatinine levels (p < 0.001; Figure 1B). Raised creatinine levels are a marker of impaired renal function. Since voriconazole is a low aqueous solubility molecule, its IV formulation includes a solubilizing agent, the sulfobutylether‐β‐cyclodextrin sodium, which accumulates in kidneys, leading to reduced renal function 27. Higher serum creatinine levels observed in patients receiving IV voriconazole could be explained by the renal accumulation of sulfobutylether‐β‐cyclodextrin.

Ultimately, no correlations were found among voriconazole exposure and BMI or ethnicity. In literature only few studies observed the effect of BMI on voriconazole pharmacokinetics, response or toxicity. Hoenigi et al., in a cohort of patients with haematological malignancies, reported BMI as significant predictor for potentially toxic drug concentrations 28. Moreover, Shao and colleagues, showed an association between BMI and lower drug through levels 29. Concerning ethnicity, related differences in voriconazole exposure are often explained by the different CYP2C19 activity among races 30.

There were some limitations in this study. First, it was observational and retrospective. Moreover, genetic analysis for polymorphisms in CYP2C19 genes was not performed. Despite being statistically significant and the large sample size, some correlations were weak; this could be due to the lack of a standardized protocol for voriconazole dosing. Thus, further works should include evaluating patients with hepatic and renal impairment and monitoring voriconazole toxicity markers, such as aminotransferases (aspartate transaminase and alanine transaminase) levels, rush and altered colour/light perception. In addition, it should be interesting to apply it to a population–pharmacokinetic model on the collected data.

Nevertheless, our findings suggest an effect of age, sex and creatinine levels on drug pharmacokinetics variability. Voriconazole TDM may be useful for predicting drug exposure and individualizing antifungal therapy in children.

Competing Interests

There are no competing interests to declare.

We thank CoQuaLab ( www.coqualab.it ) for methodological support and assistance in the preparation and execution of the pharmacogenetic analysis. This research did not receive any specific grant from funding agencies in the public, commercial, or not‐for‐profit sectors. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

Allegra, S. , Fatiguso, G. , De Francia, S. , Favata, F. , Pirro, E. , Carcieri, C. , De Nicolò, A. , Cusato, J. , Di Perri, G. , and D'Avolio, A. (2018) Therapeutic drug monitoring of voriconazole for treatment and prophylaxis of invasive fungal infection in children. Br J Clin Pharmacol, 84: 197–203. doi: 10.1111/bcp.13401.