Therapeutic drug monitoring of voriconazole: validation of a high performance liquid chromatography method and comparison with an ARK immunoassay

Abstract

Objective

Voriconazole is an antifungal agent used in the treatment of aspergillosis and fluconazole-resistant Candida infections. Therapeutic drug monitoring (TDM) of voriconazole is recommended to optimise clinical results. The aim of this study was the development and validation of a high performance liquid chromatography (HPLC) method for measuring voriconazole in human serum, and comparison with an ARK immunoassay method.

Methods

Linearity, precision, accuracy and stability of the HPLC method were validated according to the US Food and Drug Administration (FDA) and European Medicines Agency (EMA) guidelines. The method was applied to the analysis of 58 trough serum samples from patients treated with voriconazole, and the HPLC-UV (ultraviolet) method was compared with an ARK immunoassay. The correlation of both methods was studied by the Pearson regression coefficient and the concordance of the values was evaluated by the Bland-Altman and Passing-Bablok methods.

Results

All validation parameters met the criteria set out in the FDA and EMA guidelines. The standard curve was linear over a concentration range of 0.25–16 µg/mL with a limit of quantification of 0.125 µg/mL. No interactions between voriconazole and other drugs was observed and voriconazole was stable after 1 month at −80°C. Comparison of the HPLC method and the enzyme immunoassay method showed a linear correlation with a systematic error of −0.61 µg/mL between both methods.

Conclusion

The method developed is simple and fast and can be easily applied for routine therapeutic drug monitoring of voriconazole. The HPLC-UV method was more sensitive than the immunoassay method and there was concordance with the immunoassay. Consequently both methods could be used, considering the correlation between them.

Introduction

Voriconazole is a triazole antifungal agent with potent broad-spectrum activity. It is currently recommended as a first-line agent for the treatment of invasive aspergillosis, fluconazole-resistant serious invasive Candida species infections, and infections caused by emerging pathogens, such as Fusarium and Scedosporium. ^1–4 Voriconazole has a narrow therapeutic index, large interpatient variability and non-linear pharmacokinetics. Several studies have found a relationship between trough plasma concentrations (Cmin) and clinical response in terms of efficacy as well as toxicity. Poor treatment outcomes have been reported in patients with Cmin <1 mg/L,^{5 6} concentrations >2 mg/L have been related to optimal responses,^7–9 and Cmin >4.5–5.5 mg/L have been associated with adverse events such as vision disturbances, rash and hepatotoxicity.^{5 10 11} In this context, therapeutic drug monitoring (TDM) represents a useful tool to optimise dosing regimens and clinical outcome.^{12 13}

Several factors contribute to the wide intervariability of voriconazole. These include genetic polymorphism in the CYPC2C19 gene, drug–drug interactions and patient characteristics such as age, weight and liver function.^{14 15} The large variability in voriconazole plasma concentrations together with the narrow therapeutic window make individualised dosing adjustments based on TDM necessary to optimise therapeutic response and to minimise the probability of toxicity.¹⁶

Different methods, including agar well diffusion bioassays, high-performance liquid chromatography (HPLC), liquid chromatography-mass spectrometry (LC-MS) and enzyme immunoassays (EIA) have been described for the determination of voriconazole in biological fluids. Table 1 shows the advantages and disadvantages of most widely used methods in clinical pharmacokinetic laboratories.

Table 1.

Methods for measuring voriconazole in human plasma

	HPLC	LC-MS	EIA
Advantages	Sensitive and specific Possibility of simultaneous analysis of multiple drugs Less expensive equipment than LC-MS	Highly sensitive and specific Possibility of simultaneous analysis of multiple drugs Small sample volume	Fast results No need for specific equipment No need for a sample preparation step
Disadvantages	Large sample volume to ensure sensitivity Subject to interference from multiple substances Requires a sample preparation step	Expensive equipment Requires a high degree of specialty and experience Not widely available Requires a sample preparation step	Low specificity High cost per determination
References	25–30	31–36	23 24

EIA, enzyme immunoassays; HPLC, high-performance liquid chromatography; LC-MS, liquid chromatography-mass spectrometry.

Currently, in clinical pharmacokinetic laboratories there is no standardisation regarding the method used for plasma voriconazole determination. The choice depends on the availability of equipment, reagents and staff experience. HPLC and EIA are the methods most commonly used. Given the recommendation to perform voriconazole TDM in clinical practice, it is essential to have chromatographic and immunoassay methods suitable for pharmacokinetic determination in clinical practice as well as for evaluating the relationship between both methods.

We therefore conducted a study to develop and validate an HPLC-UV method for measuring voriconazole in human plasma samples and to perform a comparison with an ARK immunoassay in order to evaluate the agreement between both methods.

Materials and methods

Development and validation of an HPLC-UV method for measuring voriconazole in human plasma

Voriconazole pure drug substance was kindly supplied by Pfizer (Pfizer SA, Madrid, Spain). HPLC grade acetonitrile was purchased from Panreac Química SAV (Castellar del Vallés, Spain). HPLC water from Millipore’s Milli-Q System was used throughout the analysis. Stock solutions of voriconazole were prepared in dimethyl sulfoxide (DMSO) (Sigma–Aldrich Quimica). Drug-free human serum from healthy donors was supplied by the blood bank department.

The HPLC system consisted of an Agilent 1260 series HPLC system (Agilent Technologies, USA) equipped with Diode Array Detector HS, a solvent delivery quaternary pump system, maximum pressure 400 bar and an autosampler with thermostat. The software model OpenLAB CDS 3D UV (PDA) was used for the data processing. The mobile phase consisted of a filtered and degassed mix of acetonitrile: ultrapure water (60:40, v/v). Chromatographic and detection conditions are described in table 2.

Table 2.

Chromatographic and instrumental conditions

Instrumental parameters	Conditions
Elution mode	Isocratic
Flow rate	0.8 mL/min
Volume of injection	50 µL
Wave length detection	255 nm
Column	SunFire C18 5 µm 4.6×150 mm
Guard column	SunFire C18 5 µm 4.6×20 mm
Temperature of the column	25°C
Temperature of the autosampler	25°C
Pressure of the system	900–1200 psi
Retention time voriconazole	3.20 min

Stock solutions of voriconazole were prepared for the calibration standards (CS) and quality control (QC) samples, respectively, by dissolving 40 mg of voriconazole in 25 mL of DMSO for each solution. Using this solution we prepared eight CS at final concentrations of 0.125, 0.25, 0.50, 1, 2, 4, 8 and 16 µg/mL by spiking the appropriate amounts of the voriconazole stock solution into drug-free human serum.

Both the CS and the samples of our patients were processed in the same way. Protein precipitation was performed by the addition of acetonitrile in a 1 to 1 ratio. The final mixture was then shaken for 30 s on a Vortex Shaker at maximum speed followed by centrifugation at 13 800 g at 25°C for 15 min.

For method comparison, the results were evaluated with the ARK Voriconazole Assay (ARK Diagnostics, Inc, Fremont, CA, USA) that is based on competition between the drug in the specimen and voriconazole labelled with the enzyme glucose-6-phosphate dehydrogenase (G6PDH) for binding to the antibody reagent. On binding to the antibody, G6PDH enzyme activity decreases. However, in the presence of the drug from the specimen, enzyme activity increases in a manner that is directly proportional to the drug concentration. Active enzyme converts the cofactor nicotinamide adenine dinucleotide (NAD) to NAD + hydrogen (NADH) that is measured spectrophotometrically at 340 nm. For this study, the ARK Voriconazole Assay was applied to an Architect c4000 clinical chemistry analyzer (Abbot Diagnostics, Illinois, USA), fully automated platform with random access (800 test/hour). CS and QC provided with the kit by ARK comprised a synthetic protein matrix with the following concentrations: 0.0, 0.5, 1.5, 4.0, 8.00 and 16.0 µg/mL (CS) and 1.0, 5.0 and 10.0 µg/mL (QS). CS and patient samples were centrifuged (1500 g for 10 min at 4°C) before analysis.

Validation procedure of the HPLC-UV method

The linearity, precision and accuracy of the assay were validated according to US Food and Drug Administration (FDA) and European Medicines Agency (EMA) guidelines.^{17 18}

Linearity was evaluated by analysing standard voriconazole solutions in the range of 0.125–16.0 µg/mL. Experiments were performed in triplicate and on two non-consecutive days. Linearity was assessed by linear regression.

Intra-/inter-day precision and accuracy of the analytical method were evaluated by triplicate processing and analysis of seven CS samples (0.25, 0.5, 1, 2, 4, 8 and 16 µg/mL). Precision was expressed as coefficient of variation, calculated as CV%=(standard deviation/mean of measured values)×100; whereas accuracy was expressed as a percentage of the relative error, determined with the formula RE%=([mean measured concentration−nominal concentration]/nominal concentration)×100. Criteria for acceptability of data included accuracy within ±15% deviation from the nominal values and precision within ±15% of CV%, except for the lower limit of quantification (LLOQ), for which values should not exceed 20% of CV%.

Limit of detection (LOD) is defined as the concentration at which the analyte can be distinguished from background signal. This was determined by measuring the peak area that was greater than or equal to the average of the blanks +3 SD. The limit of quantification (LOQ) was defined as the analyte response that was at least five times the response of a blank sample and whose precision and accuracy were within 20%.

Stability of voriconazole was tested in low (0.5 µg/mL), medium (2 µg/mL) and high (4.0 µg/mL) concentrations, after long-term storage (freeze for 1 month at −80°C). Percentage deviations of measured peak areas were compared with those obtained at the beginning of the study.

Clinical application and comparison with an ARK immunoassay method

The previously described method was applied to the analysis of 58 trough serum samples from patients treated with voriconazole. Using these 58 human samples, the HPLC method developed was compared with the routinely used ARK immunoassay carried out at the General University Hospital Castellón.

Blood samples of voriconazole were collected at steady state just before the next dose of voriconazole (trough level). After centrifugation (1500 g for 10 min at 4°C), the serum was further aliquoted into two polypropylene Eppendorf tubes and then stored at −80°C until analysis.

Correlation between HPLC and immunoassay methods

The correlation or linearity of both methods was studied by the Pearson regression coefficient. Subsequently, concordance of the values ​​obtained with the two methods was evaluated by the Bland-Altman method, which graphically represents the differences of the analysed concentrations with respect to their mean value. Finally, the concordance was also evaluated by the Passing-Bablok method, the coefficient of concordance of Lin (rc) was estimated and deviations from linearity were determined by the CUSUM test. Results were expressed as the median and IQR; a statistical study was performed using the Mann-Whitney U test for quantitative tests using the STATA/IC-14.1 program.

Patient care relied solely on the ARK method. The study was approved by the local ethics committee and written informed consent was obtained from each participant.

Results

Development and validation of an HPLC-UV method for measuring voriconazole in human plasma

The standard curve was linear over a concentration range of 0.25–16.0 µg/mL for voriconazole, with a correlation coefficient of 0.9999 (figure 1).

Figure 1.

Linearity of the assay. The plot represents standard concentrations of voriconazole (0.25–16 μg/mL) versus area under curve observed with a correlation coefficient of 0.9999.

The LLOQ and LOD of voriconazole was determined to be 0.25 µg/mL and 0.125 µg/mL, respectively. The precision and accuracy ranged from 5.19% to 8.96% and from −13.12% to 8.04%, respectively. The intra-day precision and inter-day precision for QC samples were <15%. The method also showed accuracy within 15%. The CV% was within the acceptable limits stated for bioanalytical method validation (table 3).

Table 3.

Intra-day and inter-day precision (expressed as coefficient of variation, CV%) and accuracy (expressed as mean percentage relative error, RE%) of determined voriconazole concentrations in serum quality controls

Cnom (μg/ml)	C (μg/ml) (mean±SD)	Precision (CV%)	Accuracy (RE%)
Intra-assay (within run) (n=3)
0.25	0.22±0.04	8.96	−12.12
0.5	0.46±0.04	6.09	−8.04
1	0.97±0.09	6.47	−2.54
2	2.01±0.17	6.51	0.15
4	4.04±0.25	5.19	1.09
8	8.10±0.71	7.39	1.27
16	15.94±0.99	5.31	−0.38
Inter-assay (between run) (n=6)
0.25	0.28±0.12	10.24	−0.58
0.5	0.56±0.19	6.92	1.90
1	1.05±0.16	12.53	2.86
2	2.08±0.21	9.01	3.96
4	4.11±0.26	5.91	4.68
8	8.15±0.58	7.02	11.03
16	15.90±0.80	5.04	11.13

Voriconazole-free serum samples did not show any interference with the signal. No interactions between voriconazole and matrix components were detected. In addition, when analysing patient samples, we did not observe any chromatographic interference with other drugs used in our patients.

Post-preparation stability testing showed a slight loss of stability after long-term storage at −80°C for 1 month in the lowest voriconazole concentrations, although in no case did the difference exceed 20% as proposed by the international guidelines (table 4).

Table 4.

Stability of voriconazole after 1 month long-term storage at −80°C

Cnom (μg/ml)	Peak area 1	Peak area 2	Peak area difference (%)
0.5	37	43	83
2	118	124	94
4	226	245	91

Clinical application and comparison with an ARK immunoassay method

The suitability of this analytical method to determine voriconazole concentration was investigated using a total of 58 trough serum samples. No concentrations below LLOQ (<0.25 µg/mL) were observed in our patients. Concentrations above 5.5 µg/mL (associated with potential toxic effects) were observed in 8.62% (5 out of 58) of samples. A high percentage of samples analysed (25.86%, 15 out of 58) showed concentrations <1 µg/mL, which corresponded to the minimal therapeutic targets for voriconazole proposed by Pascual et al.⁵

Correlation between the HPLC and ARK immunoassay method

These 58 serum samples were determined with both methods, obtaining a median concentration of 2.80 µg/mL (IQR 0.97–4.01) and of 3.52 µg/mL (95% CI 1.47–5.08) (p=0.082) in the determinations with HPLC and ARK immunoassay, respectively.

Comparison of the HPLC and ARK immunoassay methods showed a significant linear correlation (Pearson’s r: 0.90, p<0.0001). Passing-Bablok regression showed an intercept of −0.22 (95% CI −0.45 to −0.03) and a slope of 0.89 (95% CI 0.77–0.97), with a substantial agreement between both methods, as indicated by a concordance correlation coefficient of 0.91. The Bland-Altman plot revealed a small systematic error of −0.61 μg/mL (95% CI −2.03 to −0.82) between both methods. The CUSUM test was indicative of non-significant linearity deviation (p>0.20) (figure 2A, B).

Figure 2.

Bland-Altman plot (A) and Passing-Bablok regression (B).

Discussion

Therapeutic failure of invasive fungal infections is potentially life threatening. Achieving optimal voriconazole concentrations is essential to achieve clinical success. In this study we found a high percentage of patients with infra-therapeutic concentrations, similar to previous studies.⁶ We also found a high interpatient variability similar to other publications.^{19 20} The importance of TDM of antifungal agents is recognised by clinical guidelines, and accumulating evidence supports TDM for voriconazole, making it necessary to have adequate analytical methods for voriconazole measurement.^{2 21 22}

A simple and fast HPLC-UV method was developed and validated to quantify voriconazole in patient samples using a small volume of human serum and providing information for clinical decisions in less than 1 hour. The mobile phase is a mixture of water and acetonitrile without buffers or gradient separation. Therefore, the method is considered easy and simple. The method has met the validation criteria of regulatory agencies (EMA and FDA) but has also been validated by a comparison with an ARK immunoassay method used for the routine determination of voriconazole in a pharmacokinetic laboratory.

The ARK immunoassay consists of convenient, liquid-stable, ready-to-use reagents for homogeneous enzyme immunoassays, which makes it very attractive for pharmacokinetic determinations. Two previous studies have validated this method for voriconazole determination in human samples. These methods showed good performance in the concentration range between 1 and 5.5 µg/mL. However, our HPLC method guarantees higher sensitivity (LLOQ=0.25 µg/mL), so it can be applied for the determination of both therapeutic and prophylactic low concentrations. For this reason, we consider HPLC as the method of choice for the pharmacokinetic determination of voriconazole whenever it is available.

Despite the advantages of chromatographic methods, many clinical pharmacokinetic laboratories employ immunoassay techniques to determine plasma concentrations of voriconazole. Being able to correlate the results obtained between both techniques is very useful due to the variability in the methods of determination according to the laboratory, as mentioned above. We made a method comparison between the previously developed and validated HPLC method and an ARK immunoassay method used routinely. This comparison showed a significant linear correlation and a slight overestimation in the determination of voriconazole by the ARK immunoassay method compared with HPLC, but with a relationship between both methods that was linear and constant.

To our knowledge, there are only two previous studies that have made a comparison between chromatographic and EIA methods used for voriconazole determination. The first one is a validation of an ARK immunoassay method and comparison with an ultra-HPLC with photodiode array detention (UPLC).²³ In this study a linear correlation was observed between both methods, with a concordance correlation coefficient of 0.96. Similar to our study, the Bland-Altman plot also revealed a slight overestimation of the EIA over the UPLC with a systematic error of −0.29 mg/L. The second study is a validation of an ARK immunoassay method and comparison with liquid chromatography-tandem mass spectrometry (LC-MS/MS).²⁴ In this second study a good correlation was also observed, with a correlation coefficient of 0.98 between both methods.

There is no study that compares ARK immunoassay and conventional HPLC. Having this comparison is very useful when considering the availability of UPLC or mass spectrometry is unusual in clinical pharmacokinetic laboratories.

This study shows the existence of concordance between simple HPLC-UV and ARK immunoassay in the determination of voriconazole in patient serum samples. Like the results of these previous studies, voriconazole concentrations determined by HPLC or immunoassay could be used indistinctly, considering the correlation between the two methods. Clinical laboratories and hospitals that use one of these two techniques for the pharmacokinetic monitoring of voriconazole can use this information in their clinical practice. However, it should be taken into account that the analytical method must be correctly validated according to international guidelines, and we must also consider the limit of quantification of the method when making analytical interpretations.

In conclusion, having a sensitive and specific method for the determination of voriconazole is essential, since TDM is recommended to guarantee optimal antifungal therapy. This chromatographic method can be applied easily for routine TDM in pharmacokinetic laboratories. This simple and fast method can be offered to physicians to optimise drug dosage and improve clinical results.

What this paper adds.

What is already known on this subject?

• EIA (enzyme immunoassays) or HPLC-UV (high-performance liquid chromatography-ultraviolet) are two analytical methods used by laboratories for the pharmacokinetic determination of voriconazole. The use of one method or the other depends on the availability of the equipment and the experience of the personnel.

What this study adds?

• In this study, we developed and validated a simple HPLC-UV method for measuring voriconazole in human serum samples and propose a correlation with an ARK immunoassay.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.