Prospective Analysis of Factors Influencing Inter-Individual Variation in Trough Plasma Voriconazole Concentrations in Older Patients—Impact of High α1-Acid Glycoprotein Levels

Abstract

Voriconazole (VRCZ), an azole-based, deep-seated antifungal agent, is used as a 1st-line treatment for aspergillosis in Japan. VRCZ exhibits nonlinear pharmacokinetic (PK) behavior with relatively large inter-individual variability in plasma concentration. Additionally, genetic polymorphisms of CYP2C19 have been reported to influence the metabolic variability of VRCZ. The purpose of this study was to search for and identify clinically relevant potential factors influencing the PK and plasma concentration of VRCZ to better inform VRCZ dosing regimens. Thirty patients receiving VRCZ were enrolled. Total (C_t) and unbound (C_u) trough plasma concentrations of VRCZ were determined by the HPLC-UV method. Univariate and multivariate correlation analyses were used to evaluate the relationships between C_t or C_t/dose per body weight (C_t/D) and individual demographic and laboratory characteristics. Since the increasing trend of C-reactive protein (CRP) inversely correlated with the classification of CYP2C19 gene polymorphisms, it was suggested that the inflammation counteracted the trend of C_t according to CYP2C19 gene polymorphisms. Spearman’s rank-order correlation analysis showed significant correlations between C_t and dose per body weight, CRP, and α1-acid glycoprotein (α1-AGP). Multivariate linear regression analysis showed that age, dose per body weight, CRP, and α1-AGP were significant explanatory factors for C_t. In particular, elevated α1-AGP levels were found to have significant explanatory value for decreased C_t. Although the present study has critical limitations, such as the patient sample was small in size and being limited to a single medical institution, this finding may explain some of the inter-individual variability in plasma VRCZ concentration.

INTRODUCTION

Voriconazole (VRCZ) is an azole-based, deep-seated antifungal agent with broad-spectrum antifungal activity and is recommended as a 1st-line treatment for aspergillosis. The clinical efficacy of VRCZ has been associated with plasma concentrations in patients with deep-seated fungal infections, including invasive pulmonary aspergillosis. VRCZ is currently the only antifungal agent in Japan for which therapeutic drug monitoring (TDM) can be performed, and dose adjustment based on trough plasma concentration measurements is recommended. The effective therapeutic range of trough plasma VRCZ concentration is recommended to be ≥1.0–2.0 μg/mL, and a trough plasma concentration of ≥4.0–5.0 μg/mL has been reported to increase the risk of visual impairment and liver damage.^1–3) In particular, the incidence of liver damage caused by VRCZ has been reported to be strongly correlated with plasma concentration.⁴⁾ On the other hand, because VRCZ exhibits nonlinear pharmacokinetic behavior, unexpectedly abnormally high plasma concentrations may be observed.⁵⁾ In addition, the enzymatic activity of the molecular species of CYP, CYP2C9, and CYP2C19, which are the main metabolizing enzymes for VRCZ, can vary significantly among individuals, resulting in a large variability in the pharmacokinetics and steady-state plasma concentrations of VRCZ among different patients.⁶⁾ Therefore, there is a characteristic drug–drug interaction between VRCZ and other medications metabolized by CYP2C9 and CYP2C19,⁷⁾ and TDM results should be considered when determining the safe concentration range and dose adjustment of VRCZ, especially when other medications metabolized by CYP2C9 and CYP2C19 are used in combination with VRCZ.

However, many Japanese facilities rely on subcontractors to measure plasma VRCZ concentrations, and it takes several days from sample submission until the measurement results are received; consequently, high plasma concentrations are sometimes detected only after the appearance of adverse effects, such as liver damage. It is important to have a system in place that allows for TDM to be performed at an early stage to assess treatment efficacy and safety. If clinical factors that influence inter-individual variability in plasma VRCZ concentration are known at the start of dosing, these factors can be considered in the initial dose setting.

Due to the existence of genetic polymorphisms in CYP2C19, a major metabolizing enzyme for VRCZ, it has been reported that poor metabolizers (PMs) with defective or reduced metabolic capacity due to genetic polymorphisms have a slower rate of VRCZ metabolism, resulting in abnormally high plasma concentrations, which is one of the major determinants of toxicity.⁸⁾ The frequency of PM is relatively higher in Asians⁹⁾; a genetic analysis of Japanese subjects revealed that the frequency of PM was as high as 18.8%,¹⁰⁾ and data collected from 100 Japanese subjects in a domestic phase III trial were used as the population data.¹⁰⁾ The VFEND®TDM calculation tool (Pfizer Japan Inc., Tokyo, Japan), a pharmacokinetic analysis software for VRCZ, uses data collected from 100 Japanese subjects as a population and creates dosing regimens that vary according to 3 types of CYP2C19 gene polymorphisms.¹¹⁾

In addition, demographic and clinical characteristics considered as covariates for steady-state plasma concentrations of VRCZ in the analysis of population pharmacokinetics in the domestic phase III study included age, body weight, dose, and serum albumin. The maximum rate of disappearance (V_max) of VRCZ decreased with a decrease in albumin, which may indicate a decrease in liver function.¹²⁾ However, the large inter-individual variability in the pharmacokinetics and steady-state plasma concentration of VRCZ is believed to be due not only to nonlinear pharmacokinetic behavior or genetic polymorphisms of CYP2C19 but also to unexpected potential factors contributing to the variability. For example, it has been reported that plasma VRCZ concentrations are extremely high in patients with severely impaired liver function^13,14) and in those with high levels of C-reactive protein (CRP) and interleukin-6 (IL-6) due to systemic inflammation.¹⁵⁾ Therefore, in the present study, we aimed to investigate other endogenous factors whose plasma concentrations are known to fluctuate during inflammation. Among these factors, we focused on α1-acid glycoprotein (α1-AGP) because, as a basic drug, VRCZ has a high affinity for α1-AGP^16,17) and may show a higher protein binding ratio to α1-AGP than to albumin in the blood circulation.¹⁸⁾ In other words, α1-AGP is an acute-phase protein whose levels increase dramatically in inflammatory conditions and diseases, such as infection and cancer, and it is believed that the pharmacokinetics and steady-state plasma concentration of VRCZ may be affected by these diseases.¹⁹⁾

With this in mind, we sought to comprehensively search for clinical factors that may influence the pharmacokinetics and inter-individual variability in the trough plasma concentrations of VRCZ and to identify statistically significant and clinically relevant factors, with the objective of using these factors to better inform future VRCZ dosing regimens.

PATIENTS AND METHODS

Patients and Data Collection

This study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Clinical Research Review Committee of Sakai City Medical Center (Project Identification Code: 20-200). Informed consent was obtained from all patients enrolled in the study. This study was performed at a single institution.

Thirty Japanese patients who had received VRCZ (VFEND®, Pfizer Japan Inc.) therapy (8 intravenously and 22 orally) for prophylaxis in cases of hematological cancer or treatment of pulmonary aspergillosis infection and underwent TDM testing at Sakai City Medical Center from July 2020 to September 2023 were enrolled in this study. The TDM data of patients taking VRCZ were collected prospectively. Patients were >40 years in age, had not received dialysis, had undergone genotyping, and were not taking drugs known to affect VRCZ pharmacokinetics.²⁰⁾ Some patients were taking proton pump inhibitors (PPIs), H₂-blockers, prednisolone (PSL), venetoclax, cyclosporine A, and tacrolimus; however, they were not taking PPIs, such as omeprazole, which affect the metabolism of VRCZ.²¹⁾

Each patient was treated with 100–300 mg of VRCZ twice daily. VRCZ was administered either orally or intravenously between meals. The initial loading dose on the 1st day and constant doses from the 2nd day were determined according to the contents of the package insert and the physician’s clinical judgment. That is, on the 1st day of administration, the intravenous loading dose was 6 mg/kg twice daily (12 mg/kg/d in total), and from the 2nd day of administration onwards, the constant dose was continued at 3–4 mg/kg twice daily (6–8 mg/kg/d in total). In the case of oral administration, the loading dose was 300 mg twice daily (600 mg/d in total) on the 1st day of administration, and from the 2nd day of administration onwards, the constant dose was 150–200 mg twice daily (300–400 mg/kg/d in total). The 1st blood samples were collected in disodium EDTA tubes just prior to the subsequent oral or intravenous administration from days 4 to 7 after initiation of VRCZ therapy, when the plasma concentrations measured as initial trough values would have reached a steady state. Because VRCZ is administered intravenously or orally between meals, the time for the 2 daily doses was set at 10 : 00 and 20 : 00. Therefore, the time for blood sampling to determine the trough concentration was set at 09 : 30, immediately before the 10 : 00 dose.

Subject of Research

Data on the initial trough plasma concentration of VRCZ (both total concentration [C_t] and unbound concentration [C_u]) measured up to 7 d after the initiation of VRCZ therapy, age, sex, body weight, dose per body weight, albumin, total bilirubin, albumin–bilirubin (ALBI) score, aspartate transaminase (AST), alanine transaminase (ALT), α1-AGP, CRP, and creatinine clearance (CL_cr) estimated by the Cockcroft–Gault equation, and the 3 genetic polymorphism groups (poor metabolizer [PM], heterozygous extensive metabolizer [HEM], extensive metabolizer [EM]) for CYP2C19 were collected. Age, body weight, and dose per body weight were considered at the initiation of VRCZ therapy. Laboratory values, such as albumin, total bilirubin, ALBI scores, AST, ALT, α1-AGP, CRP, and CL_cr, were assessed on the day of TDM for VRCZ.

The genotypes of patients were determined by BML Inc. (Osaka, Japan), where the CYP2C19 genotype is a phenotype predicted based on the following genotypes: EM: CYP2C19*1/*1 or CYP2C19*1/*17; HEM: CYP2C19*1/*2 or CYP2C19*1/*3; and PM: CYP2C19*2/*2, CYP2C19*2/*3, or CYP2C19*3/*3.

As an assessment of hepatic reserve capacity, the ALBI scores were calculated based on albumin and total bilirubin using the following formula^14,22) and were evaluated in 3 grades: the higher the ALBI score (i.e., ALBI grade 3), the poorer the hepatic reserve capacity.

  

$$\begin{array}{ll}\text{ALBI} \text{score} \text{=}& \text{0.66}\times {\text{Log}}_{\text{10}} \text{(total} \text{bilirubin} \text{[mg/dL]}\times \text{17.1)}\\ & \text{– (0.085}\times \text{albumin [g/dL]}\times \text{10)}\end{array}$$

The scores were graded as follows: grade 1, ALBI score ≤ −2.60; grade 2, −2.60 < ALBI score ≤ −1.39; and grade 3, −1.39 < ALBI score.

Measurement of C_t and C_u of VRCZ

Blood samples were centrifuged at 3000 × g for 10 min at 4°C, and the separated plasma was stored at −30°C until analysis. Plasma concentrations of VRCZ were determined using an HPLC-UV system (Prominence-i LC-2030; Shimadzu, Kyoto, Japan). For determining C_t, 200 μL of acetonitrile was added to a 100-μL plasma sample, and the mixture was stirred for 1 min by vortexing, followed by centrifugation at 12000 × g for 10 min at 4°C. The supernatant was then collected, stirred again for 30 s by vortexing, and 100 μL was injected into the HPLC-UV system.²³⁾ For measuring C_u, Centrifree-30K tubes (Millipore, Bedford, MA, U.S.A.) were prepared by adding 500 μL of distilled water to the sample reservoir and centrifuging for 30 min at 2000 × g and 20°C to remove glycine, as recommended by the supplier. Subsequently, a 300-μL plasma sample was added to the sample reservoir of the Centrifree-30K unit. The Centrifree-30K tubes were then centrifuged for 30 min at 2000 × g and 37°C. The solution in the filtration collection cup was collected and stirred again for 30 s by vortexing, and 100 μL was injected into the HPLC-UV system.²⁴⁾ The column used was COSMOSIL 5C18-AR-II (4.6 mm id × 250 mm; Nacali Tesque Ltd., Kyoto, Japan), and the mobile phase consisted of acetonitrile and 10 mmol/L sodium phosphate buffer (pH 3.8) in a 45 : 55 ratio. The flow rate was 1.0 mL/min, and the column temperature was 40°C. The UV detector was set at a detection wavelength of 262 nm. Calibration curves for C_t and C_u were prepared using human pooled blank plasma and phosphate-buffered saline at 6 concentrations of 0.2, 0.5, 1.0, 2.0, 5.0, and 10.0 μg/mL, respectively, and showed good linearity in this range with R² > 0.999. The intra- and inter-day accuracy values, expressed as percent coefficient of variation (CV), were all within ±10%, and precision values (as percent CV) were all <10% in each calibration curve.

Statistical Analysis

The normality of distribution for all data was assessed and confirmed using the Shapiro–Wilk normality test. Normally distributed data for both demographic and laboratory values were presented as the mean ± standard deviation (S.D.). Since C_t was a variable with a normal distribution, an independent-samples 2-tailed Student’s t-test was used for comparison of C_t of VRCZ between 2 groups treated with intravenous or oral routes, and one-way ANOVA/Tukey’s multiple comparisons test was used for multiple comparisons of C_t among 3 groups taking PPIs or PSL in combination with VRCZ, not taking them in combination with VRCZ, and, in the same way, C_t among 3 CYP2C19 genotypes. In the case of a skewed distribution, data were presented as the medians (interquartile ranges).

Spearman’s rank-order correlation analysis was employed to assess univariate correlations between C_t and each factor among the demographic and laboratory values, and the results were expressed as Spearman’s p values. Multiple regression analyses were performed to identify potential factors that have a statistically significant effect on C_t among all the factors listed in Supplementary Table S2 as explanatory variables. That is, factors with low statistical contribution were sequentially excluded from the explanatory variables based on the results of each analysis, and multiple regression analysis was repeated to narrow down the number of explanatory variables. The processes in which the number of explanatory variables was 7 or fewer are shown in the results. Among all possible regression models, the best multiple regression equation was adopted, with which significant differences (p value of coefficient) were found in all explanatory variables, and both the degree of freedom-adjusted coefficient of determination (R²) and Akaike’s information criterion (AIC) were considered for comparing the goodness of fit to the data among models.

Additionally, the predicted C_t by the multiple regression equation was compared for one-to-one correspondence with the observed C_t. The predictability was examined by comparing the observed and predicted C_t using the mean prediction error (ME) (μg/mL) as a measure of bias and root mean squared error (RMSE) (μg/mL) as a measure of precision, which were calculated using Eqs. (1) and (2), respectively.²⁵⁾

  

$$ME=\frac{1}{N}\sum _{i=1}^N(C_{Pred,i}-C_{Obs,i})$$(1)

  

$$RMSE=\sqrt{\frac{1}{N}\sum _{i=1}^N{(C_{Pred,i}-C_{Obs,i})}^2}$$(2)

In Eqs. (1) and (2), N is the number of data points, and C_Pred,i and C_Obs,i are the predicted and observed values of C_t, respectively, in the ith patient.

The results with p values of <0.05 were considered statistically significant. Statistical analyses were all performed with Bell Curve for Excel, version 4.08 (Social Survey Research Information Co., Ltd., Tokyo, Japan).

RESULTS

Clinical Characteristics of Enrolled Patients

The 30 enrolled patients were predominantly older adults; 23 were men, and the median age was 73.0 (63.5–76.0) years (Table 1). Thirteen patients had hematological cancer, and the other 17 mainly had pulmonary aspergillosis and were being treated with prednisone or other steroids. No adverse events related to the present patients were observed during the course of this study. Body weight and dose/body weight were 57 ± 12 and 3.0 ± 0.6 mg/kg, respectively. C_t and C_u, measured as the initial trough levels, were 4.7 ± 2.3 and 2.0 ± 1.1 μg/mL, respectively, and the unbound fraction ratio was 43.0 ± 9.1%.

Table 1. Demographic, Clinical, and Laboratory Characteristics of All Patients in This Study

Indicators
Number of patients	30
Male-to-female ratio	Male : female = 23 : 7
Administration route	Intravenous: 8; per oral: 22
Purpose of administration	Treatment: 17; prevention: 13
Age (year)	73.0 (63.5–76.0)
Body weight (kg)	57 ± 12
Dose per body weight (m/kg)	3.0 ± 0.6
Total plasma trough concentration (μg/mL)	4.7 ± 2.3
Unbound plasma trough concentration (μg/mL)	2.0 ± 1.1
Unbound fraction ratio (%)	43.0 ± 9.1
Ct/D ([μg/mL]/[mg/kg])	1.5 ± 0.7
Albumin (g/dL)	2.7 ± 0.7
α1-AGP (mg/dL)	168 ± 33
Total bilirubin (mg/dL)	0.50 (0.39–0.76)
AST (IU/L)	26 (18–34)
ALT (IU/L)	22 (13–31)
CRP (mg/dL)	4.9 (3.9–9.0)
CLcr (mL/min)	63 ± 32
ALBI score	−1.8 ± 0.6
Number of patients by ALBI grade	Grade 1 : 5; grade 2 : 18; grade 3 : 7
Ct in PM (μg/mL), n = 4	4.8 ± 1.2
Ct in HEM (μg/mL), n = 19	4.4 ± 2.7
Ct in EM (μg/mL), n = 7	5.6 ± 1.5
CRP in PM (mg/dL), n = 4	4.1 (3.0–5.5)
CRP in HEM (mg/dL), n = 19	4.8 (3.9–7.4)
CRP in EM (mg/dL), n = 7	12.7 (8.2–17.3)
Ct after intravenous route (μg/mL), n = 8	5.0 ± 2.1
Ct after oral route (μg/mL), n = 22	4.0 ± 2.1
Ct treated with VRCZ only (μg/mL), n = 15	5.1 ± 2.4
Ct treated with VRCZ + PPIs (μg/mL), n = 8	4.4 ± 1.7
Ct treated with VRCZ + PSL (μg/mL), n = 7	4.3 ± 2.8

ALBI scores were calculated from the albumin and total bilirubin levels to determine hepatic reserve capacity, as follows: [0.66 × Log₁₀ (17.1 × total bilirubin) − 0.085 × 10 × albumin]; therefore, ALBI grades were evaluated as follows: grade 1, ALBI score ≤ −2.60; grade 2, −2.60 < ALBI score ≤ −1.39; and grade 3, −1.39 < ALBI score. The CYP2C19 genotypes included 3 phenotypes predicted based on the genotypes, as follows: poor metabolizer (PM): CYP2C19*2/*2, CYP2C19*2/*3, or CYP2C19*3/*3; heterozygous extensive metabolizer (HEM): CYP2C19*1/*2 or CYP2C19*1/*3; and extensive metabolizer (EM): CYP2C19*1/*1 or CYP2C19*1/*17. α1-AGP, α1-acid glycoprotein; ALBI, albumin–bilirubin; ALT, alanine transaminase; AST, aspartate transaminase; C_t/D, [total trough plasma concentration (μg/mL)]/[dose per body weight (mg/kg)]; CL_cr, creatinine clearance; CRP, C-reactive protein; PPIs, proton pump inhibitors; PSL, prednisolone.

Albumin and total bilirubin levels were 2.7 ± 0.7 g/dL (reference range: 3.9–4.9 g/dL) and 0.50 (0.39–0.76) mg/dL (reference range: 0.4–1.5 mg/dL), respectively. The ALBI grades with VRCZ treatment were as follows: grade 1, 5 patients; grade 2, 18 patients; and grade 3, 7 patients. These grades showed that a relatively larger number of patients had low hepatic reserve capacity. On the other hand, AST and ALT levels were 26 (18–34) and 22 (13–31) IU/L, respectively.

The α1-AGP level was 168 ± 33 mg/dL (reference range: 42–93 mg/dL), and the CRP level was 4.9 (3.9–9.0) mg/dL (reference range: 0.3–0.9 mg/dL); a relatively larger number of patients exhibited hypoalbuminemia and hyper-α1-AGPemia, potentially due to inflammatory conditions. The CL_cr was 63 ± 32 mL/min (reference range: 100–120 mL/min); CL_cr values were reduced because the study group included relatively older patients.

Inflammation Effect on Trend According to CYP2C19 Gene Polymorphism

The 3 types of CYP2C19 gene polymorphisms were as follows: PM, 4 patients; HEM, 19 patients; and EM, 7 patients. Although it was expected that C_t and C_t/D would be higher in patients with PM and lower in those with EM, the ANOVA did not reveal any significant differences in C_t (mean ± S.D.) among the 3 CYP2C19 genotypes (Table 1 and Supplementary Table S1). Therefore, Spearman’s rank-order correlation analysis was employed to find out which demographic and laboratory values were effective in eliminating the trends of C_t dependent on CYP2C19 genetic polymorphisms in the present patient group, indicating that there was a significant correlation (p = 0.0253) only between CRP and the classification (PM, HEM, and EM) of CYP2C19 gene polymorphisms (Supplementary Table S2). Additionally, the Kruskal–Wallis test/Scheffe’s multiple comparisons test was used for multiple comparisons of both demographic and laboratory values among the 3 CYP2C19 genotypes. From the results, the CRP levels were the only clinical information found to significantly increase in accordance with and be inversely correlated with the classification of CYP2C19 gene polymorphisms; p = 0.7783 between PM and HEM, p = 0.0454 between PM and EM, and p = 0.0344 between HEM and EM (Table 1 and Supplementary Table S1). Thus, an inflammatory condition identified by an increase in CRP was suggested to have a strong influence on the trend in the change of C_t according to CYP2C19 gene polymorphisms in the present patient group.

Factors Affecting Inter-Individual Variations in Trough Plasma Concentration

Difference in Administration Routes and Effect of Concomitant Medications

Since 8 patients had received VRCZ intravenously and 22 patients had taken it orally, statistical analysis was performed to confirm whether there was a difference in C_t between the 2 groups. From the result, no significant difference was revealed by the 2-tailed Student’s t-test (p = 0.1039) in C_t between the intravenous (5.0 ± 2.1 μg/mL) and oral (4.0 ± 2.1 μg/mL) routes of VRCZ (Table 1). Therefore, bias dependent on the administration route was not a concern in this study. Additionally, since 8 and 7 patients took PPIs and PSL independently with VRCZ, respectively, and 15 patients took only VRCZ, ANOVA analysis was performed to confirm whether there were effects of concomitant medicines on C_t of VRCZ among the 3 groups (Table 1). From the results, no significant differences were revealed by the one-way ANOVA/Tukey test in C_t among the 3 groups [p = 0.7597 between VRCZ (5.1 ± 2.4 μg/mL) and VRCZ + PPIs (4.4 ± 1.7 μg/mL); p = 0.7593 between VRCZ (5.1 ± 2.4 μg/mL) and VRCZ + PSL (4.3 ± 2.8 μg/mL); p = 0.9996 between VRCZ + PPIs (4.4 ± 1.7 μg/mL) and VRCZ + PSL(4.3 ± 2.8 μg/mL)]. Therefore, the effects of PPI and PSL medication on C_t of VRCZ did not need to be considered in the present study.

Correlations in the Univariate Analysis

Spearman’s rank-order correlation analyses were performed to examine the correlations of C_t with body weight, gender, age, dose per body weight, albumin, total bilirubin, ALBI scores, α1-AGP, CL_cr, genetic polymorphism of CYP2C19, and CRP (Supplementary Table S2). C_t was significantly correlated with dose per body weight (Fig. 1), and among the demographic and clinical characteristics, it was significantly correlated with α1-AGP and CRP levels (Fig. 2). Similarly, C_t/D was significantly correlated only with α1-AGP among the demographic and clinical characteristics (Fig. 3). Particularly, C_t and C_t/D were significantly negatively correlated with α1-AGP (Figs. 2 and 3). Accordingly, on examining the relationships between α1-AGP and other demographic and clinical characteristics while considering covariates, α1-AGP was significantly negatively correlated with albumin (Fig. 4); therefore, there was a significant positive correlation between the α1-AGP level and ALBI score (Fig. 4). However, no significant correlations were found between C_t and albumin or ALBI. On the other hand, CRP was significantly correlated with both albumin and total bilirubin levels as covariates (Fig. 5), suggesting that the liver function of these patients was severely affected by acute inflammation.

Fig. 1. Correlations between Total Trough Plasma Concentrations and Dose per Body Weight of VRCZ

Statistical significance: p = 0.0019 by Spearman’s rank-order correlation analysis and Spearman correlation coefficient = 0.544. C_t, total trough plasma concentration.

Fig. 2. Correlations of Total Trough Plasma Concentrations of VRCZ with Age, CRP, α1-AGP, ALB, T-Bil, and ALBI Score

Statistical significance for CRP and α1-AGP: p = 0.0359 and p = 0.0106 by Spearman’s rank-order correlation analysis and Spearman correlation coefficients = 0.385 and −0.460, respectively. α1-AGP, α1-acid glycoprotein; ALB, albumin; ALBI, albumin–bilirubin; C_t, total trough plasma concentration; CRP, C-reactive protein; T-Bil, total bilirubin.

Fig. 3. Correlations between Total Trough Plasma Concentrations/(Dose per Body Weight) of VRCZ and Age, CRP, α1-AGP, ALB, T-Bil, and ALBI Score

Statistical significance for α1-AGP: p = 0.0357 by Spearman’s rank-order correlation analysis and Spearman correlation coefficient = −0.385. C_t, total trough plasma concentration.

Fig. 4. Correlations between α1-AGP and Unbound Fraction of VRCZ, Age, CRP, ALB, T-Bil, and ALBI Score

Statistical significance for ALB (p = 0.0202) and ALBI score (p = 0.0326), respectively, indicating that the correlation between α1-AGP and ALBI scores was highly dependent on that between α1-AGP and ALB.

Fig. 5. Correlations between CRP and Age, ALB, T-Bil, and ALBI Score

Statistical significance for ALB (p = 0.0061) and T-Bil (p = 0.0427), indicating that the correlations between CRP and ALB or T-Bil were strongly related with impaired liver function in patients receiving VRCZ.

Multivariate Analysis to Identify Explanatory Variables

In the multiple regression analysis, 7 explanatory variables, including age, dose per body weight, α1-AGP, CRP, albumin, total bilirubin, and ALBI scores, were selected as most suitable for accurately assessing C_t of VRCZ. Among several explanatory variables in Models 1–5 (Table 2), multiple regression analysis showed that age, dose per body weight, α1-AGP, and CRP (Model 4) were the most plausible explanatory variables with a significant influence based on the p value of the coefficient of each variable, degree of freedom-adjusted coefficient of determination (R²), and AIC. Specifically, since CRP and total bilirubin were covariates, one of them had to be adopted.

Table 2. Estimated Parameters Based on Multiple Linear Regression for Accurately Assessing Total Trough Plasma Concentrations of Voriconazole Using 4 to 7 Explanatory Variables, Including Age, Dose/Body Weight, CRP, α1-AGP, Albumin, Total Bilirubin, and ALBI Score among All Patients (n = 30)

Model 1: Seven explanatory variables for the objective variable of total trough plasma concentrations

Explanatory variable	Regression coefficient	S.E.	Standardized coefficient	95% CI of coefficient	p Value of coefficient
Age	0.053	0.019	0.718	0.014–0.093	0.0105*
Dose/body weight	1.405	0.412	0.816	0.553–2.257	0.0024**
α1-AGP	−0.034	0.007	−1.106	−0.047 to −0.020	<0.001***
CRP	0.121	0.055	0.222	0.007–0.235	0.0387*
Total bilirubin	1.525	1.180	0.189	−0.917–3.966	0.2093
Albumin	0.840	0.652	0.447	−0.508–2.188	0.2102
ALBI score	0.868	0.753	0.317	−0.690–2.426	0.2610
Intercept	0.000

Statistical significance of regression coefficients: *p < 0.05, **p < 0.01, ***p < 0.001.

R² = 0.9319, R = 0.9653, AIC = 24.69, significance: p < 0.001 for the multiple regression equation.

Model 2: Six explanatory variables for the objective variable of total trough plasma concentrations

Explanatory variable	Regression coefficient	S.E.	Standardized coefficient	95% CI of coefficient	p Value of coefficient
Age	0.058	0.019	0.774	0.018–0.097	0.0057**
Dose/body weight	1.269	0.397	0.737	0.449–2.088	0.0039**
α1-AGP	−0.033	0.007	−1.088	−0.047 to −0.020	<0.001***
CRP	0.107	0.054	0.195	−0.005–0.218	0.0605
Total bilirubin	2.084	1.083	0.259	−0.151–4.320	0.0662
Albumin	0.190	0.328	0.101	−0.488–0.867	0.5688
Intercept	0.000

Statistical significance of regression coefficients: **p < 0.01, ***p < 0.001.

R² = 0.9309, R = 0.9649, AIC = 24.37, significance: p < 0.001 for the multiple regression equation.

Model 3: Five explanatory variables for the objective variable of total trough plasma concentrations

Explanatory variable	Regression coefficient	S.E.	Standardized coefficient	95% CI of coefficient	p Value of coefficient
Age	0.060	0.018	0.808	0.023–0.097	0.0029**
Dose/body weight	1.352	0.365	0.785	0.601–2.104	0.0011**
α1-AGP	−0.033	0.006	−1.065	−0.046 to −0.020	<0.001***
CRP	0.096	0.050	0.176	−0.007 to 0.199	0.0674
Total bilirubin	2.154	1.062	0.267	−0.034 to 4.341	0.0533
Intercept	0.000

Statistical significance of regression coefficients: **p < 0.01, ***p < 0.001.

R² = 0.9328, R = 0.9658, AIC = 22.78, significance: p < 0.001 for the multiple regression equation.

Model 4: Four explanatory variables including age, dose/body weight, α1-AGP, and CRP for the objective variable of total trough plasma concentrations

Explanatory variable	Regression coefficient	S.E.	Standardized coefficient	95% CI of coefficient	p Value of coefficient
Age	0.054	0.019	0.727	0.015–0.093	0.0086**
Dose/body weight	1.628	0.358	0.946	0.892–2.365	<0.001***
α1-AGP	−0.030	0.007	−0.976	−0.043 to −0.016	<0.001***
CRP	0.147	0.046	0.270	0.053–0.241	0.0034**
Intercept	0.000

Statistical significance of regression coefficients: **p < 0.01, ***p < 0.001.

R² = 0.9259, R = 0.9622, AIC = 24.88, significance: p < 0.001 for the multiple regression equation.

Model 5: Four explanatory variables including age, dose/body weight, α1-AGP, and total bilirubin for the objective variable of total trough plasma concentrations

Explanatory variable	Regression coefficient	S.E.	Standardized coefficient	95% CI of coefficient	p Value of coefficient
Age	0.062	0.019	0.840	0.023–0.101	0.0030**
Dose/body weight	1.286	0.381	0.747	0.502–2.069	0.0023**
α1-AGP	−0.032	0.007	−1.032	−0.045 to −0.018	<0.001***
Total bilirubin	3.184	0.961	0.395	1.208–5.159	0.0027**
Intercept	0.000

Statistical significance of regression coefficients: **p < 0.01, ***p < 0.001.

R² = 0.9247, R = 0.9616, AIC = 25.35, significance: p < 0.001 for the multiple regression equation.

Age, dose per body weight, and CRP were identified as factors in the positive direction, and α1-AGP as a factor in the negative direction (Table 2). The multiple regression equation for predicting C_t using the regression coefficients of the explanatory variables is shown as follows:

  

$$\begin{array}{rl}\text{Predicted}{\text{C}}_{\text{t}}= & 1.628\times (\text{dose/body weight})+0.054\times \text{age}\\ & + 0.147\times \text{CRP}-0.030\times \alpha 1\text{-AGP}\end{array}$$

The bias and precision of predicted C_t were 0.005 μg/mL (ME) and 1.325 μg/mL (RMSE), respectively, among all patients using the multiple regression equation. These results indicated that the predictability by the multiple regression equation in this study was close to a 1 : 1 relationship despite a certain degree of variation (Supplementary Fig. S1). Additionally, the validation of this prediction formula was performed for a different patient group at the Sakai City Medical Center, namely, a group of 10 similar new patients who were admitted after the study period for the current patients (admitted within the entire approved study period). The result of the validation was added as Supplementary Fig. S1 in the Supplementary Materials, where the ME and RMSE for the new data set were 0.205 and 1.593 μg/mL, respectively, and no significant deviation was observed (Supplementary Fig. S1). Therefore, the validity and applicability of the current prediction formula were likely to be sufficient; however, continuous verification of the formula is being carried out using data from the patient group accumulated at our hospital.

DISCUSSION

There was no significant difference in C_t between the intravenous and oral routes of VRCZ, as revealed by the 2-tailed Student’s t-test (p = 0.1039), indicating that the bioavailability of VRCZ after the oral route was relatively high in the present patients, which was consistent with the previous report.²⁶⁾

Additionally, there were no significant differences in the C_t of VRCZ among the 3 groups of CYP2C19 genetic polymorphisms. Specifically, the patients in this study were relatively older, and a larger proportion of them had reduced liver function based on the ALBI grade.^12–14) Additionally, the high levels of both CRP and α1-AGP suggested that VRCZ treatment was initiated under conditions involving an induced inflammatory status.¹⁵⁾ Increase in CRP due to acute inflammation, which was correlated with both the albumin and total bilirubin levels, was suggested to impair the liver function of these patients because CRP was produced in the hepatocytes of the liver by IL-6 activity and was increased in the blood during inflammatory reactions.^13,15) From the previous studies, inflammatory conditions identified by an increase in CRP were also suggested to counteract the trend in the change of C_t according to CYP2C19 gene polymorphisms in the present patients.^27–32) The overlap of these characteristics and backgrounds suggests that the CYP2C19 genetic polymorphisms did not affect the rate of VRCZ metabolism and C_t in the patients in the present study.^{12,13,15,28,32)}

On the other hand, the C_t of VRCZ was significantly correlated with dose per body weight, α1-AGP, and CRP (Figs. 1 and 2), and C_t/D was significantly correlated with α1-AGP (Fig. 3), with CRP as a covariate for albumin and total bilirubin (Fig. 5). These results suggest that a persistent inflammatory state in an older patient population leads to a decline in liver function.^13,15) Additionally, both C_t and C_t/D exhibited significant negative correlations with α1-AGP (Figs. 2, 3), and there was also a significant negative correlation between α1-AGP and albumin (Fig. 4). These results suggest that increased α1-AGP in conjunction with decreased albumin affected the pharmacokinetics of VRCZ through its stronger protein-binding affinity with VRCZ.^16,18)

On the other hand, the results of the multiple regression analysis showed that 4 factors—dose per body weight, age, CRP, and α1-AGP—had a significant effect on C_t. Relationships between trough plasma concentrations of VRCZ and dose per body weight, age, and CRP have been reported by several previous studies.^27–35) However, in the present study, higher α1-AGP levels were newly identified as a factor resulting in lower trough plasma concentrations.

The plasma protein binding ratio of VRCZ is generally considered to be approximately 50–60%,^17,36) although it varies under certain different conditions, such as pathological conditions and assay methods, as well as between humans and animals. In the present study, the plasma protein binding ratio of VRCZ was 57.0 ± 12.0%, which was in good agreement with that in previous reports^17,36) despite the large fluctuations in α1-AGP and albumin levels in individual patients. Recent reports, examining the relationship between adverse events and C_t or C_u of VRCZ, have shown that VRCZ has a higher affinity for α1-AGP than for albumin and that there is a significant correlation with the binding property of VRCZ, particularly in liver disease.^16,17,37) However, the number of reports on the C_u of VRCZ remains small, while the C_t of VRCZ is often assessed in clinical practice; therefore, the present study focused on the assessment of C_t.

As an acute phase protein, α1-AGP is synthesized mainly in the liver, and its plasma concentrations remain at 50–100 mg/dL under physiological conditions, but they increase 2–4-fold in the blood and tissues during inflammation or infection, as in the present group of patients.^19,38) The induction and expression of α1-AGP are partially controlled by inflammatory cytokines, such as IL-1, IL-6, and tumor necrosis factor-α (TNF-α), which play an important role in immune and inflammatory responses.^{15,19,38–40)} It has also been shown that α1-AGP is synthesized and localized in inflammatory sites not only in the liver but also in monocytes, lymphocytes, and vascular endothelial cells, all of which are involved in inflammatory reactions.^38,41) Therefore, it is believed that α1-AGP is induced by inflammatory cytokines during acute inflammation and that it has some function at the site of lesions. However, the exact biological significance and physiological effects of α1-AGP and their mechanisms of action have not been fully elucidated. On the other hand, in terms of the mechanism of CRP production, monocytes and macrophages are 1st activated in response to infected microorganisms and tissue damage, and inflammatory cytokines such as TNF-α and IL-1 are produced. These act on Kupffer cells in the liver to induce the production of IL-6, and ultimately CRP is synthesized and secreted by hepatocytes.^15,42) In the condition of elevated CRP, it is predicted that the activity of hepatic metabolic enzymes such as CYP2C19 will decrease due to a decline in the liver function, suggesting that the plasma concentration of VRCZ will increase.^13,15) Conversely, it has been reported that α1-AGP is a binding protein for basic drugs, such as VRCZ, in blood^43–46) and that a drug molecule bound to α1-AGP might be transferred from blood to tissues, such as the liver, via its receptor-mediated endocytosis.^47–52) Previous studies have suggested that α1-AGP is cleared via receptors present in liver parenchymal cells, with 2 routes of elimination from circulation being strongly considered: direct receptor-mediated uptake of α1-AGP itself into the liver and further uptake into the liver via asialoglycoprotein receptors after conversion to asialo-AGP by sialidase.^50–52) These experimental findings of previous studies suggest that increased protein binding of VRCZ to α1-AGP increases the volume of distribution of VRCZ to organs and thus reduces the plasma concentrations of VRCZ in a steady-state condition through receptor-mediated tissue distribution of the drug molecule bound to α1-AGP^50–52) and its alternative interaction with the organ surface membrane.^53,54)

These findings suggest that the α1-AGP level is a factor with a negative directional influence on the variation of the C_t of VRCZ. In addition, the specific patient population in the present study may have been more susceptible to the influence of α1-AGP due to a background involving poor hepatic reserve capacity, older age, and low albumin and high α1-AGP levels in many patients. For these reasons, we considered that the C_t predicted by multiple regression analysis using the data of patients in the present study exhibited a closer correlation with the observed C_t values.

The present study has some limitations. First, the details of the causal relationships between the binding capacity of α1-AGP and the pharmacokinetics and plasma concentrations of VRCZ remain unclear and need to be fundamentally investigated in the future. Second and third, the patient sample in the current study was small in size and was limited to a single medical institution. In addition, these patients formed a biased population sample because they were relatively older and had impaired liver function and chronic inflammatory conditions. In the future, it will also be necessary to verify the new findings of the present study in other patient populations with larger samples and from other institutions. However, our study revealed that the impact of high α1-AGP levels on VRCZ plasma concentrations differed significantly from that of other demographic and clinical laboratory values. This finding may explain some of the large inter-individual variability in VRCZ plasma concentrations noted in previous reports.

CONCLUSION

We evaluated patient-specific factors influencing trough plasma VRCZ concentration and showed that an increased α1-AGP level was a significant explanatory variable for decreased plasma VRCZ concentration. Although the biological significance and physiological effects of α1-AGP are not fully clarified, its levels are known to increase 2–4-fold in blood and tissues during inflammation and infection. Since α1-AGP is a binding plasma protein for VRCZ, drug molecules of VRCZ bound to α1-AGP might increase the volume of its distribution while reducing its steady-state plasma concentration through receptor-mediated tissue distribution. This new finding may explain some of the large inter-individual variability in plasma VRCZ concentrations.

Ackowledgments

The authors are grateful to all the medical staff of Sakai City Medical Center and to the patients who participated in this study. The authors would also like to thank the native speaker for English language editing.

Funding

This research has not received any specific Grants from funding agencies in the public, commercial, or not-for-profit sectors.

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

This article contains supplementary materials.

REFERENCES

• 1)  Ueda K, Nannya Y, Kumano K, Hangaishi A, Takahashi T, Imai Y, Kurokawa M. Monitoring trough concentration of voriconazole is important to ensure successful antifungal therapy and to avoid hepatic damage in patients with hematological disorders. Int. J. Hematol., 89, 592–599 (2009).
• 2)  Pascual A, Calandra T, Bolay S, Buclin T, Bille J, Marchetti O. Voriconazole therapeutic drug monitoring in patients with invasive mycoses improves efficacy and safety outcomes. Clin. Infect. Dis., 46, 201–211 (2008).
• 3)  Hamada Y, Seto Y, Yago K, Kuroyama M. Investigation and threshold of optimum blood concentration of voriconazole: a descriptive statistical meta-analysis. J. Infect. Chemother., 18, 501–507 (2012).
• 4)  Matsumoto K, Abematsu K, Shigemi A, Kanazawa N, Watanabe E, Yokoyama Y, Ikawa K, Morikawa N, Takeda Y. Therapeutic drug monitoring of voriconazole in Japanese patients: analysis based on clinical practice data. J. Chemother., 28, 198–202 (2016).
• 5)  Purkins L, Wood N, Greenhalgh K, Allen MJ, Oliver SD. Voriconazole, a novel wide-spectrum triazole: oral pharmacokinetics and safety. Br. J. Clin. Pharmacol., 56(Suppl. 1), 10–16 (2003).
• 6)  Matsumoto K, Ikawa K, Abematsu K, Fukunaga N, Nishida K, Fukamizu T, Shimodozono Y, Morikawa N, Takeda Y, Yamada K. Correlation between voriconazole trough plasma concentration and hepatotoxicity in patients with different CYP2Cl9 genotypes. Int. J. Antimicrob. Agents, 34, 91–94 (2009).
• 7)  Kakuda TN, Van Solingen-Ristea R, Aharchi F, De Smedt G, Witek J, Nijs S, Vyncke V, Richard MW, Hoetelmans RMW. Pharmacokinetics and short-term safety of etravirine in combination with fluconazole or voriconazole in HIV-negative volunteers. J. Clin. Pharmacol., 53, 41–50 (2013).
• 8)  Li X, Yu C, Wang T, Chen K, Zhai S, Tang H. Effect of cytochrome P450 2C19 polymorphisms on the clinical outcomes of voriconazole: a systematic review and meta-analysis. Eur. J. Clin. Pharmacol., 72, 1185–1193 (2016).
• 9)  Mikus G, Schöwel V, Drzewinska M, Rengelshausen J, Ding R, Riedel KD, Burhenne J, Weiss J, Thomsen T, Haefeli WE. Potent cytochrome P450 2C19 genotype-related interaction between voriconazole and the cytochrome P450 3A4 inhibitor ritonavir. Clin. Pharmacol. Ther., 80, 126–135 (2006).
• 10)  Kubota T, Chiba K, Ishizaki T. Genotyping of S-mephenytoin 4′-hydroxylation in an extended Japanese population. Clin. Pharmacol. Ther., 60, 661–666 (1996).
• 11)  Umemura T, Kakizaki H, Mutoh Y, Mizuno T, Ito Y, Hioki T, Kato H, Hagihara M, Yamada T, Ikeda Y, Mikamo H, Ichihara T, Hamada Y. Effectiveness and safety of the simulation-based first-dose design of voriconazole. J. Infect. Chemother., 31, 102453 (2025).
• 12)  Chantharit P, Tantasawat M, Kasai H, Tanigawara Y. Population pharmacokinetics of voriconazole in patients with invasive aspergillosis: serum albumin level as a novel marker for clearance and dosage optimization. Ther. Drug Monit., 42, 872–879 (2020).
• 13)  Wang T, Yan M, Tang D, Xue L, Zhang T, Dong Y, Zhu L, Wang X, Dong Y. A Retrospective, multicentre study of voriconazole trough concentrations and safety in patients with Child–Pugh class C cirrhosis. J. Clin. Pharm. Ther., 43, 849–854 (2018).
• 14)  Nashimoto S, Imai S, Sugawara M, Takekuma Y. Usefulness of the albumin-bilirubin score in determining the initial dose of voriconazole for patients with liver cirrhosis. Biol. Pharm. Bull., 46, 230–236 (2023).
• 15)  Vreugdenhil B, van der Velden WJFM, Feuth T, Kox M, Pickkers P, van de Veerdonk FL, Blijlevens NMA, Brüggemann RJM. Moderate correlation between systemic IL-6 responses and CRP with trough concentrations of voriconazole. Br. J. Clin. Pharmacol., 84, 1980–1988 (2018).
• 16)  Vanstraelen K, Wauters J, De Loor H, Vercammen I, Annaert P, Lagrou K, Spriet I. Protein-binding characteristics of voriconazole determined by high-throughput equilibrium dialysis. J. Pharm. Sci., 103, 2565–2570 (2014).
• 17)  Yuan ZQY, Qiao C, Yang ZC, Yu L, Sun LN, Qian Y, Zhang XH, Meng L, Zhang XY, Wang YQ. The impact of plasma protein binding characteristics and unbound concentration of voriconazole on its adverse drug reactions. Front. Pharmacol., 11, 505 (2020).
• 18)  Di L. An update on the importance of plasma protein binding in drug discovery and development. Expert Opin. Drug Discov., 16, 1453–1465 (2021).
• 19)  Ceciliani F, Lecchi C. The immune function of α₁ acid glycoprotein. Curr. Protein Pept. Sci., 20, 505–524 (2019).
• 20)  Nivoix Y, Levêque D, Herbrecht R, Koffel JC, Beretz L, Ubreaud-Sequier G. The enzymatic basis on drug–drug interactions with systemic triazole antifungals. Clin. Pharmacokinet., 47, 779–792 (2008).
• 21)  Qi F, Zhu L, Li N, Ge T, Xu G, Liao S. Influence of different proton pump inhibitors on the pharmacokinetics of voriconazole. Int. J. Antimicrob. Agents, 49, 403–409 (2017).
• 22)  Johnson PJ, Berhane S, Kagebayashi C, Satomura S, Teng M, Reeves HL, O’Beirne J, Fox R, Skowronska A, Palmer D, Yeo W, Mo F, Lai P, Iñarrairaegui M, Chan SL, Sangro B, Miksad R, Tada T, Kumada T, Toyoda H. Assessment of liver function in patients with hepatocellular carcinoma: a new evidence-based approach-the ALBI grade. J. Clin. Oncol., 33, 550–558 (2015).
• 23)  Gu P, Li Y. Development and validation of a stability-indicating HPLC method for determination of voriconazole and its related substances. J. Chromatogr. Sci., 47, 594–598 (2009).
• 24)  Florent A, Gandia P, Seraissol P, Chatelut E, Houin G. Determination of plasma unbound fraction of voriconazole in patients treated with a prophylactic or a curative treatment. Ther. Drug Monit., 36, 752–758 (2014).
• 25)  Shimamura K, Wajima T, Yano Y. Pharmacokinetic prediction for intravenous β-lactam antibiotics in pediatric patients. J. Pharm. Sci., 96, 3125–3139 (2007).
• 26)  Theuretzbacher U, Ihle F, Derendorf H. Pharmacokinetic/pharmacodynamic profile of voriconazole. Clin. Pharmacokinet., 45, 649–663 (2006).
• 27)  van Wanrooy MJP, Span LFR, Rodgers MGG, van den Heuvel ER, Uges DRA, van der Werf TS, Kosterink JGW, Alffenaar JWC. Inflammation is associated with voriconazole trough concentrations. Antimicrob. Agents Chemother., 58, 7098–7101 (2014).
• 28)  Aiuchi N, Nakagawa J, Sakuraba H, Takahata T, Kamata K, Saito N, Ueno K, Ishiyama M, Yamagata K, Kayaba H, Niioka T. Impact of polymorphisms of pharmacokinetics-related genes and the inflammatory response on the metabolism of voriconazole. Pharmacol. Res. Perspect., 10, e00935 (2022).
• 29)  Miyakawa-Tanaka K, Suzuki J, Hirasawa Y, Nakamura S, Takeda K, Narumoto O, Matsui H. Positive correlation between voriconazole trough concentrations and C-reactive protein levels in patients with chronic pulmonary aspergillosis: a retrospective cohort study. J. Infect. Chemother., 29, 683–687 (2023).
• 30)  van den Born DA, Märtson AG, Veringa A, Punt NC, van der Werf TS, Alffenaar JWC, Sturkenboom MGG, Touw DJ. Voriconazole exposure is influenced by inflammation: a population pharmacokinetic model. Int. J. Antimicrob. Agents, 61, 106750 (2023).
• 31)  Hao X, Li Y, Zhang Y, Bian J, Zhao J, Zhao Y, Hu L, Luo X, Yang C, Feng Y, Huang L. Individualized treatment with voriconazole in the Chinese population: inflammation level as a novel marker for dose optimization. Br. J. Clin. Pharmacol., 90, 440–451 (2024).
• 32)  Klomp SD, Veringa A, Alffenaar JWC, de Boer MGJ, Span LFR, Guchelaar HJ, Swen JJ. Inflammation altered correlation between CYP2C19 genotype and CYP2C19 activity in patients receiving voriconazole. Clin. Transl. Sci., 17, e13887 (2024).
• 33)  Hamada Y, Kawasumi N, Hirai J, Yamagishi Y, Mikamo H. Evaluation of voriconazole oral dosage in Japan. Jpn. J. Antibiot., 67, 279–284 (2014).
• 34)  Dote S, Sawai M, Nozaki A, Naruhashi K, Kobayashi Y, Nakanishi H. A retrospective analysis of patient-specific factors on voriconazole clearance. J. Pharm. Health Care Sci., 2, 10 (2016).
• 35)  Liang Z, Yu M, Liu Z, Liu F, Jia C, Xiong L, Dai Q, Qin S, Cheng L, Sun F. Inflammation affects liver function and the metabolism of voriconazole to voriconazole-N-oxide in adult and elderly patients. Front. Pharmacol., 13, 835871 (2022).
• 36)  Roffey SJ, Cole S, Comby P, Gibson D, Jezequel SG, Nedderman ANR, Smith DA, Walker DK, Wood N. The disposition of voriconazole in mouse, rat, rabbit, guinea pig, dog, and human. Drug Metab. Dispos., 31, 731–741 (2003).
• 37)  Vanstraelen K, Wauters J, Vercammen I, de Loor H, Maertens J, Lagrou K, Annaert P, Spriet I. Impact of hypoalbuminemia on voriconazole pharmacokinetics in critically ill adult patients. Antimicrob. Agents Chemother., 58, 6782–6789 (2014).
• 38)  Fournier T, Medjoubi NN, Porquet D. Alpha-1-acid glycoprotein. Biochim. Biophys. Acta Protein Struct. Mol. Enzymol., 1482, 157–171 (2000).
• 39)  Baumann H, Firestone GL, Burgess TL, Gross KW, Yamamoto KR, Held WA. Dexamethasone regulation of α1-acid glycoprotein and other acute phase reactants in rat liver and hepatoma cells. J. Biol. Chem., 258, 563–570 (1983).
• 40)  Mafuru M, Wu S, He S, Lu X, Huang J, Jiang H. The influence of proinflammatory cytokines on voriconazole trough concentration in patients with different forms of hematologic disorders. J. Clin. Pharmacol., 59, 1340–1350 (2019).
• 41)  Shibata K, Okubo H, Ishibashi H, Tsuda-Kawamura K, Yanase T. Rat alpha 1-acid glycoprotein: uptake by inflammatory and tumor tissues. Br. J. Exp. Pathol., 59, 601–608 (1978).
• 42)  Volanakis JE. Human C-reactive protein: expression, structure, and function. Mol. Immunol., 38, 189–197 (2001).
• 43)  Routledge PA. The plasma protein binding of basic drugs. Br. J. Clin. Pharmacol., 22, 499–506 (1986).
• 44)  Matsumoto K, Sukimoto K, Nishi K, Maruyama T, Suenaga A, Otagiri M. Characterization of ligand binding sites on the alpha1-acid glycoprotein in humans, bovines and dogs. Drug Metab. Pharmacokinet., 17, 300–306 (2002).
• 45)  Nishi K, Ueno M, Murakami Y, Fukunaga N, Akuta T, Kadowaki D, Watanabe H, Suenaga A, Maruyama T, Otagiri M. A site-directed mutagenesis study of drug-binding selectivity in genetic variants of human α1-acid glycoprotein. J. Pharm. Sci., 98, 4316–4326 (2009).
• 46)  Nishi K, Ono T, Nakamura T, Fukunaga N, Izumi M, Watanabe H, Suenaga A, Maruyama T, Yamagata Y, Curry S, Otagiri M. Structural insights into differences in drug-binding selectivity between two forms of human α1-acid glycoprotein genetic variants, the A and F1*S forms. J. Biol. Chem., 286, 14427–14434 (2011).
• 47)  Morell AG, Gregoriadis G, Scheinberg IH, Hickman J, Ashwell G. The role of sialic acid in determining the survival of glycoproteins in the circulation. J. Biol. Chem., 246, 1461–1467 (1971).
• 48)  Ishibashi S, Hammer RE, Herz J. Asialoglycoprotein receptor deficiency in mice lacking the minor receptor subunit. J. Biol. Chem., 269, 27803–27806 (1994).
• 49)  Braun JR, Willnow TE, Ishibashi S, Ashwell G, Herz J. The major subunit of the asialoglycoprotein receptor is expressed on the hepatocellular surface in mice lacking the minor receptor subunit. J. Biol. Chem., 271, 21160–21166 (1996).
• 50)  Matsumoto K, Nishi K, Kikuchi M, Watanabe H, Nakajou K, Komori H, Kadowaki D, Suenaga A, Maruyama T, Otagiri M. Receptor-mediated uptake of human α1-acid glycoprotein into liver parenchymal cells in mice. Drug Metab. Pharmacokinet., 25, 101–107 (2010).
• 51)  Nishi K, Komori H, Kikuchi M, Uehara N, Fukunaga N, Matsumoto K, Watanabe H, Nakajou K, Misumi S, Suenaga A, Maruyama T, Otagiri M. Characterization of the hepatic cellular uptake of α1-acid glycoprotein (AGP), part 1: a peptide moiety of human AGP is recognized by the hemoglobin β-chain on mouse liver parenchymal cells. J. Pharm. Sci., 101, 1599–1606 (2012).
• 52)  Komori H, Nishi K, Uehara N, Watanabe H, Shuto T, Suenaga A, Maruyama T, Otagiri M. Characterization of hepatic cellular uptake of α1-acid glycoprotein (AGP), part 2: involvement of hemoglobin β-chain on plasma membranes in the uptake of human AGP by liver parenchymal cells. J. Pharm. Sci., 101, 1607–1615 (2012).
• 53)  Nishi K, Maruyama T, Halsall HB, Handa T, Otagiri M. Binding of alpha1-acid glycoprotein to membrane results in a unique structural change and ligand release. Biochemistry, 43, 10513–10519 (2004).
• 54)  McCurdy TR, Bhakta V, Eltringham-Smith LJ, Gataiance S, Fox-Robichaud AE, Sheffield WP. In vivo clearance of alpha-1 acid glycoprotein is influenced by the extent of its N-linked glycosylation and by its interaction with the vessel wall. J. Biomed. Biotechnol., 2012, 292730 (2012).