Combined Effect of CYP2C19 Genetic Polymorphisms and C-Reactive Protein on Voriconazole Exposure and Dosing in Immunocompromised Children

Background Pediatric patients have significant interindividual variability in voriconazole exposure. The aim of the study was to identify factors associated with voriconazole concentrations and dose requirements to achieve therapeutic concentrations in pediatric patients. Methods Medical records of pediatric patients were retrospectively reviewed. Covariates associated with voriconazole plasma concentrations and dose requirements were adjusted by using generalized linear mixed-effect models. Results A total of 682 voriconazole steady-state trough concentrations from 91 Chinese pediatric patients were included. Voriconazole exposure was lower in the CYP2C19 normal metabolizer (NM) group compared with the intermediate metabolizer (IM) group and the poor metabolizer (PM) group (p = 0.0016, p < 0.0001). The median daily dose of voriconazole required to achieve therapeutic range demonstrated a significant phenotypic dose effect: 20.8 mg/kg (range, 16.2–26.8 mg/kg) for the CYP2C19 NM group, 18.2 mg/kg (range, 13.3–21.8 mg/kg) for the CYP2C19 IM group, and 15.2 mg/kg (range, 10.7–19.1 mg/kg) for the CYP2C19 PM group, respectively. The extent of impact of C-reactive protein (CRP) levels on voriconazole trough concentrations and dose requirements varied between CYP2C19 phenotypes. Increases of 20, 120, 245, and 395 mg/L from 5 mg/L in CRP levels were associated with increases in voriconazole trough concentration by 22.22, 50, 64.81, and 75% respectively, in the NM group; by 39.26, 94.48, 123.93, and 146.63%, respectively, in the IM group; and by 17.17, 37.34, 46.78, and 53.65%, respectively, in the PM group. Meanwhile, increases of 20, 120, 245, and 395 mg/L from 5 mg/L in CRP levels were associated with increases in voriconazole dose requirements by 7.15, 14.23, 17.35, and 19.43%, respectively, in the PM group; with decreases in voriconazole dose requirements by 3.71, 7.38, 8.97, and 10.03%, respectively, in the NM group; and with decreases by 4, 9.10, 11.05, and 12.35%, respectively, in the IM group. In addition, age and presence of immunosuppressants had significant effects on voriconazole exposure. Conclusions Our study suggests that CYP2C19 phenotypes, CRP concentrations, age, and the presence of immunosuppressants were factors associated with the pharmacokinetic changes in voriconazole. There was heterogeneity in the effect of CRP on voriconazole plasma concentrations across different CYP2C19 genotypes. Combining relevant factors with dose adaptation strategies in therapeutic drug monitoring may help to reduce the incidence of subtherapeutic and supratherapeutic concentrations in clinical practice.

Results: A total of 682 voriconazole steady-state trough concentrations from 91 Chinese pediatric patients were included. Voriconazole exposure was lower in the CYP2C19 normal metabolizer (NM) group compared with the intermediate metabolizer (IM) group and the poor metabolizer (PM) group (p = 0.0016, p < 0.0001). The median daily dose of voriconazole required to achieve therapeutic range demonstrated a significant phenotypic dose effect: 20.8 mg/kg (range, 16.2-26.8 mg/kg) for the CYP2C19 NM group, 18.2 mg/kg (range, 13.3-21.8 mg/kg) for the CYP2C19 IM group, and 15.2 mg/kg (range, 10.7-19.1 mg/kg) for the CYP2C19 PM group, respectively. The extent of impact of C-reactive protein (CRP) levels on voriconazole trough concentrations and dose requirements varied between CYP2C19 phenotypes. Increases of 20, 120, 245, and 395 mg/L from 5 mg/L in CRP levels were associated with increases in voriconazole trough concentration by 22.22, 50, 64.81, and 75% respectively, in the NM group; by 39.26, 94.48, 123.93, and 146.63%, respectively, in the IM group; and by 17.17, 37.34, 46.78, and 53.65%, respectively, in the PM group. Meanwhile, increases of 20, 120, 245, and 395 mg/L from 5 mg/L in CRP levels were associated with increases in voriconazole dose requirements by 7.15, 14.23, 17.35, and 19.43%, respectively, in the PM group; with decreases in voriconazole dose requirements by 3.71, 7.38, 8.97, and 10.03%, respectively, in the NM group; and with decreases by 4, 9.10, 11.05, and 12.35%, respectively, in the IM group. In addition, age and presence of immunosuppressants had significant effects on voriconazole exposure.

INTRODUCTION
Invasive fungal infections are significant causes of morbidity in immunocompromised pediatric patients with myelosuppressive chemotherapy or hematopoietic stem cell transplantation (HSCT) (1,2). Voriconazole is a broad-spectrum antifungal agent used for invasive fungal infections prophylaxis and treatment (3,4). Pediatric patients have larger interindividual and intraindividual variability compared with adults (5, 6); thus, therapeutic drug monitoring (TDM) is essential for optimizing voriconazole dosing regimens in pediatric patients. Whereas most of the published studies were focused on adult patients, there are few studies of voriconazole TDM and dosage regimens in pediatric patients, especially for the Asian population. The drug package leaflet recommends a weight-based dose of 9 mg/kg per 12 h intravenous (IV) or oral (PO) for 2-to 12-year-old children. The loading dose is 6 mg/kg per 12 h IV for patients older than 12 years. To our knowledge, there is rarely evidence on the optimal dose regimens for pediatric patients, although the pharmacokinetics of voriconazole in such a population may be influenced by more factors in different degrees or mechanisms compared with adults.
Inflammations are frequently observed in patients with myelosuppressive chemotherapy or HSCT. The inflammation, reflected by C-reactive protein (CRP) concentrations, was associated with elevated voriconazole trough concentration in adults (21)(22)(23); the effect of inflammation on voriconazole concentrations in pediatric patients was rarely studied. A previous study demonstrated that the CRP value seems to be associated with higher voriconazole trough concentrations only in children ≥12 years old (24,25). In the present study, the CRP levels in pediatric patients were hypothesized to impact voriconazole exposure in a CYP2C19 phenotype-dependent way. Moreover, the pharmacokinetics of voriconazole were reported to be influenced by other factors including age, gender, route of administration, drug-drug interactions, and liver and renal function (26), whose effects have also been considered in the analysis in this study.
Therefore, a retrospective study was performed in immunocompromised pediatric patients who were treated with voriconazole. Sociodemographic, clinical, voriconazole TDM, dose adjustment, and CYP2C19 genetic data were collected. The aim of the study was to describe the TDM and dose requirements of voriconazole, as well as to identify the determinants of the variability in voriconazole exposure and hence dose requirement to achieve the therapeutic range in Asian children.

Study Design and Patient Population
A single-center retrospective study of immunocompromised children <18 years of age who were treated with voriconazole as an inpatient over a 3-year period (May 2017-May 2020) at Department of Pediatric Hematology, Nanfang Hospital of Southern Medical University, China, was performed. Inclusion criteria were aged <18 years with routine TDM of voriconazole and taken as trough concentrations at steady state, which was defined as after 3 days of treatment (without loading dose) or dose adjustment (27,28). Patients with potential interacting drugs used concomitantly (such as strong inhibitor or inducer of CYP450 described in the summary of product information), liver cirrhosis, chronic renal failure (with estimated creatine clearance ≤ 60 ml/min) were excluded. Individuals were genotyped for CYP2C19 at the discretion of physicians. The following data were extracted from the clinical chart: (1) demographic characteristics, including gender, age, body weight, underlying disease; (2) voriconazole treatment and TDM data, including daily dosage (mg/kg), route of administration, and plasma trough concentrations; (3) other factors that could potentially influence the voriconazole trough concentration, laboratory parameters such as CRP; total protein (TP); albumin (ALB); total bilirubin (TBIL); direct bilirubin (DBIL); alanine aminotransferase (ALT); aspartate transaminase (AST); serum creatinine (CR), which were measured within the same day; and the concomitant medication. The concomitant medications were immunosuppressants used including cyclosporine, tacrolimus, and sirolimus; glucocorticoids used including prednisone, methylprednisolone, and dexamethasone; and proton pump inhibitors (PPIs) used including omeprazole and esomeprazole.

Determination of Voriconazole Plasma Concentration
For each patient, the voriconazole trough concentrations were determined on blood sample withdraw within a 1-h period before the next administration. Plasma concentrations of voriconazole were measured by high-performance liquid chromatographytandem mass spectrometry method. Sample preparation included protein precipitation by mixing 100 µl of plasma with 300 µl of acetonitrile containing 6,7-dimethyl-2, 3-di(2-pyridyl) quinoxaline (internal standard, 500 ng/ml). Two microliters of the supernatant was injected into the chromatographic system after centrifugation at 14,000 revolutions/min for 15 min. Separation conditions were the following: Agilent Poroshell 120 EC-C18 column (3.0 × 50 mm, 2.5-µm); mobile phase composed initially of 20:80 acetonitrile with formic acid (0.1%)/water with formic acid (0.1%) using isocratic elution as the mode of separation; flow rate of 0.5 ml/min; column temperature of 30 • C. The plasma concentration of voriconazole was linear (r > 0.999) over the range of 0.1-10 µg/ml. The intraday and interday accuracy and precision data did not exceed 15% and could be used to determine voriconazole trough concentration accurately. The therapeutic range of voriconazole was between 1 and 5 µg/ml (30,31). The CDR (concentration-to-dose ratio) value was applied to standardize voriconazole concentration.

Statistical Analysis
The Hardy-Weinberg equilibrium (HWE) test was performed using an appropriate χ 2 test, with p < 0.05 indicating a lack of agreement with HWE. To carry out the descriptive analysis, continuous and nonnormal variables were expressed as median and interquartile range. Categorical variables were summarized as frequency and percentage. Demographics (age, gender), clinical characteristics (route of administration, voriconazole dose, laboratory parameters reflected inflammation status, liver and kidney function, coadministered medications), and CYP2C19 genotype were possible factors considered in our study. The crude correlation between CRP levels and voriconazole exposure was analyzed by the Spearman rank correlation test. A univariate linear mixed-effects regression analysis was conducted to identify covariates associated with voriconazole, and p = 0.15 was used as the entry threshold. Then, a multivariate generalized linear mixed-effects model was analyzed based on the results of univariate analyses to assess the effect of covariates on voriconazole concentration and dose requirement, with the patient effect specified as a random effect. Nonnormal quantitative variables were transformed by using their logarithmic value (voriconazole CDR, CRP, and weight-corrected dose). The following confounding factors were introduced into the final model for adjustment: age, gender, weight-corrected dose, presence of immunosuppressants, CRP, CYP2C19 phenotypes, and the interaction term of CRP and phenotypes. A p < 0.05 was considered to be statistically significant, except for subgroup analyses where a p-value threshold of 0.017 was considered (adjustment for multiple comparisons). All statistical analyses were performed with SAS 9.3 (SAS Institute Inc., Cary, NC, USA). Graphic operations were completed by R Language (version 4.1.2).

Population Characteristics
A total of 682 voriconazole steady-state trough concentrations from 91 Chinese pediatric patients were qualified for this study. The demographic characteristics, type of HSCT, underlying diseases, CYP2C19 genotypes, laboratory parameters, and coadministered medication are summarized in Table 1. The majority was male (59.3%) and had a recent history of HSCT (75.8%). The main underlying disease in our study was β-thalassemia major followed by hematologic malignances. CYP2C19 allele frequencies were 29.7% for * 2, 3.8% for * 3 and 0.5% for * 17. None of the genotypes deviated from HWE.  (Figure 1;     (Figure 2; Table 3.2). The effects of CYP2C19 diplotypes on voriconazole CDR and dose requirements are depicted in Supplementary Figures 1, 2. There were some visual trends between different CYP2C19 diplotypes; the statistical analysis was not performed for limited sample size.

Influence of CRP Varied by CYP2C19 Phenotypes on Voriconazole Concentrations and Dose Requirements
The effects of CRP on voriconazole concentrations and dose requirements were identified. The interactive effects of CRP and CYP2C19 phenotypes on voriconazole exposure and dose requirements were statistically significant in the multivariate linear mixed-effects regression analysis (Tables 3.1, 3.2). Further subgroups analysis confirmed the positive association between CRP level and voriconazole exposure in the CYP2C19 NM group (p < 0.0001), IM group (p < 0.0001), and PM group (p = 0.0026). Moreover, the impact of CRP on voriconazole concentration was slightly greater in the IM group (coefficient of 0.2605) than in the NM (coefficient of 0.2400) group and PM group (coefficient of 0.2307) (

Other Factors Affecting Voriconazole Concentrations and Dose Requirements
The effects of other factors including age, gender, route of administration, different drug-drug interactions, and liver and renal function on the pharmacokinetics of voriconazole were analyzed by univariate linear mixed-effects regression analysis in advance (data was not shown), affecting factors with p < 0.15 included for further multivariate linear mixedeffects regression analysis. No significant effect was observed of gender on voriconazole concentrations and dose requirements in multivariate analysis, but age and immunosuppressant comedication did have a significant effect (Tables 3.1, 3.2).

DISCUSSION
In the present study, voriconazole trough concentrations demonstrate high interindividual and intraindividual variability in pediatric patients.  (28,34). Our study showed that 28.4% of 682 voriconazole trough concentrations are below the therapeutic range. In cases of trough concentration < 1 and >5 µg/ml, 56.7% of dose increases and 65.9% of dose decreases result in the optimal therapeutic level of 1-5 µg/ml, respectively. The results demonstrate that  TDM-based dose adjustment is sorely needed to optimize drug concentrations. The study of clinical importance of CYP2C19 genotype on voriconazole pharmacokinetics in pediatric patients is controversial. It has been proposed that voriconazole exposure in immunocompromised children and adolescents could not be predicted based on CYP2C19 genotype (5,6). In this study, voriconazole exposure was lower in the CYP2C19 NM group compared with the IM group and the PM group. Our results corroborate previous findings in pediatric patients (15)(16)(17)19). Walsh et al. (15) and Karlsson et al. (16) showed significantly higher elimination capacity in the NM group than in the IM and PM groups in pediatric population pharmacokinetic analysis. Narita et al. (17) reported higher trough plasma concentrations of voriconazole in the PM and IM groups compared with the NM and UM groups. A recent study demonstrated that patients aged no more than 12 years and more than 12 years required doses of 6.53 ± 2.08 and 3.95 ± 0.85 mg/kg of body weight twice daily (p = 0.007) in the CYP2C19 UM or NM group, patients younger than 12 and older than 12 years required     reported that inflammation severely affects midazolam clearance in pediatric patients (35). In in vitro studies, it was demonstrated that proinflammatory cytokines, such as interleukin 2 (IL-2), IL-6, and tumor necrosis factor α, may negatively regulate the CYP2C19 enzyme (36). Moreover, it has been reported that cytokines may stimulate the production of acute-phase proteins, such as CRP (37). Only two previous studies in pediatric patients showed that CRP levels were significantly associated with voriconazole pharmacokinetics in children 12 years or older, and no effect of inflammation on voriconazole concentration was observed in younger children (24,25). The CRP concentrations had a significant impact on voriconazole exposure and dose requirement in our study. Moreover, the impact of CRP on the trough concentration of voriconazole was modulated by CYP2C19 phenotypes, and the extent of increase in voriconazole trough concentration was larger for the IM group than the PM group and the NM group (  (38), whereas a recent meta-analysis demonstrated a smaller effect of inflammation for adult patients with decreased metabolic capacity for CYP2C19 (IM and PM) than those with normal (NM) or elevated metabolic capacity (RM and UM) (39). The study by Gautier-Veyret et al. (40) was inconsistent with ours. In their study, CRP levels tended to positively influence voriconazole concentration in adults with an increased metabolic activity, but had no impact on adults with normal metabolic activity. The conflicting results could be explained by the genetic factors that integrated both CYP2C19 and CYP3A polymorphisms in the latter study. In our study, we also presented the interactive effect between CRP and CYP2C19 phenotypes on daily dose of voriconazole required to achieve therapeutic range. The dosages required were decreased in the NM group and the IM group by different degrees with the same increase in the CRP level, but increased in the PM group with the same increase in the CRP level (Table 4.2; Figure 4). This may be explained by the phenomenon of inflammationinduced phenoconversion of polymorphic drug-metabolizing enzymes (36).
Age was an influencing factor contributing to the interindividual variability in voriconazole exposures in our population. Hicks et al. (13) reported a correlation between age and higher voriconazole through concentrations corrected for daily dose in pediatric patients, and age also did have a significant impact in a multivariate analysis. In vitro and in vivo studies demonstrated that voriconazole appears to be more rapid metabolized in pediatric patients compared with that in adults; the maintenance doses/weights needed in pediatric patients were higher (41,42). Mechanistically, age is closely related to the activity of liver microsomes, the body fat, and total body water, which potentially leads to the difference of metabolism or distribution of voriconazole.
Voriconazole is often treated in combination with immunosuppressant drugs in patients after HSCT. Immunosuppressant therapies, including cyclosporine, tacrolimus, and sirolimus, are substrates of CYP3A4 enzyme. The drug-drug interactions have high clinical relevance because of the inhibition effect of voriconazole on CYP3A4 enzyme, whereas the effect of immunosuppressants on voriconazole exposure is rarely reported. In our study, immunosuppressant treatments were statistically significant covariates explaining interindividual variability in voriconazole exposure and dose requirement in multivariate linear mixed-effects regression analysis. Zeng et al. (43) reported that the cyclosporine concentration was an independent factor of voriconazole concentration variation, but combination of cyclosporine did not affect voriconazole concentration. Hashemizadeh et al. (44) showed no significant correlation between immunosuppressant treatments and voriconazole concentration. Cyclosporine administration was significantly associated with an increase in voriconazole exposure in the plasma, cerebrospinal fluid, and brain in an animal model of cerebral scedosporiosis (45). An in vitro study demonstrated that low concentrations of voriconazole was mainly (∼93%) metabolized by CYP2C19, whereas CYP3A4 was mainly (∼73%) responsible for metabolism in the case of high concentrations of voriconazole (46). As cyclosporine, tacrolimus, and sirolimus are substrates of CYP3A4, voriconazole disposition at high concentration might be modified by competition at the CYP3A4 catalytic site, whereas more in vitro and in vivo studies are still needed for the conflicting issue about the effect of immunosuppressants on voriconazole exposure. This study has several limitations. First, this is a retrospective observation study in a single tertiary center. Second, genetic polymorphisms of other proteins (CYP3A4, CYP3A5, FMO) that may influence voriconazole pharmacokinetics were not analyzed. Further prospective studies are still needed to validate our findings and to explore the effect of other possible covariates not included in the present study.

CONCLUSION
In this study, we found that influencing factors such as CYP2C19 phenotype, CRP concentration, age, and presence of immunosuppressants were significantly associated with voriconazole trough concentrations and dose requirements to reach therapeutic range in Chinese pediatric patients. The combination of influencing factors with routine TDM may help to individualize voriconazole dosing and improve clinical outcomes.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Nanfang Hospital of Southern Medical University. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
JC, YW, and SL performed and designed the research. JC and YW analyzed the data and wrote the article. JC, YH, XF, and YR collected the children's data. All authors contributed to the article and approved the submitted version.