Impact of Obesity on Voriconazole Pharmacokinetics among Pediatric Hematopoietic Cell Transplant Recipients

Voriconazole (VCZ) is an antifungal agent with wide inter- and intrapatient pharmacokinetic (PK) variability and narrow therapeutic index. Although obesity was associated with higher VCZ trough concentrations in adults, the impact of obesity had yet to be studied in children. We characterized the PK of VCZ in obese patients by accounting for age and CYP2C19 phenotype. We conducted intensive PK studies of VCZ and VCZ N-oxide metabolite in 44 hematopoietic stem cell transplantation (HSCT) recipients aged 2 to 21 years who received prophylactic intravenous VCZ every 12 hours (q12h).

ABSTRACT

Voriconazole (VCZ) is an antifungal agent with wide inter- and intrapatient pharmacokinetic (PK) variability and narrow therapeutic index. Although obesity was associated with higher VCZ trough concentrations in adults, the impact of obesity had yet to be studied in children. We characterized the PK of VCZ in obese patients by accounting for age and CYP2C19 phenotype. We conducted intensive PK studies of VCZ and VCZ N-oxide metabolite in 44 hematopoietic stem cell transplantation (HSCT) recipients aged 2 to 21 years who received prophylactic intravenous VCZ every 12 hours (q12h). Blood samples were collected at 5 and 30 minutes; at 1, 3, 6, and 9 hours after infusion completion; and immediately before the next infusion start. We estimated PK parameters with noncompartmental analysis and evaluated for an association with obesity by multiple linear regression analysis. The 44 participants included 9 (20%) with obesity. CYP2C19 metabolism phenotypes were identified as normal in 22 (50%), poor/intermediate in 13 (30%), and rapid/ultrarapid in 9 patients (21%). Obesity status significantly affects the VCZ minimum concentration of drug in serum (C_min) (higher by 1.4 mg/liter; 95% confidence interval [CI], 0.0 to 2.8; P = 0.047) and VCZ metabolism ratio (VCZ_RATIO) (higher by 0.4; 95% CI, 0.0 to 0.7; P = 0.03), while no association was observed with VCZ area under the curve (AUC) (P = 0.09) after adjusting for clinical factors. A younger age and a CYP2C19 phenotype were associated with lower VCZ AUC. Obesity was associated with decreased metabolism of VCZ to its inactive N-oxide metabolite and, concurrently, increased VCZ C_min, which is deemed clinically meaningful. Future research should aim to further characterize its effects and determine a proper dosing regimen for the obese.

TEXT

Voriconazole (VCZ) is widely used for prophylaxis and treatment of invasive fungal infection (IFI) among children and adults after hematopoietic stem cell transplantation (HSCT) (1). Clinical pharmacology is an important consideration in the clinical use of VCZ because of its narrow therapeutic concentration range and large intra- and interpatient pharmacokinetic (PK) variability (2). Although optimal dosing of VCZ has been studied in children (3), neither official recommendations nor high-quality studies are published on VCZ dosing in obese children. This gap in knowledge can potentially pose serious harm to obese children either by insufficient efficacy or increased risk of adverse drug effects in the current era of ever-increasing pediatric obesity prevalence (4).

Unique physiology in obese individuals can potentially affect PK of medications via various mechanisms, such as increased fat-lean tissue balance, higher whole-blood volume, increased liver and kidney perfusion, and glomerular hyperfiltration (5). Obese people are also known to be under subclinical chronic inflammation (6), which could potentially affect various cytochrome P450 (CYP) drug-metabolizing enzyme activities (7). The impact of obesity on adult patients treated with VCZ has been reported in several settings. Adults with obesity were shown to have a higher VCZ concentration than those without when VCZ dose is based on total body weight (8–10). The obese had a better attainment of target VCZ concentrations when either adjusted or ideal body weight was used for its dosing. An intensive PK study on healthy obese volunteers reported that fixed VCZ dose, regardless of body weight, can be used among the obese because of their comparable volume of distribution and clearance with those of the nonobese (11). Although dosing VCZ based on total body weight is recommended for children (12), the effect of obesity is unknown.

First, the activities of drug-metabolizing enzymes of VCZ, namely, CYP2C19 and 3A4, present dynamic changes over the childhood age range. In vivo studies showed that both CYP2C19 and 3A4 activities increase up to 200% of the adult value between the age of 1 to 2 years and gradually reach the adult levels around 12 years (13). In addition, CYP2C19 phenotype-based dosing of VCZ is recommended by the Clinical Pharmacogenetics Implementation Consortium (CPIC) because of the substantial effects of genetic polymorphism on VCZ metabolism (14). Furthermore, there are other factors that can affect the PK of VCZ, such as fasting-fed status, drug-drug interactions, and hepatic impairment (2).

There is a need for characterization of obesity’s impact on the PK of VCZ by a dedicated pediatric study that employs intensive PK studies. The adult data on obesity and VCZ should not be generalized to the pediatric population because of the unique physiology among children. Physiologic changes seen in obesity, such as increased body fat-water balance or higher organ perfusions, also show substantial developmental changes throughout childhood (15). Therefore, PK studies of VCZ and its main metabolite, VCZ N-oxide (VCZ_N-O), are helpful to delineate the PK profiles of obesity among children across developmental stages. In addition, the previously conducted adult studies on obesity and VCZ did not account for CYP2C19 phenotypes (8–11).

To address this gap in knowledge, we retrospectively analyzed PK of VCZ in a prospectively designed study among children who received prophylactic VCZ after HSCT. We aimed to identify the clinical impact of obesity on VCZ exposure by comparing the VCZ trough concentration (minimum concentration of drug in serum [C_min]) between the obese and nonobese. We also characterized the effects of obesity on the systemic exposure of VCZ area under the curve (AUC) and VCZ metabolic ratio.

RESULTS

The 44 patients analyzed included 9 (20%) obese with a mean body mass index (BMI) percentile of 96.9 (Table 1). Among them, four were aged 2 to 12 years and the rest were aged ≥12. Non-Hispanic white was the most common race/ethnicity (32 patients, 73%). There were similar distributions of the CYP2C19 phenotypes between the obese and nonobese (P = 0.35); approximately half of the phenotypes were determined as normal metabolism, and the majority of the other half were intermediate and rapid metabolizers in both groups. There was one patient with an ultrarapid phenotype in the nonobese group and one with a poor metabolizer phenotype in the obese group. Approximately half of the patients had acute lymphoblastic or myeloid leukemia (n = 21, 48%). There was no significant difference in the distribution of age, age group, gender, race, or HSCT indication. One-fourth of the nonobese group was classified as overweight.

TABLE 1.

Patient characteristics

Characteristica	Values by patient category			P value
	All (n = 44)	Nonobese (n = 35)	Obese (n = 9)
Age (yrs), mean (SD)	11.7 (5.3)	11.5 (5.4)	12.8 (4.8)	0.50
Age group (yrs), n (%)				1.00
2–12	22 (50)	18 (51)	4 (44)
≥12	22 (50)	17 (49)	5 (56)
Male sex, n (%)	29 (66)	22 (63)	7 (78)	0.65
wt (kg), mean (SD)	46.5 (24.0)	41.4 (19.8)	66.5 (29.5)	0.004
BMI (kg/m2), mean (SD)	20.5 (4.8)	18.8 (0.8)	26.9 (5.4)	<0.0001
BMI (%tile), mean (SD)	64.9 (29.0)	56.4 (26.8)	96.9 (1.5)	<0.0001
BMI category, n (%)				<0.0001
Underweight (<5th %tile)	1 (2)	1 (3)	0
Normal wt (5th–85th %tile)	26 (59)	26 (74)	0
Overweight (85th–95th %tile)	8 (18)	8 (23)	0
Obesity (>95th %tile)	9 (20)	0	9 (100)
Race, n (%)				0.21
White, non-Hispanic	32 (73)	26 (74)	6 (67)
White, Hispanic	3 (7)	3 (9)	0
Black, non-Hispanic	3 (7)	3 (9)	0
Native American	3 (7)	1 (3)	2 (22)
Others/unknown	3 (7)	2 (6)	1 (11)
CYP2C19 phenotype, n (%)				0.35
Poor	1 (2)	0	1 (11)
Intermediate	12 (27)	10 (29)	2 (22)
Normal	22 (50)	18 (51)	4 (44)
Rapid	8 (18)	6 (17)	2 (22)
Ultrarapid	1 (2)	1 (3)	0
Drug-drug interactions, n (%)
Pantoprazole	41 (93)	33 (94)	8 (89)	1.00
Corticosteroids	1 (2)	0	1 (11)	0.46
Cyclosporine	40 (91)	32 (91)	8 (89)	1.00
Drugs listed in FDA labelb	0	0	0
Indications, n (%)				0.42
ALL	10 (23)	7 (19)	3 (33)
AML	11 (25)	10 (28)	1 (13)
MDS	4 (9)	4 (11)	0
Aplastic anemia	8 (18)	5 (14)	3 (38)
Others	11 (25)	9 (25)	2 (25)

^aBMI, body mass index; FDA, Food and Drug Administration; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; MDS, myelodysplastic syndrome.

^bIncludes the drugs in the list of “effects of other drugs on voriconazole pharmacokinetics” in the FDA package insert of voriconazole.

Scatterplots and simple linear regression lines are shown in Fig. 1, which display the association between PK parameters and the BMI percentile as a continuous variable. There was no significant correlation between BMI percentile and any of the PK parameters (i.e., dose-normalized VCZ C_min, dose-normalized VCZ AUC, or VCZ_RATIO) (P = 0.05, 0.24, and 0.37, respectively).

FIG 1.

Relationship between BMI percentile and PK parameters. The dashed lines represent the threshold for obesity (i.e., 95th BMI percentile). “VCZ trough, dose-normalized,” VCZ trough (mg/liter)/VCZ dose (mg/kg); “VCZ AUC, dose-normalized,” VCZ AUC (mg h/liter)/VCZ dose (mg/kg); VCZ, voriconazole; AUC, area under the curve; VCZ N-O, voriconazole N-oxide metabolite; BMI, body mass index.

The results of PK studies are shown in Table 2. There was no significant difference in any of the studied PK parameters (i.e., VCZ AUC, VCZ_N-O AUC, VCZ_RATIO, VCZ Vss, and VCZ half-life) between the obese and the nonobese patients. Although the mean C_min (2.1 mg/liter; standard deviation [SD], 2.0) was in the target range with the mean VCZ dose of 7.7 mg/kg of body weight (SD, 1.9), only 16 patients (36%) achieved the target VCZ C_min. A significantly higher VCZ dose was given to the nonobese than to the obese (8.0 versus 6.6 mg/kg, P = 0.046). No significant difference was observed by obesity status in VCZ C_min and target C_min achievement.

TABLE 2.

Pharmacokinetic results

Pharmacokinetic measurea	Values by patient category			P value
	All (n = 44)	Nonobese (n = 35)	Obese (n = 9)
VCZ AUC (mg·h/liter) (mean [SD])	45.9 (38.9)	44.3 (41.5)	52 (27.2)	0.60
VCZN-O AUC (mg·h/liter)b (mean [SD])	71.4 (17.0)	72.4 (14.4)	67.5 (25.4)	0.45
VCZRATIOb ,c (mean [SD])	0.6 (0.5)	0.6 (0.5)	0.9 (0.6)	0.13
VCZ Vss (liter)b (mean [SD])	78.2 (53.8)	78.5 (55.8)	77 (48.1)	0.94
VCZ half-life (hour)b (mean [SD])	8.5 (7.5)	7.4 (6.9)	12.6 (8.7)	0.06
VCZ dose (mg/kg) (mean [SD])	7.7 (1.9)	8 (1.7)	6.6 (2.5)	0.046
VCZ Cmin (mg/liter) (mean [SD])	2.1 (2.0)	1.9 (1.9)	2.8 (2.1)	0.19
Achievement of target Cmin (n[%])				0.23
Low (<1.5 mg/liter)	21 (48)	19 (54)	2 (22)
Target (1.5–5.0 mg/liter)	16 (36)	11 (31)	5 (56)
High (>5.0 mg/liter)	7 (1)	5 (14)	2 (22)

^aVCZ, voriconazole; AUC, area under the curve; VCZ_N-O, voriconazole N-oxide metabolite; C_min, minimum concentration of drug in serum.

^bTotal number of patients was 43.

^cVCZ_RATIO is the ratio of VCZ AUC to VCZ N-oxide metabolite AUC.

Multiple linear regression analysis was conducted to ascertain the relationship between clinical covariates, including obesity status, age, dose (mg/kg), and CYP2C19 phenotype, and VCZ PK parameters (Table 3). Obesity status was associated with higher VCZ C_min (estimate, 1.4 mg/liter; 95% CI, 0.0 to 2.8; P = 0.047) and higher VCZ metabolism ratio (VCZ_RATIO) (estimate, 0.4; 95% CI, 0.0 to 0.7; P = 0.03). Age and VCZ dose (mg/kg) had a significant association with VCZ AUC and VCZ_RATIO. CYP2C19 normal and rapid/ultrarapid phenotypes compared with poor/intermediate were associated with significantly lower values in VCZ C_min, VCZ AUC, and VCZ_RATIO.

TABLE 3.

Multiple linear regression analysis adjusted by obesity status, age, VCZ dose, and CYP2C19 phenotype

Predictora	VCZ Cmin (mg/liter)			VCZ AUC (mg·h/liter)			VCZ/VCZ N-O AUC ratiob
	Estimate	95% CI	P value	Estimate	95% CI	P value	Estimate	95% CI	P value
Intercept	−2.4	−7.2 to 2.5	0.33	−89.7	−182.3 to 2.9	0.06	−0.6	−1.8 to 0.6	0.33
Obese	1.4	0.0 to 2.8	0.047	22.7	−3.6 to 49.0	0.09	0.4	0.0 to 0.7	0.03
Age (yr)	0.1	−0.0 to 0.3	0.06	3.4	0.7 to 6.1	0.02	0.04	0.0 to 0.1	0.02
Dose (mg/kg)	0.5	0.0 to 0.9	0.03	14.5	6.1 to 22.8	0.001	0.1	0.0 to 0.2	0.02
CYP2C19 phenotype vs poor/intermediate
Normal	−1.5	−2.8 to −0.3	0.02	−27.7	−51.0 to −4.5	0.02	−0.5	−0.8 to −0.2	0.004
Rapid/ultrarapid	−2.0	−3.6 to −0.4	0.02	−32.5	−62.8 to −2.3	0.04	−0.6	−0.9 to −0.2	0.006
Observations	44			44			43
R2	0.33			0.38			0.39
Adjusted R2	0.24			0.30			0.31

^aVCZ, voriconazole; AUC, area under the curve.

^bVCZ/VCZ N-O AUC ratio, VCZ AUC/VCZ N-oxide metabolite AUC.

DISCUSSION

In this analysis of VCZ PK among children who underwent HSCT, we demonstrated no significant difference in the attainment of therapeutic C_min by obesity status. However, a higher VCZ C_min and higher VCZ_RATIO was associated with obesity status after adjusting for age, VCZ dose (mg/kg), and CYP2C19 phenotype. The present study employed intensive PK studies of VCZ and its main metabolite VCZ_N-O, which allowed an in-depth PK analysis of VCZ and N-oxide systemic exposure. Intensive PK studies also enabled an extrapolation of C_min at a precise trough timing, which would have been otherwise difficult to obtain in patients on VCZ, who are likely under a complex medical care plan.

Obesity status showed a decreased metabolism of VCZ and concurrently higher VCZ C_min in our cohort after adjusting for clinical factors. Higher VCZ exposures in our obese children are consistent with those previously reported in adult studies. VCZ C_min was significantly higher in obese adults dosed by total body weight than by normal body weight, obese dosed by adjusted body weight, or obese dosed by ideal body weight (6.2 ± 2.09 mg/liter versus 3.5 ± 2.86, 3.3 ± 1.32, or 3.95 ± 2.81, respectively; P = 0.001) (8). The prevalence of obesity in our patients is comparable to that of the U.S. national data in 2015 to 2016 (20% versus 18.5%, respectively) (16). It is possible that increased VCZ C_min in the obese is directly attributed to decreased metabolism by CYP2C19. Without measurement of concentrations of other metabolites, we were not able to assess the activities of other metabolizing pathways.

Several explanations are possible for the decreased metabolism of VCZ to VCZ_N-O in the obese children when they received VCZ based on total body weight. First of all, obese children may have relatively decreased physiologic functions involved in CYP2C19 activity compared with those of the nonobese with the same weight. For instance, obese children may have a relatively small liver size and decreased hepatic blood flow. Unlike adults, children continue to grow until they reach the age of final height; thus, those nonobese patients would be older and maybe physiologically more mature than obese at the same weight. In adults, a healthy volunteer study reported no substantial increase in effective body mass contributing to VCZ metabolism with obesity (11). Intensive PK studies with oral 300 mg VCZ every 12 hours (q12h) in healthy obese adult volunteers showed comparable apparent clearance of VCZ compared with the nonobese (10.1 versus 8.36 liters/h). Besides increased body adipose tissue, obesity is known to cause a subclinical chronic inflammatory state (6). Decreased activity of CYP2C19 is observed using a human hepatocyte model in the presence of inflammatory cytokines, such as interleukin-6 (17). In vivo studies reported an association between higher levels of inflammatory markers and increased systemic VCZ concentration along with decreased metabolism to VCZ_N-O (18–20). However, it is not known how much the subclinical, low-level inflammation by obesity affects CYP activities.

We confirmed previously reported demographic and pharmacogenomic effects on the VCZ PK. An increase in age was linearly associated with higher VCZ AUC and VCZ_RATIO (P = 0.02 and 0.02, respectively), and it was approaching significance with VCZ C_min (P = 0.06). CYP2C19 poor or intermediate phenotype (compared with normal or rapid/ultrarapid) was identified as an independent predictor of higher VCZ C_min, higher VCZ AUC, and higher VCZ_RATIO. Similarly, a lower clearance per body weight of VCZ in older children was suggested by a previous study in pediatric HSCT recipients; lower intravenous VCZ doses were required to achieve the target C_min among older children (31.5, 15.5, and 11.9 mg/kg/day for ages 0 to 2 years, 2 to 12, and >12, respectively; P = 0.035) (21). It is also supported by the decreased CYP2C19 and CYP3A4 activities with increasing age in childhood (13). A correlation between CYP2C19 phenotypes and VCZ exposure has been reported in various pediatric cohorts (22–25).

Our study results call for vigilance when VCZ is used in obese children. Similar to other medications used in pediatrics, VCZ dose is based on total body weight. This dosing practice assumes that PK profiles change proportionally with the changes in body weight. This is unlikely to hold true when a weight increase is not accompanied by a proportional increase in physiological functions, as occurs in obesity. We demonstrated an increased VCZ C_min and a decreased VCZ metabolism in the obese children when they were dosed according to total body weight after adjusting for clinical factors. However, we also showed a better C_min goal achievement, although it is statistically not significant, among the obese because of an overall underexposure of VCZ in our cohort. Moreover, our obesity cohort inadvertently received lower VCZ doses in the phase 1 study process than that of the nonobese. This obese population should be monitored carefully to avoid overexposure when VCZ dose is to be increased to correct this overall underexposure.

A notable increase in VCZ C_min warrants an investigation for an altered dosing regimen for obese children. Simulations of obesity-based dosing would need to account for partial nonlinear voriconazole PK that has been reported in children. The clearance of VCZ in children is most commonly explained by an elimination model, including saturable Michaelis-Menten kinetics (26, 27), which precludes the use of the simple linear pharmacokinetics formula for different dose simulations (i.e., AUC = dose/clearance). Thus, alternative dose simulation methods should be considered in future research that can accommodate nonlinear pharmacokinetics, such as the population pharmacokinetics method by nonlinear mixed effect modeling. A possible dose modification for obese children is the use of adjusted weight, which was successfully implemented in adults (8, 10).

The present study is subject to several limitations. First, our small sample size is of statistical concern, due to the wide 95% CI ranges and high type 1 and 2 errors. Second, our definition of obesity could falsely classify individuals with a high proportion of lean body mass as obese. However, our definition is in accordance with other PK studies conducted in obese children (28). In addition, patients having a higher lean body mass is unlikely in our population who generally had chronic and/or serious conditions that warranted HSCT. Third, we did not account for the degree of obesity because of the limited sample size. Future study with a larger sample size may further classify those with a BMI of ≥95th percentile into obesity and severe obesity, which would allow an assessment of the dose-response relationship by obesity. In addition, measurement of other VCZ metabolites might have allowed a further understanding of the impact of obesity on the PKs of VCZ. Although CYP2C19 is recognized as the major metabolic pathway of VCZ (14), adjustment of phenotypes for other drug-metabolizing enzymes might have possibly improved the regression model.

In conclusion, we demonstrated that VCZ pharmacokinetics were affected by obesity in children who underwent HSCT; its decreased metabolism likely resulted in increased VCZ C_min, which is deemed clinically meaningful. Careful monitoring is needed to dose VCZ based on the actual body weight of obese children, especially when higher doses are given. Further studies should investigate a proper dosing regimen for obese children by model-based dose simulation and confirm its effect in a prospective study.

MATERIALS AND METHODS

Study population.

This study was conducted with data collected as part of a single-institution, phase I dose-finding study of VCZ in children undergoing HSCT, and details of the primary study are described by Knight-Perry et al. (26). The study was approved by the University of Minnesota Institutional Review Board and the Masonic Cancer Center Protocol Review Committee. Briefly, the primary objective of the main study was to define the maximum tolerated/minimum efficacious dose of voriconazole for three age-specific groups (<2 years, ≥2 to <12 years, and ≥12 to ≤21 years). The study consisted of an initial VCZ dose-determination phase using a 3 + 3 dose escalation design and a subsequent expansion phase. Informed consent was obtained from the patient or legal guardian, depending on age, after confirmation of study eligibility. Inclusion criteria were children aged ≤21 years who received VCZ to prevent IFI after HSCT and had adequate organ function within 14 days of enrollment defined by serum creatinine of <1.5 times the upper limit of normal (ULN) and aspartate transaminase (AST), alanine aminotransferase (ALT), and/or total bilirubin of <3 times ULN.

The present study initially analyzed 46 patients aged 2 to 21 years old by grouping them based on obesity status. The body weight before the start of conditioning chemotherapy regimen for HSCT was used to categorize obesity according to the Centers for Disease Control and Prevention based on age- and gender-specific body mass index (BMI) percentile as follows: underweight, <5th; normal weight, ≥5th and <85th; overweight, ≥85th and <95th; and obesity, ≥95th (29). The participants aged <2 years were excluded from the present study because no validated definition of obesity exists in that age group.

VCZ dosing and therapeutic drug monitoring.

The study participants were initiated on intravenous VCZ at a rate of 3 mg/kg/hour every 12 hours on the day after the stem cell infusion. The initial VCZ doses in the dose-determination phase were decided based on the enrolling dose level in the following range: 8 to 12 mg/kg q12h (age, ≥2 to <12 years) and 6 to 8 mg/kg q12h (age, ≥12 years) (26). According to a standard 3 + 3 dose escalation design, the initial VCZ doses were escalated from the dose level 1 to 3 after enrolling 3 patients within each age group without minimum efficacious dose (MED) or dose limiting toxicity (DLT). MED is defined by the starting VCZ dose that achieved a therapeutic C_min range of 1.5 to 5 mg/liter within a 7-day period prior to any dose adjustment. DLT included any one of the following within the first week of VCZ: liver or kidney dysfunctions (i.e., ALT, AST, and/or bilirubin at >5 times ULN; serum creatinine at >3 times ULN), any vision changes, any visual/auditory hallucinations. Patients enrolled in the expansion phase received the age-specific initial VCZ dose according to the results of the dose-finding phase.

PK analysis.

Observational PK studies were performed on days 5 to 7. Blood samples (approximately 3 ml) were collected at patient bedside at the following 6 time points: 5 and 30 minutes; 1, 3, 6, and 9 hours after infusion completion; and immediately before the next infusion start. Samples were processed to isolate plasma, and VCZ and VCZ_N-O plasma concentrations were measured by a validated assay using liquid chromatography-tandem mass spectrometry (LC-MS/MS) as described previously (26). VCZ_N-O is a main, inactive metabolite of VCZ that is formed via CYP2C19 and represents over 20% of voriconazole dose excreted in the urine (30). The assay limit of quantitation was 0.025 mg/liter for both VCZ and VCZ_N-O, and all concentrations were above the limit. PK assessment of steady-state profiles was performed with noncompartmental analysis as implemented in R (version 3.6.0) PKNCA package (version 0.8.5). Because of the nature of this study being conducted in patients undergoing complex medical care, there were margins of difference between the intended and actual sample collection timings (i.e., not all 12-hour concentration samples were collected precisely at 12.0 hours). Therefore, we calculated the VCZ and VCZ_N-O C_min concentrations by extrapolating/interpolating concentrations at 12 hours after the start of VCZ infusion (i.e., immediately prior to the next dose in a q12h interval) when the last VCZ concentration was drawn earlier than or later than 30 minutes from the time point 12 hours. Extrapolation and interpolation were based on the linear best fit regression estimate for elimination rate from at least 3 concentration time points. Estimates for AUC were generated with the trapezoidal method and linear-up log-down. Among the PK studies that required intra-/extrapolation, those with AUC intra-/extrapolation of >20% were excluded from analyses. As a result, we excluded two cases of PK studies for VCZ and three for VCZ_N-O. Therefore, we excluded two patients from the study population who did not have eligible PK studies for either VCZ or VCZ_N-O. The final study population included 44 patients. The VCZ metabolic ratio is calculated to estimate the CYP2C19 activity by using the ratio of VCZ AUC to VCZ_N-O AUC.

CYP2C19 genotyping and phenotyping.

Genomic DNA in blood was collected prior to the transplant, and genotypes of the variants of interest were determined with a custom amplicon single nucleotide polymorphism (SNP) panel at University of Minnesota Genomics Center. We defined the phenotypes of CYP2C19 according to the CPIC guideline based on the diplotypes (i.e., ultrarapid [*17/*17], rapid [*1/*17], normal [*1/*1], intermediate [*1/*2, *1/*3, or *2/*17], and poor [*2/*2, *2/*3, or *3/*3]) by genotyping the following SNPs (alleles): rs12248560 (*17), rs4244285 (*2), and rs4986893 (*3) (14). The genotype distributions of the tested SNPs were in Hardy-Weinberg equilibrium (significance, P < 0.05), and the call rates were ≥98%.

Outcomes.

The primary outcome was the achievement of the therapeutic VCZ C_min, which is the most commonly used measure of VCZ exposure in therapeutic drug monitoring and correlates with clinical outcomes (12, 31, 32). Secondary outcomes included characteristics of other PK parameters of VCZ AUC, VCZ_N-O AUC, and VCZ_RATIO (= VCZ AUC/VCZ_N-O AUC). We also assessed the steady-state volume of distribution and half-life of VCZ because these parameters were potentially subject to change by obesity given the moderately lipophilic nature of VCZ. The degree of VCZ metabolism via CYP2C19 was estimated by VCZ_N-O AUC and VCZ_RATIO. Correlations between the BMI percentile (as a continuous variable) and VCZ PK parameters were determined to examine a linear relationship between them.

Statistical analysis.

All statistical significance was assessed at alpha of 0.05. We used the chi-square test for categorical and t test for continuous variables by the obesity status. Graphical analysis (scatterplots and simple linear regression lines) with the Pearson correlation coefficient was assessed between BMI percentile and the PK parameters (i.e., dose-normalized VCZ C_min [mg/liter]/[mg/kg], dose-normalized VCZ AUC [mg·h/liter]/[mg/kg], and VCZ_RATIO). Multiple linear regression analysis was performed to ascertain whether obesity (binary), age (continuous), VCZ dose in mg/kg (continuous), and CYP2C19 phenotype (categorical in the following 3 groups: poor/intermediate, normal, and rapid/ultrarapid) were associated with VCZ C_min, VCZ AUC, and VCZ_RATIO. We adjusted for VCZ dose (mg/kg) as well as age and CYP2C19 phenotypes because of their well-known effects on the PK of VCZ (13, 14).