CYP3A5 influences oral tacrolimus pharmacokinetics and timing of acute kidney injury following allogeneic hematopoietic stem cell transplantation

Introduction: Polymorphisms in genes responsible for the metabolism and transport of tacrolimus have been demonstrated to influence clinical outcomes for patients following allogeneic hematologic stem cell transplant (allo-HSCT). However, the clinical impact of germline polymorphisms specifically for oral formulations of tacrolimus is not fully described. Methods: To investigate the clinical impact of genetic polymorphisms in CYP3A4, CYP3A5, and ABCB1 on oral tacrolimus pharmacokinetics and clinical outcomes, we prospectively enrolled 103 adult patients receiving oral tacrolimus for the prevention of graft-versus-host disease (GVHD) following allo-HSCT. Patients were followed in the inpatient and outpatient phase of care for the first 100 days of tacrolimus therapy. Patients were genotyped for CYP3A5 *3 (rs776746), CYP3A4 *1B (rs2740574), ABCB1 exon 12 (rs1128503), ABCB1 exon 21 (rs2032582), ABCB1 exon 26 (rs1045642). Results: Expression of CYP3A5 *1 was highly correlated with tacrolimus pharmacokinetics in the inpatient phase of care (p < 0.001) and throughout the entirety of the study period (p < 0.001). Additionally, Expression of CYP3A5 *1 was associated with decreased risk of developing AKI as an inpatient (p = 0.06). Variants in ABCB1 were not associated with tacrolimus pharmacokinetics in this study. We were unable to discern an independent effect of CYP3A4 *1B or *22 in this population. Conclusion: Expression of CYP3A5 *1 is highly influential on the pharmacokinetics and clinical outcomes for patients receiving oral tacrolimus as GVHD prophylaxis following allo-HSCT.


Introduction
Allogeneic hematopoietic stem cell transplantation (allo-HSCT), a mainstay treatment for a variety of hematologic malignancies (Sureda et al., 2015), is associated with a significant risk of graft-versus-host disease (GVHD) (Ziakas et al., 2014).Therapeutic immune suppression, often utilizing tacrolimus, is necessary to reduce the risk of GVHD (Hiraoka et al., 2001).Significant inter-patient variability in the pharmacokinetics of tacrolimus, along with a narrow therapeutic window, makes dosing of tacrolimus difficult.Sub-therapeutic levels of tacrolimus are associated with an increased risk of GVHD, while supratherapeutic levels are associated with severe toxicity (Prezepiorka et al., 1999;Jacobson et al., 2001;Lee et al., 2002;Musuamba et al., 2009;Levine et al., 2010).Clinically, therapeutic drug monitoring of tacrolimus has been employed to personalize tacrolimus doses for patients and account for the known inter-patient pharmacokinetic and pharmacodynamic variability of tacrolimus (Elens et al., 2007;Thervet et al., 2008;Brunet et al., 2019).
Out of many factors associated with inter-patient pharmacokinetic and pharmacodynamics variabilities of tacrolimus, single nucleotide polymorphisms (SNPs) in cytochrome (CYP) 3A4 and 3A5 enzymes have been associated with clinical outcomes associated with tacrolimus (Kuehl et al., 2001;Thervet et al., 2010;Elens et al., 2011;Wang et al., 2011).The CYP3A5 *3 allele, the most frequently occurring CYP3A5 allele in populations of European ancestry, references an intronic SNP rs776746 that creates a cryptic splice site and an aberrant splice isoform with partial intron insertion, resulting in a premature stop codon and eliminating the expression of the functional CYP3A5 enzyme (Kuehl et al., 2001;Thervet et al., 2010).While the CYP3A5 *3 allele is the most prevalent allele, individuals with the wild type allele (designated *1) are considered CYP3A5 *1 expressers (*1/*1 or *1/*3).CYP3A5 *1 expressers required increased tacrolimus dose to maintain therapeutic levels of the drug compared to patients with the CYP3A5 *3/*3 genotype (Thervet et al., 2010;Khaled et al., 2016;Hamadeh et al., 2019).CYP3A4 also plays a critical role in tacrolimus metabolism (Elens et al., 2011;Wang et al., 2011).Two specific CYP3A4 alleles have been linked to variable tacrolimus metabolism: CYP3A4 *22, which refers to a polymorphism in intron 6 of CYP3A4 that reduces expression of the CYP3A4 mRNA and protein (Wang et al., 2011;Wang and Sadee, 2016), and CYP3A4 *1B which refers to a SNP on the promoter region of CYP3A4 whose impact remains to be further elucidated (Garcia-Martin et al., 2002;Spurdle et al., 2002;Gervasini et al., 2012;Zochowska et al., 2012).Adenosine triphosphate-binding cassette B1 (ABCB1) is a cell membrane transporter responsible for the active transport of tacrolimus (Op den Buijish et al., 2007).It has been proposed that ABCB1 plays a role in the bioavailability of tacrolimus in the setting of oral administration; however, little is known about the specific impacts of ABCB1 genomic variants on the pharmacokinetic and pharmacodynamic properties of tacrolimus in transplant patients (Mourad et al., 2006;Hawwa et al., 2009;Herrlinger et al., 2011;Kim et al., 2012;Tapiradamaz et al., 2014;Li et al., 2015;Debette-Gratien et al., 2016).
In the 2015 guidelines for the use of pharmacogenomic data to guide tacrolimus dosing, the Clinical Pharmacogenetics Implementation Consortium (CPIC) recommended that patients expressing CYP3A5 *1 should receive doses 1.5-2-fold higher compared to standard dosing (Birdwell et al., 2015).While this recommendation is made for kidney, heart, lung, and allo-HSCT patients, it is notable that such recommendation in the setting of allo-HSCT is made based on evidence provided for intravenous tacrolimus (Woodahl et al., 2008;Yanagimachi et al., 2010;Onizuka et al., 2011;Yanagisawa et al., 2011;Iwamoto et al., 2015;Khaked et al., 2016;Hamadeh et al., 2019;Suztsugu et al., 2019).Very little data is available for patient receiving allo-HSCT alongside oral tacrolimus (Zhu et al., 2020).As such, clinicians have to extrapolate the findings of intravenous tacrolimus studies when considering this clinical problem.In addition to hepatic expression, CYPs in the 3A family are highly expressed in the intestine and are detectable in some brain regions (McFadyen et al., 1998;Paine et al., 2006).Given the potential for intestinal metabolism, as well as firstpass hepatic metabolism prior to reaching systemic circulation, the effect of genomic variants on tacrolimus administered orally poses an additional level of complexity in comparison to intravenous tacrolimus.While significant data exists demonstrating the role of pharmacogenomic variants on oral tacrolimus outside of allo-HSCT (Goto et al., 2004;Kato et al., 2018;Pasternak et al., 2019;Pasternak et al., 2022), this data may not fully apply in the setting of allo-HSCT given the differences in tacrolimus dosing and a desire for less aggressive early tacrolimus levels to confer a graft-versus-leukemia effect (Imado et al., 2004).It is therefore possible that existing findings using intravenous formulations of tacrolimus in the setting of allo-HSCT, and those of oral tacrolimus outside of allo-HSCT, may not fully reflect the pharmacogenomic implications of tacrolimus metabolism specific to the oral route in an allo-HSCT population.In this study, we present a prospective analysis of 103 patients who received oral tacrolimus for the prevention of GVHD following allo-HSCT.

Study population
A cohort of 103 adult patients receiving oral tacrolimus for the prevention of GVHD following allo-HSCT at The Ohio State University Wexner Medical Center (OSUMC) were prospectively enrolled into this study following approval from the Cancer Institutional Research Board's guidance (IRB# 2012C0021) from 06/2012 until 07/2014.Inclusion criteria included age great than 18 years, patient to receive their first allo-HSCT during the study period, and absence of renal or hepatic dysfunction at baseline.Exclusion criteria included HIV positivity, presence of any medical conditions that would interfere with the patient's ability to sign the informed consent, hypersensitivity to any component of tacrolimus, clinical history of any solid organ transplantation, previous chronic use of tacrolimus, or pregnancy.

Study procedures
Following patient consent, but prior to transplantation, whole blood samples were obtained to determine the patient's CYP3A4, CYP3A5, ABCB1 genotypes.Clinical data was collected for patients for up to the first 100 days post-transplant or until the patient was lost to follow-up.As a standard of care in HSCT patients treated at OSUMC, tacrolimus therapeutic drug monitoring was employed using liquid chromatography/tandem mass spectrometry at least three times per week to measure trough drug concentration until goal steady state concentrations were reached.Standard of care at our institution includes starting immediate release oral tacrolimus 2 days prior to transplantation at a dose of 0.03 mg/kg.Twice daily dosing was considered standard; however, clinicians could elect to use daily dosing if necessary for the patient.Tacrolimus levels were normalized by both tacrolimus doses received as well as patient's actual body weight as described in previous studies as follows: Normalized Tacrolimus level (ng/dL/mg/kg) = Tacrolimus level (ng/dL)/Total daily dose of Tacrolimus (mg)/Actual body weight (kg) (Sy et al., 2013;Cusinato et al., 2014;Ito et al., 2017).
To assess the proportion of patients who were within therapeutic range for each group during this study, we binned tacrolimus levels every 5 days starting from the day of transplant.We collected records of held tacrolimus doses due to supratherapeutic levels.We also collected the number of tacrolimus dose was changes during the inpatient stay.Concerning the number of dose changes a patient experienced, we considered held doses as a change in the dose.Due to the imbalance in the numbers of patients with different genotypes, we normalized both total dose changes and total doses held by the number of patients with a specific genotype.We also assessed the percentage of total days of therapy each patient had to hold at least one dose of tacrolimus.In addition, dose changes for each patient during the inpatient stay were also assessed as normalized by each inpatient length of stay (LOS).

Study endpoints
Pharmacokinetic endpoints were based on tacrolimus levels normalized to both the total daily dose of tacrolimus and the patient's actual body weight.The goal range of serum tacrolimus levels for this study was defined as between 8 ng/mL and 12 ng/mL for the first 60 days after transplant as per our institution's standard.After 60 days post-transplant, goal trough levels were determined by the preference of the treating physician.Analysis of time to therapeutic range was a simple comparison between the day number when the first therapeutic trough level was achieved as a continuous variable.Adverse event rates, including both acute kidney injury (AKI), tacrolimus-related neurotoxicity, and GVHD were collected.AKI was defined using KDIGO criteria (Khwaja, 2012).Neurotoxicity was defined by documentation of tremor or confusion attributed by the clinical team to be related to tacrolimus.GVHD was defined by documentation by the clinical team.For this study, only acute GVHD was included in our analysis.Patients without documentation of neurotoxicity or GVHD were assumed to not have experienced these adverse events.

Statistical analyses
All data analyses were conducted using R 3.3.2(The R Foundation; https://www.r-project.org).Patients' characteristics were analyzed using descriptive statistics.Briefly, categorical data were presented as count and frequency; continuous data were summarized as mean and standard deviation (SD) except for age and The European Group for Blood and Marrow Transplantation (EBMT) risk score, which were summarized as median and range.Differences in categorical variables were analyzed using χ 2 tests or Fisher's exact tests.For continuous variables, Student's t tests and one-way ANOVA were used to compare differences between two groups and among multiple groups where appropriate.For each SNP, the consistency between its distribution and the hyphenated Hardy-Weinberg principle was analyzed using χ 2 tests.Potential statistical associations between different SNPs were also analyzed using χ 2 tests using the R package genetics.Statistical tests were two-sided, and p values < 0.05 were considered statistically significant.

Cohort demographics and genotype frequencies
In the present study, we assessed the impacts of various SNPs of CYP3A4, CYP3A5, and ABCB1 on clinical outcomes of a cohort of 103 patients who were given oral tacrolimus for the prevention of GVHD after allo-HSCT.Full demographic features of these patients listed in Table 1.Median follow up was 98 days (interquartile range 92.5-100 days).In this cohort, tacrolimus levels were not associated with the use of voriconazole (p = 0.59) or steroids (p = 0.36).

CYP3A5 *1 expressers exhibited reduced normalized tacrolimus levels
To assess the effect of our tested genomic variants with tacrolimus levels across the full study period, we calculated the median normalized tacrolimus level for each patient for the first 100 days of therapy.Univariate analyses demonstrated that median normalized tacrolimus level for each individual patient over the entire study period was statistically associated with each of CYP3A4 *1B, CYP3A4 *22, and CYP3A5 *1 expression (p = 0.004, p = 0.03, p = 0.0009, respectively; Table 3).Multivariate analyses confirmed the association between normalized tacrolimus levels and CYP3A5 *1 expression alone (Table 3; Figure 1).Based on these results, we focused further study on CYP3A5 *1 expression.
We then analyzed both weight normalized tacrolimus doses and tacrolimus levels at different time points during the inpatient stay of patients, including the first drawn level, day 6, day 10, and at discharge (±2 days).This analysis revealed a clinically important trend in dose requirements for CYP3A5 *1 expressers.While the weight normalized tacrolimus dose at the time of the first level did not differ significantly based on CYP3A5 genotype (p = 0.26; Figure 2A), the tacrolimus level at first assessment varied significantly (p < 0.001; Figure 2B).Over the course of the inpatient stay, CYP3A5 *1 expressers received increasing tacrolimus doses compared to non-expressers (day 6: p = 0.17; day 10: p = 0.008; Discharge: p < 0.001) while the discrepancy in tacrolimus levels slowly diminished (day 6: p < 0.001; day 10: p = 0.32.Discharge: p = 0.37).CYP3A5 genotype was clearly associated with changes in total daily dose of tacrolimus between the start of therapy and discharge, with the CYP3A5 *3/*3 genotype associated with a decreased dose from drug initiation to discharge (p = 0.0002) (Figure 2C).The CPIC guidelines for CYP3A5 genotype and tacrolimus dosing suggest a 1.5-2 times higher dose for patients who express CYP3A5 *1 (Birdwell et al., 2015).At time of discharge, with statistically equivalent tacrolimus drug concentrations, patients who express CYP3A5 *1 were stabilized on a median normalized tacrolimus dose 1.98 times higher than those who did not express CYP3A5 *1.

CYP3A5 *1 expressers experienced similar, but delayed rates of AKI
To assess the effect of the tested genomic variants on clinical outcomes, we focused on the development of AKI and neurotoxicity in patients during the inpatient and outpatient stays.Expressers of CYP3A5 *1 showed a clinical trend towards lower rates of AKI compared to non-expressers during the inpatient stay (7.7% vs. 25%; p = 0.06); however, rates of AKI over the entire period of this study does not differ by CYP3A5 genotype (100% vs. 91.2%;p = 0.28).CYP3A5 *1 expressers exhibited statistically similar rates of neurotoxicity (33.3% vs. 20.1%;p = 0.33) and GVHD (50.0% vs. 49.5%;p = 0.97) as non-expressers.

Discussion
Allo-HSCT is an integral therapy modality for the treatment of a number of hematologic malignancies requiring accurate and precise use of immune suppressing agents to prevent graft mediated harm to the patient.Multiple recent studies have confirmed the effect of CYP3A5, CYP3A4, and ABCB1 in patients receiving tacrolimus via intravenous infusion in the setting of allo-HSCT (Hamadeh et al., 2019;Yoshikawa et al., 2021).However, very little data is available in the setting of oral tacrolimus.In this study, we prospectively enrolled 103 patients who received oral tacrolimus for GVHD prophylaxis following an allo-HSCT, as is standard of care at our institution.
Previous studies have demonstrated that CYP3A5 *1 expressers and CYP3A4* 1B SNPs are associated with the pharmacokinetics and pharmacodynamic properties of tacrolimus (Semiz et al., 2011;Hamadeh et al., 2019;Zhu et al., 2020;Yoshikawa et al., 2021;Liu et al., 2022).Due to the potential linkage between the CYP3A5 *1 expressers and CYP3A4* 1B SNP, it is difficult to fully characterize the effect of each individual allele (Sinues et al., 2007;Estrela et al., 2008).In our study, we determined that CYP3A5 *1 expression was associated with an increased  metabolism of tacrolimus leading to an increased dose requirement and time to reach therapeutic goals, while we were not able to identify an independent role for CYP3A4* 1B in tacrolimus metabolism (Musuamba et al., 2009;Liu et al., 2022).While most CYP3A5 *1 expressers required doses increases following the first tacrolimus level to reach therapeutic goals, overall, these patients received less dose adjustments than those with the CYP3A5 *3/ *3 genotype.This observation could at least partially be attributed to clinicians' reaction to high versus low tacrolimus levels.The clinician's reaction to subtherapeutic tacrolimus levels, to an extent, dictates how much time is required to bring CYP3A5 *1 expressers to therapeutic tacrolimus levels.Proactive genotyping of CYP3A5 may allow clinicians to feel more confident in increasing the tacrolimus dose for carriers.
Genomic variations in ABCB1 as well as the CYP3A4 *22 SNP were not associated with tacrolimus pharmacokinetics.While ABCB1 is not directly implicated in the metabolism of tacrolimus, it plays a vital role in the active transport of tacrolimus (Brunet et al., 2019;Hamadeh et al., 2019).Since our study included only patients who were initiated on oral tacrolimus, this gene was anticipated to be of great importance in tacrolimus absorption.However, there is no evidence from our study in support of the association between the selected three ABCB1 SNPs and any clinically important endpoint as were found in previous studies (Hamadeh et al., 2019).We did observe a linkage association between SNPs in ABCB1 exon 12 and both exon 21 and exon 26, which has been previously reported, therefore complicating our analysis (Allabi et al., 2005;Qiu et al., 2012).Further study is needed to better understand the clinical actionability of ABCB1 in this setting.
AKI is one of the most common adverse events for patients prescribed tacrolimus and is associated with high drug concentrations.Here, CY3A5 *1 expressers showed a clinical trend toward deceased rates of AKI during the inpatient stay, which dissipated over the course of the study.Expressers of CYP3A5 *1 spend significantly more time at subtherapeutic tacrolimus levels during the inpatient stay; however, once these patients reached therapeutic goals, the difference in AKI rates among patients with different genotypes disappeared.This finding is of significance clinically as proactive genotyping may help to identify patients who are at higher risk of early AKI development (Seligson et al., 2023).Patients who are not expressers of CY3A5 *1 may require closer renal monitoring during the inpatient stay while expressers of CY3A5 *1 should be followed closely in the outpatient setting as they reach therapeutic goals.
Finally, it is important to note key limitations to this study.First, this was a single center study that, despite strong enrollment, resulted in a relatively small number of patients with variant SNPs.For example, our study cohort did not have many patients who expressed CYP3A5 *1 or the CYP3A4 *1B allele, which compromised our effort to separate the potential impacts of CYP3A4 *1B.Our study also did not include SNPs in CYP3A5, notably *6 and *7, among other relevant SNPs in CYP3A4 and ABCB1 noted by the AMP and CPIC guidelines (Birdwell et al., 2015;Pratt et al., 2023).These SNPs were not well associated with tacrolimus metabolism when this study was designed.Further study should be sure to include these relevant SNPs in their analysis.From a clinical perspective, our study did not include any exclusions or measures of adherence to tacrolimus dosing after discharge.Data following discharge should then be considered as real-world data without a guarantee of adherence.Lastly, our study was reliant on chart review for many of the clinical variables assessed in this study, thus this study was not immune from errors in charting of the clinic course of a patient.
In conclusion, we provide evidence to support the role of CYP3A5 *1 expression in the pharmacokinetics of oral tacrolimus in a large cohort of adult patients with allo-HSCT and reflect other recent findings in similar patient populations (Yamashita et al., 2016;Zhu et al., 2020).Patients expressing CYP3A5 *1 required higher doses of tacrolimus to maintain therapeutic concentrations, suggesting that pharmacogenomics testing of CYP3A5 may be used to assist in the dosing of oral tacrolimus in this clinical setting.Prospective studies in large cohorts of patients at multiple clinical sites are warranted to verify the findings presented here.

FIGURE 2 CYP3A5
FIGURE 2 CYP3A5 *1 expressers experienced lower initial tacrolimus levels and required increased doses of tacrolimus at discharge.(A) The weight adjusted tacrolimus dose increased during the inpatient stay for CYP3A5 *1 expressers while the opposite was true for CYP3A5 *3/*3 patients.(B) There was a significant difference in the tacrolimus levels achieved in patients grouped by CYP3A5 genotype up until day 10 of therapy.(C) Patients with a CYP3A5 *3/*3 genotype were discharged on lower weight adjusted doses of tacrolimus to their starting dose.ns, not significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001, ****, p < 0.0001.

TABLE 1
Demographic characteristics of patients included in this Study (N = 103).

TABLE 1 (
Continued) Demographic characteristics of patients included in this Study (N = 103).

TABLE 2
Genotypes of selected SNPs in CYP3A5, CYP3A4, and ABCB1 in patients (N = 103).p-value was from χ2 test based on the Hardy Weinberg Equilibrium (HWE) assumption.p-value <0.05 indicates that the genotype distribution in the study cohort significantly deviates from HWE., MAF, the minor allele frequency. *

TABLE 3
Correlation between alleles and tacrolimus concentrations.
*p-value was from linear regression on all normalized tacrolimus concentrations across the study period.