Composite CYP3A (CYP3A4 and CYP3A5) phenotypes and influences on tacrolimus dose adjusted concentration in adult heart transplant recipients

CYP3A5 genetic variants are associated with tacrolimus metabolism. Controversy remains on whether CYP3A4 increased [* 1B (rs2740574), *1G (rs2242480)] and decreased function [*22 (rs35599367)] genetic variants provide additional information. This study aims to address whether tacrolimus dose-adjusted trough concentrations differ between combined CYP3A (CYP3A5 and CYP3A4) phenotype groups. Significant differences between CYP3A phenotype groups in tacrolimus dose-adjusted trough concentrations were found in the early postoperative period and continued to 6 months post-transplant. In CYP3A5 nonexpressers, carriers of CYP3A4*7Bor *7G variants (Group 3) compared to CYP3A4*1/*1 (Group 2) patients were found to have lower tacrolimus dose-adjusted trough concentrations at 2 months. In addition, significant differences were found among CYP3A phenotype groups in the dose at discharge and time to therapeutic range while time in therapeutic range was not significantly different. A combined CYP3A phenotype interpretation may provide more nuanced genotype-guided TAC dosing in heart transplant recipients.


INTRODUCTION
A calcineurin inhibitor-based regimen is generally considered the cornerstone of immunosuppression following heart transplantation (HTx). 1 Currently, the calcineurin inhibitor tacrolimus (TAC) is most often used; however, inter-individual variability and a narrow therapeutic window lead to challenges with both drug toxicities and e cacy. 2 Despite frequent TAC monitoring and dose adjustment during the early postoperative period, many patients fail to maintain therapeutic TAC levels. 3,4 Decreased time in the therapeutic range (TTR) has been associated with an increased risk of rejection, highlighting the importance of personalizing TAC dosing beyond standardized dosing and drug monitoring. [3][4][5] One strategy that may improve patient outcomes is to use pharmacogenetic test results to optimize TAC dose and, ideally, attain a higher TTR. 6-8 To effectively use this strategy, a better understanding of the impact of variants in two key pharmacokinetic genes on variability in TAC concentrations is needed.
TAC is metabolized in the liver and intestines by cytochrome P450 3A5 (CYP3A5) and 3A4 (CYP3A4) enzymes, and in solid organ transplant recipients CYP3A5 variants can account for up to 45% of the variability in TAC dose requirements. 9,10 Patients who are CYP3A5 expressers (*1 carriers) are predicted to have signi cantly lower doseadjusted trough concentrations (C 0 /D) and require higher doses of TAC compared to patients who are CYP3A5 nonexpressers (*3,*6, and *7 homozygotes). 11 Although CYP3A5 variants are known to impact TAC pharmacokinetics, variants in CYP3A4 may also play an important role, particularly for CYP3A5 nonexpressers. 11 While there are no guidelines for CYP3A4-guided TAC dosing, CYP3A4 variants predicted to increase enzyme function [*1B (currently known as c.-392G > A) and *1G (currently known as c.1026 + 12G > A)], and decrease enzyme function (*22) were associated with TAC metabolism in renal and lung transplant patients. [12][13][14][15][16][17][18] Our current knowledge on the effect of the combined CYP3A phenotype on TAC concentrations in HTx recipients is limited. 8, 19,20 In addition, controversy remains on whether CYP3A4 variants are clinically relevant in HTx recipients due to con icting and limited literature as well as linkage disequilibrium between CYP3A5*3 and CYP3A4*1B (rs776746 and rs2740574) reported in some populations. 14,20 A combined CYP3A (CYP3A4 and CYP3A5) phenotype can potentially provide a more nuanced metabolic pro le and thus more accurately aid in individualizing TAC dosing post-transplantation.
To our knowledge no HTx study has evaluated CYP3A phenotypes using both CYP3A5 (*3, *6, *7) and CYP3A4 (*1B, *1G, *22) genetic variants. Thus, we sought to determine the effect of a predicted CYP3A composite phenotype on TAC pharmacokinetic outcomes in the HTx population. We hypothesize that CYP3A composite phenotypes in uence TAC C 0 /D in HTx recipients in the early postoperative period and up to 6 months after HTx.

Study design and population
We conducted a single-center, retrospective study using Vanderbilt University Medical Center's (VUMC) BioVU, a DNA repository linked to de-identi ed electronic health records (EHR). 21,22 Patients initially transplanted in the adult HTx program at VUMC were screened for inclusion. Patients were included if they received oral TAC postoperatively and had genotyping data available [CYP3A4 (*1B, *1G, *22) and CYP3A5 (*3, *6, *7)]. Sublingual TAC doses were converted to oral equivalents (1:2 ratio) to account for differences in absorption. 23 Patients were excluded if they received a combined HTx with another organ, were administered intravenous TAC, or had incomplete data (< 5 days of inpatient TAC data). TAC (dose and concentration), other medications (i.e. immunosuppressants, prophylactic antimicrobials, CYP3A inducers and inhibitors), and demographic data were abstracted from the EHR by manual review. Medication data were con rmed through inpatient administration data and manual review of notes. This study was reviewed and approved by the VUMC Institutional Review Board.
Tacrolimus trough data TAC monitoring and dose adjustments were performed per institutional protocol with no major changes throughout the study period. TAC trough concentrations (C 0 ) were collected daily postoperatively during a patient's index stay, and then at each subsequent HTx clinic visit thereafter. Dose adjustments were made by treating clinicians to achieve a TAC C 0 of either 8-10 ng/ml or 10-12 ng/ml taking into account patient comorbidities, perceived rejection risks and potential drug-drug interactions. TAC C 0 were retrospectively captured from the de-identi ed EHR. For the primary outcome analysis, missing troughs were estimated using the median of the previous and subsequent troughs. 18 TAC concentrations measured < 10 or > 14 hours after the previous dose were excluded as non-true troughs.

Outcome de nitions
The primary outcome was median dose-and weight-adjusted TAC trough concentration (C 0 /D) from postoperative day (POD) 2 to discharge among CYP3A groups, capturing TAC pharmacokinetics in the inpatient setting. Post hoc analyses on CYP3A4 and CYP3A5 phenotypes were conducted to investigate impact of individual genes. C 0 /D from POD 14 to day 30 was also assessed to explore the early and late post HTx setting. The daily TAC C 0 /D was calculated as follows: C 0 /D = (TAC trough, ng/ml)/(TAC dose, mg/kg/d). The TAC dose was de ned as the average of doses administered during the 48 hours (~ 4 half-lives) before the trough. If no doses were given in the preceding 48 hours (e.g. when doses were held due to elevated troughs), the daily C 0 /D was unde ned and excluded. During a patient's index admission, weight was de ned as the recorded weight on the day of HTx or within 7 days prior when available. Following hospital discharge, weight was de ned by weight on the day of discharge.
Additional pharmacokinetic outcomes of interest included median TAC C 0 /D at two, three-, and six-months posttransplant, TAC C 0 /D and dose at discharge, time in therapeutic range (TTR), time to therapeutic range (TtTR), percent of troughs in range, percent of subtherapeutic (≤ 8 ng/ml) and supratherapeutic levels (≥ 12 ng/ml) during POD 2 to 30. The Rosendaal linear interpolation method was used to calculate TTR using trough goal ≥ 8 and ≤ 12 ng/ml. 24 The TtTR was assessed as the number of days between the rst TAC dose and two consecutive troughs ≥ 8 and ≤ 12 ng/ml (where median of previous and post troughs were used for days of no trough levels). 18 Given that moderate and strong CYP3A4 inducers and inhibitors are known to affect CYP3A, we conducted post hoc analysis to investigate the impact of concomitant drugs (identi ed by FDA and Flockhart tables) on the median TAC Page 4/18 C 0 /D. 25,26 We also assessed acute kidney injury (AKI) incidence and stage using the Kidney Disease Improving Global Outcomes guideline criteria from day of HTx to POD 30. 27 Genotyping and haplotype inference We selected CYP3A single nucleotide variants (SNVs) previously associated with TAC dose and clinical response in solid organ transplant populations [CYP3A5*3 (rs776746), *6 (rs10264272), *7 (rs41303343); CYP3A4*1B (rs2740574), *1G (rs2242480), and *22 (rs35599367, rs2740574)]. Patient genotypes were obtained from BioVU and previously assayed using the Illumina Expanded Multi-Ethnic Genotyping Array in the Vanderbilt Technologies for Advanced Genomics laboratory using manufacturer-speci ed reagents and protocols. All samples with ≥ 99% average call rates and passing quality control were included. CYP3A4*1G and CYP3A5*7 were imputed through statistical inference to the 1000G Phase3v5 reference panel using Minimac4 on the Michigan Imputation Server. Imputed variants had Rsq ≥ 0.97, indicating high-quality imputation.

Study Population
In all, 213 patients were screened for eligibility, 36 of whom were eliminated by inclusion or exclusion criteria (Fig. 1). Thus, a total of 177 patients HTx between March 2008 and 2020 were assessed for the primary analysis with characteristics described in  Observed genotypes and assigned phenotypes Most patients (n = 82, 46%) were in CYP3A Group 2 as expected given the higher prevalence of self-identi ed White patients ( Table 2). All genotype frequencies were in Hardy-Weinberg equilibrium (HWE), except for CYP3A4*1B, CYP3A4*1G, and CYP3A5*3 (Table S1). When taking the population into consideration and analyzed separately, only CYP3A4*1B deviated from HWE in the White population. The diplotypes and allele frequencies are found in Table S2.
Pharmacokinetic outcomes CYP3A, CYP3A4 and CYP3A5 tacrolimus concentration/dose ratio A total of 2,708 TAC troughs were included in the analysis. Median TAC C 0 /D from POD 2 to discharge for the whole cohort was 109 (ng/mL)/(mg/kg/d) [IQR 78-182]. There were statistically signi cant differences between the four CYP3A groups in median TAC C 0 /D from POD 2 to discharge (P = 0.001, Table 3). CYP3A Group 1 had the highest median TAC C 0 /D followed by Group 2, Group 3, and Group 4 (Fig. 2, Table 3). This difference was also sustained from POD 14 to 30 which had similar median TAC C 0 /D trends (P = 0.0001, Table S3). Group 4 had median TAC C 0 /D which was 59%, 46%, and 34% lower compared to Group 1, Group 2, and Group 3, respectively, during POD 2 to discharge. A similar pattern was found when the CYP3A4 and CYP3A5 phenotypes were analyzed separately in our post hoc analysis. CYP3A4 rapid expressers had 44% and 57% lower median TAC C 0 /D compared to normal expressers and poor/intermediate expressers, respectively, from POD 2 to discharge (P < 0.0005) (Fig. 3, Table S4). In the CYP3A5 phenotypes, intermediate and normal metabolizers had 45% and 56% lower C 0 /D compared to poor metabolizers, respectively, from POD 2 to discharge (P < 0.0005) (Fig. 4, Table S5).  Dose at discharge and ambulatory tacrolimus concentration/dose ratio At discharge, CYP3A Group 1 had the highest and Group 4 had the lowest median TAC C 0 /D -at 2 and 6 months the trends were similar ( Table 3). The opposite trend was seen for the dose at discharge where Group 4 had 2-, 1.7-, and 1.5-fold higher dose requirements compared to Group 1, Group 2, and Group 3, respectively (P < 0.0005, P < 0.0005, P = 0.0053, respectively, Fig. 5). At 2 months post-HTx, pairwise post hoc analysis showed the median TAC C 0 /D between Group 2 and Group 3 was signi cantly different (P = 0.02). Group 2 and Group 3 are CYP3A5 poor metabolizers, with the differences being that Group 3 are CYP3A4*1B or *1G carriers (rapid expressers) while Group 2 does not carry any CYP3A4 variants. Additional outcomes for C 0 /D at discharge, 2 months, 3 months, and 6 months post-HTx are summarized in S4 and S5.
Time in therapeutic range, time to therapeutic range, and percent of therapeutic levels TTR ranged from 30%-57% and the percent of troughs in therapeutic range ranged from 29-52% among the CYP3A groups; however, there were no signi cant differences across groups for either outcome ( Table 3, Table S3). The TtTR was statistically different across groups (P = 0.008); CYP3A Group 1 took the fewest number of days, while Group 4 took the greatest number of days, to reach therapeutic range (7.5 vs 10.5 days, respectively, post-hoc P = 0.001).
Outcomes for percent of subtherapeutic and supratherapeutic levels are summarized in Table S3. Drug-drug interactions and AKI outcomes analysis are available in the supplementary document (Table S6, Table S7).

DISCUSSION
To our knowledge, this is the rst study to evaluate the effect of combined CYP3A phenotypes in HTx recipients. We found differences in early post operative TAC C 0 /D between predicted CYP3A phenotype groups. In addition, these differences remained signi cant at 2, 3 and 6 months. We also found evidence suggesting differences in TAC C 0 /D are driven by CYP3A4 variants in CYP3A5 nonexpressers, similar to a recent HTx study using CYP3A5 genotyping and CYP3A4 phenotyping via mRNA expression. 20 Our ndings were evident at 2 months post-HTx, where a signi cant difference was observed in TAC C 0 /D between CYP3A Group 2 and Group 3 (P = 0.02). Given both groups are CYP3A5 nonexpressers, differences are likely attributable to CYP3A4*1B and *1G carrier status in CYP3A Group 3. However, differences driven by CYP3A4*22 in CYP3A5 nonexpressers (Group 1 vs Group 2) were not found potentially due to our small sample size (Group 1, n = 14) and the lack of CYP3A4*22 homozygous carriers in our cohort.
Prior to this study, combined CYP3A phenotypes with increased (CYP3A4*1B and *1G) and decreased (CYP3A4*22) function CYP3A4 alleles had only been assessed in our previous study of lung transplant recipients. 18 Similar to the lung transplant cohort, CYP3A4*22 and CYP3A5 variant carriers (Group 1) in our HTx population had the highest TAC C 0 /D while CYP3A4 *1B,*1G variant carriers and CYP3A5 expressers (Group 4) had the lowest TAC C 0 /D. 13,17,18 Our previous work in lung transplant recipients showed that patients with the lowest predicted CYP3A activity (Group 1) had two-fold higher median C 0 /D and 51% lower TAC dose requirements at discharge while patients with the highest CYP3A activity (Group 4) had 43% lower median C 0 /D and 1.7 fold higher dose requirements at discharge compared to patients who were CYP3A5 nonexpressers and CYP3A4 normal expressers (Group 2). 18 In our HTx cohort, Group 1 had 1.3-fold higher median C 0 /D and 14% lower TAC dose requirements at discharge while Group 4 had 46% lower median C 0 /D and 1.7 fold higher dose requirements at discharge when compared to Group 2. The less variable difference between Group 1 and Group 2 in our HTx cohort compared to lung cohort suggests additional validation in differences between transplant populations.
Previously published transplant studies have investigated in vivo CYP3A4 activity as well as CYP3A4 and CYP3A5 variants with increased or decreased function. 8,13,17,19,28 Luo et al found endogenous CYP3A4 phenotype (assessed by urinary metabolic ratio) signi cantly correlated with TAC C 0 /D and weight-corrected daily dose in renal transplant recipients. 28 After regression analysis, endogenous CYP3A4 phenotype, CYP3A5*3, and post-operative period accounted for 60% of the variability in C 0 /D. In CYP3A5 nonexpressers (CYP3A5*3/*3), CYP3A4 phenotype was responsible for 52% and 40% of the variability in TAC C 0 /D and dose, respectively; however, CYP3A4 phenotype accounted for only 15% and 11% of the variability, respectively, among CYP3A5 expressers. Our study also supports TAC pharmacokinetic differences in CYP3A4 variant carriers among CYP3A5 nonexpressers (Group 2 vs 3) at 2 months. Aouam et al. found that in renal transplant recipients, CYP3A4*1B carriers had signi cantly lower TAC C 0 /D and required higher doses to maintain therapeutic C 0 compared to CYP3A4*1/*1. 17 After a stepwise regression model, only CYP3A4 polymorphism correlated signi cantly with C 0 /D variation regardless of the post-transplant phase and explained 22% of the variability in the late transplant phase (over 90 days). These renal studies show similar ndings of CYP3A4 variants accounting for signi cant C 0 /D and dose requirement variation.

Limitations
Given the retrospective and single-centered nature, our study has limitations and is most applicable to populations similar to ours. Additional genetic factors, drug-drug interactions, as well as differences in oral formulations could also in uence our ndings. TAC is a substrate of P-glycoprotein (encoded by ABCB1) and in uenced by P-glycoprotein inhibitors and inducers, which we did not assess given the con icting evidence on utility of ABCB1 variants. 18,29 CYP3A4 rare and unknown function variants were not included in this study which may decrease accuracy of the phenotype prediction compared to comprehensive genotyping and phasing methods. Although we assessed for differences in TAC C 0 /D in patients who were concomitantly taking CYP3A inhibitors, there was a small sample size, therefore, our study is not powered to nd differences pertaining to drug-drug-gene interactions. Patients were administered TAC capsules and suspension orally or through a nasogastric tube, which may affect the TAC pharmacokinetic pro le. However, the relative bioavailability is found to be comparable and our outpatient results (where all patients were on capsules) were similar to those in the inpatient setting. 30 Patient adherence and administration of TAC could not be veri ed in the outpatient setting; however, notes were reviewed to assess accuracy of doses. CYP3A4*1B likely deviated from HWE due to differences in allele frequencies in White populations. 11,23,24 Lastly, there was a small sample size of CYP3A4*22 carriers (n = 14) and no homozygous carriers, limiting our analysis of the impact of CYP3A4*22 (CYP3A4 Group 1) on TAC C 0 /D.
In conclusion, a combined CYP3A phenotype signi cantly impacted TAC C 0 /D both in the early postoperative setting and up to 6 months following HTx. Consequently, dose requirements at discharge, and TtTR signi cantly differed between the groups. Post hoc analysis showed these differences were driven mainly by CYP3A4 Group 4 (CYP3A4*1B,*1G carriers) differing from other groups. Genotyping for both CYP3A5 and CYP3A4 variants can have the potential to provide more nuanced genotype guided TAC dosing to optimize patient outcomes. Further investigations are planned to assess the association of ndings from this study with clinical outcomes including acute rejection, hospital readmission, development of de novo donor speci c antibodies and graft vasculopathy, as well as to assess the effect of a combined CYP3A phenotype on patients of various ancestries. Our ndings are most pertinent to populations similar to ours, enriched with CYP3A5*3 carriers. Future research in larger HTx populations across multiple centers are necessary to con rm our results and to assess the effect of a combined CYP3A phenotype on patients of diverse ancestries.  Analysis of tacrolimus C 0 /D from POD 2 to discharge by CYP3A4 phenotype

Declarations
Page 18/18 Figure 4 Analysis of tacrolimus C 0 /D from POD 2 to discharge by CYP3A5 phenotype Figure 5 Tacrolimus dose at discharge by CYP3A phenotype

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.