Evaluation of clinical and genetic factors in the population pharmacokinetics of carbamazepine

Aims Carbamazepine can cause hypersensitivity reactions in ~10% of patients. An immunogenic effect can be produced by the electrophilic 10,11‐epoxide metabolite but not by carbamazepine. Hypothetically, certain single nucleotide polymorphisms might increase the formation of immunogenic metabolites, leading ultimately to hypersensitivity reactions. This study explores the role of clinical and genetic factors in the pharmacokinetics (PK) of carbamazepine and 3 metabolites known to be chemically reactive or formed through reactive intermediates. Methods A combination of rich and sparse PK samples were collected from healthy volunteers and epilepsy patients. All subjects were genotyped for 20 single nucleotide polymorphisms in 11 genes known to be involved in the metabolism or transport of carbamazepine and carbamazepine 10,11‐epoxide. Nonlinear mixed effects modelling was used to build a population‐PK model. Results In total, 248 observations were collected from 80 subjects. A 1‐compartment PK model with first‐order absorption and elimination best described the parent carbamazepine data, with a total clearance of 1.96 L/h, central distribution volume of 164 L and absorption rate constant of 0.45 h−1. Total daily dose and coadministration of phenytoin were significant covariates for total clearance of carbamazepine. EPHX1‐416G/G genotype was a significant covariate for the clearance of carbamazepine 10,11‐epoxide. Conclusion Our data indicate that carbamazepine clearance was affected by total dose and phenytoin coadministration, but not by genetic factors, while carbamazepine 10,11‐epoxide clearance was affected by a variant in the microsomal epoxide hydrolase gene. A much larger sample size would be required to fully evaluate the role of genetic variation in carbamazepine pharmacokinetics, and thereby predisposition to carbamazepine hypersensitivity.

Aims: Carbamazepine can cause hypersensitivity reactions in $10% of patients. An immunogenic effect can be produced by the electrophilic 10,11-epoxide metabolite but not by carbamazepine. Hypothetically, certain single nucleotide polymorphisms might increase the formation of immunogenic metabolites, leading ultimately to hypersensitivity reactions. This study explores the role of clinical and genetic factors in the pharmacokinetics (PK) of carbamazepine and 3 metabolites known to be chemically reactive or formed through reactive intermediates.
Methods: A combination of rich and sparse PK samples were collected from healthy volunteers and epilepsy patients. All subjects were genotyped for 20 single nucleotide polymorphisms in 11 genes known to be involved in the metabolism or transport of carbamazepine and carbamazepine 10,11-epoxide. Nonlinear mixed effects modelling was used to build a population-PK model.

Results:
In total, 248 observations were collected from 80 subjects. A 1-compartment PK model with first-order absorption and elimination best described the parent carbamazepine data, with a total clearance of 1.96 L/h, central distribution volume of 164 L and absorption rate constant of 0.45 h −1 . Total daily dose and coadministration of phenytoin were significant covariates for total clearance of carbamazepine. EPHX1-416G/G genotype was a significant covariate for the clearance of carbamazepine 10,11-epoxide.
Conclusion: Our data indicate that carbamazepine clearance was affected by total dose and phenytoin coadministration, but not by genetic factors, while carbamazepine 10,11-epoxide clearance was affected by a variant in the microsomal epoxide hydrolase gene. A much larger sample size would be required to fully evaluate the role of genetic variation in carbamazepine pharmacokinetics, and thereby predisposition to carbamazepine hypersensitivity.  1 Therapy with CBZ is complicated because of the drug's complex pharmacokinetic (PK) profile. [2][3][4] CBZ is almost completely metabolised in the liver and the major oxidation route is conversion to carbamazepine 10,11-epoxide (CBZE; Figure 1), which is pharmacologically active 2 and electrophilic. 5 Other 3-hydroxycarbamazepine (3OH-CBZ), 4 2,3-dihydroxycarbamazepine, and the o-quinone of the catechol. 2 Multiple cytochrome P450 (CYP) isoforms are involved in the formation of these metabolites. 4,6 Other enzymes involved in CBZ metabolism include uridine diphosphate glucuronosyltransferase (UGT2B7), 7 microsomal epoxide hydrolase and myeloperoxidase. 8 Clinically, CBZ has a narrow therapeutic index. It induces multiple CYP isoforms and transporters, 9 and also induces its own metabolism. There is large interindividual variability in plasma levels of CBZ with poor correlation to dose. Patient factors that influence the PK of CBZ include sex, age and total body weight. [10][11][12] Concomitant medications such as valproic acid, phenytoin, felbamate and phenobarbital have also been associated with variation in metabolism of CBZ. [13][14][15] CBZ is generally well tolerated, but up to 10% of patients experience a hypersensitivity reaction. 1 Carriage of HLA-B*15:02 has been associated with increased risk of CBZ-induced Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) in patients from South-East Asia while carriage of HLA-A*31:01 has been associated with increased susceptibility to all phenotypes of CBZ hypersensitivity in European, Japanese and Korean populations. 16 Phenytoin is an aromatic anticonvulsant that is also a known cause of SJS/TEN. Similar to CBZ, carriage of HLA-B*15:02 is associated with increased susceptibility to hypersenstivity. 17 Conventionally, it is thought that chemically reactive drug metabolites, such as arene oxides and epoxides, can be cytotoxic or form neo-antigens with cellular proteins, which activate the immune system, resulting in hypersensitivity reactions. 21 Recently, however, carbamazepine 10,11-epoxide has also been shown to have a specific immunogenic effect, namely alteration of selective peptide presentation by HLA-B*15:02, through binding to the protein, a property not demonstrated by CBZ. 22 Theoretically, genetic variants in CBZ metabolism could not only alter the routes of metabolism between different individuals, but they could also increase the formation of chemically reactive and otherwise immunogenic metabolites. 23 Table 1 outlines the studies that have investigated effects of genetic

| Selection of genetic polymorphisms
The effect of genetic variation on the PK of CBZ was investigated by analysis of single nucleotide polymorphisms (SNPs). The selection of genes and SNPs (

| Population PK modelling
Where:  The known auto-induction effect of CBZ, was incorporated in the base model as a binary, categorical induction status effect on CL TOT and CL 3E . Data from PICME I and the first visit of the PICME II autoinduction group were categorised as having come from a noninduced period while data from the second visit of the PICME II autoinduction group and the PICME II maintenance group were categorised as from an induced period (i.e. a binary, noncontinuous time function)-reflecting the length of CBZ dosing up to the point of PK observation, with the latter groups having received at least 14 days daily dosing of CBZ, which would be expected to increase clearance. More sophisticated models for autoinduction as a continuous function of time (and drug exposure, taking the form of e.g. an autoinduction enzyme turnover model used previously for rifampicin were attempted in exploratory analysis, but failed to minimise successfully and/or give acceptable parameter estimate precision. 54 We believe that this is at least in Data from PICME I and PICME II autoinduction visit 1 were IND = 0, while data from PICME II autoinduction visit 2 and PICME II maintenance were IND = 1. Further equations for covariate models are expressed in typical value format, omitting the random effect component for clarity.
Exploratory analyses investigating incorporation of interoccasion variability as a random effect in the model failed to produce acceptable fittings, despite the potential for this to be required to describe the PICME II autoinduction group data in particular, where patients' PK was monitored on 2 occasions. These analyses required only a single θ COV degree of freedom parameter: Covariate model selection adopted a standard forward addition/backward deletion approach. 55 The covariates were initially examined in

| Nomenclature of targets and ligands
Key protein targets and ligands are hyperlinked to corresponding entries in http://www.guidetopharmacology.com, 56 and are permanently archived in the Concise Guide to Pharmacology 2019/20. 57 3 | RESULTS

| Subject demographics and genetic polymorphisms
In total, 80 subjects were recruited into the studies (Table 3). Rich PK sampling data were obtained from 8 healthy volunteers who completed the PICME I study. Rich PK sampling data were also obtained from 3 patients with a new diagnosis of epilepsy who completed dose titration of CBZ as part of the autoinduction group of PICME II. Eighty sparse PK samples were collected from 69 patients who were recruited to the maintenance group of PICME II. In total, 248 sets of drug and metabolite assays were used in the analysis.
The distribution of the 20 SNPs amongst these subjects is recorded in Table 4. No subjects were carriers for CYP2C19*17 (rs192154563). Each genotype frequency was consistent with HWE, and minor allele frequencies ranged from 4 to 55% in keeping with polymorphism frequencies reported in the literature. No demographic covariates were significant in the final PK model.

| Base model and covariate fitting
As illustrated by the VPC and diagnostic plots (Figures 3 and 4), the PK profiles for CBZ and its 3 assayed metabolites, and their variability,  The population mean estimates for CL TOT , V1 and KA of CBZ are provided in Table 5. The CL TOT value in our model was 1.96 L/h which falls in the centre of previously published CL TOT values which range between 1.15 and 3.58 L/h. 10,11,13,14 The CL value for CBZE in our model was 9.71 L/h. In a previous population PK model, which included CBZE the CL estimate for CBZE in a 70 -kg patient prescribed 400 mg CBZ was 28 L/h. 58   The gene EPHX1 encodes microsomal epoxide hydrolase, which catalyses hydrolysis of the electrophilic CBZE to DiOH-CBZ. 6 The  61 The authors postulate that the polymorphism may contribute to the risk of CBZ-induced SJS/TEN by increasing the plasma concentration of CBZE. The same SNP in EPHX1 was associated with increased levels of CBZE in 1 study 24 but no differences were  were not significant covariates of the PK of CBZ.

| DISCUSSION
The influence of phenytoin therapy on CBZ's metabolism has been recognised in several other population PK models. [11][12][13]15 Phenytoin increases the metabolism of many drugs, including immunosuppressants, 62 chemotherapeutic agents 63 and antiretroviral drugs, 64 through induction of multiple CYP isoforms and upregulation of P-gp. 65 Patients who are coprescribed CBZ with phenytoin potentially require larger doses of CBZ to maintain plasma levels.
The total daily dose of CBZ was another significant covariate in the population PK model. All subjects in the study were prescribed controlled release formulations of CBZ. There was a positive correlation between total daily dose and clearance of the drug. The effect of dose on clearance may be explained by a reduction in bioavailability and increase in clearance through greater autoinduction at higher doses. 66 Several other population PK models of CBZ have reported that the total daily dose is a significant covariate. 10,[67][68][69][70] Concomitant treatments with sodium valproate, 11,12,14,69 phenobarbital 11-13,67,69,71 and felbamate 13 have been reported to be significant covariates in population PK models of CBZ. None of those antiepileptic drugs (AEDs) achieved a significant effect in the current model. This is most likely to have been due to the small numbers of subjects receiving any of these drugs. No other AEDs were significant in our model. Concomitant treatment with omeprazole (n = 6) and statins (n = 13) were also investigated as covariates in the population PK model. Omeprazole is a proton-pump inhibitor that is associated with drug-drug interactions secondary to its inhibition of CYP2C19 and CYP3A4. 72 Statins have been reported to inhibit CYP enzymes, 73 and they act as substrates of the organic anion transporters OATP1B1 and OATP1B3 and of P-gp. 74 Neither was found to affect metabolism of CBZ significantly.
There is considerable uncertainty regarding the influence of genetic variation on metabolism of CBZ, with conflicting results from several studies (Table 1). Up to 1/3 of patients with epilepsy do not respond to AED therapy and the transporter hypothesis proposes that over expression of efflux transporters such as ABCB1 in the bloodbrain barrier limits access of AEDs to the epileptic focus. 75 Earlier studies have reported the requirement for higher CBZ maintenance doses, 24 increased clearance of CBZ 29 as well as both higher 32  Similarly, combined HLA and CYP2C9*3 pharmacogenetic screening for phenytoin hypersensitivity improved the sensitivity and specificity of predictive testing. 20 We have previously reported that chemically reactive metabolites of CBZ, namely CBZE and arene oxides, are able to form covalent adducts with human serum albumin. 5 We hypothesise that reduced clearance of CBZE in subjects with the EPHX1-416G/G genotype leads to greater formation of covalent protein adducts, ultimately resulting in an increased frequency of immune-mediated hypersensitivity reactions, especially in those expressing the HLA risk allele.
Limitations of the current study include the relatively small number of subjects and the limited number of elderly patients.
Consequently, the power of the study might have been insufficient to (c.-163C > A) genotype has been shown to affect CBZ's PK in children. 70 Finally, the recent discovery of 856 SNPs in human CYP3A4, 77 which codes the principal monoxygenase catalysing CBZ 10,11-epoxidation, 76 indicates PK might be influenced by many more genetic polymorphisms than were included in this or any other study.
To make significant progress, future studies should attempt to characterise the genetic polymorphisms on a much larger scale in patients with CBZ hypersensitivity reactions in order to determine the influence of many more variants on metabolism pathways within a relevant, pathological, context.
In conclusion, a population PK model has been developed for CBZ which successfully incorporates certain clinical parameters of adult epilepsy patients. We identified carriage of the EPHX1-416G/G genotype as being a significant covariate of CBZE clearance and concomitant treatment with phenytoin and CBZ dose as significant covariates in affecting the PK of CBZ.