Plasma and Hepatic Exposures of Celecoxib and Diclofenac Prescribed Alone in Patients with Cytochrome P450 2C9*3 Modeled after Virtual Oral Administrations and Likely Associated with Adverse Drug Events Reported in a Japanese Database

Abstract

The impacts of polymorphic cytochrome P450 (P450 or CYP) 2C9 on drug interactions and the pharmacokinetics of cyclooxygenase inhibitors have attracted considerable attention. In this survey, the prescribed dosage was reduced or discontinued in 150 and 56 patients, respectively, receiving celecoxib and diclofenac prescribed alone, as recorded in a Japanese database of adverse drug events. Among the factors underlying adverse events, intrinsic drug clearance rates may be a contributing factor. The pharmacokinetically modeled plasma concentrations of celecoxib after an oral 200-mg dose increased in CYP2C9*3 homozygotes: the area under the plasma concentration curve was 4.7-fold higher than that in CYP2C9*1 homozygotes. In patients with CYP2C9*3/*3, the virtual hepatic concentrations of diclofenac after three daily 25-mg doses for a week were 11-fold higher than the plasma concentrations in subjects with CYP2C9*1/*1. The in vivo and in vitro fractions of the victim drug metabolized by a specific polymorphic P450 form is an important determining factor for estimating drug–drug interactions. Virtual hepatic and plasma exposures estimated by pharmacokinetic modeling in patients harboring the impaired CYP2C9*3 allele could represent a causal factor for adverse events induced by celecoxib or diclofenac in a manner similar to that for drug interactions.

INTRODUCTION

Individual drug therapies are important to ensure safe medical outcomes. The potential for polymorphic cytochrome P450 (P450 or CYP) 2C9 to cause drug interactions and the corresponding labeling recommendations for many clinically important drugs such as phenytoin, valproic acid, and warfarin are well known.¹⁾ Patients receiving phenytoin or valproic acid are generally controlled by therapeutic drug monitoring based on plasma concentrations,^2,3) whereas those receiving warfarin are controlled using therapeutic prothrombin time–international normalized ratios.^4,5) Although CYP2C9 genotypes have been introduced on the package insert for the new drug siponimod,^6,7) there are few other cases of information relating to the influence of CYP2C9 polymorphism on drug labeling recommendations in Japan.

We focused on investigating events reported in the Japanese Adverse Drug Event Report (JADER) database associated with the prescription of single drugs only.^8,9) Among the various factors underlying the mechanisms related to these adverse events associated with the prescription of celecoxib or diclofenac alone, variations in the in vivo intrinsic clearance of the drug, mediated mainly by P450 enzymes, in these individuals may be a causal factor. We recently reported that a high virtual hepatic and plasma exposure (generated using pharmacokinetic modeling) in subjects harboring impaired CYP3A4*16 may indicate a causal factor for statin intolerance.⁹⁾ The aim of the current study was to estimate virtual hepatic and plasma exposures to celecoxib and diclofenac, as model P450 2C9 substrates, in subjects with impaired. CYP2C9 The in vitro fractions of intrinsic clearance values for diclofenac and celecoxib and diclofenac metabolized to 4′-hydroxydiclofenac by liver microsomal P450 2C9 were high (0.9–1.0).¹⁰⁾ The P450 2C9 Ile359Leu missense variant, namely CYP2C9*3 (rs1057910), has been found in East Asians and had allele frequencies of 2.4% in a Japanese population (686/28258, Tohoku Medical Megabank database 14KJPN, https://www.ncbi.nlm.nih.gov/snp/rs1057910). Using simplified physiologically based pharmacokinetic (PBPK) modeling, we estimated the elevation of virtual hepatic and plasma exposures of celecoxib and diclofenac mediated by the impaired variant CYP2C9.3.

MATERIALS AND METHODS

Numbers and Types of Adverse Events with Celecoxib and Diclofenac Prescribed Alone

The number of patients recorded in the JADER database experiencing adverse drug events between April 2004 and March 2022 associated with celecoxib or diclofenac prescribed alone was investigated in the same manner as that used for atorvastatin cases described previously.⁹⁾ Diclofenac cases were extracted for oral administration only. The number of drug regimen adjustments recorded as “reduced” or “discontinued” and the number of adverse event cases were analyzed. The times to the onset of events of interest, such as hepatic disorders after celecoxib or diclofenac administration, were calculated as previously described,⁹⁾ and the frequencies and types of adverse events are shown in Table 1. Cumulative incidences of adverse events were evaluated using the Kaplan–Meier method with JMP Pro 13.2.1 (SAS Institute, Cary, NC, U.S.A.) and Prism 9 (GraphPad Software, La Jolla, CA, U.S.A.).

Table 1. Adverse Events Associated with Celecoxib or Diclofenac Therapy in Subjects Not Receiving Any Other Prescription Drugs

Adverse event	Number		Percentage, %
	Celecoxib	Diclofenac	Celecoxib	Diclofenac
Total	1005 (76)	691 (48)	100	100
Cerebral infarction	23 (4)	3	2.3	0.4
Renal impairment	21 (1)	6	2.1	0.9
Gastric ulcer	19 (1)	33 (2)	1.9	4.8
Drug-induced liver injury	16 (2)	4	1.6	0.6
Abnormal hepatic function	14 (3)	4 (3)	1.4	0.6
Renal disorder	8	2	0.8	0.3
Gastric ulcer hemorrhage	8 (2)	24	0.8	3.5
Acute kidney injury	6	15 (1)	0.6	2.2
Liver disorder	6 (1)	12 (4)	0.6	1.7
Acute hepatitis	2	2 (1)	0.2	0.3
Alanine aminotransferase increased	2	3	0.2	0.4
Aspartate aminotransferase increased	2	3	0.2	0.4
Aspirin-exacerbated respiratory disease	2	10	0.2	1.4
Others	876 (62)	570 (37)	87.2	82.5

Data were obtained from the JADER database. There were 788 and 444 subjects for celecoxib and diclofenac, respectively (Fig. 1). Some subjects had multiple adverse events. The numbers in parentheses are adverse events used to calculate the time to onset in 67 and 37 patients, respectively.

Virtual Plasma and Hepatic Concentrations of Celecoxib and Diclofenac

Plasma concentrations versus time datasets for celecoxib and diclofenac after oral administration have been recorded in subjects reported/estimated to harbor CYP2C9*1/*1 and CYP2C9*3/*3. The procedure for establishing simplified human PBPK models has been described previously^9,11) and is based on the intrinsic clearance mediated by polymorphic CYP2C9. The input parameters required for PBPK models, such as the in vivo hepatic intrinsic clearance (CL_h,int), are generally computed to provide the best fit to reported plasma concentrations,⁹⁾ as briefly outlined in the footnote to Table 2. The input parameters and output results, respectively, for PBPK modeling in subjects harboring CYP2C9*1 and CYP2C9*3 alleles are shown in Tables 3 and 4.

Table 2. Chemical Properties and Calculated Parameters for PBPK Modeling of Celecoxib and Diclofenac Based on Reported Pharmacokinetic Data

Parameter	Abbreviation (unit)	Drug
		Celecoxib	Diclofenac
Octanol–water partition coefficient	Log P	4.37	4.58
Plasma unbound fraction	fu,p	0.0317	0.00812
Blood–plasma concentration ratio	Rb	0.740	0.664
Liver (kidney)–plasma concentration ratio	Kp,h/Kp,r	2.96	2.90
Fraction absorbed × intestinal availability	Fa·Fg	0.804	0.883
Absorption rate constant	ka (1/h)	0.988 ± 0.095a)	0.422 ± 0.049a)
Volume of systemic circulation	V1 (L)	280 ± 1a)	73.9 ± 0.4a)
Hepatic intrinsic clearance	CLh,int (L/h)	1380 ± 10a)	4030 ± 10a)
Hepatic clearance	CLh (L/h)	30.1	24.4
Renal clearance	CLr (L/h)	0.608	2.52
Maximum concentration in plasma	Cmax (ng/mL)	345 (0.94)b)	402 (0.96)b)
Area under the concentration curve from 0 to 24 h in plasma	AUC24 (ng h/mL)	3460 (1.10)b)	2690 (1.16)b)

The acid dissociation constant, plasma unbound fraction (f_u,p), and octanol–water partition coefficient (log P) for celecoxib and diclofenac (molecular weight 381 and 296, respectively) were obtained by in silico estimation using ACD/Percepta, Simcyp, and Chem Draw software, respectively.^21–23) The liver (kidney) to plasma concentration ratios (K_p,h/K_p,r) and blood-to-plasma concentration ratio (R_b) were calculated from the f_u,p, and log P values as follows:

The reported basic pharmacokinetic data for celecoxib¹³⁾ and diclofenac¹⁹⁾ were used. a) Data of fitting estimations (mean ± standard deviation) were modified by the reported information for 2% renal elimination of celecoxib.²⁴⁾ A set of differential equations was solved for the amounts and concentrations:

where X_g, V_h, V_r, C_h, C_r, and C_b are the amount of drug in the gut compartment; liver and kidney volumes; and hepatic, renal, and blood substrate concentrations, respectively.²⁵⁾ V_h, V_r, and Q_h/Q_r are the liver (1.5 L) and kidney (0.28 L) volumes and the blood flow rates of the systemic circulation to the hepatic/renal compartments (96.6 L/h) in humans (70 kg body weights).²⁵⁾ b) Estimated PBPK model outputs for virtual oral administration of 200 mg celecoxib¹³⁾ and 100 mg diclofenac.¹⁹⁾ The values in parentheses are the ratios to the observed values. C_max, maximum concentration; AUC₂₄, area under the concentration curve from 0 to 24 h.

Table 3. Reported in Vitro Hepatic Intrinsic Clearance Values of Celecoxib and Diclofenac by P450 2C9 for Estimating in Vivo Hepatic Intrinsic Clearance Values as Input Parameters for PBPK Modeling in Subjects Harboring CYP2C9*1 or CYP2C9*3

Parameter	Celecoxib	Diclofenac
Fitted in vivo hepatic intrinsic clearance for CYP2C9.1, L/h, taken from the literature	1380 ± 10 from literature13) and 894 ± 114)	4030 ± 10 from literature19)
Fitted in vivo hepatic intrinsic clearance for CYP2C9.3, L/h, taken from the literature	201 ± 113) and 122 ± 514)	Not available
Reported in vitro fraction metabolized by CYP2C9	0.926)	1.010)
Reported in vivo fraction metabolized by CYP2C9	Not available	0.515)
In vitro intrinsic clearance
Vmax and Km for 2C9.1	8.9 min−1 and 3.3 µM27)	13 min−1 and 1.8 µM20)
Vmax and Km for 2C9.3	0.9 min−1 and 3.6 µM27)	8.1 min−1 and 11 µM20)
2C9.3/2C9.1 ratio of Vmax/Km	0.093	0.10
Estimated in vivo hepatic intrinsic clearance for CYP2C9.3, L/h	254	403 and 2360

The absorption rate constant (1/h) and volume of the systemic circulation (L) of celecoxib for the mean of two subjects with CYP2C9*3/*3¹³⁾ were modified from the basic values for 26 Korean subjects with CYP2C9*1/*1, as shown in Table 2, to 0.664 ± 0.068 (1/h) and 260 ± 1 (L), respectively. To fit the other reported pharmacokinetic data for celecoxib in 13 Caucasian subjects with CYP2C9*1/*1 and one subject with CYP2C9*3/*3¹⁴⁾, the absorption rate constants (1/h) and volumes of systemic circulation (L) were modified to 1.01 ± 0.12, 0.989 ± 0158 (1/h), and 172 ± 1 and 168 ± 5 (L), respectively.

Table 4. Estimated Plasma and Liver C_max and AUC₁₆₈ Values Obtained Using Human PBPK Models after Virtual Oral Doses of 200 mg Celecoxib Twice Daily for 7 d or 25 mg Diclofenac Three Times Daily for 7 d in Virtual Subjects Harboring CYP2C9*1/*1 or CYP2C9*3/3.

Plasma/liver	Genotype	Celecoxib		Diclofenac
		Cmax, µg/mL or µg/g	AUC168, µg h/mL or µg h/g	Cmax, µg/mL or µg/g	AUC168, µg h/mL or µg h/g
Plasma	CYP2C9*1/*1	0.456 (1.0)	50.1 (1.0)	0.119 (1.0)	14.0 (1.0)
	CYP2C9*3/*3
	with in vitro fm	1.85 (4.1)	237 (4.7)	0.884 (7.4)	126 (9.0)
	with in vivo fm	Not available	Not available	0.184 (1.5)	23.8 (1.7)
Liver	CYP2C9*1/*1	8.06 (18)	338 (6.7)	0.862 (7.2)	93.0 (6.6)
	CYP2C9*3/*3
	with in vitro fm	18.7 (41)	1620 (32)	5.91 (50)	838 (60)
	with in vivo fm	Not available	Not available	1.28 (11)	158 (11)

The modified in vivo hepatic intrinsic clearance values were used (Table 2). C_max, maximum concentration; AUC₁₆₈, area under the concentration curve from 0 to 168 h. The numbers in parentheses indicate the percentages of plasma C_max or AUC₁₆₈ values with respect to subjects with CYP2C9*1/*1.

Docking Simulation of Celecoxib and Diclofenac into CYP2C9.1 and CYP2C9.3

Docking simulations were performed using the reported crystal structures of the CYP2C9.1 and CYP2C9.3 hemes¹²⁾ with probe substrates. The results were ranked according to the ligand interaction energies (U values, kcal/mol) using the ASEDock program in the MOE software package (ver. 2022.02; Chemical Computing Group, Montreal, Canada).⁹⁾ Lower U values indicate more effective interactions of the substrates with the P450s.

RESULTS

Adverse Events with Celecoxib and Diclofenac Prescribed Alone

Only cases in which celecoxib (n = 788) or diclofenac (n = 444) was prescribed alone were analyzed in this study (Figs. 1A, B). As recorded in the JADER database, the drug dosage was reduced or discontinued because of adverse events (other than allergies) in 150 and 56 patients receiving celecoxib or diclofenac, respectively; the breakdown of the types of adverse events is summarized in Table 1. Limited but adequate data were available to calculate the time to the onset of adverse events, as shown in Table 1. The median number of days to the onset of adverse events was 18, 8, and 1 for 100-, 200-, and 400-mg daily doses of celecoxib, respectively (Fig. 1C). In the present cohort using celecoxib, there were no differences in the time to onset between the 100-mg and 200-mg doses of celecoxib (p = 0.8). The median number of days (with interquartile ranges) to the onset of adverse events in patients prescribed diclofenac was 2 (0–10) days; because the data available regarding the dosage were limited, the time to onset of adverse events (n = 48) for diclofenac were combined for doses of 25–75 mg and for cases in which the dosage was unknown (Fig. 1D).

Fig. 1. The Case Selection Process (A and B) and Time to Onset of Selected Events in Patients Receiving No Other Prescribed Drugs (C and D) Associated with Celecoxib and Diclofenac for Data Taken from the Japanese Adverse Drug Event Report Database

(A, B) Patient numbers (n) are shown. (C, D) The number of events (n) are indicated.

Virtual Plasma and Hepatic Concentrations of Celecoxib

Among the various factors underlying the mechanisms (other than allergies) related to these adverse events associated with the prescription of celecoxib or diclofenac alone, variations in the in vivo intrinsic clearance of the drug, mediated mainly by P450 2C9, in these individuals may be a causal factor. The reported plasma concentrations of celecoxib versus time after a single oral dose of 200 mg in Korean¹³⁾ and Caucasian¹⁴⁾ subjects harboring CYP2C9*1 or CYP2C9*3 are shown in Figs. 2A and B. The lines in these figures were generated using the current PBPK models with the input parameters shown in Tables 2 and 3 for a virtual oral dose of 200 mg celecoxib. The directly calculated in vivo hepatic intrinsic clearance value of 201 L/h for Korean subjects was similar to the in vivo intrinsic clearance value of 254 L/h estimated using a combination of the reported in vitro fraction metabolized by CYP2C9 (0.9) and the ratio of the in vitro intrinsic clearance of recombinant CYP2C9.3 to that of CYP2C9.1 (0.093). The estimated virtual time-dependent plasma concentrations of celecoxib in this study were similar to those reported for two- or one-subject data (Fig. 2B) and increased after a virtual dose of 200 mg in CYP2C9*3 homozygotes (Fig. 2C) compared with CYP2C9*1 homozygotes.

Fig. 2. Reported and Estimated Human Plasma and Liver Concentrations of Celecoxib

(A, B) Mean reported plasma concentrations after oral administration of 200 mg celecoxib (circles,¹³⁾ and triangles,¹⁴⁾ are shown with standard deviations (bars) for (A) subjects homozygous for CYP2C9*1 (n = 26 and 13) and (B) subjects homozygous for CYP2C9*3 (n = 2 and 1). The lines represent the PBPK model results. (C) Estimated virtual plasma concentrations of celecoxib after a virtual dose of 200 mg in CYP2C9*3 homozygotes in a combination of the in vitro fraction metabolized by polymorphic P450 2C9 and the reported impairment ratio of in vitro CL_h,int values. (D, E) Lines show the PBPK model results for plasma (solid lines) and liver (dotted lines) concentrations of celecoxib in subjects harboring CYP2C9*1/*1 (in black) or CYP2C9*3/*3 (in red) after 200 mg twice daily for 7 d.

The time-dependent plasma and hepatic concentrations of celecoxib after multiple daily doses of 200 mg (the standard Japanese recommended dose) twice a day for 7 d were estimated for sets of wild and mutant genotypes, as shown in Figs. 2D and E. The maximum plasma and hepatic concentration (C_max) values generated by the PBPK models were 4.1- and 2.3-fold (1.85 µg/mL and 18.7 µg/g) higher, respectively, in subjects homozygous for CYP2C9*3 (Table 4). The areas under the plasma and hepatic concentration curves (AUCs) up to 168 h also increased to 237 µg h/mL and 1620 µg h/g, respectively, in subjects homozygous for CYP2C9*3 during 7 d of the above regimen. These high hepatic and plasma celecoxib exposures in subjects with CYP2C9*3 (Fig. 2E) modeled after a week of virtual administration (Table 4) indicated that genetic mutations resulting in impaired CYP2C9 activity could be a causal factor associated with adverse events, as shown in Fig. 1C. The effects of genetic polymorphisms such as CYP2C9*3 might result in modification of the catalytic function similar to that occurring in the co-administration drugs that result in drug–drug interactions.

We recently demonstrated that the ratios of the impaired metabolic capacities of CYP3A4.16 to CYP3A4.1 and the ratios of ligand-interaction energies (U values) were significantly correlated.⁹⁾ To assess the metabolic capacity of the reported CYP2C9.1 and CYP2C9.3 structures, docking simulations were performed for celecoxib. As shown in Fig. 3, celecoxib fits well in the active site of CYP2C9.1; the lowest ligand-interaction energy (U value) of this interaction was −45.1 kcal/mol for possible oxidation reactions (Fig. 3A). The U value of the interaction of celecoxib with CYP2C9.3 was −38.8 kcal/mol (Fig. 3B). These results imply that docking simulations of P450 2C9 substrates into variant P450 2C9 structures may potentially be predictive of changes in metabolic capacity resulting from genetic polymorphisms.

Fig. 3. Docking Simulations of Celecoxib into the Reported Structures of CYP2C9.1 (A) and CYP2C9.3 (B)

Ligand interaction energies (U) are given in kcal/mol.

Virtual Plasma and Hepatic Concentrations of Diclofenac

The in vitro fraction of diclofenac metabolized to 4′-hydroxydiclofenac by liver microsomal P450 2C9 was 1.0.¹⁰⁾ However, the in vivo fractions of diclofenac metabolized to 4′-hydroxydiclofenac by P450 2C9 in a humanized-liver mouse model (either pretreated with the time-dependent P450 2C9 inactivator tienilic acid or left untreated¹⁵⁾ and by P450 2C in marmosets genotyped for wild-type and mutant P450 2C19¹⁶⁾ was approximately 0.5. The average value of 0.54 was based on 0.62 (from C_max) and 0.47 (from AUC) in the humanized-liver mouse model and 0.54 (from AUC) in marmosets. CYP2C9*3/*3 had minimal effects on the pharmacokinetics of diclofenac in one¹⁷⁾ and three¹⁸⁾ subjects compared with six and three subjects harboring CYP2C9*1/*1, respectively. The U values for the interaction of CYP2C9.1 and CYP2C9.3 with diclofenac were similar (approximately −40 kcal/mol, results not shown), suggesting that there are only slight differences in the oxidation reactions. In this study, the virtual plasma and liver concentrations of diclofenac were generated for two parameter values: one based on the in vitro fraction and one based on the in vivo fraction of diclofenac metabolized to 4′-hydroxydiclofenac by P450 2C9.

By applying a similar approach to that used for celecoxib (Fig. 2A), we set up a PBPK model based on the reported plasma concentrations of diclofenac in subjects harboring CYP2C9*1¹⁹⁾ (Fig. 4A). The in vivo hepatic intrinsic clearance value for subjects with CYP2C9*3/*3 was calculated once using the in vitro fraction metabolized by CYP2C9 (1.0)¹⁰⁾ and once using in vivo fraction metabolized by CYP2C9 (0.5).¹⁵⁾ The ratio of the in vitro intrinsic clearance of recombinant CYP2C9.3 to CYP2C9.1 (0.10) was also used in the model for subjects with CYP2C9*3/*3²⁰⁾ (Table 3). The estimated virtual plasma concentrations of diclofenac increased after a virtual dose of 100 mg in CYP2C9*3 homozygotes (Fig. 4B) compared to CYP2C9*1 homozygotes for both scenarios. The modeled time-dependent concentrations of diclofenac for virtual administrations of three 25-mg doses per day for 7 d are shown in Figs. 4C and D. The plasma and hepatic C_max values generated by PBPK models were 0.119 µg/mL and 0.862 µg/g, respectively, in subjects harboring CYP2C9*1/*1 (Table 4), whereas the plasma and hepatic C_max values increased to 0.184 µg/mL and 1.28 µg/g, respectively, in subjects homozygous for CYP2C9*3, even when using the conservative in vivo fraction of diclofenac metabolized to 4′-hydroxydiclofenac by polymorphic P450 2C9.

Fig. 4. Reported and Estimated Human Plasma and Liver Concentrations of Diclofenac

(A) Mean reported plasma concentrations after an oral 100-mg dose of diclofenac in CYP2C9*1 homozygotes (n = 6)¹⁹⁾ are shown. The line represents the PBPK model results. (B) Modeled plasma concentrations of diclofenac after a virtual dose of 100 mg in CYP2C9*3 homozygotes based on the in vitro f_m (black) and the in vivo f_m (red). (C, D) Lines show the PBPK model results for plasma (solid lines) and liver (dotted lines) concentrations of diclofenac after 25 mg three times daily for 7 d in subjects harboring CYP2C9*1/*1 (C) or CYP2C9*3/*3 (D) based on the in vitro f_m (black) and the in vivo f_m (red).

DISCUSSION

The in vivo and in vitro fractions of the victim drug metabolized by a specific polymorphic P450 form are important determining factors for estimating impaired P450 variant activity as a causal intrinsic clearance factor (Tables 3, 4), in a manner similar to drug–drug interactions caused by the co-administration of P450-specific substrates/inhibitors. Virtual hepatic and plasma exposures estimated by pharmacokinetic modeling in patients harboring the impaired CYP2C9*3 allele could represent a causal factor for adverse events induced by celecoxib or diclofenac in a manner similar to that for drug interactions. We succeeded in estimating the plasma concentrations of celecoxib by modifying the in vivo CL_h,int input parameter value in the simplified PBPK model using a combination of the in vitro fraction metabolized by polymorphic P450 2C9 and the reported impairment ratio of in vitro CL_h,int values of the mutant P450 to the wild type. Because actual expression levels of CYP2C9.1 and CYP2C9.3 in individual patients harboring the heterozygous CYP2C9*1/*3 could not be determined, the virtual hepatic or plasma exposures caused by CYP2C9.1 and CYP2C9.3 in combination were not estimated under the present conditions. In terms of the similarity of in vivo and in vitro fractions of drug metabolized by polymorphic P450 2C9, the in vivo and in vitro fractions of S-warfarin metabolized by P450 2C9 were both 0.8 (0.75 in suppressed human liver microsomes¹⁰⁾ and an average of 0.81 based on 0.78 for the C_max and 0.84 for the AUC values for the humanized-liver mouse model.¹⁵⁾

For diclofenac, the in vivo and in vitro fractions metabolized by P450 2C9 were different, resulting in minimal and extensive effects, respectively, of CYP2C9*3 on plasma pharmacokinetics (Fig. 4B). Even based on the conservative (in vivo) estimation of the effects of CYP2C9*3, virtual hepatic exposure to diclofenac in subjects harboring CYP2C9*3 would be elevated compared with the modeled plasma concentrations (Table 4). The relationship between drug exposure and adverse events in actual subjects and patients will be investigated in future studies of typical CYP2C9 substrates.

In conclusion, the virtual hepatic and plasma exposures estimated by PBPK modeling for patients harboring the CYP2C9*3 allele may indicate impaired CYP2C9 activity as a causal factor for adverse events in a manner similar to drug interactions caused by the co-administration of P450 2C9-specific substrates/inhibitors. The in vivo and in vitro fractions of the victim drug metabolized by a specific polymorphic P450 form is an important determinant factor for these estimations in PBPK modeling. The ratios of ligand-interaction energies in in silico docking simulation would be a simple marker for impaired metabolic capacities of CYP2C9.3 to CYP2C9.1 or CYP3A4.16 to CYP3A4.1 in vivo. Accumulation of pharmacogenomic information on these CYP2C9-dependent drugs in clinical practice in Japan is needed and recommended.

Acknowledgments

The authors would like to thank Haruka Nishimura, Masayoshi Utsumi, and Tomonori Miura for their assistance. This work was supported partly by the Japan Agency for Medical Research and Development (AMED) under Grant No. 23mk0101253h0102. We are also grateful to David Smallbones for copyediting a draft of this article.

Conflict of Interest

The authors declare no conflict of interest.

REFERENCES

• 1) Sangkuhl K, Claudio-Campos K, Cavallari LH, Agundez JAG, Whirl-Carrillo M, Duconge J, Del Tredici AL, Wadelius M, Rodrigues Botton M, Woodahl EL, Scott SA, Klein TE, Pratt VM, Daly AK, Gaedigk A. PharmVar GeneFocus: CYP2C9. Clin. Pharmacol. Ther., 110, 662–676 (2021).
• 2) Fleming J, Chetty M. Therapeutic monitoring of valproate in psychiatry: how far have we progressed? Clin. Neuropharmacol., 29, 350–360 (2006).
• 3) Peterson GM, Khoo BH, von Witt RJ. Clinical response in epilepsy in relation to total and free serum levels of phenytoin. Ther. Drug Monit., 13, 415–419 (1991).
• 4) Miyagata Y, Nakai K, Sugiyama Y. Clinical significance of combined CYP2C9 and VKORC1 genotypes in Japanese patients requiring warfarin. Int. Heart J., 52, 44–49 (2011).
• 5) Kato Y, Ichida F, Saito K, Watanabe K, Hirono K, Miyawaki T, Yoshimura N, Horiuchi I, Taguchi M, Hashimoto Y. Effect of the VKORC1 genotype on warfarin dose requirements in Japanese pediatric patients. Drug Metab. Pharmacokinet., 26, 295–299 (2011).
• 6) Huth F, Gardin A, Umehara K, He H. Prediction of the impact of cytochrome P450 2C9 genotypes on the drug-drug interaction potential of siponimod with physiologically-based pharmacokinetic modeling: a comprehensive approach for drug label recommendations. Clin. Pharmacol. Ther., 106, 1113–1124 (2019).
• 7) Glaenzel U, Jin Y, Nufer R, Li W, Schroer K, Adam-Stitah S, Peter van Marle S, Legangneux E, Borell H, James AD, Meissner A, Camenisch G, Gardin A. Metabolism and disposition of siponimod, a novel selective S1P1/S1P5 agonist, in healthy volunteers and in vitro identification of human cytochrome P450 enzymes involved in its oxidative metabolism. Drug Metab. Dispos., 46, 1001–1013 (2018).
• 8) Tsuchiya M, Obara T, Sakai T, Nomura K, Takamura C, Mano N. Quality evaluation of the Japanese Adverse Drug Event Report database (JADER). Pharmacoepidemiol. Drug Saf., 29, 173–181 (2020).
• 9) Adachi K, Ohyama K, Tanaka Y, Sato T, Murayama N, Shimizu M, Saito Y, Yamazaki H. High hepatic and plasma exposures of atorvastatin in subjects harboring impaired cytochrome P450 3A4 *16 modeled after virtual administrations and possibly associated with statin intolerance found in the Japanese adverse drug event report database. Drug Metab. Pharmacokinet., 49, 100486 (2023).
• 10) Murayama N, Yajima K, Hikawa M, Shimura K, Ishii Y, Takada M, Uno Y, Utoh M, Iwasaki K, Yamazaki H. Assessment of multiple cytochrome P450 activities in metabolically inactivated human liver microsomes and roles of P450 2C isoforms in reaction phenotyping studies. Biopharm. Drug Dispos., 39, 116–121 (2018).
• 11) Adachi K, Shimizu M, Yamazaki H. Updated in silico prediction methods for fractions absorbed and key input parameters of 355 disparate chemicals for physiologically based pharmacokinetic models for time-dependent plasma concentrations after virtual oral doses in humans. Biol. Pharm. Bull., 45, 1812–1817 (2022).
• 12) Maekawa K, Adachi M, Matsuzawa Y, Zhang Q, Kuroki R, Saito Y, Shah MB. Structural Basis of Single-Nucleotide Polymorphisms in Cytochrome P450 2C9. Biochemistry, 56, 5476–5480 (2017).
• 13) Kim SH, Kim DH, Byeon JY, Kim YH, Kim DH, Lim HJ, Lee CM, Whang SS, Choi CI, Bae JW, Lee YJ, Jang CG, Lee SY. Effects of CYP2C9 genetic polymorphisms on the pharmacokinetics of celecoxib and its carboxylic acid metabolite. Arch. Pharm. Res., 40, 382–390 (2017).
• 14) Prieto-Pérez R, Ochoa D, Cabaleiro T, Román M, Sánchez-Rojas SD, Talegón M, Abad-Santos F. Evaluation of the relationship between polymorphisms in CYP2C8 and CYP2C9 and the pharmacokinetics of celecoxib. J. Clin. Pharmacol., 53, 1261–1267 (2013).
• 15) Miura T, Uehara S, Shimizu M, Murayama N, Utoh M, Suemizu H, Yamazaki H. Different roles of human cytochrome P450 2C9 and 3A enzymes in diclofenac 4′- and 5-hydroxylations mediated by metabolically inactivated human hepatocytes in previously transplanted chimeric mice. Chem. Res. Toxicol., 33, 634–639 (2020).
• 16) Nakanishi K, Uehara S, Kusama T, Inoue T, Shimura K, Kamiya Y, Murayama N, Shimizu M, Uno Y, Sasaki E, Yamazaki H. In vivo and in vitro diclofenac 5-hydroxylation mediated primarily by cytochrome P450 3A enzymes in common marmoset livers genotyped for P450 2C19 variants. Biochem. Pharmacol., 152, 272–278 (2018).
• 17) Yasar U, Eliasson E, Forslund-Bergengren C, Tybring G, Gadd M, Sjoqvist F, Dahl ML. The role of CYP2C9 genotype in the metabolism of diclofenac in vivo and in vitro. Eur. J. Clin. Pharmacol., 57, 729–735 (2001).
• 18) Kirchheiner J, Meineke I, Steinbach N, Meisel C, Roots I, Brockmoller J. Pharmacokinetics of diclofenac and inhibition of cyclooxygenases 1 and 2: no relationship to the CYP2C9 genetic polymorphism in humans. Br. J. Clin. Pharmacol., 55, 51–61 (2003).
• 19) Dammann HG, Simon-Schultz J, Steinhoff I, Damaschke A, Schmoldt A, Sallowsky E. Differential effects of misoprostol and ranitidine on the pharmacokinetics of diclofenac and gastrointestinal symptoms. Br. J. Clin. Pharmacol., 36, 345–349 (1993).
• 20) Yamazaki H, Inoue K, Chiba K, Ozawa N, Kawai T, Suzuki Y, Goldstein JA, Guengerich FP, Shimada T. Comparative studies on the catalytic roles of cytochrome P450 2C9 and its Cys- and Leu-variants in the oxidation of warfarin, flurbiprofen, and diclofenac by human liver microsomes. Biochem. Pharmacol., 56, 243–251 (1998).
• 21) Kato M, Shitara Y, Sato H, Yoshisue K, Hirano M, Ikeda T, Sugiyama Y. The quantitative prediction of CYP-mediated drug interaction by physiologically based pharmacokinetic modeling. Pharm. Res., 25, 1891–1901 (2008).
• 22) Uchimura T, Kato M, Saito T, Kinoshita H. Prediction of human blood-to-plasma drug concentration ratio. Biopharm. Drug Dispos., 31, 286–297 (2010).
• 23) Takano R, Murayama N, Horiuchi K, Kitajima M, Kumamoto M, Shono F, Yamazaki H. Blood concentrations of acrylonitrile in humans after oral administration extrapolated from in vivo rat pharmacokinetics, in vitro human metabolism, and physiologically based pharmacokinetic modeling. Regul. Toxicol. Pharmacol., 58, 252–258 (2010).
• 24) Davies NM, McLachlan AJ, Day RO, Williams KM. Clinical pharmacokinetics and pharmacodynamics of celecoxib: a selective cyclo-oxygenase-2 inhibitor. Clin. Pharmacokinet., 38, 225–242 (2000).
• 25) Kamiya Y, Handa K, Miura T, Ohori J, Kato A, Shimizu M, Kitajima M, Yamazaki H. Machine learning prediction of the three main input parameters of a simplified physiologically based pharmacokinetic model subsequently used to generate time-dependent plasma concentration data in humans after oral doses of 212 disparate chemicals. Biol. Pharm. Bull., 45, 124–128 (2022).
• 26) Rodrigues AD. Impact of CYP2C9 genotype on pharmacokinetics: are all cyclooxygenase inhibitors the same? Drug Metab. Dispos., 33, 1567–1575 (2005).
• 27) Tang C, Shou M, Rushmore TH, Mei Q, Sandhu P, Woolf EJ, Rose MJ, Gelmann A, Greenberg HE, De Lepeleire I, Van Hecken A, De Schepper PJ, Ebel DL, Schwartz JI, Rodrigues AD. In-vitro metabolism of celecoxib, a cyclooxygenase-2 inhibitor, by allelic variant forms of human liver microsomal cytochrome P450 2C9: correlation with CYP2C9 genotype and in-vivo pharmacokinetics. Pharmacogenetics, 11, 223–235 (2001).