Cardiovascular diseases (CVDs) are the leading cause of mortality and morbidity worldwide, necessitating effective and comprehensive therapeutic strategies to manage these complex conditions. Often accompanied by comorbidities such as hyperlipidemia, diabetes, and hypertension, CVDs typically require a multi-therapy, or polypharmacy, approach (Marengoni and Onder 2015; Sheikh-Taha and Asmar 2021). This strategy involves the concurrent use of multiple medications, targeting various aspects of the disease process, potentially enhancing therapeutic outcomes. However, it also significantly increases the potential for drug-drug interactions (DDIs), which may lead to reduced drug effectiveness or increased risk of adverse events (Liu et al. 2017).

Recent advancements in techniques for evaluating in vitro disposition pathways, along with tools for quantitatively predicting pharmacokinetic (PK) changes resulting from in vivo DDIs, have become crucial in the drug development process. Physiologically-based pharmacokinetic (PBPK) modeling, in particular, serves as an invaluable tool for quantitatively predicting the in vivo potential of DDIs (Sager et al. 2015). By integrating comprehensive drug characteristics with human physiology, PBPK modeling can simulate a wide range of clinical scenarios (Hanke et al. 2018).

Atorvastatin, a reversibly inhibitor of 3-hydroxy-3- methlyglutaryl-coenzyme A (HMG-CoA) reductase, is a first-line agent for the treatment of hypercholesterolemia (Li et al. 2019). It is extensively metabolized by CYP3A4, and it is also known as a substrate of organic anion transporting polypeptide 1B1 (OATP1B1) (Duan et al. 2017). Amlodipine is calcium channel blockers (CCBs) used for the treatment of hypertension. It is primarily metabolized by CYP3A4, and according to Norvasc^® label, the exposure of amlodipine may increase to a greater extent with concomitant administration of strong CYP3A4 inhibitors (e.g., ketoconazole, itraconazole, ritonavir) compared to its administration alone. However, data regarding the quantitative impact of CYP3A4 inducers on amlodipine is limited.

The objectives of this study were to: 1) develop PBPK models of two representatives of CVD drugs, atorvastatin and amlodipine, 2) evaluate and verify the predictive performance of the developed models by comparing their predictions with observed PK data from Asian populations, and 3) explore the potential for DDIs of atorvastatin and amlodipine with strong CYP3A4 index perpetrators, including itraconazole, clarithromycin, and rifampicin.

Software

PBPK model development, model modification, parameter optimization, simulation, and local sensitivity analysis were conducted using PK-Sim^® (version 11 Update 2, www.open-systems-pharmacology.org, 2024), and the PK profiles of drugs were digitized using PlotDigitizer (Online App, www.plotdigitizer.com). Figure creations from the simulated data were primarily performed using ‘ggplot2’ and ‘coveffectsplot’ packages in R (version 3.4.0, The R Foundation for Statistical Computing) in RStudio (version 4.1.1, RStudio Inc., Boston, MA, USA).

Virtual populations

The development and simulation of the drug models employed a virtual Asian population, constructed based on the East Asian (Tanaka, 1996) incorporated in PK-Sim^®. Individuals embedded within PK-Sim^®. The virtual Asian population consisted of 100 individuals, randomly selected to fit within specific distribution criteria. The age of these individuals ranged from 20 to 50 years, with weights starting from 50 kg. The body mass index (BMI) was maintained within a range of 20 and 25 kg/m², ensuring a balanced gender distribution with females representing 50% of the population. Physiological parameters and enzyme expression information were sourced from the database provided in PK-Sim^®.

PBPK model development

1) Atorvastatin and amlodipine

PBPK models for atorvastatin and amlodipine were developed, with all input parameters were obtained from published data. Some parameters describing metabolism and elimination were optimized when the initial parameters could not adequately explain the reported observed PK profiles.

2) Itraconazole, clarithromycin, and rifampicin

Since atorvastatin and amlodipine are substrate of CYP3A4, index probe drugs utilized were itraconazole (an inhibitor of CYP3A4), clarithromycin (an inhibitor of CYP3A4), and rifampicin (an inducer of both CYP3A4 and CYP2C19). The PBPK models for these perpetrators were developed by adapting templates from Open-Systems-Pharmacology PBPK library (https://github.com/Open-Systems-Pharmacology/OSP-PBPK-Model-Library) incorporating modifications to better reflect observations in an Asian population.

Simulations

1) Drugs as a single agent for model verification

To verify the adequacy of each PBPK model, simulations were conducted to generate plasma concentration – time profiles of atorvastatin and amlodipine. These simulated PK profiles were then compared with observed PK profiles for each compound, as sourced from literature.

The simulations were conducted following the dosing regimen from which the observed data were derived. Atorvastatin, which is a substrate for CYP3A4, follows a dosing regimen of a once daily (QD) administration of 40 mg for 7 days or 14 days, and a single oral administration of 80 mg in tablet form. The dosing regimen of amlodipine includes a single oral administration of 5 mg or 10 mg.

Itraconazole, a strong index inhibitor of CYP3A4, is administered 200 mg QD. Clarithromycin, an index inhibitor of CYP3A4, is also administered 500 mg QD. Rifampicin, a strong inducer of CYP3A4 and CYP2C19 enzymes, is administered orally 600 mg QD.

2) DDI simulations

The simulations were conducted to generate concentration - time profiles of a substrate with or without a perpetrator, and the DDI potentials were assessed by comparing the changes of area under the curve (AUC) and maximum plasma concentration (C_max) of the substrate when administered with the drug alone or in combination with the perpetrator. Perpetrators were daily administered for 7 days to reach steady-state and substrates were administered once at day 7. The doses of amlodipine, atorvastatin, itraconazole, rifampicin, and clarithromycin were 10 mg, 40 mg, 200 mg, 600 mg, and 500 mg, respectively.

Model evaluation

The developed model performance of atorvastatin and amlodipine were evaluated by comparing simulated data with observed data. The PK profiles of drugs were simulated using the same dosing regimens as the clinical data referenced. The arithmetic mean ratio (AMR, predicted/observed) of AUC and C_max values were compared, and in cases where the observed data reported geometric means, the geometric mean ratios (GMRs) were compared instead. Ratios within a 2-fold range were considered indicative of adequate model performance.

To evaluate the DDI potentials of atorvastatin and amlodipine with strong index perpetrators, PK profiles of atorvastatin and amlodipine following single administration were simulated (n = 100) both in the presence and absence of each perpetrator. The GMRs of AUC_inf and C_max for atorvastatin and amlodipine, with or without each perpetrator, were then calculated.

Local sensitivity analysis was also performed on each model calculating sensitivity coefficient (S) as the following equation (1):

$$S=\frac{\Delta AUC}{AUC}\cdot \frac{p}{\Delta p}$$(1)

where ΔAUC is the change of the AUC, AUC is the simulated AUC with the original parameter value, Δp is the change of the examined model parameter value, and p is the original model parameter value. A sensitivity value of +1.0 signifies that a 10% increase in the examined parameter causes a 10% increase in the simulated AUC (Hanke et al. 2018).

Virtual populations

The details of height, age, weight, BMI, and BSA of a virtual Asian population used for the simulations are listed in Table 1. The average body weights of the virtual Asian population were 58.3 kg overall, 62.2 kg for males, and 54.4 kg for females.

Table 1 . Demographics characteristics of virtual Asian population.

Variables	Total (n = 100)	Male (n = 50)	Female (n = 50)
Body weight (kg)	58.3 (50.0-78.4)	62.2 (53.3-78.4)	54.4 (50.0-64.7)
Age (year)	35.3 (20.0-49.8)	36.2 (20.0-49.5)	34.4 (20.4-49.8)
BMI (kg/m2)	22.3 (20.0-25.0)	21.7 (20.0-24.7)	22.9 (20.4-25.0)
BSA (m2)	1.62 (1.44-1.97)	1.71 (1.54-1.97)	1.53 (1.44-1.70)

BMI, body mass index; BSA, body surface area. Data were expressed as mean (range)..

PBPK model development

The PBPK models of atorvastatin and amlodipine were developed and the model input parameters including their physicochemical properties, absorption, distribution, and elimination were summarized in Table 2.

Table 2 . The input parameters of atorvastatin and amlodipine PBPK models.

Parameters	Units	Atorvastatin	Amlodipine
Physicochemical properties
Molecular weight	g/mol	558.64	408.88
Lipophilicity		4.06a	3.43e
Fraction unbound		0.2a	0.07e
pKa		4.31 (acid)b	9.40 (base), 1.90 (base)e
Solubility	mg/mL	0.0212 (pH 2.1)c 0.0321 (pH 3.1) 0.0796 (pH 4.1) 0.127 (pH 5.0) 0.227 (pH 5.4) 1.220 (pH 6.0)	4.23 (pH 4)e
Absorption
Specific intestinal permeability	10-6 cm/sec	7.90d	19.5f
Distribution
Partition coefficient		Schmitt	PK-Sim standard
Cellular permeabilities		PK-Sim standard	PK-Sim standard
Metabolism
CYP3A4
Specific CL	L/µmol/min	8.00g	1.24g
Transport & excretion
OATP1B1
Km	μM	0.77d
Vmax	μM/min	100.00e
Renal CL
GFR			1.00 (assumed)
Interactions
CYP3A4
Ki	μM		20.0e
BCRP
Ki	μM		78.2e

^aCorsini et al. 1999, ^bDrugbank 2005, ^cMorse et al. 2019, ^dDuan et al. 2017, ^eRhee et al. 2018, ^fKadono et al. 2010, ^gParameters were optimized based on the observation data..

BCRP, breast cancer resistance protein; CL, clearance; CYP, cytochrome P450; GFR, glomerular filtration rate; K_i, inhibition constant; OATP, organic-anion-transporting polypeptide; pK_a, acid dissociation constant..

1) Atorvastatin

Most of the physicochemical properties for atorvastatin were obtained from the literatures (Corsini et al. 1999; Morse et al. 2019; Drugbank 2005). The values of intrinsic clearance were optimized by fitting them to observed data. Specific CL by CYP3A4 was finally determined as 8.00 L/μmol/min. Since atorvastatin is also known as a substrate of OATP1B1, kinetic parameters including K_m and V_max were applied in the model. K_m as 0.77 μM was from Duan’s paper (Duan et al. 2017) and V_max as 100 μM/min was fitted to observed PK data. Since atorvastatin was not considered perpetrators of metabolizing enzymes, neither inhibition nor induction kinetic parameters were included in the model.

2) Amlodipine

Given that amlodipine is extensively metabolized by CYP3A4, the CL_int for CYP3A4 in human liver microsome (HLM) data for hepatic metabolism of amlodipine (Rhee et al. 2018) and total hepatic CL for additional hepatic metabolism (Mukherjee et al. 2018) was firstly incorporated to account for the elimination of amlodipine. However, the specific fractions of amlodipine eliminated by CYP3A4 metabolism and other pathways have not been reported, these proportions were optimized and determined by simulating the PK changes, such as AUC ratio (AUCR), during interactions with various perpetrators. The appropriate elimination fractions were then established by comparing these simulation results to published values. In the final amlodipine model, specific CL by CYP3A4 was 1.24 L/μmol/min, GFR was assumed as 1 for renal CL, and additional hepatic CL were not considered. The value of specific intestinal permeability was applied from Kadono’s paper (Kadono et al. 2010). The kinetic parameters related to the inhibition of CYP3A4 and BCRP by amlodipine were also integrated into the model.

Simulations

1) Drugs as a single agent

The PK profiles of atorvastatin and amlodipine were generated and compared with observed data (Park et al. 2004; Kim et al. 2010; Kim et al. 2013; Birmingham et al. 2015; Park et al. 2017; Woo et al. 2017; Wang et al. 2020; Kim et al. 2022a; Kim et al. 2022b; Tang et al. 2024). The simulations employed the same dosing regimens and trial designs as those used in the selected observations. The good model performance is shown in Fig. 1. The predicted profiles were overlaid to the observations for comparisons. The predicted AUC and C_max were also compared with the observations, and these results are listed in Table 3. A local sensitivity analysis for each model was conducted and illustrated in Fig. 2. Lipophilicity for atorvastatin, and liver volume for amlodipine are the parameters most sensitive to changes in the AUC_inf.

Figure 1. The predicted and observed plasma concentration ‒ time profiles of (a), (b), (c) atorvastatin and (d), (e) amlodipine model. Simulations are shown in lines, and observed data are shown in dots, or dots ± standard deviation, if available. Shaded area are 90 percentile prediction intervals of simulated profile.

Figure 2. Local sensitivity analysis of (a) atorvastatin and (b) amlodipine model.

Table 3 . Comparison of PK parameters between observed and predicted mean of each substrate.

Study	Dosing regimen	Race	N	Parameters	Observed mean (a)	Predicted mean (b)	Ratio (b/a)
Atorvastatin
Birmingham et al. 2015	Single, 40 mg	Chinese	32	Cmax (ng/mL)	21.8a	30.8a	1.41a
				AUC24 (ng·h/mL)	111a	121a	1.09a
		Japanese	31	Cmax (ng/mL)	23.3a	30.8a	1.32a
				AUC24 (ng·h/mL)	123a	121a	0.98a
Park et al. 2017	Once-daily, 40 mg	Koreans	33	Cmax,ss (ng/mL)	27.5	31.9	1.16
				AUCtau,ss (ng·h/mL)	90.4	133	1.47
Woo et al. 2017	Single, 80 mg	Koreans	50	Cmax (ng/mL)	36.2	63.6	1.76
				AUCinf (ng·h/mL)	172	265	1.54
Kim et al. 2022a	Once-daily, 40 mg	Koreans	18	Cmax,ss (ng/mL)	34.7	31.9	0.92
				AUCtau,ss (ng·h/mL)	171	133	0.78
Kim et al. 2022b	Single, 80 mg	Koreans	35	Cmax (ng/mL)	84.3	63.6	0.75
				AUCinf (ng·h/mL)	279	265	0.95
Amlodipine
Park et al. 2004	Single, 5 mg (test)	Koreans	18	Cmax (ng/mL)	3.70	2.69	0.73
				AUCinf (ng·h/mL)	188	140	0.75
	Single, 5 mg (reference)		18	Cmax (ng/mL)	3.60	2.69	0.75
				AUCinf (ng·h/mL)	170	140	0.83
Kim et al. 2010	Single, 5 mg (test)	Koreans	24	Cmax (ng/mL)	2.60	2.69	1.03
				AUCinf (ng·h/mL)	159	140	0.88
	Single, 5 mg (reference)		24	Cmax (ng/mL)	2.74	2.69	0.98
				AUCinf (ng·h/mL)	159	140	0.88
Kim et al. 2013	Single, 5 mg (test)	Koreans	20	Cmax (ng/mL)	3.54	2.69	0.76
				AUCinf (ng·h/mL)	149	140	0.94
	Single, 5 mg (reference)		20	Cmax (ng/mL)	3.28	2.69	0.82
				AUCinf (ng·h/mL)	142	140	0.99
Wang et al. 2020	Single, 5 mg (test)	Chinese	22	Cmax (ng/mL)	3.79	2.69	0.71
			20	AUCinf (ng·h/mL)	167	140	0.84
	Single, 5 mg (reference)		22	Cmax (ng/mL)	3.87	2.69	0.69
			20	AUCinf (ng·h/mL)	167	140	0.84
Han et al. 2021 (cohort 11)	Single, 5 mg	Chinese	24	Cmax (ng/mL)	4.37	2.69	0.62
				AUCinf (ng·h/mL)	193	140	0.73
Han et al. 2021 (cohort 12)	Single, 5 mg	Chinese	24	Cmax (ng/mL)	2.23	2.69	1.21
				AUCinf (ng·h/mL)	103.2	140	1.34
Han et al. 2021 (cohort 1)	Single, 10 mg	Chinese	16	Cmax (ng/mL)	7.30	5.38	0.74
				AUCinf (ng·h/mL)	250	277	1.11
Han et al. 2021 (cohort 2)	Single, 10 mg	Chinese	20	Cmax (ng/mL)	6.13	5.38	0.88
				AUCinf (ng·h/mL)	306	277	0.90
Han et al. 2021 (cohort 3)	Single, 10 mg	Chinese	20	Cmax (ng/mL)	8.00	5.38	0.67
				AUCinf (ng·h/mL)	283	277	0.98
Han et al. 2021 (cohort 4)	Single, 10 mg	Chinese	22	Cmax (ng/mL)	5.34	5.38	1.01
				AUCinf (ng·h/mL)	295	277	0.94
Han et al. 2021 (cohort 5)	Single, 10 mg	Chinese	18	Cmax (ng/mL)	5.55	5.38	0.97
				AUCinf (ng·h/mL)	240	277	1.15
Han et al. 2021 (cohort 6)	Single, 10 mg	Chinese	18	Cmax (ng/mL)	7.60	5.38	0.71
				AUCinf (ng·h/mL)	275	277	1.01
Han et al. 2021 (cohort 7)	Single, 10 mg	Chinese	20	Cmax (ng/mL)	9.57	5.38	0.56
				AUCinf (ng·h/mL)	372	277	0.75
Han et al. 2021 (cohort 8)	Single, 10 mg	Chinese	18	Cmax (ng/mL)	6.27	5.38	0.86
				AUCinf (ng·h/mL)	209	277	1.32
Han et al. 2021 (cohort 9)	Single, 10 mg	Chinese	18	Cmax (ng/mL)	6.02	5.38	0.89
				AUCinf (ng·h/mL)	288	277	0.96
Han et al. 2021 (cohort 10)	Single, 10 mg	Chinese	18	Cmax (ng/mL)	5.00	5.38	1.08
				AUCinf (ng·h/mL)	243	277	1.14
Tang et al. 2024	Single, 10 mg (test)	Chinese	13	Cmax (ng/mL)	6.71	5.38	0.80
				AUC72 (ng·h/mL)	216	160	0.74
	Single, 10 mg (reference)	Chinese	13	Cmax (ng/mL)	6.93	5.38	0.78
				AUC72 (ng·h/mL)	220	160	0.73

^aGeometric mean..

2) DDI simulations

The DDI simulations of atorvastatin and amlodipine with strong index perpetrators of CYP3A4, itraconazole, clarithromycin, and rifampicin. The perpetrator models were adopted from freely open-source data, modified CL_int to better describe the observations. The interaction parameters to explain DDIs remained the same in the provided model. The predicted plasma concentration–time profiles of atorvastatin and amlodipine with or without perpetrators are shown in Fig. 3 and predicted C_max and AUC_inf, of atorvastatin and amlodipine with or without perpetrators are summarized in Table 4. For atorvastatin, in the presence of itraconazole, the GMR for AUC_inf was 2.64 (90% CI: 2.54-2.75) and for C_max was 1.34 (90% CI: 1.33-1.36). With clarithromycin, the GMR for AUC_inf was 2.37 (90% CI: 2.25-2.50) and for C_max was 1.34 (90% CI: 1.32-1.36). In the presence of rifampicin, the GMR for AUC_inf was 0.43 (90% CI: 0.42-0.44) and for C_max was 0.58 (90% CI: 0.57-0.59). For amlodipine, in the presence of itraconazole, the GMR for AUC_inf was 2.06 (90% CI: 1.96-2.16) and for C_max was 1.58 (90% CI: 1.55-1.61). With clarithromycin, the GMR for AUC_inf was 1.69 (90% CI: 1.65-1.74) and for C_max was 1.49 (90% CI: 1.49-1.52). In the presence of rifampicin, the GMR for AUC_inf was 0.27 (90% CI: 0.265-0.272) and for C_max was 0.38 (90% CI: 0.376-0.383). Forest plots for the changes of exposure by perpetrators are shown in Fig. 4.

Figure 3. The predicted plasma concentration ‒ time profiles of (a) atorvastatin 40 mg and (b) amlodipine 10 mg with or without clarithromycin 500 mg, itraconazole 200 mg, and rifampicin 600 mg.

Figure 4. Forest plots showing the comparison of predicted geometric mean ratio (90% CI) of C_max and AUC_inf of atorvastatin and amlodipine with or without clarithromycin, itraconazole, and rifampicin.

Table 4 . Predicted exposures (arithmetic mean ± SD) of atorvastatin and amlodipine after administration of each drug alone or combination with CYP3A4 inhibitor/inducer.

Substrates	PK parameters	Alone	With itraconazole	With clarithromycin	With rifampicin
Atorvastatin (40 mg)	AUCinf (ng·h/mL)	131.05 ± 38.16	366.55 ± 166.87	358.00 ± 277.60	56.47 ± 14.46
	Cmax (ng/mL)	32.22 ± 7.42	42.98 ± 8.74	42.82 ± 8.92	18.89 ± 5.25
Amlodipine (10 mg)	AUCinf (ng·h/mL)	283.46 ± 104.36	627.44 ± 362.93	491.20 ± 197.64	74.96 ± 24.11
	Cmax (ng/mL)	5.39 ± 1.11	8.57 ± 1.92	8.10 ± 2.03	2.04 ± 0.42

Atorvastatin and amlodipine are among the most commonly prescribed medications for CVDs. Due to the nature of the disease, these drugs are frequently used long-term and in combination with other medications, significantly increasing the potential for DDIs. In addition, Asians generally have lower body weights and BMI than Caucasians, which can lead to higher systemic drug exposures (Gandelman et al. 2012; Kario et al. 2013). There have been limited data reported to support racial difference exposures of both atorvastatin and amlodipine between Asians and Caucasians (Gandelman et al. 2012), therefore, we created virtual Asian populations from the built-in Asian Individual (Tanaka, 1996) in PK-Sim without further modifications.

In the case of atorvastatin, several papers have been published on the topic of PBPK model development (Zhang 2015; Duan et al. 2017; Li et al. 2019; Morse et al. 2019; Reig-López et al. 2021). In all these papers, Simcyp^® was used for the model development. Zhang (2015) established atorvastatin PBPK model incorporating two metabolites, atorvastatin-lactone and 2-hydroxy-atorvastatin (Zhang 2015). In Zhang’s model, enzyme and transporter kinetic parameters including CYP3A4, CYP2C8, UGT1A1, UGT1A3, P-gp, BCRP, and OATP1B1, as well as passive diffusion clearance (CL_PD) in liver were incorporated. Duan et al. (2017) developed atorvastatin PBPK model focusing exclusively on the parent drug, excluding its metabolite models (Duan et al. 2017). In the model, enzyme and transporter kinetic parameters for CYP3A4, BCRP, and OATP1B1, along with CL_PD in liver were utilized for the disposition of atorvastatin. Additionally, the model facilitated the evaluation of exposure differences attributed to SLCO1B1 polymorphisms. Li et al. (2019) refined the Zhang’s and Duan’s model, focusing on establishing the lactone form of atorvastatin and atorvastatin acyl glucuronide, which are implicated in statin-induced myopathy (Hermann et al. 2006; Li et al. 2019). This model elaborated on the UGT metabolism of atorvastatin and incorporated the contributions of OATP1B3 and bile excretion for the elimination of atorvastatin in liver. Morse et al. (2019) developed comprehensive atorvastatin model including atorvastatin, atorvastatin lactone, hydroxy-atorvastatin, and hydroxy-atorvastatin lactone. In Morse’s model, the conversion of atorvastatin acid to atorvastatin lactone in gastric fluid was illustrated to explain the decreased absorption of atorvastatin under conditions of delayed gastric emptying.

Regarding amlodipine, various studies of PBPK models have been also published (Rhee et al. 2018; Mukherjee et al. 2018; Han et al. 2021; Ryu et al. 2021). Rhee et al. (2018) developed amlodipine, fimasartan, and hydrochlorothiazide using Simcyp^® to predict the potentials of DDI among these drugs. Not only HLM CL_int by CYP3A4 but also additional human intestinal microsome (HIM) CL_int are reflected in the amlodipine model (Rhee et al. 2018). Mukherjee et al. (2018) also developed amlodipine model using Simcyp^® to predict the changes of amlodipine exposure by ritonavir, and PBPK/pharmacodynamic (PBPK/PD) model was finally developed to suggest dose adjustment for amlodipine during ritonavir treatment (Mukherjee et al. 2018). In this model, distribution of amlodipine was explained using a minimal PBPK model, and systolic blood pressure (SBP) data were used as PD model. Han et al. (2021) developed PBPK model using GastroPlus^TM for pediatrics and simulated amlodipine concentration at different age groups to suggest the optimal dosing regimen in pediatrics (Han et al. 2021). Ryu et al. (2021) developed amlodipine model using PK-Sim^® to predict drug interactions between amlodipine and MT921. They developed amlodipine model focused on drug transporter system including apical sodium dependent bile acid transporter (ASBT), because MT921 is found to be both a substrate and an inhibitor of ASBT, whereas amlodipine is an inhibitor of ASBT in vitro. The metabolism by CYP3A4 was not incorporated into this amlodipine model (Ryu et al. 2021).

In the development of the atorvastatin and amlodipine models presented here, significant emphasis was placed on exploring the potential for DDI. It is crucial to note that the magnitude of exposure changes of a substrate caused by perpetrators can significantly depend on the fraction metabolism by specific enzyme(s) for each drug (Greenblatt et al. 2000; Kirby and Unadkat 2010). Both atorvastatin and amlodipine are metabolized by CYP3A4, and considerable effort was made to accurately incorporate the fraction metabolized by CYP3A4 (f_m,CYP3A4) into our model. This precise incorporation aims to enhance the reliability of out predictions concerning DDIs. The simulated AUCRs of atorvastatin and amlodipine with or without perpetrators were compared with reported observed data and based on these comparison results, additional hepatic clearance rather than by CYP3A4 were not included in both atorvastatin and amlodipine final models. Atorvastatin was known as a substrate of OATP1B1, the K_m and V_max of OATP1B1 were reflected in the atorvastatin model. On the other hand, amlodipine was known as inhibitors or CYP3A4 and BCRP, the K_i values of these enzyme and transporter were incorporated into the amlodipine model. DDI simulation results were reasonably explain the changes of AUC and C_max of atorvastatin and amlodipine, these developed models can be effectively used to explore further the potential DDIs of atorvastatin and amlodipine with various comedications.

The authors declare that they have no conflict of interest.

None.