Integration of a Physiologically Based Pharmacokinetic and Pharmacodynamic Model for Tegoprazan and Its Metabolite: Application for Predicting Food Effect and Intragastric pH Alterations

A physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) model for tegoprazan and its major metabolite M1 was developed to predict PK and PD profiles under various scenarios. The PBPK model for tegoprazan and M1 was developed and predicted using the SimCYP® simulator and verified using clinical study data obtained after a single administration of tegoprazan. The established PBPK/PD model was used to predict PK profiles after repeated administrations of tegoprazan, postprandial PK profiles, and intragastric pH changes. The predicted tegoprazan and M1 concentration–time profiles fit the observed profiles well. The arithmetic mean ratios (95% confidence intervals) of the predicted to observed values for the area under the curve (AUC0–24 h), maximum plasma drug concentration (Cmax), and clearance (CL) for tegoprazan and M1 were within a 30% interval. Delayed time of maximum concentration (Tmax) and decreased Cmax were predicted in the postprandial PK profiles compared with the fasted state. This PBPK/PD model may be used to predict PK profiles after repeated tegoprazan administrations and to predict differences in physiological factors in the gastrointestinal tract or changes in gastric acid pH after tegoprazan administration.


Introduction
Proton-pump inhibitors (PPIs) are widely used for treating gastroesophageal reflux disease (GERD) and peptic ulcers or for eradicating Helicobacter pylori [1]. GERD is caused by the regurgitation of stomach contents into the esophagus, resulting from relaxation of the lower esophageal sphincter or an increase in gastric pressure [2]. PPIs, including omeprazole, pantoprazole, and esomeprazole, are primarily used to prevent damage to the esophageal epithelial tissues, caused by gastric acid reflux, in patients with GERD, Barrett's esophagus, and peptic ulcer disease [3]. The plasma half-life of most PPIs is less than 2 h; however, the half-life of tenatoprazole is up to 9 h [1]. Although PPIs are widely used to treat gastric acid-related disorders, they have several limitations. PPIs, after their conversion to an active form in a low-pH environment in the parietal cell canaliculus, can inhibit H + /K + -ATPase; however, it takes 2-3 days to reach the steady-state of acid inhibition [1,4]. In addition, most PPIs exhibit large inter-individual variability and potential for high drug-drug or drug-genetic interactions because they are primarily A recent study predicted the DDI of tegoprazan using a PBPK model [18] that was established using the advanced dissolution, absorption, and metabolism (ADAM) absorption model and the minimal PBPK distribution model. They evaluated the developed model using clinical data from single and repeated administrations of tegoprazan, and the model predicted the observed PK profiles well. When clarithromycin, which is used for Helicobacter pylori eradication therapy, was co-administered with tegoprazan, the established model could predict the potential for DDI. The predicted systemic exposure of tegoprazan increased by approximately three-fold when co-administered with the maximum recommended dose of clarithromycin or ketoconazole, compared with that when administered without a CYP3A4 modulator. Furthermore, when the CYP3A4 inducer rifampicin was administered with tegoprazan, systemic exposure of tegoprazan was predicted to decrease by up to 30%. Although the study predicted and validated CYP3A4-mediated DDI using this model, the effect of metabolites of tegoprazan and food were not considered.
The objectives of this study were (1) to establish a PBPK/pharmacodynamic (PD) model of tegoprazan and its major metabolite M1 using their physicochemical characteristics, in vitro experiments for absorption and metabolism, and clinical data; (2) to identify the factors that have a significant influence on the postprandial PK profiles of tegoprazan; and (3) to predict the change in intragastric pH following the administration of tegoprazan.

Assessment of Microsomal Metabolic Stability for Tegoprazan and M1
The NADPH-generating system, glucose 6-phosphate (3.3 mM), 0.1 M potassium phosphate (pH 7.4), and human liver microsomes (HLMs; 0.5 mg/mL) were combined and incubated at 37 • C for 5 min. The organic content in the reaction mixture was 0.01% DMSO and less than 1% acetonitrile. Then, 1 µM tegoprazan and M1 were added to the reaction tube (total volume: 160 µL) and incubated at 37 • C for 0, 5, 15, 30, and 60 min. To terminate the reaction, 320 µL of acetonitrile containing an internal standard (100 ng/mL carbamazepine) was added to each sample. All experiments were performed in triplicate. The samples were vortexed for 5 min and centrifuged at 3000 rpm at 4 • C for 10 min. The concentrations of tegoprazan and M1 were determined by liquid chromatography-tandem mass spectrometry (LC-MS/MS). The concentrations of tegoprazan and M1 at 0 min were used to assess metabolism by HLMs based on changes in drug concentration over time. The terminal slope of linear regression was used to calculate the intrinsic clearance (CL int ) of tegoprazan and M1. For the in vitro HLM system, CL int was calculated using Equation (1). CL int,mic (µL/min/mg protein) = k × V incubation /C incubation (1) where V incubation is the incubation volume and C incubation is the concentration of the microsomal protein.

Caco-2 Permeability Assay
Caco-2 cells were cultured in DMEM supplemented with 20% FBS, 1% non-essential amino acids, and 1% penicillin-streptomycin. Caco-2 cells were seeded onto collagencoated 12-Transwell membranes at a density of 5 × 10 5 cells/mL and incubated under optimal conditions (maintained at 37 • C in a humidified atmosphere with 5% CO 2 /95% air) for 21 days. The culture medium was replaced every second day. For the experiments, after the growth medium was removed, the attached cells were washed with pre-warmed HBSS (pH 7.4) and pre-incubated with HBSS for 20 min at 37 • C. The integrity of the cell monolayers was evaluated before and after the permeability experiments by measuring transepithelial electrical resistance (TEER) values using an epithelial volt/ohm meter (World Precision Instruments, Sarasota, FL, USA). Caco-2 cells with TEER values higher than 300 Ω·cm 2 were used for permeability experiments, and alterations in TEER values were less than 10% in all cases (i.e., 417.3 ± 6.3 Ω·cm 2 and 400.8 ± 3.5 Ω·cm 2 , before and after the permeability experiments, respectively).
To determine the apical to basal (A to B) permeability of tegoprazan, 0.5 mL of HBSS containing 10 µM tegoprazan and 1.5 mL of fresh HBSS were added to the apical and basal side of the insert, respectively. The insert was transferred to a new well that contained 1.5 mL of fresh HBSS every 15 min, for 1 h. Aliquots (0.3 mL) on the basal side were transferred to fresh tubes and stored at −80 • C until further analysis. To investigate B-to-A permeability, 1.5 mL of HBSS containing 10 µM tegoprazan and 0.5 mL of fresh HBSS were added to the basal and apical sides of the insert, respectively. Aliquots (0.3 mL) on the apical side were transferred to fresh tubes, and 0.3 mL of pre-warmed fresh HBSS was replaced every 15 min for 1 h. The transport study was performed for 1 h based on the linearity of the cumulative transported tegoprazan amount versus transport time. The cumulative transported amount was calculated as the sum of transported amounts for 0-15 min, 15-30 min, 30-45 min, and 45-60 min, and the transport rate of tegoprazan was the slope of the regression line derived from the cumulative transported amount versus time plot. The apparent permeability (P app ) was calculated from the following Equation (2): P app 10 −6 cm/s = transport rate (nmol/min) concentration (µM) × area(cm 2 ) × 60 s (2)

Evaluation of Intrinsic Clearance Using Recombinant CYPs (rCYPs)
The elimination profiles of tegoprazan and M1 were assessed using rCYPs. All the assay procedures were performed according to the manufacturer's instructions. Master mix (1.399 mL), containing the NADPH-generating system, 3.3 mM of glucose 6-phosphate, PBS buffer, deionized water, and each rCYP (25 pmol/µL of CYP2C8, 2C19, and 2E1 and 50 pmol/µL of CYP3A4, 2C9, and 2D6), was added to a microtube and pre-incubated at 37 • C for 5 min. Next, 1 µM tegoprazan and M1 were added and incubated at 37 • C for 0, 5, 10, 20, 30, 45, and 60 min. A total of 100 µL of the sample, incubated for predefined duration, was transferred to a fresh e-tube. Next, 200 µL of acetonitrile containing tegoprazan-d 6 (0.5 µM) was added, vortexed for 5 min, and centrifuged at 16,200× g and 4 • C for 10 min. The concentrations of the remaining tegoprazan and M1 were determined by LC-MS/MS. All experiments were performed in duplicate. CL int,rCYP levels were assessed using the results of the disappearance test. The terminal slope of the linear regression was used to determine the CL int for tegoprazan. The CL int resulting from rCYPs was calculated using Equation (3): CL int,rCYP (µL/min/pmol CYP) = 0.693

Determination of Tegoprazan and M1 Concentration in In Vitro Samples
Caco-2 and rCYP samples were quantified using LC-MS/MS (UPLC ® I-class coupled with TQ-MS (Waters, Milford, MA, USA). The mobile phase comprised (a) 0.1% formic Pharmaceutics 2022, 14, 1298 5 of 18 acid and (b) acetonitrile containing 0.1% formic acid. Gradient elution at a flow rate of 0.3 mL/min was applied. Tegoprazan, M1, and the internal standard (tegoprazan-d 6 ) were optimized in electrospray ionization positive mode. Acetonitrile with 0.1 µM of tegoprazand 6 was added to the collected Caco-2 or rCYP samples. Then, processed samples were centrifuged at 13,000× g for 20 min. After centrifugation, 100 µL of supernatant was transferred to an LC-MS vial and analyzed using LC-MS/MS. The analytical method was validated based on the Bioanalytical Method Validation Guidance for Industry published by the US Food and Drug Administration and the Guideline on Bioanalytical Method Validation published by the Korean Ministry of Food and Drug Safety (MFDS) [32,33]. Linearity was in the range of 0.1-20 µM, and the inter-batch accuracy and precision were in the ranges of 91.8-101.3% and 2.3-8.9%, respectively. The stability of tegoprazan was evaluated, and it satisfied the acceptance criteria (coefficient of variances was within 15%) under variable conditions including reinjection, processed sample stability (PSS), and short-term stability for the stock and sample.

Clinical PK Dataset
Plasma tegoprazan and M1 concentration-time profiles after a single oral administration were obtained from a previous comparative pharmacokinetic study comparing two formulations of tegoprazan in healthy Korean male subjects (IRB number: CUH 2015-03-007) [34]. This study was designed as a randomized, open-label, single-dose, two-sequence, and two-period crossover study. A total of 12 volunteers participated, and the mean ± standard deviation of age, height, and weight were 23.9 ± 1.3 years, 173.1 ± 7.6 cm, and 68.4 ± 8.2 kg, respectively. All subjects were administered a single tegoprazan film-coated 100 mg tablet orally of formulation type 1 or 2 (HK inno.N Corp., Seoul, Korea), and pharmacokinetic profiles were analyzed. For pharmacokinetic analysis, 8 mL of blood was collected to measure the plasma tegoprazan and M1 concentrations during each period at the following time points: pre-dose (0 h) and 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 12, 24, and 48 h post-dose. Plasma tegoprazan and M1 concentrations were measured by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Individual concentration-time profiles were used to develop and verify tegoprazan and M1 PBPK models. The PK parameters were calculated using a non-compartmental analysis (NCA) performed by Phoenix ® WinNonlin ® 6.2 (Certara, Princeton, NJ, USA).

PBPK Model Development
The tegoprazan and M1 PBPK models were developed using SimCYP ® simulator, version 20 (Certara, Princeton, NJ, USA). Compound libraries were incorporated into an in vitro dataset, including the microsomal clearance of tegoprazan and M1. Observed PK parameters were obtained from the aforementioned clinical studies (Tables 1 and 2).
The absorption of tegoprazan was applied to the advanced dissolution, absorption, and metabolism (ADAM) model, and the diffusion layer model (DLM) built into SimCYP ® was applied to the considered formulation and pH-dependent solubility of tegoprazan. The permeability of tegoprazan was determined by a Caco-2 permeability assay and optimized by parameter estimation. Caco-2 permeability was optimized such that the predicted concentration-time profiles were fitted to clinical study data. The metabolite was generated by the metabolism of the parent drug; therefore, absorption was not considered.
The tissue distribution in the PBPK model was predicted by the distribution model (e.g., Rodger and Rowland method, Poulin and Theil method) and the tissue:plasma partition coefficient (Kp) scalar for specific organ or global values. Predicted systemic distribution was expressed as the volume of distribution in a steady state (V ss ). The V d values of the clinical study were applied to the initial model of tegoprazan, and parameter estimation was used to determine the optimized Kp scalar fitted to the observed data. The Rodger and Rowland method (method 2) was applied to consider the weak base property and distribution of the neutral and ionic forms [35]. The minimal PBPK model was applied to the M1 distribution module to reduce the complexity. The volume of a single adjusting compartment (V sac ), non-physiological compartment, input rate to SAC (k in ), and elimination rate from SAC (k out ) were estimated using the parameter estimation module within SimCYP ® .
The elimination profiles of tegoprazan and M1 were obtained from in vitro, in vivo, and in silico predictions. Optimized in-house intrinsic clearance for CYPs (CL int,rCYPs ), and human liver microsomal and cytosolic (HLM and HLC) clearance were applied to the elimination module of the tegoprazan model. Observed or estimated renal and biliary clearance were applied to the tegoprazan and M1 models [36]. Whole-organ metabolic clearance, using the optimized value for the in vitro assay results, was applied to the M1 model. ADAM, advanced dissolution, absorption, and metabolism; B/P, blood-to-plasma ratio; f u , unbound fraction; P app , caco-2 , apparent permeability of caco-2; V ss , volume of distribution at steady state; CL int , intrinsic clearance; CL R , renal clearance.

Model Verification
The virtual dosing regimen was set to a single oral dose of 100 mg tegoprazan, to match that of the observed data. The simulation conditions were set for 100 virtual subjects who participated in 10 virtual clinical studies (1000 volunteers). The age range was 20 to 40 years, and the male-to-female ratio was 1:1. The predicted and observed PK parameters were compared to evaluate the model prediction results. The AUC 0-24 h , maximum plasma concentration (C max ), time required to reach the C max (T max ), and CL for tegoprazan and AUC 0-48 h and C max for M1 are presented as results, and the predicted/observed ratios for AUC, C max , and CL were calculated to estimate the predictive power of the model. If the Pharmaceutics 2022, 14, 1298 7 of 18 predicted vs. observed ratio and its 95% CI were within the 30% range (0.7-1.3), the model was considered to fit well. Using the extracted plasma tegoprazan concentration-time profile reported in the literature [15], the predicted parameters (AUC 0-24 h and C max ) under conditions of repeated administrations for 7 days (100 mg of tegoprazan q.d.) were compared with the results of the clinical studies.

Prediction and Comparison of Postprandial PK Profiles Using the Developed Model
Postprandial PK profiles were predicted using an established model. Concentrationtime profiles were obtained following a single oral dose (50 mg) of tegoprazan in the fasting state or 30 min after a high-fat diet. The predicted AUC last , C max , and T max values were compared with those of a previous clinical study [16].

Incorporation of a PD Model to Predict Tegoprazan-Induced Gastric pH Changes
To simulate the change in gastric acid pH after the administration of tegoprazan, an intragastric PD model following the administration of tegoprazan was incorporated into the developed PBPK model. The trend of the 24-h gastric acid pH baseline was described using a sixth order Fourier series, and the sigmoid E max model was applied to the indirect response model based on the baseline intragastric pH model used to predict gastric acid pH alterations after tegoprazan administration. The equations for the indirect response-sigmoid E max model with the baseline intragastric pH model were as follows: where Base is the baseline intragastric pH model; a 0 , a i and b i are the Fourier coefficients; i is an integer sequence; w is the fundamental frequency; x is time; E max is the maximum effect; C p is plasma tegoprazan concentration; EC 50 is the concentration at half of the maximum effect; k in is the predicted plasma tegoprazan concentration; k out is the output rate constant; R is the response; and γ is the Hill coefficient. Furthermore, the predicted PD response was reflected into the individual gastric acid profiles in the population library used in the PBPK simulation to apply the gastric acid pH-lowering effect (Figure 1). A PD model with gastric pH feedback function was generated using a custom Lua script within the SimCYP ® simulator.
Pharmaceutics 2022, 14, x FOR PEER REVIEW 9 of 20 maximum effect; Cp is plasma tegoprazan concentration; EC50 is the concentration at half of the maximum effect; kin is the predicted plasma tegoprazan concentration; kout is the output rate constant; R is the response; and γ is the Hill coefficient. Furthermore, the predicted PD response was reflected into the individual gastric acid profiles in the population library used in the PBPK simulation to apply the gastric acid pH-lowering effect ( Figure  1). A PD model with gastric pH feedback function was generated using a custom Lua script within the SimCYP ® simulator. The established PBPK/PD model was evaluated using observed plasma tegoprazan concentrations and gastric acid pH-time profiles after a single administration of 50 mg tegoprazan. This model was used to predict the PK and gastric acid pH profiles after repeated administrations of 100 mg tegoprazan once daily for 7 days. The predicted PD response was evaluated from the extracted gastric acid pH profiles following the administration of tegoprazan [15]. The results of the PD model were evaluated at a holding rate of pH ≥ 4, which is the surrogate endpoint of the GERD healing rate [37,38].

Caco-2 Permeability Assay
The mean ± standard deviation of apical to basolateral (A to B) permeability was 12.53 ± 1.32 × 10 −6 cm/s ( Table 3). The recovery calculated through an analysis of the results and the Papp value satisfied the range of ±20%; therefore, we confirmed that this was not an error during the experimental processes. Moreover, since the efflux ratio was lower than 2, the effect of transport via p-glycoprotein is expected to be insignificant. Mean caco-2 permeability was used to develop the initial tegoprazan PBPK model and subsequent model improvement. The established PBPK/PD model was evaluated using observed plasma tegoprazan concentrations and gastric acid pH-time profiles after a single administration of 50 mg tegoprazan. This model was used to predict the PK and gastric acid pH profiles after repeated administrations of 100 mg tegoprazan once daily for 7 days. The predicted PD response was evaluated from the extracted gastric acid pH profiles following the administration of tegoprazan [15]. The results of the PD model were evaluated at a holding rate of pH ≥ 4, which is the surrogate endpoint of the GERD healing rate [37,38].

Caco-2 Permeability Assay
The mean ± standard deviation of apical to basolateral (A to B) permeability was 12.53 ± 1.32 × 10 −6 cm/s ( Table 3). The recovery calculated through an analysis of the results and the P app value satisfied the range of ±20%; therefore, we confirmed that this was not an error during the experimental processes. Moreover, since the efflux ratio was lower than 2, the effect of transport via p-glycoprotein is expected to be insignificant. Mean caco-2 permeability was used to develop the initial tegoprazan PBPK model and subsequent model improvement.

Assessment of Microsomal Metabolic Stability for Tegoprazan and M1
In the control sample without NADPH, the percentages of tegoprazan and M1 remaining after incubation were 99.54% ± 5.32% and 102.2% ± 6.52%, respectively, indicating that the disappearance of tegoprazan and M1 primarily occurred via hepatic CYPs. The HLM intrinsic clearance (CL int,HLM ) of tegoprazan and M1 was 15.96 and 2.44 µL/min/mg protein, respectively ( Figure 2). These values were applied to the PBPK model.

Assessment of Microsomal Metabolic Stability for Tegoprazan and M1
In the control sample without NADPH, the percentages of tegoprazan and M1 remaining after incubation were 99.54% ± 5.32% and 102.2% ± 6.52%, respectively, indicating that the disappearance of tegoprazan and M1 primarily occurred via hepatic CYPs. The HLM intrinsic clearance (CLint,HLM) of tegoprazan and M1 was 15.96 and 2.44 µ L/min/mg protein, respectively ( Figure 2). These values were applied to the PBPK model.

Model Verification
The final model was used to predict the PK profiles of tegoprazan and M1. Most of the observed mean plasma concentration-time profiles were included in the 5th and 95th percentiles of the predicted mean plasma concentration-time profiles (Figure 3). The ratios of predicted vs. observed AUClast, Cmax, and CL, and 95% confidence interval values were included in the 30% range of the mean ratio (Table 4).

Model Verification
The final model was used to predict the PK profiles of tegoprazan and M1. Most of the observed mean plasma concentration-time profiles were included in the 5th and 95th percentiles of the predicted mean plasma concentration-time profiles (Figure 3). The ratios of predicted vs. observed AUC last , C max , and CL, and 95% confidence interval values were included in the 30% range of the mean ratio (   All parameters are presented as the arithmetic mean (5th-95th percentile), except Tmax, which is presented as the median (minimum-maximum value). The acceptance criteria for Cmax and AUClast ratio were within 30%. AUC: area under the curve from time 0 to the last sampling time point (for tegoprazan and M1, these AUCs were for 0-24 h and 0-48 h, respectively); Cmax, maximum concentration; CL, clearance; Tmax, time of maximum drug concentration  All parameters are presented as the arithmetic mean (5th-95th percentile), except T max , which is presented as the median (minimum-maximum value). The acceptance criteria for C max and AUC last ratio were within 30%. AUC: area under the curve from time 0 to the last sampling time point (for tegoprazan and M1, these AUCs were for 0-24 h and 0-48 h, respectively); C max , maximum concentration; CL, clearance; T max , time of maximum drug concentration

Prediction of Systemic Exposure to Tegoprazan and M1 after Repeated Administrations
Since it is common for gastric acid pH to increase throughout repeated tegoprazan administrations for gastroesophageal reflux disease, gastric ulcer, and Helicobacter pylori eradication, which are the primary indications for tegoprazan [39], the plasma concentrationtime profiles were predicted for repeated administrations using the established model. The predicted mean plasma tegoprazan concentration-time profiles at a dose of 100 mg were fitted to the observed profiles ( Figure 4). However, the developed model overpredicted the C max on Day 7. The predicted C max of tegoprazan was 1.46-fold higher than the observed C max (predicted C max 1257.8 ± 250.5 ng/mL vs. observed C max 845.2 ± 40.7 ng/mL).

Administrations
Since it is common for gastric acid pH to increase throughout repeated tegoprazan administrations for gastroesophageal reflux disease, gastric ulcer, and Helicobacter pylori eradication, which are the primary indications for tegoprazan [39], the plasma concentration-time profiles were predicted for repeated administrations using the established model. The predicted mean plasma tegoprazan concentration-time profiles at a dose of 100 mg were fitted to the observed profiles ( Figure 4). However, the developed model overpredicted the Cmax on Day 7. The predicted Cmax of tegoprazan was 1.46-fold higher than the observed Cmax (predicted Cmax 1257.8 ± 250.5 ng/mL vs. observed Cmax 845.2 ± 40.7 ng/mL).

Prediction and Comparison of Postprandial PK Profiles Using the Developed Model
The ratios of predicted vs. observed tegoprazan AUC0-24 h and Cmax values were within the 30% range of the mean ratio, which is consistent with previously reported results (Table 5) [16]. For M1, the ratio of the predicted vs. observed AUC0-48 h ratio was within the 30% range of the mean ratio; however, the predicted vs observed Cmax ratio exceeded the 30% range. For drug administered after a high-fat diet, the predicted mean Cmax of tegoprazan decreased by 31.3% (fasted state: 684.9 ng/mL; at 30 min after a highfat meal: 470.7 ng/mL) and the predicted median Tmax was delayed from 0.6 h (0.3-1.3) to 1.05 h (0.3-3.5). The predicted AUC0-24 h of tegoprazan was not significantly different from that in the fasted state. The predicted mean AUC0-48 h and Cmax of M1 did not show large differences compared to fasting, and the predicted median Tmax of M1 was delayed from 5.8 h (1.5-23) to 7.5 h (2.3-24).

Prediction and Comparison of Postprandial PK Profiles Using the Developed Model
The ratios of predicted vs. observed tegoprazan AUC 0-24 h and C max values were within the 30% range of the mean ratio, which is consistent with previously reported results (Table 5) [16]. For M1, the ratio of the predicted vs. observed AUC 0-48 h ratio was within the 30% range of the mean ratio; however, the predicted vs observed C max ratio exceeded the 30% range. For drug administered after a high-fat diet, the predicted mean C max of tegoprazan decreased by 31.3% (fasted state: 684.9 ng/mL; at 30 min after a high-fat meal: 470.7 ng/mL) and the predicted median T max was delayed from 0.6 h (0.3-1.3) to 1.05 h (0.3-3.5). The predicted AUC 0-24 h of tegoprazan was not significantly different from that in the fasted state. The predicted mean AUC 0-48 h and C max of M1 did not show large differences compared to fasting, and the predicted median T max of M1 was delayed from 5.8 h (1.5-23) to 7.5 h (2.3-24). Table 5. The predicted and observed AUC last , C max , and T max for tegoprazan and M1 after a single administration of 50 mg tegoprazan in the fasted state or at 30 min after a high-fat meal.  All parameters are presented as arithmetic mean ± standard deviation, except for T max , which is presented as the median (minimum-maximum value). The ratios of each parameter were calculated as follows: predicted mean/observed mean. The observed parameters were obtained from [34]. AUC, area under the concentrationtime curve; C max , maximum concentration; T max , time of maximum drug concentration.

Incorporation of a PD Model to Predict Tegoprazan-Induced Gastric pH Changes
The established PBPK/PD model successfully predicted plasma tegoprazan concentrations and gastric acid pH-time profiles after a single administration of 50 mg tegoprazan ( Figure 5).
AUC, area under the concentration-time curve; Cmax, maximum concentration; Tmax, time of maximum drug concentration.

Incorporation of a PD Model to Predict Tegoprazan-Induced Gastric pH Changes
The established PBPK/PD model successfully predicted plasma tegoprazan concentrations and gastric acid pH-time profiles after a single administration of 50 mg tegoprazan ( Figure 5).  Plasma tegoprazan concentrations and gastric acid pH-time profiles after repeated administrations of 100 mg tegoprazan once daily for 7 days were predicted using the established PBPK/PD model. Most of the observed concentration profiles on Day 1 and 7 fit well into the 5th and 95th percentile intervals of the predicted mean concentration (Figure 6a,c). When the PD response was not applied, the intragastric pH was fixed at 1.49, which was the default baseline pH (Figure 6b). However, when the PD response with feedback function was applied, the intragastric pH changed over time (Figure 6d). The observed gastric acid pH profiles were also well fitted to the predicted gastric acid pH-time profiles (Figure 6d). Plasma tegoprazan concentrations and gastric acid pH-time profiles after repeated administrations of 100 mg tegoprazan once daily for 7 days were predicted using the established PBPK/PD model. Most of the observed concentration profiles on Day 1 and 7 fit well into the 5th and 95th percentile intervals of the predicted mean concentration (Figure  6a,c). When the PD response was not applied, the intragastric pH was fixed at 1.49, which was the default baseline pH (Figure 6b). However, when the PD response with feedback function was applied, the intragastric pH changed over time (Figure 6d). The observed gastric acid pH profiles were also well fitted to the predicted gastric acid pH-time profiles (Figure 6d).  When evaluating the percent holding rate at pH ≥ 4 for 24 h, which is a PD parameter, the holding rate after a single administration of 50 mg tegoprazan or repeated administrations of 100 mg tegoprazan once daily for 7 days matched the results of the clinical studies (Table 6). When evaluating the percent holding rate at pH ≥ 4 for 24 h, which is a PD parameter, the holding rate after a single administration of 50 mg tegoprazan or repeated administrations of 100 mg tegoprazan once daily for 7 days matched the results of the clinical studies (Table 6).  Observed parameters were obtained from [15]. q.d.: once daily.

Discussion
A PBPK/PD model for tegoprazan and M1 was successfully established. In vitro permeability and metabolism data and clinical study results were used to develop and evaluate the model. Plasma tegoprazan and M1 concentration-time profiles and gastric acid pH-time profiles after single or repeated administrations were predicted using the developed model. The ratios of predicted vs. observed AUC 0-24 h , C max , and CL (95% confidence intervals) values for tegoprazan were 0.90 (0.85-0.96), 0.88 (0.80-1.00), and 1.25 (1.17-1.34), respectively. The predicted results of the M1 model revealed that the arithmetic mean ratios (95% confidence interval) of the predicted AUC 0-48 h and C max to the observed plasma concentration-time profile were 0.88 (0.84-0.94) and 1.01 (0.95-1.08), respectively (Table 4). According to the simulation results, the developed model accurately predicted the PK profiles of tegoprazan and M1. The PBPK model was applied to predict PK profiles following repeated administrations of tegoprazan, postprandial PK profiles, and alteration of gastric acid pH profiles after tegoprazan administration. When the clinical study and predicted results were compared, good prediction results were evident under all conditions. Thus, we confirmed that the developed PBPK/PD model had sufficient predictive power.
The elimination profiles of tegoprazan were reflected in the PBPK model through a metabolism assay using recombinant CYPs. Based on previous reports, tegoprazan is primarily metabolized by CYP3A4 [18,40]. According to the information leaflet for K-CAB ® tablets (tegoprazan 50 mg), although the metabolism of tegoprazan is decreased in the presence of ketoconazole, other CYP1A2, 2C9, 2C19, and 2D6 inhibitors did not significantly decrease the in vitro metabolism of tegoprazan (exact value not presented) [36]. When comparing the results of applying only the intrinsic clearance of CYP3A4 and 2C19 to the elimination module of the tegoprazan model with those of the intrinsic clearance of CYP2C8, 2C9, 2D6, and 2E1 together with CYP3A4 and 2C19, no significant differences were observed (Supplementary Figure S1). Both CYP3A4-and 2C19-mediated metabolism of tegoprazan are thought to be major metabolic pathways. Based on these results, the CYP3A4-and 2C19-mediated metabolism of tegoprazan is likely to be a major metabolic enzyme involved in tegoprazan metabolism.
The PBPK/PD model developed in this study successfully predicted the PK profile of M1, the major metabolite. In a preclinical study, M1 showed reversible inhibitory potential against porcine H + /K + -ATPase, with 10-fold less potency than tegoprazan [41]. The experimental half maximal inhibitory concentration (IC 50 ) values for tegoprazan and M1 were 0.53 µM and 6.19 µM, respectively [40,41]. M1 is also expected to exhibit an acid suppression effect based on preclinical studies and in vitro assay results; however, no studies have evaluated the efficacy of M1 in humans. In the present study, a PBPK model for M1 was developed using minimal PBPK distribution and clearance profiles in HLM to evaluate PK profiles. The developed M1 model may contribute to the mechanistic interpretation of DDI in the PK profiles of M1 and tegoprazan.
Integrating the PBPK model with an intragastric PD model, gastric pH changes after tegoprazan administration were obtained. To apply the circadian rhythm of intragastric pH, a Fourier series was fitted to the baseline pH profile and the profile was applied to the basic structure of the PD model. The indirect response-sigmoid E max model with the baseline intragastric pH model and gastric acid pH feedback function of the ADAM model was integrated into the PBPK model. The PD response was successfully predicted with the changes of intragastric pH in virtual subjects after tegoprazan administration. Using the integrated model, PD responses under various clinical settings could be predicted.
The developed PBPK/PD model may be used to evaluate the effect of tegoprazaninduced intragastric pH increase on the pharmacokinetics of drugs co-administered with tegoprazan. The predicted results using the established model indicated that the observed plasma tegoprazan concentration and gastric acid pH-time profiles after a single administration of 50 mg tegoprazan and repeated administrations of 100 mg tegoprazan once daily for 7 days were well fitted to the prediction profiles. The percentage holding rates of pH ≥ 4 for 24 h after a single administration of 50 mg tegoprazan or repeated administrations of 100 mg tegoprazan for 7 days were similar to those of the clinical study. The percent holding rate at pH ≥ 4 for 24 h after a single administration of 50 mg tegoprazan was predicted to be approximately 49.6%. The percent holding rates at pH ≥ 4 for 24 h after repeated administrations of 100 mg tegoprazan once daily for 7 days were predicted to be approximately 60.3% on Day 1 and Day 7.
According to solubility profiles, repeated administrations of tegoprazan may affect its absorption. The pKa of tegoprazan is 5.1, which is indicative of a weak base [40,41]. Due to this physicochemical property, tegoprazan may be specifically highly distributed in the parietal cells of the stomach, where the pH is 1-2. This indicates that tegoprazan has the potential to effectively suppress gastric acid secretion [17,42]. In the clinical study, after repeated administrations of 100 mg tegoprazan once daily, the observed C max at Day 7 decreased by approximately 40.2% compared with Day 1 (Day 1: 1413.3 ± 24.7 ng/mL; Day 7: 845.2 ± 40.7 ng/mL) [15]. When 100 mg of tegoprazan was repeatedly administered to 10 healthy Chinese subjects, AUC and CL on Day 7 of administration did not show a significant difference compared with Day 1 after administration of tegoprazan; however, V d /F increased significantly on Day 7 compared with that on Day 1 [39], whereas the predicted C max at Day 7 increased by approximately 1.84% compared with Day 1 (Day 1: 1235.1 ± 241.9 ng/mL; Day 7: 1257.8 ± 250.5 ng/mL). Based on these results, it was suggested that repeated administrations of tegoprazan might increase intragastric pH, thereby reducing the absorption of tegoprazan itself.
The decrease in C max after repeated administrations of tegoprazan appears to be influenced not only by the decrease in acidity of gastric juice but also by other physiological or drug-specific factors. The developed model also overpredicted the C max on Day 7 compared with the observed profiles after repeated administrations (observed value: 845.2 ± 40.7 ng/mL; predicted value: 1257.8 ± 250.5 ng/mL). Based on the predicted results using the PBPK/PD model, gastric acid pH increased after taking tegoprazan; however, the systemic exposure or C max of tegoprazan was barely affected (C max of Day 7 when gastric acid pH was fixed at 1.49: 1266.1 ± 251.2 ng/mL; C max of Day 7 when dynamic gastric acid pH by PD model: 1257.8 ± 250.5 ng/mL). In a previous study, it was estimated that the cause of increase in the volume of distribution following repeated administrations of tegoprazan was not only due to ionized molecules by gastric acid accumulated in the acidic secretory canaliculi of parietal cells, but also because the ionized form has poor cell membrane permeability [39]. The volume of distribution could be increased after repeated administrations of tegoprazan. Moreover, dissolution-related parameters, such as supersaturation ratio and precipitation rate constant, might influence the decrease in C max values. Tegoprazan showed high permeability (experimental Caco-2 permeability: 12.53 ± 1.32 × 10 −6 cm/s) and low solubility (0.02 mg/mL in pH 6.8), and it was considered a biopharmaceutics classification system (BCS) class II drug. In a PBPK study for BCS class II drugs, the colonic absorption, bile micelle solubilization, and unbound fraction in gut enterocytes (fu gut ) were identified as important factors for the absorption process. In another PBPK study using a compound with pH-dependent solubility, particle size was considered a significant factor for drug exposure [42]. Although DLM was applied in the absorption module to reflect the effect of the formulation in the current model, the default values provided in SimCYP ® for formulation-related parameters, such as particle size, radius, density, and supersaturation ratio, were applied owing to a lack of information. This might lead to some discrepancies in the absorption process and related parameters.
The differences in residence time resulting from the diet, along with pH changes in the gastrointestinal tract, were supposed to affect the postprandial PK profiles of tegoprazan. According to a study that assessed differences in tegoprazan pharmacokinetics when administered in the fasting state, 30 min before a high-fat meal and 30 min after a high-fat meal, the C max and T max of tegoprazan were affected by food consumption, although systemic exposure of tegoprazan did not change. The C max for 30 min after a high-fat meal decreased to 38.7% compared with that after a fasting state (fasting state: 803 µg/L; administered at 30 min after high-fat meal: 492 µg/L). The median T max values were 1 h in the fasting state, 0.48 h at 30 min before a high-fat meal, and 3 h at 30 min after a high-fat meal [16]. These results indicated that the difference in postprandial PK profiles resulted from the physicochemical characteristics of tegoprazan and the time delay in reaching the small intestine [16]. Postprandial PK profiles of tegoprazan were predicted using an established model. When a high-fat diet was applied to the simulation, T max was delayed and C max was decreased, which is consistent with the results of the clinical study ( Table 5). The predicted absorbed fraction in the duodenum upon administration of tegoprazan at 30 min after a high-fat meal was increased by approximately 72.7%, however, it decreased in other segments, compared with that in the group administered while fasting (Supplementary Table S2). The predicted surface solubility increased by 14.8% and 40.3% in the stomach and duodenum, respectively, and there was a slight decrease or no difference in solubility in the remaining segments in this group compared to that in a fasted state (Supplementary Table S2). The predicted mean residence time in the stomach upon administration of tegoprazan at 30 min after a high-fat meal was increased by approximately 170.4% compared to that administered while fasting (Supplementary  Table S3). The predicted mean gastric residence time was delayed; therefore, it was assumed that tegoprazan was more distributed to parietal cells in the fed state than in the fasting state. Increases in the intragastric pH after tegoprazan administration or food intake are thought to be the major cause of a decrease in the C max of tegoprazan. However, physiological factors (i.e., gastric and small intestine mean gastric residence times) and formulation-related parameters (i.e., particle size, bile micelle-mediated solubility parameters) can also affect postprandial PK profiles.
The limitation of this study is that the food effect was not considered in the PD model. In the SimCYP ® simulator, physiological factors in the gastrointestinal tract, such as intragastric/gastrointestinal (GI) tract pH and gastric emptying time, are applied differently in the fasted and fed states. If the drug response predicted by the PD model is applied to the intragastric pH of the virtual population, the dynamics of intragastric pH in the fed state predicted by the internal algorithm may not be reflected. If additional data on the actual conditions are available, such as the food effect or impact of the co-administered drug, and the circadian rhythm associated with intragastric pH can be reflected, the PBPK/PD model may predict a more realistic intragastric pH profile. The current model was built with extensive parameter optimization. When in vitro Caco-2 permeability and the intrinsic clearance of rCYPs were applied to the model, predicted concentration-time profiles were different from those of clinical study data. Caco-2 permeability and rCYP intrinsic clearance were estimated using parameter estimation, and the optimized value was fitted to the observed clinical study data. Although the final model was improved with extensive parameter optimization, it was validated using separate clinical study data that were not used for model development and model improvement. Therefore, the tegoprazan-M1 PBPK model developed in this study is considered a reasonable model.

Conclusions
In conclusion, a PBPK/PD model for tegoprazan and its metabolite M1 was developed by incorporating experimental in vitro metabolism and absorption profiles and clinical data. This model was successfully verified using the observed data following a single administration of tegoprazan. Moreover, the developed model fit the observed PK profiles well after repeated administrations. With respect to predicting postprandial PK profiles, T max was delayed, and C max was decreased. Although the absorbed fraction of tegoprazan increased in some GI tract segments in the postprandial state, it was expected that the delay in gastric and small intestine mean residence time would have a greater effect. The incorporated PD model provides a basis for evaluating changes in intragastric pH following tegoprazan administration. These PBPK/PD approaches may provide fundamental concepts for interpreting postprandial PK profiles and the impact of intragastric pH alterations on the co-administration of drugs with tegoprazan.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/pharmaceutics14061298/s1, Table S1: Intrinsic clearances using recombinant CYPs for tegoprazan and M1; Table S2: The predicted arithmetic mean metabolized fraction, absorbed fraction, and surface solubility of tegoprazan after a single administration of 50 mg tegoprazan in the fasting state or at 30 min after a high-fat meal; Table S3: The predicted gastric lag time and mean residence time for tegoprazan after a single administration of 50 mg tegoprazan in a fasting state or at 30 min after a high-fat meal; Figure S1: Predicted plasma (a,b) tegoprazan and (c,d) M1 concentration-time profiles after a single administration of 100 mg of tegoprazan. Blue solid lines and black dotted lines represent the predicted mean concentrations and the 95th and 5th percentiles, respectively. Two major CYPs represent CYP3A4 and 2C19, and six CYPs represent CYP3A4, 2C19, 2C8, 2C9, 2D6, and 2E1.  Informed Consent Statement: Informed consent was obtained from all the subjects involved in the study.