Prediction of combination effect of quinidine on the pharmacokinetics of tipepidine using a physiologically based pharmacokinetic model

Abstract Tipepidine, an antitussive drug, has been reported to have central pharmacological effects and can be expected to be safely repositioned as treatment for psychiatric disorders. Since tipepidine requires three doses per day, development of a once-daily medication would be highly beneficial. Previously, we reported that combination use with quinidine, a CYP2D6 inhibitor, prolongs the half-life of tipepidine in chimeric mice with humanised liver. In this study, to predict this combination effect in humans, a physiologically based pharmacokinetic (PBPK) model was developed, and quantitative simulation was conducted. The simulation results indicated that concomitant administration of tipepidine with quinidine increased the predicted Cmax, AUC, and t1/2 of tipepidine in the Japanese population by 3.4-, 6.6-, and 2.4-fold, respectively. Furthermore, to compare with another approach that aims to prolong the half-life, the PK profile of tipepidine administered in hypothetical extended-release form was simulated. Extended-release form was predicted to be more influenced by CYP2D6 genotype than combination with quinidine, and the predicted plasma exposure was markedly increased in poor metabolizers, potentially leading to adverse effects. In conclusion, quantitative simulation using the PBPK model suggests the feasibility of the safe repositioning of tipepidine as a once-daily medication in combination with quinidine.


Introduction
Tipepidine is a non-narcotic antitussive drug approved in 1959 in Japan.It has been widely used for more than 60 years and its safety has already been established in patients of all ages ranging from children to the elderly.Recently, tipepidine has been reported to have various central pharmacological effects in addition to antitussive effects and is expected to be effective in a variety of psychiatric disorders such as depression, attention-deficit/hyperactivity disorder (ADHD), obsessive-compulsive disorder, and schizophrenia (Kawaura et al. 2009;Honda et al. 2011;Soeda et al. 2016;Miki et al. 2019).Several clinical open studies have reported the effects of tipepidine for ADHD and depression (Sasaki et al. 2014a(Sasaki et al. , 2014b;;Dehbozorghi et al. 2019;Hoobehfekr et al. 2021).Thus, tipepidine could be safely repositioned as treatment for psychiatric disorders.
When considering the treatment of chronic psychiatric disorders, it is desirable for a therapeutic drug to have a stable and sustained effect, and to improve patient adherence and quality of life by reducing the dosing frequency.However, tipepidine has a very short half-life, approximately 1.8 h in humans (Asverin package insert), and requires three doses per day.Therefore, when repositioning tipepidine as treatment for psychiatric disorders, development of a oncedaily medication would be highly beneficial.In fact, methylphenidate, a therapeutic drug for ADHD, has been developed as an extended-release formulation to overcome the issue of its short half-life (Maldonado 2013).
One approach to prolong the half-life is increasing metabolic stability using enzyme inhibitors.Previously, we reported that cytochrome P450 (CYP) 2D6 is the main enzyme catalysing tipepidine metabolism, and that oral coadministration of quinidine, a CYP2D6 inhibitor, with tipepidine to chimeric mice with humanised liver increased the plasma exposure and prolonged the half-life of tipepidine (Hayashi et al. 2023).This study provided supportive evidence for the usefulness of combination use with CYP2D6 inhibitors for repositioning tipepidine as treatment for psychiatric disorders.However, further studies are necessary to confirm this combination effect in humans.
Physiologically based pharmacokinetic (PBPK) modelling is a mechanistic modelling combining the concept of the anatomical and physiological properties of the human body and the physicochemical and biological properties of a drug to simulate the pharmacokinetic (PK) profile of the drug (Yoon et al. 2021).PBPK modelling is a tool for quantitative predicting the effects of drug interactions or interindividual variation on PK profiles (Sychterz et al. 2022).The aim of this study was to develop and verify a PBPK model for tipepidine by using commercially available PBPK modelling and simulation software and to apply the model to quantitatively predict the effect on tipepidine of combination with quinidine.Furthermore, to compare this approach with another approach that aims to prolong the half-life of tipepidine, the PK profile of tipepidine assuming it was administered in hypothetical extended-release form was also simulated.

Clinical study data used for development and verification of PBPK models
For the development and verification of the tipepidine PBPK model, we used the clinical PK data obtained from the drug interview form for Asverin tablets (tipepidine hibenzate tablets) (PMDA 2023) and the oral administration study of Asverin in healthy Japanese subjects (Saito et al. 2020).For the development and verification of the quinidine PBPK model, we used the clinical PK data obtained from the concomitant administration study of quinidine and dextromethorphan, which is metabolised mainly by CYP2D6, in healthy adults (Pope et al. 2004).

Computer software and model verification
All PBPK model development and simulations were performed using the population-based PBPK simulator Simcyp (Version 21, Certara UK Ltd., Sheffield).The verification of the developed model was conducted by comparing the PBPK modelpredicted values with the observed clinical values.For the assessment, the absolute average fold error (AAFE), which represents the average absolute error of each data point, was utilised as the evaluation index.When the AAFE was 2 or less, it was judged that the PBPK model was developed appropriately.The AAFE was calculated using Equation 1.
where n is number of samples.

Tipepidine PBPK model development and verification
The input parameters and sources for the tipepidine PBPK model are summarised in Table 1.Physicochemical properties were predicted by ChemDraw (Ver18.2,PerkinElmer, Inc., Waltham, MA) or Simcyp.Absorption, distribution, and elimination parameters were obtained by a top-down approach using clinical PK data from the interview form for Asverin tablets (PMDA 2023).Each parameter was calculated by fitting values of the time to reach the maximum plasma concentration (t max ) of approximately 1.3 h, the maximal plasma concentration (C max ) of approximately 37 ng/mL, and the apparent half-life (t 1/2 ) of approximately 1.8 h to the following 1-compartment model equations 2 to 6.
C max ¼ F � Dose=Vss � ðka=keÞ ðk e =ðk e −k a ÞÞ (3) where k a is the absorption rate constant, k e is the elimination rate constant, F is the bioavailability, CL is the total systemic clearance, V ss is the steady-state volume of distribution, and Q is the hepatic blood flow.Since most of the administered tipepidine is excreted in the urine as metabolites (PMDA 2023), the elimination of tipepidine from the body is mainly owing to hepatic metabolism.Tipepidine is considered to have sufficient permeability and solubility, so it was assumed that tipepidine has good gastrointestinal absorption.The hepatic blood flow and body weight were set at 84 L/h and 70 kg, respectively (Hosea et al. 2009).The intrinsic clearance (CL int ) of CYP2D6 was back-calculated from the oral clearance using the retrograde calculator in Simcyp.Based on data from our previous study (Hayashi et al. 2023), the metabolic contribution ratio of CYP2D6 was assumed to be 85.4%.In addition, hepatic metabolic clearance other than by CYP2D6 was included in the elimination parameters to estimate 14.6% of the oral clearance.Intestinal availability (Fg) was calculated by Qgut model in Simcyp based on the CL int of each enzyme and P450 abundance in the small intestine.
The developed tipepidine PBPK model was verified by comparing the predicted values with the observed clinical values from the clinical study (Saito et al. 2020), in which 42 Japanese male subjects aged 20-39 years were enrolled and received a single oral dose of 40 mg of tipepidine (Asverin tablets).The distribution of CYP2D6 genotypes in the study population was as follows: one poor metabolizer (PM), 5 intermediate metabolizers (IM), 35 extensive metabolizers (EM), and one ultra-rapid metabolizer (UM).For the PBPK model verification, the Sim-Japanese file built in Simcyp was used as the virtual Japanese population reflecting the gene frequency.Furthermore, to simulate each genotype of CYP2D6, virtual populations of PM, IM, EM, and UM were created by changing the gene frequency in the Sim-Japanese file.Simulations were conducted with 200 virtual subjects in each group.To verify the PBPK model, the model-predicted C max and the area under the plasma concentration versus time curve (AUC) were compared with the C max and AUC observed in the clinical study.

Quinidine PBPK model development and verification
Quinidine was selected as a CYP2D6 inhibitor, and the compound file of quinidine built in Simcyp was used as the default quinidine PBPK model.The input parameters and sources for the quinidine PBPK model are summarised in Table 2.
The developed quinidine PBPK model was verified by comparing the predicted values with the observed clinical values from a concomitant oral administration study of dextromethorphan and quinidine in healthy adults (Pope et al. 2004).The clinical study consisted of two parts: one involving 46 healthy adult EM of CYP2D6 who received 30 mg of dextromethorphan and 0-75 mg of quinidine twice daily for 7 days, and another involving 7 EM and 2 PM who received 30 mg of dextromethorphan and 30 mg of quinidine twice daily for 7 days.The compound file of dextromethorphan built in Simcyp was used as the dextromethorphan PBPK model for the simulation of concomitant administration.The input parameters and sources for the dextromethorphan PBPK model are summarised in Table 3.For the PBPK model verification, the Sim-Healthy Volunteers file built in Simcyp was used to create virtual populations of EM and PM by changing the gene frequency.Simulations were conducted with 200 virtual subjects in each group.To verify the PBPK model, the model-predicted C max and AUC were compared with the C max and AUC observed in the clinical study.We also developed a refined quinidine PBPK model by a parameter estimation approach that optimises the inhibition constant (Ki) value for CYP2D6 based on the observed C max and AUC.

Simulation of the PK of tipepidine in combination with quinidine
A simulation of concomitant administration was performed using the PBPK model developed for tipepidine and quinidine.The Sim-Japanese file built in Simcyp was used as the virtual Japanese population reflecting the gene frequency.Simulations were conducted with 200 virtual subjects in each group.The dosing regimen consisted of oral administration of 40 mg of tipepidine with or without 50 mg of quinidine once daily for 7 days.The PK profile of tipepidine was simulated after the last oral dose, and PK parameters such as C max , AUC, and t 1/2 were calculated.

Simulation of the PK of tipepidine in extended-release formulation
The PK profile of tipepidine assuming it was administered in hypothetical extended-release formulation was simulated by changing the absorption rate constant of tipepidine from 1.35 h −1 to 0.225 h −1 to mimic the PK profile of tipepidine when concomitantly administered with quinidine.The Sim-Japanese file built in Simcyp was used as the virtual Japanese population and used to create the virtual PM population of CYP2D6 by changing the gene frequency.
Simulations were conducted with 200 virtual subjects in each group.The dosing regimen consisted of oral administration of 40 mg of tipepidine with 50 mg of quinidine, or 320 mg of the extended-release form of tipepidine alone once daily for 7 days.The PK profile of tipepidine was simulated after the last oral dose, and PK parameters such as C max and AUC were calculated.

Development and verification of the tipepidine PBPK model
A simulation was performed using the PBPK model developed for tipepidine.Figure 1 shows the observed and PBPK model-predicted values and fold errors of PK parameters of tipepidine for each genotype of CYP2D6.Comparing each genotype, the C max and AUC of tipepidine were higher in the order of PM, IM, EM, and UM, with accurate predictions for all genotypes except UM.The AAFEs of C max and AUC of tipepidine were 1.85 and 1.93, respectively; since both parameters were 2 or less, it was considered that an appropriate model could be developed.

Development and verification of the quinidine PBPK model
The quinidine PBPK model was developed using the compound file of quinidine built in Simcyp as a default.To verify the PBPK model developed for quinidine including the Ki value for CYP2D6, we performed coadministration simulations of quinidine and dextromethorphan, which is metabolised mainly by CYP2D6.Figure 2 shows the observed and PBPK model-predicted values and fold errors of the PK parameters of quinidine.The C max and AUC of quinidine in each dose group were accurately predicted, and there was no difference in plasma exposure between EM and PM.The AAFEs of C max and AUC of quinidine were 1.14 and 1.21,  respectively; since both parameters were 2 or less, it was considered that an appropriate model could be developed for all the input parameters of quinidine except the Ki value for CYP2D6.Figure 3 shows the observed and PBPK model-predicted values and fold errors of PK parameters of dextromethorphan.As the concomitant dose of quinidine increased, the C max and AUC of dextromethorphan increased due to stronger inhibition of CYP2D6.However, using the default quinidine model, the C max and AUC of dextromethorphan in EM co-administered with quinidine were underpredicted, with AAFE values of 2.10 and 2.96, respectively.To solve the underprediction, we developed a refined quinidine PBPK model by the parameter estimation approach that optimises the Ki value for CYP2D6 based on the observed C max and AUC of dextromethorphan.By optimising the Ki value from the default value of 0.0119 lM (from Simcyp) to 0.0007 lM, which is a refined value obtained by the parameter estimation approach, the AAFEs of C max and AUC of dextromethorphan were reduced to 1.14 and 1.09, respectively.The results indicated that the refined model, compared to the default model, could accurately predict the inhibitory activity of quinidine on CYP2D6.Therefore, we decided to use the refined model.

Simulation of the effects of combination with quinidine on the PK of tipepidine
To quantitatively predict the effects of combination with quinidine on the PK of tipepidine, a simulation of concomitant administration was performed using the PBPK model developed for tipepidine and quinidine.Figure 4 shows the mean PBPK model-predicted plasma concentration time profiles of tipepidine in the Japanese population after the last oral dose of 40 mg of tipepidine with or without 50 mg of quinidine once daily for 7 days, and Table 4 shows the PBPK model-predicted PK parameters.The complete PK profiles for all 7 days are shown in Figure S1.The predicted plasma concentration of tipepidine did not accumulate significantly.The results of the simulation showed that concomitant administration of tipepidine with quinidine increased the predicted C max , AUC, and t 1/2 of tipepidine by 3.4-, 6.6-, and 2.4-fold, respectively.

Comparison with the extended-release formulation of tipepidine
To compare with another approach that aims to prolong the half-life of tipepidine, the PK profile of tipepidine administered in hypothetical extended-release form was also simulated.We conducted simulations of oral administration of 40 mg of tipepidine with 50 mg of quinidine, or 320 mg of the extended-release form of tipepidine once daily for 7 days, in the Japanese population and in the PM population of CYP2D6, assuming a high plasma exposure scenario.Figure 5 shows the mean PBPK model-predicted plasma concentration time profiles of tipepidine after the last oral dose of 7-days administration, and Table 5 shows the PBPK model-predicted PK parameters.The complete PK profiles for all 7 days are shown in Figure S2.The predicted plasma concentration of tipepidine did not accumulate significantly.There was little difference in predicted PK parameters of tipepidine when concomitantly administered with quinidine between the PM population and the Japanese population.On the other hand, the predicted C max and AUC of tipepidine when administered in extended-release form were 4.8-and 7.3-fold higher in the PM population than the Japanese population, respectively.

Discussion
In this study, to quantitatively predict the PK of tipepidine concomitantly administered with quinidine, we developed PBPK models for tipepidine and quinidine using Simcyp and performed simulations.The tipepidine PBPK model was developed using data obtained from the interview form for Asverin tablets (PMDA 2023) and our previous study (Hayashi et al. 2023).The tipepidine PBPK model was validated by comparing the predicted values with the observed clinical values from an oral administration study of tipepidine in Japanese subjects (Saito et al. 2020) (Figure 1).The distribution of CYP2D6 genotypes in this clinical study was as follows: one PM (2.4%), 5 IM (11.9%), 35 EM (83.3%), and one UM (2.4%), which was almost the same as the reported gene frequencies in the Japanese population (PM: 0.5%, IM: 18.3%, EM: 77.5%, UM: very rare) (Ebisawa et al. 2005).The AAFEs of C max and AUC of tipepidine were both 2 or less, confirming that an appropriate model could be developed.When comparing the data for each genotype, the fold error was very small for all genotypes except UM, but the fold error for UM was more than 10.The observed value for UM was based on the data from only one subject, so it is possible that the observed value deviated significantly from the actual average value.In addition, while several studies have reported that the difference in the metabolic activity of CYP2D6 between EM and UM is approximately 2-fold (Chiba et al. 2012;Nakamura et al. 2018;Frederiksen et al. 2021), the observed exposure difference between EM and UM in this clinical study was approximately 20-fold.These data suggest that factors other than being a CYP2D6 UM may have contributed to the markedly low plasma exposure in this subject.Although the reason for prediction inaccuracy in UM remains unclear, we considered that a sufficient model has been developed to evaluate the effects of CYP2D6 inhibition.
The quinidine PBPK model was developed using the compound file of quinidine built in Simcyp as a default, and the concomitant oral administration of dextromethorphan and quinidine in healthy adults was simulated for model verification.Since the AAFEs of C max and AUC of quinidine were 2 or less, an appropriate model was considered to be developed for the all the input parameters of quinidine except the Ki value for CYP2D6 (Figure 2).However, the AAFEs of C max and AUC of dextromethorphan exceeded 2, indicating that the simulation was not appropriate (Figure 3).Two  potential reasons for the incorrect simulation were considered: 1) the inappropriateness of the PBPK model for dextromethorphan and 2) inaccuracy of the Ki value for CYP2D6 in the PBPK model for quinidine.To verify the dextromethorphan PBPK model, we validated the C max and AUC of dextromethorphan in two different genotype groups that were not affected by quinidine: the EM group that received dextromethorphan alone, and the PM group that received both dextromethorphan and quinidine.The effect of quinidine on CYP2D6 activity in the PM group is negligible because of the inherent deficiency of CYP2D6.The C max and AUC of dextromethorphan in both genotype groups were accurately predicted, suggesting that the dextromethorphan PBPK model would be appropriate.Next, we verified the Ki value for CYP2D6 in the quinidine PBPK model.Based on the data from the EM group that were co-administered with quinidine, the C max and AUC of dextromethorphan were underpredicted, suggesting that the inhibitory effect of quinidine on CYP2D6 was underestimated.By setting Ki to its optimal value to 0.0007 lM using parameter estimates based on the observed clinical data, the AAFE values were reduced to 2 or less, and the model was refined with sufficient accuracy.The refined Ki value of 0.0007 lM is lower than the Ki values 0.008 to 0.04 lM, which were reported by a conventional method (Abdel-Rahman et al. 1999;VandenBrink et al. 2012).However, the Ki value obtained from an in vitro study using human hepatocytes suspended in plasma was reported as 0.0013 lM (Mao et al. 2012), which is close to our refined Ki value.They reported that the Ki value provided more accurate predictions of metabolic inhibition in vivo than the Ki values obtained using conventional protein-free methods.In addition, other studies using the protein-containing method similarly reported good prediction accuracy (Uchaipichat et al. 2006;Lu et al. 2008).Thus, these studies provide additional supports indicating that the refined Ki value would be acceptable.
By the simulation using the developed PBPK models for tipepidine and quinidine, it was demonstrated that concomitant administration of tipepidine with quinidine increased the predicted C max , AUC, and t 1/2 of tipepidine (Table 4).In addition, the predicted plasma concentration of tipepidine at 24 h after concomitant administration with quinidine exceeded that at 8 h (assuming a dosing frequency of three times per day) after administration of tipepidine alone.The result suggests that once-daily concomitant administration of tipepidine and quinidine could consistently exceed the effective plasma concentration and maintain therapeutic effects.
Furthermore, to compare with another approach that aims to prolong the half-life, the PK profile of tipepidine assuming it was administered in hypothetical extendedrelease form was simulated by changing the absorption rate constant and dosage of tipepidine.The predicted C max and AUC of tipepidine in the Japanese population when administered in hypothetical extended-release form were comparable to the C max and AUC of tipepidine when concomitantly administered with quinidine, but showed greater individual differences (Table 5).To elucidate the influence of CYP2D6 genotypes on the individual differences, further simulations were conducted in the PM population, assuming a high plasma exposure scenario.There was little difference in the predicted PK parameters of tipepidine when concomitantly administered with quinidine between the PM population and the Japanese population.The explanation for these results could be that the inhibition of CYP2D6 by quinidine makes all subjects in the Japanese population PM-like.In contrast, the predicted C max and AUC of tipepidine when administered in extended-release form were 4.8-and 7.3-fold higher in the PM population than the Japanese population, respectively.The results suggest the possibility of a marked increase in plasma exposure of tipepidine above the therapeutic range when administered in extended-release form in the PM population.Tipepidine has been used for over 60 years and has been generally considered as a safe drug when used at the recommended dosage.However, it has been reported that adverse effects such as delirium, excitement, and confusion may occur when plasma concentration are more than 10-fold higher than the normal therapeutic level (Imai et al. 2011).The reported lowest plasma concentration at which these adverse effects appeared is 527.7 ng/mL (Imai et al. 2011).Since the predicted plasma concentration of tipepidine in the PM population when administered in extended-release form exceeds 527.7 ng/mL, use of the extended-release form of tipepidine could potentially lead to the occurrence of adverse effects.Eliglustat is an example of a drug that may cause adverse effects due to individual differences in CYP2D6 genotypes (Balwani et al. 2016).To avoid serious adverse effects of eliglustat by adjusting the dosage according to the genotype, prior genetic testing is required.If prior genetic testing were also required for the extended-release form of tipepidine, it would negatively impact its usability.Since the predicted plasma concentration of tipepidine after concomitant administration with quinidine did not exceed 527.7 ng/ mL, the combination of tipepidine and quinidine, by reducing individual differences, would be safer to use than the extended-release form.
In this study, we selected quinidine as a candidate of CYP2D6 inhibitor, because it potently inhibits CYP2D6 at the concentrations lower than that exhibiting pharmacological effects in humans (VandenBrink et al. 2012).However, quinidine also exhibits P-glycoprotein inhibition at the same concentrations for CYP2D6 inhibition, so caution is required when concomitantly using with other drugs that are P-glycoprotein substrates.Other candidates for potent CYP2D6 inhibitors include paroxetine and bupropion.Since these drugs show sufficient CYP2D6 inhibition at the concentrations that exhibit pharmacological effects, their priority as a concomitant use would be low in this study.
In conclusion, the developed PBPK model was able to quantitatively simulate the feasibility of once-daily medication in combination use of tipepidine with quinidine.In addition, the simulation considering the genotype suggested that combination with quinidine could achieve safe repositioning of tipepidine for the treatment of psychiatric disorders.

Figure 1 .
Figure 1.Observed and PBPK model-predicted C max (A) and AUC (B) of tipepidine in healthy Japanese subjects after single oral dose of 40 mg of tipepidine.Each point represents the mean ± SD.PM, poor metabolizers; IM, intermediate metabolizers; EM, extensive metabolizers; UM, ultra-rapid metabolizers.The observed data were adapted from a reference (Saito et al. 2020).

Figure 2 .
Figure2.Observed and PBPK model-predicted C max (A) and AUC (B) of quinidine in extensive metabolizers (EM) or poor metabolizers (PM) after the last oral dose of 30 mg of dextromethorphan and 10 to 75 mg of quinidine (Q) twice daily for 7 days.Each point represents the mean.The observed data were adapted from a reference(Pope et al. 2004).a AUC 0-12h .

Figure 3 .
Figure 3. Observed and PBPK model-predicted C max (A) and AUC (B) of dextromethorphan in extensive metabolizers (EM) or poor metabolizers (PM) after the last oral dose of 30 mg of dextromethorphan (DM) and 0 to 75 mg of quinidine (Q) twice daily for 7 days.Each point represents the mean.The observed data were adapted from a reference (Pope et al. 2004).a AUC 0-12h .

Figure 4 .
Figure 4. Mean PBPK model-predicted plasma concentration time profiles of tipepidine in the Japanese population after the last oral dose of 40 mg of tipepidine with (dotted lines) or without (solid lines) 50 mg of quinidine once daily for 7 days.

Figure 5 .
Figure 5. Mean PBPK model-predicted plasma concentration time profiles of tipepidine after the last oral dose of 40 mg of tipepidine with 50 mg of quinidine (A), or 320 mg of the extended-release form of tipepidine (B) once daily for 7 days, in the Japanese population (solid lines) and the poor metabolizers population (dotted lines).

Table 1 .
Input parameters for the tipepidine PBPK model.V ss : volume of distribution at steady state; CL po : oral clearance; CL int : intrinsic clearance; HLM: human liver microsomes; IF: interview form; fm: fraction metabolised.

Table 2 .
Input parameters for the quinidine PBPK model.
PBPK: physiologically-based pharmacokinetic; MW: molecular weight; LogP; partition coefficient; pKa: acid dissociation constant; B:P: blood-to-plasma ratio; f u : unbound fraction in plasma; Fa: fraction absorbed; k a : absorption rate constant; V ss : volume of distribution at steady state; CL int : intrinsic clearance; CL r : renal clearance; Ki: inhibition constant.

Table 3 .
Input parameters for the dextromethorphan PBPK model.
PBPK: physiologically-based pharmacokinetic; MW: molecular weight; LogP: partition coefficient; pKa: acid dissociation constant; B:P: blood-to-plasma ratio; f u : unbound fraction in plasma; Fa: fraction absorbed; k a : absorption rate constant; V ss : volume of distribution at steady state; CL int : intrinsic clearance; CL r : renal clearance.

Table 4 .
PBPK model-predicted pharmacokinetic parameters of tipepidine in the Japanese population after the last oral dose of 40 mg of tipepidine with or without 50 mg of quinidine once daily for 7 days.