Ex) Article Title, Author, Keywords
Ex) Article Title, Author, Keywords
DTT 2024; 3(2): 140-148
Published online September 30, 2024
https://doi.org/10.58502/DTT.24.0004
Copyright © The Pharmaceutical Society of Korea.
Jee Sun Min1*, Chae Bin Lee1,2*, Sangyoung Lee1, Seong Jun Jo1, Da Hyun Kim1, Duk Yeon Kim1, Sabin Shin1, Soo Kyung Bae1
Correspondence to:Soo Kyung Bae, baesk@catholic.ac.kr
*The authors contributed equally to this work.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Lower respiratory tract infections are prevalent in South Korea, and clarithromycin, amoxicillin, clavulanate, and loxoprofen are frequently prescribed for these infections. In addition, co-administration of these small molecule drugs and Korean traditional medicine (KTM) are common. Despite a growing number of patients administering these drugs in combination, their drug-drug interaction (DDI) potentials have not been examined. Therefore, our goal was to assess the potential DDIs among clarithromycin, amoxicillin, clavulanate, and loxoprofen when administered together using physiologically based pharmacokinetic (PBPK) modeling approach. The PBPK models of the drugs were constructed and validated based on the previously reported studies using SimCYPTM. Subsequently, the DDI potentials were evaluated in scenarios with one drug as the substrate (clarithromycin 250 mg, amoxicillin 250 mg, clavulanate 125 mg, and loxoprofen 60 mg) and the others as inhibitors. The simulated pharmacokinetic parameters were validated against observed data, showing good agreement within predefined criteria (0.5-2.0). The observed data fell within the 5th to 95th confidence interval of the simulated data, indicating accurate description of plasma concentration-time profiles. Assessing DDI potentials through AUC ratios (with inhibitors/without inhibitors), the predicted AUC ratios ranged from 0.89 to 1.03, indicating no significant DDIs (criteria: 0.80-1.25). In summary, our study supports the safe use of a combination of clarithromycin, amoxicillin, clavulanate, and loxoprofen.
Keywordsclarithromycin, amoxicillin, clavulanate, loxoprofen, physiologically based pharmacokinetic modeling, drug-drug interactions
Respiratory tract infections are divided into upper respiratory tract infections and lower respiratory tract infections. Specifically, Lower respiratory tract infections, including pneumonia, acute bronchitis, and influenza, pose a significant burden in terms of mortality, morbidity, and economic impact (Safiri et al. 2023). Furthermore, in 2022, the number of patients diagnosed with lower respiratory tract infections in South Korea exceeded 10,000,000 (Health Insurance Review and Assessment Service [HIRA] 2022).
Current guidelines for lower respiratory tract infections recommend first-line treatment with a combination of penicillin and a β-lactamase inhibitor, along with macrolide antibiotics (Song et al. 2009; Woodhead et al. 2011). Notably, an increasing number of patients are using Korean traditional medicine in conjunction with conventional small molecule drugs. However, despite this growing trend, there are currently no guidelines available for managing potential interactions between these two treatment modalities.
Therefore, we aimed to evaluate drug-drug interactions (DDIs) between small molecule drugs and Korean traditional medicine for the treatment of lower respiratory tract infections. To achieve this, we have designed cocktails of small molecules, including clarithromycin, amoxicillin, clavulanate, and loxoprofen, which are among the most frequently prescribed medications for lower respiratory tract infections in South Korea (HIRA 2022).
The cocktail approach allows for the evaluation of DDIs mediated by multiple metabolic enzymes or transporters. However, certain conditions must be met to ensure that there are no interactions among the substrates (Food and Drug Administration [FDA] 2020A; Zhou et al. 2004).
Physiologically based pharmacokinetic (PBPK) modeling is one of the mechanistic methods employed to describe the pharmacokinetic properties of the drug by integrating drug related factors (e.g., physicochemical properties and intrinsic solubility) and physiology related factors. PBPK modeling serves as an assessment tool for DDIs, as recommended in the guidelines by the US Food and Drug Administration (FDA). It is widely utilized to assess the potential for DDIs without the need for clinical studies (FDA 2020B).
In this study, our primary objective was to evaluate the compatibility of a combination of cocktail drugs, namely clarithromycin, amoxicillin, clavulanate, and loxoprofen, by predicting pharmacokinetic DDI potentials among them using PBPK modeling.
PBPK models were developed using population-based PBPK software, SimCYPTM (SimCYP Ltd, a Certara company, Sheffield, UK, version 19). The simulations were conducted with a virtual healthy population implemented in SimCYPTM, and each simulation was performed in 10 trials for 10 healthy volunteers.
The clarithromycin compound file was developed by incorporating published information on its physicochemical properties, absorption, distribution, metabolism, and excretion. Briefly, the PBPK model of clarithromycin consists of the advanced dissolution, absorption, and metabolism (ADAM) model with full PBPK model (Table 1). Clarithromycin is a basic compound with a pKa value of 8.99, and its unbound fraction (fu) was set to 0.3 (Chu et al. 1993A). The absorption related parameters were defined by using in vitro Caco-2 permeability results and solubility in phosphate buffer, as reported in previous studies (Morakul et al. 2014; Togami et al. 2014). The volume of distribution at a steady state (Vss) was obtained from pharmacokinetic data in healthy subjects who received 250 mg of clarithromycin via intravenous infusion (Chu et al. 1992).
Table 1 Summary of input parameters used for the PBPK models of clarithromycin, amoxicillin, clavulanate and loxoprofen
Parameters (Units) | Clarithromycin | Amoxicillin | Clavulanate | Loxoprofen | Reference |
---|---|---|---|---|---|
Molecular weight (g/mol) | 747.950 | 365 | 199.16 | 246.302 | |
pKa | 8.99a | 2.8 (pKa1), 7.2 (pKa2), 9.55 (pKa3)b | 2.7 | 4.19 | aMcFarland et al. (1997) bTsuji et al. (1978) |
Log P | 2.30c | 0.87d | –1.07e | 0.82f | cLappin et al. (2011) dSangster (1994) eLakshminarayana et al. (2015) fSAJA Pharmaceuticals (2012) |
B/P | 1.83* | 0.55g | 1 | gMathialagan et al. (2017) | |
fu | 0.3h | 0.8 | 0.78i | 0.03c | hChu et al. (1993B) iWatson et al. (1987) cLappin et al. (2011) |
Absorption | ADAM | First-order | ADAM | First-order | |
ka (1/h) | 1.23j | 11.25f | jSpyker et al. (1977) fSAJA Pharmaceuticals (2012) | ||
Caco-2, Papp (10–6 cm/s) | 3.5k,l | 37.5e | 30m | kTogami et al. (2014) lMorakul et al. (2014) eLakshminarayana et al. (2015) mNarumi et al. (2016) | |
Distribution | Full | Full | Full | Minimal | |
Vss (L/kg) | 1.75n | 0.346j,o | 0.17p | 0.16 | nChu et al. (1992) jSpyker et al. (1977) oZarowny et al. (1974) pBolton et al. (1986) |
Kp scalar | 0.573* | 0.55* | 0.65 | 0.33 | |
Elimination | Enzyme kinetics | In vivo clearance | Enzyme kinetics | ||
CYP3A4 | |||||
Km (pmol/min/mg protein) | 48.7 (14-hydroxy)q 59.0 (N-desmethyl)q | qRodrigues et al. (1997) | |||
Vmax (μM) | 206 (14-hydroxy)q 189 (N-desmethyl)q | qRodrigues et al. (1997) | |||
Additional HLM CLint (μL/min/mg protein) | 36.1r | 1.46p | 264s | rSuzuki et al. (2003) pBolton et al. (1986) sShrestha et al. (2018) | |
IV clearance (L/h) | 25.6j,o | jSpyker et al. (1977) oZarowny et al. (1974) | |||
CLR (L/h) | 7.2t | 8j | 6.06p | 0.062u | tDavey (1991) jSpyker et al. (1977) pBolton et al. (1986) uKim et al. (2002) |
Interactions | |||||
CYP3A4 | |||||
KI (μM) | 29.5v | vPolasek & Miners (2006) | |||
kinact (1/min) | 0.05v | vPolasek & Miners (2006) | |||
Ki (μM) | OATP1B1 (48)w OATP1B3 (32)w P-gp (4.1)y | CYP2C8 (415)x | wSeithel et al. (2007) xNiwa et al. (2016) yEberl et al. (2007) |
ADAM, advanced dissolution, absorption, and metabolism; Log P, octanol: water partition coefficient; pKa, acid dissociation constant; B/P, blood-to-plasma ratio; fu, unbound fraction; ka, first-order absorption rate constant; Papp, apparent permeability; Vss, volume of distribution at steady state; Kp, tissue to plasma coefficient; Km, Michaelis-Menten constant; Vmax, maximum rate of metabolism; HLM, human liver microsome; KI, apparent inactivation constant; kinact, maximum inactivation rate constant; Ki, inhibition constant.
*Predicted value using SimCYPTM prediction tool.
The elimination process of clarithromycin was described by using enzyme kinetics. Metabolism of clarithromycin is primarily mediated by CYP3A4, resulting in the formation of 14-hydroxy clarithromycin and N-desmethyl clarithromycin (Rodrigues et al. 1997). Based on these information, in vitro enzyme kinetic data from Suzuki et al. (2003) were used to calculate intrinsic clearance of clarithromycin in the human liver microsomes and optimized to capture the observed data. Notably, clarithromycin is recognized as a time-dependent inhibitor (TDI) of CYP3A4 and an inhibitor of P-gp, OATP1B1 and OATP1B3 transporters (Polasek and Miners 2006; Eberl et al. 2007; Seithel et al. 2007). To evaluate DDI potentials by using PBPK modeling, we utilized these in vitro DDI data obtained from previous literature.
The amoxicillin compound file was developed by incorporating previously published information on its physicochemical properties, absorption, distribution, metabolism, and excretion. Briefly, the amoxicillin model consists of first-order absorption model with full PBPK model (Table 1). Physicochemical properties of amoxicillin were obtained from previous studies (Tsuji et al. 1978). We employed middle-out approach to estimate absorption of amoxicillin. Absorption rate constant (ka) and lag-time values were obtained from pharmacokinetic data in healthy subjects who receive 250, 500, 1,000 mg of amoxicillin via intravenous and oral administration (Zarowny et al. 1974; Spyker et al. 1977). The Vss value was estimated from the previously mentioned pharmacokinetic data, along with estimated Kp values based on the Log p-value. The elimination process was defined by in vivo clearance using pharmacokinetic data in healthy subjects who receive amoxicillin via intravenous administration (Zarowny et al. 1974; Spyker et al. 1977). Additionally, amoxicillin was reported to exhibit weak inhibition of CYP2C8 with an IC50 value of 0.83 mM, and incorporated into the amoxicillin compound file (Niwa et al. 2016).
The clavulanate PBPK model consists of ADAM model with full PBPK model (Table 1). To define the absorption process, the Caco-2 permeability results from previous study were used (Lakshminarayana et al. 2015). The Vss value was determined from pharmacokinetic data, along with estimated Kp values based on the Log p-value (Bolton et al. 1986). The enzyme kinetic module was used to describe elimination process with the intrinsic clearance value calculated through retrograde prediction method implemented in SimCYPTM, utilizing human pharmacokinetic data obtained from intravenous administration.
The parameters for loxoprofen PBPK model are shown in the Table 1. To briefly summarize, absorption of loxoprofen was defined by first-order absorption model, incorporating Caco-2 permeability data from a prior study (Narumi et al. 2016). The distribution of loxoprofen was described by using a minimal PBPK model. As no intravenous data were available, the Vss value was optimized based on clinical results from oral administration. The primary metabolic pathway for loxoprofen involves carboxyl reductase and CYP3A4 mediated metabolism is negligible (Sawamura et al. 2015). Therefore, we determined the intrinsic clearance of loxoprofen by using in vitro human liver microsome and cytosol clearance data according to the previous study (Shrestha et al. 2018).
To verify the ability of the PBPK models, simulations were carried out with the same clinical trial designs including age range, gender proportion, and dosage, as the observed data (Table 2). The predicted plasma concentration profiles of the drug were compared with the observed data, which were extracted from the literature using GetData Graph Digitizer (version 2.26, S. Fedorov). The criteria for comparing the observed and simulated data involved ensuring that the observed plasma concentration profiles fell within the 5th and 95th percentiles of the simulated data (Ezuruike et al. 2022). Furthermore, predicted pharmacokinetic parameters were compared, including area under the curve (AUC) and maximum plasma concentration (Cmax) values, by calculating the ratio of the mean observed to mean predicted values. We predefined successful model performance as achieving ratio of AUC and Cmax within a two-fold range (0.5-2.0), in line with previous studies (Galetin et al. 2005; Wang 2010).
Table 2 Observed data of clarithromycin, amoxicillin, clavulanate, and loxoprofen for the PBPK model verification
Substrates | Route of administration | Dose (mg) | Males (%) | n | Age ranges (mean) (years) | Weight ranges (mean) (kg) | Reference |
---|---|---|---|---|---|---|---|
Clarithromycin | IV infusion | 250 | 100 | 19 | 18-40 (29.0) | 58-88 (71.5) | Chu et al. (1992) |
PO | 250 | 100 | 17 | 18-40 (29.0) | 58-88 (70.8) | Chu et al. (1993A) | |
PO | 250 | 100 | 12 | 24-38 (26.5) | 65-88 (79.5) | Kees et al. (1995) | |
Amoxicillin | IV infusion | 250 | 100 | 8 | 20-30 | 59-91 (74.5) | Zarowny et al. (1974) |
PO | 250 | 100 | 8 | 20-30 | 59-91 (74.5) | Zarowny et al. (1974) | |
PO | 250 | 100 | 24 | 18-32 (23.5) | 62-82 (70.1) | Idkaidek et al. (2004) | |
PO | 250 | 100 | 12 | 20-25 (22.4) | 60-76 (66.2) | Zhang et al. (2013) | |
Clavulanate | PO | 125 | 50 | 12 | (26.1) | 58.0 (F), 69.0 (M) | Adam et al. (1982) |
PO | 125 | 100 | 4 | 45-53 | 70-82 | Bolton et al. (1986) | |
Loxoprofen | PO | 60 | 100 | 6 | 24-28 | 58-70 | Kim et al. (2002) |
PO | 60 | 100 | 14 | 21-26 | 43-86 | Choi et al. (1998) | |
PO | 60 | NA | 16 | NA | NA | Insert paper |
IV, intra venous; PO, per os; NA, not applicable.
To predict the DDI potentials among clarithromycin, amoxicillin, clavulanate, and loxoprofen, a simulation was conducted as follows: set 1 (substrate: clarithromycin 250 mg, inhibitor 1: amoxicillin 250 mg, inhibitor 2: clavulanate: 125 mg and inhibitor 3: loxoprofen 60 mg), set 2 (substrate: amoxicillin 250 mg, inhibitor 1: clarithromycin 250 mg, inhibitor 2: clavulanate: 125 mg and inhibitor 3: loxoprofen 60 mg), set 3 (substrate: clavulanate: 125 mg, inhibitor 1: clarithromycin 250 mg, inhibitor 2: amoxicillin 250 mg and inhibitor 3: loxoprofen 60 mg) and set 4 (substrate: loxoprofen 60 mg, inhibitor 1: clarithromycin 250 mg, inhibitor 2: amoxicillin 250 mg and inhibitor 3: clavulanate: 125 mg). To evaluate DDI potentials among 4 compounds, we used AUC ratios (with/without co-medication). As recommended in the FDA guidance, if the AUC ratios are greater than 1.25 or less than 0.8, it suggests the potential for DDIs mediated by metabolic enzymes or transporters (FDA 2020B).
The validity of the developed clarithromycin PBPK model was confirmed using data from previous literature. Fig. 1 shows that the predicted plasma concentration-time profiles closely captured the observed profiles for both 250 mg of IV infusion and 250 mg of oral clarithromycin administration. Additionally, the confidence intervals of the predicted plasma concentration-time profiles provided good coverage of the observed data. The ratios of the predicted to observed Cmax and AUC values were well within the two-fold criteria range (0.76-1.17), demonstrating the successful validation of the clarithromycin PBPK model (Table 3).
Table 3 Predicted and observed PK parameters of clarithromycin, amoxicillin, clavulanate, and loxoprofen
Substrates | Route of administration | Dose (mg) | Observed mean ± SD | Predicted mean ± SD | Fold-error (predicted/observed) | |||
---|---|---|---|---|---|---|---|---|
Cmax (μg/mL) | AUC0−t (μg/mL·h) | Cmax (μg/mL) | AUC0−t (μg/mL·h) | Cmax | AUC0−t | |||
Clarithromycin | IV infusion | 250 | 2.78 ± 0.5 | 8.41 ± 2.44 | 3.26 | 8.08 | 1.17 | 0.76 |
PO | 250 | 0.85 | 4.71 | 0.72 | 5.39 | 0.86 | 1.14 | |
Amoxicillin | IV | 250 | NA | 10.4 ± 0.34 | NA | 11.1 | NA | 1.07 |
PO | 250 | 4.35 | 10.6 | 3.27 | 10.4 | 0.75 | 0.98 | |
Clavulanate | PO | 125 | 3.23 | 6.20 | 3.64 | 11.34 | 1.13 | 1.83 |
Loxoprofen | PO | 60 | 5.51 | 7.98 | 4.35 | 4.75 | 0.79 | 0.60 |
IV, intra venous; PO, per os; NA, not applicable.
The developed amoxicillin PBPK model was verified using previously reported clinical pharmacokinetic data for both IV infusion and oral administration of 250 mg amoxicillin. As depicted in Fig. 1, the observed data fell within the 95th confidence interval of the simulated plasma concentration-time profiles, indicating that the simulation results well captured the observed data. Therefore, it is concluded that the developed PBPK model accurately describes the pharmacokinetic properties of amoxicillin. Moreover, the ratios of predicted to observed Cmax and AUC values were well within the two-fold criteria range (0.75-1.07).
The initial clavulanate PBPK model was optimized for oral administration and subsequently validated using clinical pharmacokinetic data from a previous study involving a 125 mg oral dose of clavulanate. As illustrated in Fig. 1, the simulated plasma concentration-time profile closely aligned with the observed data, within the 95th confidence interval of the simulation. Additionally, the ratio of predicted to observed Cmax and AUC values met the two-fold criteria range (1.13 and 1.83, respectively), indicating the successful validation of the clavulanate PBPK model.
Fig. 1 presents the simulated plasma concentration-time profiles and observed data following the oral administration of 60 mg of loxoprofen. The observed data aligns well within the 95th confidence interval of the simulation, indicating an accurate representation of loxoprofen pharmacokinetic profiles by the PBPK model. Additionally, the ratios of predicted to observed Cmax and AUC values are well within the two-fold criteria range (0.79 and 0.60, respectively), further confirming the model’s validity.
As described in the Table 4, DDI potentials were assessed by calculating the AUC ratio (AUCR). For set 1 (substrate: clarithromycin), the predicted mean AUC ratio for clarithromycin was 0.89, falling within the range of 0.8-1.25. For set 2 (substrate: amoxicillin), the predicted mean AUC ratio of amoxicillin was 0.99. In set 3 (substrate: clavulanate), the predicted mean AUC ratio of clavulanate was 1.03, and in set 4 (substrate: loxoprofen), the predicted mean AUC ratio of loxoprofen is 1.01. These AUCR values, when comparing single treatment to co-treatment with inhibitors, all fell within the range of 0.8-1.25. These results indicate that there were no significant DDIs among clarithromycin, amoxicillin, clavulanate and loxoprofen.
Table 4 The predicted AUCs and AUCR of clarithromycin, amoxicillin, clavulanate, and loxoprofen after oral doses of each drug either individually or in combination, including clarithromycin 250 mg, amoxicillin 250 mg, clavulanate 125 mg, and loxoprofen 60 mg
Substrates | Dose (mg) | AUC0−t (μg/mL·h) | ||
---|---|---|---|---|
Without co-administration (a) | Co-administration (b) | AUCR (b/a) | ||
Clarithromycin | 250 | 7.14 ± 2.42 | 6.36 ± 2.31 | 0.89 |
Amoxicillin | 250 | 7.87 ± 2.04 | 7.79 ± 2.56 | 0.99 |
Clavulanate | 125 | 11.3 ± 0.71 | 11.6 ± 0.67 | 1.03 |
Loxoprofen | 60 | 4.68 ± 2.55 | 4.72 ± 3.23 | 1.01 |
AUC, area under the curve; AUCR, area under the curve ratio.
This study evaluated the potential DDIs among clarithromycin, amoxicillin, clavulanate, and loxoprofen to determine their suitability as cocktail drugs for a clinical trial.
The clarithromycin PBPK model was constructed using ADAM model and full PBPK model, including in vitro enzyme kinetic parameters. Specifically, clarithromycin is known as a substrate and a time-dependent inhibitor of CYP3A4 (Polasek and Miners 2006). The clarithromycin PBPK model was verified using various observed data, including scenarios such as single and multiple doses administration, and DDI with amoxicillin (data are not shown). The clarithromycin PBPK model accurately described the pharmacokinetic properties of clarithromycin after various dosing scenarios. The amoxicillin PBPK model was optimized using first-order absorption model and full PBPK model. The elimination of amoxicillin was estimated using in vivo clearance module because amoxicillin was mainly eliminated by renal excretion unlike clarithromycin (Zarowny et al. 1974). The developed amoxicillin PBPK model reasonably described its pharmacokinetic characteristics when compared with the observed pharmacokinetic data. Additionally, the PBPK models for clavulanate and loxoprofen were well-validated by comparing them to the observed data.
To assess the suitability of these drugs as cocktail components for clinical trials, we conducted simulations involving four different scenarios to evaluate potential DDIs. The simulated results showed no significant DDI potentials, as the AUCR remained within the acceptable criteria range of 0.8 to 1.25. Clarithromycin is one of the potent CYP3A4 inhibitors explicitly mentioned in the FDA guidelines (FDA 2023). Consequently, numerous studies have investigated DDI potentials involving clarithromycin as a perpetrator. Furthermore, clarithromycin was shown to inhibit various drug transporters, including OATP1B1, OATP1B3, and P-gp, as demonstrated by previous research (Eberl et al. 2007; Seithel et al. 2007). Despite its inhibitory effects on multiple metabolic enzymes and transporters, the simulation results did not reveal any significant DDIs when clarithromycin was employed as a perpetrator. This can be attributed to the fact that amoxicillin, clavulanate, and loxoprofen were not the substrates of those metabolic enzymes and transporters. Furthermore, several studies have demonstrated the absence of significant DDIs between clarithromycin and amoxicillin, which are commonly co-administered for Helicobacter pylori eradication, consistent with our simulation results (Mainz et al. 2002; Jin et al. 2018; Liang et al. 2023).
Amoxicillin and clavulanic acid are commonly used in combination as an antibacterial agent. In this study, we observed no significant pharmacokinetic differences when they were administered as either victims or perpetrators. These results align with previous literature, which reported no alterations in the pharmacokinetics of either drug when compared with separate administration (Todd and Benfield 1990). Additionally, amoxicillin exhibits a weak inhibitory effect on CYP2C8 (Niwa et al. 2016). However, among the other drugs tested, none of the drugs were substrates of CYP2C8, resulting in no significant DDI potentials mediated by amoxicillin.
Like other drugs, loxoprofen exhibited no changes in its pharmacokinetic properties when administered as either a victim or a perpetrator in this study. A previous study by Uwai et al. (2004) suggested that loxoprofen and its trans-OH metabolite inhibit human organic anion transporters, specifically OAT1 and OAT3, which are involved in renal tubular secretion. Furthermore, Mathialagan et al. (2017) demonstrated that amoxicillin undergoes renal clearance via OAT1 and OAT3. Inhibition of these renal transporters could potentially reduce drug elimination, making the evaluation of DDI potentials mediated by renal transporters an essential consideration. One limitation of our study is that we did not incorporate renal transporter kinetics into the loxoprofen and amoxicillin PBPK models, especially in terms of renal clearance due to lack of the in vitro kinetic data. Although Uwai et al. (2004) demonstrated IC50 values of loxoprofen and its metabolite using methotrexate as substrates for OAT1 and OAT3, methotrexate exhibited low-affinity for OAT1 with its high Km values. Thus, in vitro inhibitory data using known substrate might be required to investigate inhibitory effect of loxoprofen (FDA 2020A). Therefore, further investigation is necessary to assess DDIs mediated by renal transporters, particularly the inhibitory effect of loxoprofen.
Another limitation of this study is that lack of the verification of PBPK model for amoxicillin in DDI evaluation. Although amoxicillin exhibited an inhibitory effect on CYP2C8, the amoxicillin PBPK model was not verified for DDI studies, only for single administration.
In conclusion, we successfully developed PBPK models for clarithromycin, amoxicillin, clavulanate, and loxoprofen to evaluate DDI potentials. It is expected that significant pharmacokinetic DDIs are unlikely to occur and pharmacodynamic DDIs have not been reported currently, suggesting co-administration of these drugs is safe. Moreover, the cocktail method combining these drugs can be further applied to evaluate DDIs between the cocktail drugs and the Korean traditional medicine for respiratory infections.
The authors declare that they have no conflict of interest.
This study was supported by a grant of Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare (HF20C0002), Republic of Korea.
DTT 2024; 3(2): 140-148
Published online September 30, 2024 https://doi.org/10.58502/DTT.24.0004
Copyright © The Pharmaceutical Society of Korea.
Jee Sun Min1*, Chae Bin Lee1,2*, Sangyoung Lee1, Seong Jun Jo1, Da Hyun Kim1, Duk Yeon Kim1, Sabin Shin1, Soo Kyung Bae1
1College of Pharmacy and Integrated Research Institute of Pharmaceutical Sciences, The Catholic University of Korea, Bucheon, Korea
2Johns Hopkins Drug Discovery, Department of Neurology, Johns Hopkins School of Medicine, Baltimore, MD, USA
Correspondence to:Soo Kyung Bae, baesk@catholic.ac.kr
*The authors contributed equally to this work.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Lower respiratory tract infections are prevalent in South Korea, and clarithromycin, amoxicillin, clavulanate, and loxoprofen are frequently prescribed for these infections. In addition, co-administration of these small molecule drugs and Korean traditional medicine (KTM) are common. Despite a growing number of patients administering these drugs in combination, their drug-drug interaction (DDI) potentials have not been examined. Therefore, our goal was to assess the potential DDIs among clarithromycin, amoxicillin, clavulanate, and loxoprofen when administered together using physiologically based pharmacokinetic (PBPK) modeling approach. The PBPK models of the drugs were constructed and validated based on the previously reported studies using SimCYPTM. Subsequently, the DDI potentials were evaluated in scenarios with one drug as the substrate (clarithromycin 250 mg, amoxicillin 250 mg, clavulanate 125 mg, and loxoprofen 60 mg) and the others as inhibitors. The simulated pharmacokinetic parameters were validated against observed data, showing good agreement within predefined criteria (0.5-2.0). The observed data fell within the 5th to 95th confidence interval of the simulated data, indicating accurate description of plasma concentration-time profiles. Assessing DDI potentials through AUC ratios (with inhibitors/without inhibitors), the predicted AUC ratios ranged from 0.89 to 1.03, indicating no significant DDIs (criteria: 0.80-1.25). In summary, our study supports the safe use of a combination of clarithromycin, amoxicillin, clavulanate, and loxoprofen.
Keywords: clarithromycin, amoxicillin, clavulanate, loxoprofen, physiologically based pharmacokinetic modeling, drug-drug interactions
Respiratory tract infections are divided into upper respiratory tract infections and lower respiratory tract infections. Specifically, Lower respiratory tract infections, including pneumonia, acute bronchitis, and influenza, pose a significant burden in terms of mortality, morbidity, and economic impact (Safiri et al. 2023). Furthermore, in 2022, the number of patients diagnosed with lower respiratory tract infections in South Korea exceeded 10,000,000 (Health Insurance Review and Assessment Service [HIRA] 2022).
Current guidelines for lower respiratory tract infections recommend first-line treatment with a combination of penicillin and a β-lactamase inhibitor, along with macrolide antibiotics (Song et al. 2009; Woodhead et al. 2011). Notably, an increasing number of patients are using Korean traditional medicine in conjunction with conventional small molecule drugs. However, despite this growing trend, there are currently no guidelines available for managing potential interactions between these two treatment modalities.
Therefore, we aimed to evaluate drug-drug interactions (DDIs) between small molecule drugs and Korean traditional medicine for the treatment of lower respiratory tract infections. To achieve this, we have designed cocktails of small molecules, including clarithromycin, amoxicillin, clavulanate, and loxoprofen, which are among the most frequently prescribed medications for lower respiratory tract infections in South Korea (HIRA 2022).
The cocktail approach allows for the evaluation of DDIs mediated by multiple metabolic enzymes or transporters. However, certain conditions must be met to ensure that there are no interactions among the substrates (Food and Drug Administration [FDA] 2020A; Zhou et al. 2004).
Physiologically based pharmacokinetic (PBPK) modeling is one of the mechanistic methods employed to describe the pharmacokinetic properties of the drug by integrating drug related factors (e.g., physicochemical properties and intrinsic solubility) and physiology related factors. PBPK modeling serves as an assessment tool for DDIs, as recommended in the guidelines by the US Food and Drug Administration (FDA). It is widely utilized to assess the potential for DDIs without the need for clinical studies (FDA 2020B).
In this study, our primary objective was to evaluate the compatibility of a combination of cocktail drugs, namely clarithromycin, amoxicillin, clavulanate, and loxoprofen, by predicting pharmacokinetic DDI potentials among them using PBPK modeling.
PBPK models were developed using population-based PBPK software, SimCYPTM (SimCYP Ltd, a Certara company, Sheffield, UK, version 19). The simulations were conducted with a virtual healthy population implemented in SimCYPTM, and each simulation was performed in 10 trials for 10 healthy volunteers.
The clarithromycin compound file was developed by incorporating published information on its physicochemical properties, absorption, distribution, metabolism, and excretion. Briefly, the PBPK model of clarithromycin consists of the advanced dissolution, absorption, and metabolism (ADAM) model with full PBPK model (Table 1). Clarithromycin is a basic compound with a pKa value of 8.99, and its unbound fraction (fu) was set to 0.3 (Chu et al. 1993A). The absorption related parameters were defined by using in vitro Caco-2 permeability results and solubility in phosphate buffer, as reported in previous studies (Morakul et al. 2014; Togami et al. 2014). The volume of distribution at a steady state (Vss) was obtained from pharmacokinetic data in healthy subjects who received 250 mg of clarithromycin via intravenous infusion (Chu et al. 1992).
Table 1 . Summary of input parameters used for the PBPK models of clarithromycin, amoxicillin, clavulanate and loxoprofen.
Parameters (Units) | Clarithromycin | Amoxicillin | Clavulanate | Loxoprofen | Reference |
---|---|---|---|---|---|
Molecular weight (g/mol) | 747.950 | 365 | 199.16 | 246.302 | |
pKa | 8.99a | 2.8 (pKa1), 7.2 (pKa2), 9.55 (pKa3)b | 2.7 | 4.19 | aMcFarland et al. (1997) bTsuji et al. (1978) |
Log P | 2.30c | 0.87d | –1.07e | 0.82f | cLappin et al. (2011) dSangster (1994) eLakshminarayana et al. (2015) fSAJA Pharmaceuticals (2012) |
B/P | 1.83* | 0.55g | 1 | gMathialagan et al. (2017) | |
fu | 0.3h | 0.8 | 0.78i | 0.03c | hChu et al. (1993B) iWatson et al. (1987) cLappin et al. (2011) |
Absorption | ADAM | First-order | ADAM | First-order | |
ka (1/h) | 1.23j | 11.25f | jSpyker et al. (1977) fSAJA Pharmaceuticals (2012) | ||
Caco-2, Papp (10–6 cm/s) | 3.5k,l | 37.5e | 30m | kTogami et al. (2014) lMorakul et al. (2014) eLakshminarayana et al. (2015) mNarumi et al. (2016) | |
Distribution | Full | Full | Full | Minimal | |
Vss (L/kg) | 1.75n | 0.346j,o | 0.17p | 0.16 | nChu et al. (1992) jSpyker et al. (1977) oZarowny et al. (1974) pBolton et al. (1986) |
Kp scalar | 0.573* | 0.55* | 0.65 | 0.33 | |
Elimination | Enzyme kinetics | In vivo clearance | Enzyme kinetics | ||
CYP3A4 | |||||
Km (pmol/min/mg protein) | 48.7 (14-hydroxy)q 59.0 (N-desmethyl)q | qRodrigues et al. (1997) | |||
Vmax (μM) | 206 (14-hydroxy)q 189 (N-desmethyl)q | qRodrigues et al. (1997) | |||
Additional HLM CLint (μL/min/mg protein) | 36.1r | 1.46p | 264s | rSuzuki et al. (2003) pBolton et al. (1986) sShrestha et al. (2018) | |
IV clearance (L/h) | 25.6j,o | jSpyker et al. (1977) oZarowny et al. (1974) | |||
CLR (L/h) | 7.2t | 8j | 6.06p | 0.062u | tDavey (1991) jSpyker et al. (1977) pBolton et al. (1986) uKim et al. (2002) |
Interactions | |||||
CYP3A4 | |||||
KI (μM) | 29.5v | vPolasek & Miners (2006) | |||
kinact (1/min) | 0.05v | vPolasek & Miners (2006) | |||
Ki (μM) | OATP1B1 (48)w OATP1B3 (32)w P-gp (4.1)y | CYP2C8 (415)x | wSeithel et al. (2007) xNiwa et al. (2016) yEberl et al. (2007) |
ADAM, advanced dissolution, absorption, and metabolism; Log P, octanol: water partition coefficient; pKa, acid dissociation constant; B/P, blood-to-plasma ratio; fu, unbound fraction; ka, first-order absorption rate constant; Papp, apparent permeability; Vss, volume of distribution at steady state; Kp, tissue to plasma coefficient; Km, Michaelis-Menten constant; Vmax, maximum rate of metabolism; HLM, human liver microsome; KI, apparent inactivation constant; kinact, maximum inactivation rate constant; Ki, inhibition constant..
*Predicted value using SimCYPTM prediction tool..
The elimination process of clarithromycin was described by using enzyme kinetics. Metabolism of clarithromycin is primarily mediated by CYP3A4, resulting in the formation of 14-hydroxy clarithromycin and N-desmethyl clarithromycin (Rodrigues et al. 1997). Based on these information, in vitro enzyme kinetic data from Suzuki et al. (2003) were used to calculate intrinsic clearance of clarithromycin in the human liver microsomes and optimized to capture the observed data. Notably, clarithromycin is recognized as a time-dependent inhibitor (TDI) of CYP3A4 and an inhibitor of P-gp, OATP1B1 and OATP1B3 transporters (Polasek and Miners 2006; Eberl et al. 2007; Seithel et al. 2007). To evaluate DDI potentials by using PBPK modeling, we utilized these in vitro DDI data obtained from previous literature.
The amoxicillin compound file was developed by incorporating previously published information on its physicochemical properties, absorption, distribution, metabolism, and excretion. Briefly, the amoxicillin model consists of first-order absorption model with full PBPK model (Table 1). Physicochemical properties of amoxicillin were obtained from previous studies (Tsuji et al. 1978). We employed middle-out approach to estimate absorption of amoxicillin. Absorption rate constant (ka) and lag-time values were obtained from pharmacokinetic data in healthy subjects who receive 250, 500, 1,000 mg of amoxicillin via intravenous and oral administration (Zarowny et al. 1974; Spyker et al. 1977). The Vss value was estimated from the previously mentioned pharmacokinetic data, along with estimated Kp values based on the Log p-value. The elimination process was defined by in vivo clearance using pharmacokinetic data in healthy subjects who receive amoxicillin via intravenous administration (Zarowny et al. 1974; Spyker et al. 1977). Additionally, amoxicillin was reported to exhibit weak inhibition of CYP2C8 with an IC50 value of 0.83 mM, and incorporated into the amoxicillin compound file (Niwa et al. 2016).
The clavulanate PBPK model consists of ADAM model with full PBPK model (Table 1). To define the absorption process, the Caco-2 permeability results from previous study were used (Lakshminarayana et al. 2015). The Vss value was determined from pharmacokinetic data, along with estimated Kp values based on the Log p-value (Bolton et al. 1986). The enzyme kinetic module was used to describe elimination process with the intrinsic clearance value calculated through retrograde prediction method implemented in SimCYPTM, utilizing human pharmacokinetic data obtained from intravenous administration.
The parameters for loxoprofen PBPK model are shown in the Table 1. To briefly summarize, absorption of loxoprofen was defined by first-order absorption model, incorporating Caco-2 permeability data from a prior study (Narumi et al. 2016). The distribution of loxoprofen was described by using a minimal PBPK model. As no intravenous data were available, the Vss value was optimized based on clinical results from oral administration. The primary metabolic pathway for loxoprofen involves carboxyl reductase and CYP3A4 mediated metabolism is negligible (Sawamura et al. 2015). Therefore, we determined the intrinsic clearance of loxoprofen by using in vitro human liver microsome and cytosol clearance data according to the previous study (Shrestha et al. 2018).
To verify the ability of the PBPK models, simulations were carried out with the same clinical trial designs including age range, gender proportion, and dosage, as the observed data (Table 2). The predicted plasma concentration profiles of the drug were compared with the observed data, which were extracted from the literature using GetData Graph Digitizer (version 2.26, S. Fedorov). The criteria for comparing the observed and simulated data involved ensuring that the observed plasma concentration profiles fell within the 5th and 95th percentiles of the simulated data (Ezuruike et al. 2022). Furthermore, predicted pharmacokinetic parameters were compared, including area under the curve (AUC) and maximum plasma concentration (Cmax) values, by calculating the ratio of the mean observed to mean predicted values. We predefined successful model performance as achieving ratio of AUC and Cmax within a two-fold range (0.5-2.0), in line with previous studies (Galetin et al. 2005; Wang 2010).
Table 2 . Observed data of clarithromycin, amoxicillin, clavulanate, and loxoprofen for the PBPK model verification.
Substrates | Route of administration | Dose (mg) | Males (%) | n | Age ranges (mean) (years) | Weight ranges (mean) (kg) | Reference |
---|---|---|---|---|---|---|---|
Clarithromycin | IV infusion | 250 | 100 | 19 | 18-40 (29.0) | 58-88 (71.5) | Chu et al. (1992) |
PO | 250 | 100 | 17 | 18-40 (29.0) | 58-88 (70.8) | Chu et al. (1993A) | |
PO | 250 | 100 | 12 | 24-38 (26.5) | 65-88 (79.5) | Kees et al. (1995) | |
Amoxicillin | IV infusion | 250 | 100 | 8 | 20-30 | 59-91 (74.5) | Zarowny et al. (1974) |
PO | 250 | 100 | 8 | 20-30 | 59-91 (74.5) | Zarowny et al. (1974) | |
PO | 250 | 100 | 24 | 18-32 (23.5) | 62-82 (70.1) | Idkaidek et al. (2004) | |
PO | 250 | 100 | 12 | 20-25 (22.4) | 60-76 (66.2) | Zhang et al. (2013) | |
Clavulanate | PO | 125 | 50 | 12 | (26.1) | 58.0 (F), 69.0 (M) | Adam et al. (1982) |
PO | 125 | 100 | 4 | 45-53 | 70-82 | Bolton et al. (1986) | |
Loxoprofen | PO | 60 | 100 | 6 | 24-28 | 58-70 | Kim et al. (2002) |
PO | 60 | 100 | 14 | 21-26 | 43-86 | Choi et al. (1998) | |
PO | 60 | NA | 16 | NA | NA | Insert paper |
IV, intra venous; PO, per os; NA, not applicable..
To predict the DDI potentials among clarithromycin, amoxicillin, clavulanate, and loxoprofen, a simulation was conducted as follows: set 1 (substrate: clarithromycin 250 mg, inhibitor 1: amoxicillin 250 mg, inhibitor 2: clavulanate: 125 mg and inhibitor 3: loxoprofen 60 mg), set 2 (substrate: amoxicillin 250 mg, inhibitor 1: clarithromycin 250 mg, inhibitor 2: clavulanate: 125 mg and inhibitor 3: loxoprofen 60 mg), set 3 (substrate: clavulanate: 125 mg, inhibitor 1: clarithromycin 250 mg, inhibitor 2: amoxicillin 250 mg and inhibitor 3: loxoprofen 60 mg) and set 4 (substrate: loxoprofen 60 mg, inhibitor 1: clarithromycin 250 mg, inhibitor 2: amoxicillin 250 mg and inhibitor 3: clavulanate: 125 mg). To evaluate DDI potentials among 4 compounds, we used AUC ratios (with/without co-medication). As recommended in the FDA guidance, if the AUC ratios are greater than 1.25 or less than 0.8, it suggests the potential for DDIs mediated by metabolic enzymes or transporters (FDA 2020B).
The validity of the developed clarithromycin PBPK model was confirmed using data from previous literature. Fig. 1 shows that the predicted plasma concentration-time profiles closely captured the observed profiles for both 250 mg of IV infusion and 250 mg of oral clarithromycin administration. Additionally, the confidence intervals of the predicted plasma concentration-time profiles provided good coverage of the observed data. The ratios of the predicted to observed Cmax and AUC values were well within the two-fold criteria range (0.76-1.17), demonstrating the successful validation of the clarithromycin PBPK model (Table 3).
Table 3 . Predicted and observed PK parameters of clarithromycin, amoxicillin, clavulanate, and loxoprofen.
Substrates | Route of administration | Dose (mg) | Observed mean ± SD | Predicted mean ± SD | Fold-error (predicted/observed) | |||
---|---|---|---|---|---|---|---|---|
Cmax (μg/mL) | AUC0−t (μg/mL·h) | Cmax (μg/mL) | AUC0−t (μg/mL·h) | Cmax | AUC0−t | |||
Clarithromycin | IV infusion | 250 | 2.78 ± 0.5 | 8.41 ± 2.44 | 3.26 | 8.08 | 1.17 | 0.76 |
PO | 250 | 0.85 | 4.71 | 0.72 | 5.39 | 0.86 | 1.14 | |
Amoxicillin | IV | 250 | NA | 10.4 ± 0.34 | NA | 11.1 | NA | 1.07 |
PO | 250 | 4.35 | 10.6 | 3.27 | 10.4 | 0.75 | 0.98 | |
Clavulanate | PO | 125 | 3.23 | 6.20 | 3.64 | 11.34 | 1.13 | 1.83 |
Loxoprofen | PO | 60 | 5.51 | 7.98 | 4.35 | 4.75 | 0.79 | 0.60 |
IV, intra venous; PO, per os; NA, not applicable..
The developed amoxicillin PBPK model was verified using previously reported clinical pharmacokinetic data for both IV infusion and oral administration of 250 mg amoxicillin. As depicted in Fig. 1, the observed data fell within the 95th confidence interval of the simulated plasma concentration-time profiles, indicating that the simulation results well captured the observed data. Therefore, it is concluded that the developed PBPK model accurately describes the pharmacokinetic properties of amoxicillin. Moreover, the ratios of predicted to observed Cmax and AUC values were well within the two-fold criteria range (0.75-1.07).
The initial clavulanate PBPK model was optimized for oral administration and subsequently validated using clinical pharmacokinetic data from a previous study involving a 125 mg oral dose of clavulanate. As illustrated in Fig. 1, the simulated plasma concentration-time profile closely aligned with the observed data, within the 95th confidence interval of the simulation. Additionally, the ratio of predicted to observed Cmax and AUC values met the two-fold criteria range (1.13 and 1.83, respectively), indicating the successful validation of the clavulanate PBPK model.
Fig. 1 presents the simulated plasma concentration-time profiles and observed data following the oral administration of 60 mg of loxoprofen. The observed data aligns well within the 95th confidence interval of the simulation, indicating an accurate representation of loxoprofen pharmacokinetic profiles by the PBPK model. Additionally, the ratios of predicted to observed Cmax and AUC values are well within the two-fold criteria range (0.79 and 0.60, respectively), further confirming the model’s validity.
As described in the Table 4, DDI potentials were assessed by calculating the AUC ratio (AUCR). For set 1 (substrate: clarithromycin), the predicted mean AUC ratio for clarithromycin was 0.89, falling within the range of 0.8-1.25. For set 2 (substrate: amoxicillin), the predicted mean AUC ratio of amoxicillin was 0.99. In set 3 (substrate: clavulanate), the predicted mean AUC ratio of clavulanate was 1.03, and in set 4 (substrate: loxoprofen), the predicted mean AUC ratio of loxoprofen is 1.01. These AUCR values, when comparing single treatment to co-treatment with inhibitors, all fell within the range of 0.8-1.25. These results indicate that there were no significant DDIs among clarithromycin, amoxicillin, clavulanate and loxoprofen.
Table 4 . The predicted AUCs and AUCR of clarithromycin, amoxicillin, clavulanate, and loxoprofen after oral doses of each drug either individually or in combination, including clarithromycin 250 mg, amoxicillin 250 mg, clavulanate 125 mg, and loxoprofen 60 mg.
Substrates | Dose (mg) | AUC0−t (μg/mL·h) | ||
---|---|---|---|---|
Without co-administration (a) | Co-administration (b) | AUCR (b/a) | ||
Clarithromycin | 250 | 7.14 ± 2.42 | 6.36 ± 2.31 | 0.89 |
Amoxicillin | 250 | 7.87 ± 2.04 | 7.79 ± 2.56 | 0.99 |
Clavulanate | 125 | 11.3 ± 0.71 | 11.6 ± 0.67 | 1.03 |
Loxoprofen | 60 | 4.68 ± 2.55 | 4.72 ± 3.23 | 1.01 |
AUC, area under the curve; AUCR, area under the curve ratio..
This study evaluated the potential DDIs among clarithromycin, amoxicillin, clavulanate, and loxoprofen to determine their suitability as cocktail drugs for a clinical trial.
The clarithromycin PBPK model was constructed using ADAM model and full PBPK model, including in vitro enzyme kinetic parameters. Specifically, clarithromycin is known as a substrate and a time-dependent inhibitor of CYP3A4 (Polasek and Miners 2006). The clarithromycin PBPK model was verified using various observed data, including scenarios such as single and multiple doses administration, and DDI with amoxicillin (data are not shown). The clarithromycin PBPK model accurately described the pharmacokinetic properties of clarithromycin after various dosing scenarios. The amoxicillin PBPK model was optimized using first-order absorption model and full PBPK model. The elimination of amoxicillin was estimated using in vivo clearance module because amoxicillin was mainly eliminated by renal excretion unlike clarithromycin (Zarowny et al. 1974). The developed amoxicillin PBPK model reasonably described its pharmacokinetic characteristics when compared with the observed pharmacokinetic data. Additionally, the PBPK models for clavulanate and loxoprofen were well-validated by comparing them to the observed data.
To assess the suitability of these drugs as cocktail components for clinical trials, we conducted simulations involving four different scenarios to evaluate potential DDIs. The simulated results showed no significant DDI potentials, as the AUCR remained within the acceptable criteria range of 0.8 to 1.25. Clarithromycin is one of the potent CYP3A4 inhibitors explicitly mentioned in the FDA guidelines (FDA 2023). Consequently, numerous studies have investigated DDI potentials involving clarithromycin as a perpetrator. Furthermore, clarithromycin was shown to inhibit various drug transporters, including OATP1B1, OATP1B3, and P-gp, as demonstrated by previous research (Eberl et al. 2007; Seithel et al. 2007). Despite its inhibitory effects on multiple metabolic enzymes and transporters, the simulation results did not reveal any significant DDIs when clarithromycin was employed as a perpetrator. This can be attributed to the fact that amoxicillin, clavulanate, and loxoprofen were not the substrates of those metabolic enzymes and transporters. Furthermore, several studies have demonstrated the absence of significant DDIs between clarithromycin and amoxicillin, which are commonly co-administered for Helicobacter pylori eradication, consistent with our simulation results (Mainz et al. 2002; Jin et al. 2018; Liang et al. 2023).
Amoxicillin and clavulanic acid are commonly used in combination as an antibacterial agent. In this study, we observed no significant pharmacokinetic differences when they were administered as either victims or perpetrators. These results align with previous literature, which reported no alterations in the pharmacokinetics of either drug when compared with separate administration (Todd and Benfield 1990). Additionally, amoxicillin exhibits a weak inhibitory effect on CYP2C8 (Niwa et al. 2016). However, among the other drugs tested, none of the drugs were substrates of CYP2C8, resulting in no significant DDI potentials mediated by amoxicillin.
Like other drugs, loxoprofen exhibited no changes in its pharmacokinetic properties when administered as either a victim or a perpetrator in this study. A previous study by Uwai et al. (2004) suggested that loxoprofen and its trans-OH metabolite inhibit human organic anion transporters, specifically OAT1 and OAT3, which are involved in renal tubular secretion. Furthermore, Mathialagan et al. (2017) demonstrated that amoxicillin undergoes renal clearance via OAT1 and OAT3. Inhibition of these renal transporters could potentially reduce drug elimination, making the evaluation of DDI potentials mediated by renal transporters an essential consideration. One limitation of our study is that we did not incorporate renal transporter kinetics into the loxoprofen and amoxicillin PBPK models, especially in terms of renal clearance due to lack of the in vitro kinetic data. Although Uwai et al. (2004) demonstrated IC50 values of loxoprofen and its metabolite using methotrexate as substrates for OAT1 and OAT3, methotrexate exhibited low-affinity for OAT1 with its high Km values. Thus, in vitro inhibitory data using known substrate might be required to investigate inhibitory effect of loxoprofen (FDA 2020A). Therefore, further investigation is necessary to assess DDIs mediated by renal transporters, particularly the inhibitory effect of loxoprofen.
Another limitation of this study is that lack of the verification of PBPK model for amoxicillin in DDI evaluation. Although amoxicillin exhibited an inhibitory effect on CYP2C8, the amoxicillin PBPK model was not verified for DDI studies, only for single administration.
In conclusion, we successfully developed PBPK models for clarithromycin, amoxicillin, clavulanate, and loxoprofen to evaluate DDI potentials. It is expected that significant pharmacokinetic DDIs are unlikely to occur and pharmacodynamic DDIs have not been reported currently, suggesting co-administration of these drugs is safe. Moreover, the cocktail method combining these drugs can be further applied to evaluate DDIs between the cocktail drugs and the Korean traditional medicine for respiratory infections.
The authors declare that they have no conflict of interest.
This study was supported by a grant of Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare (HF20C0002), Republic of Korea.
Table 1 Summary of input parameters used for the PBPK models of clarithromycin, amoxicillin, clavulanate and loxoprofen
Parameters (Units) | Clarithromycin | Amoxicillin | Clavulanate | Loxoprofen | Reference |
---|---|---|---|---|---|
Molecular weight (g/mol) | 747.950 | 365 | 199.16 | 246.302 | |
pKa | 8.99a | 2.8 (pKa1), 7.2 (pKa2), 9.55 (pKa3)b | 2.7 | 4.19 | aMcFarland et al. (1997) bTsuji et al. (1978) |
Log P | 2.30c | 0.87d | –1.07e | 0.82f | cLappin et al. (2011) dSangster (1994) eLakshminarayana et al. (2015) fSAJA Pharmaceuticals (2012) |
B/P | 1.83* | 0.55g | 1 | gMathialagan et al. (2017) | |
fu | 0.3h | 0.8 | 0.78i | 0.03c | hChu et al. (1993B) iWatson et al. (1987) cLappin et al. (2011) |
Absorption | ADAM | First-order | ADAM | First-order | |
ka (1/h) | 1.23j | 11.25f | jSpyker et al. (1977) fSAJA Pharmaceuticals (2012) | ||
Caco-2, Papp (10–6 cm/s) | 3.5k,l | 37.5e | 30m | kTogami et al. (2014) lMorakul et al. (2014) eLakshminarayana et al. (2015) mNarumi et al. (2016) | |
Distribution | Full | Full | Full | Minimal | |
Vss (L/kg) | 1.75n | 0.346j,o | 0.17p | 0.16 | nChu et al. (1992) jSpyker et al. (1977) oZarowny et al. (1974) pBolton et al. (1986) |
Kp scalar | 0.573* | 0.55* | 0.65 | 0.33 | |
Elimination | Enzyme kinetics | In vivo clearance | Enzyme kinetics | ||
CYP3A4 | |||||
Km (pmol/min/mg protein) | 48.7 (14-hydroxy)q 59.0 (N-desmethyl)q | qRodrigues et al. (1997) | |||
Vmax (μM) | 206 (14-hydroxy)q 189 (N-desmethyl)q | qRodrigues et al. (1997) | |||
Additional HLM CLint (μL/min/mg protein) | 36.1r | 1.46p | 264s | rSuzuki et al. (2003) pBolton et al. (1986) sShrestha et al. (2018) | |
IV clearance (L/h) | 25.6j,o | jSpyker et al. (1977) oZarowny et al. (1974) | |||
CLR (L/h) | 7.2t | 8j | 6.06p | 0.062u | tDavey (1991) jSpyker et al. (1977) pBolton et al. (1986) uKim et al. (2002) |
Interactions | |||||
CYP3A4 | |||||
KI (μM) | 29.5v | vPolasek & Miners (2006) | |||
kinact (1/min) | 0.05v | vPolasek & Miners (2006) | |||
Ki (μM) | OATP1B1 (48)w OATP1B3 (32)w P-gp (4.1)y | CYP2C8 (415)x | wSeithel et al. (2007) xNiwa et al. (2016) yEberl et al. (2007) |
ADAM, advanced dissolution, absorption, and metabolism; Log P, octanol: water partition coefficient; pKa, acid dissociation constant; B/P, blood-to-plasma ratio; fu, unbound fraction; ka, first-order absorption rate constant; Papp, apparent permeability; Vss, volume of distribution at steady state; Kp, tissue to plasma coefficient; Km, Michaelis-Menten constant; Vmax, maximum rate of metabolism; HLM, human liver microsome; KI, apparent inactivation constant; kinact, maximum inactivation rate constant; Ki, inhibition constant.
*Predicted value using SimCYPTM prediction tool.
Table 2 Observed data of clarithromycin, amoxicillin, clavulanate, and loxoprofen for the PBPK model verification
Substrates | Route of administration | Dose (mg) | Males (%) | n | Age ranges (mean) (years) | Weight ranges (mean) (kg) | Reference |
---|---|---|---|---|---|---|---|
Clarithromycin | IV infusion | 250 | 100 | 19 | 18-40 (29.0) | 58-88 (71.5) | Chu et al. (1992) |
PO | 250 | 100 | 17 | 18-40 (29.0) | 58-88 (70.8) | Chu et al. (1993A) | |
PO | 250 | 100 | 12 | 24-38 (26.5) | 65-88 (79.5) | Kees et al. (1995) | |
Amoxicillin | IV infusion | 250 | 100 | 8 | 20-30 | 59-91 (74.5) | Zarowny et al. (1974) |
PO | 250 | 100 | 8 | 20-30 | 59-91 (74.5) | Zarowny et al. (1974) | |
PO | 250 | 100 | 24 | 18-32 (23.5) | 62-82 (70.1) | Idkaidek et al. (2004) | |
PO | 250 | 100 | 12 | 20-25 (22.4) | 60-76 (66.2) | Zhang et al. (2013) | |
Clavulanate | PO | 125 | 50 | 12 | (26.1) | 58.0 (F), 69.0 (M) | Adam et al. (1982) |
PO | 125 | 100 | 4 | 45-53 | 70-82 | Bolton et al. (1986) | |
Loxoprofen | PO | 60 | 100 | 6 | 24-28 | 58-70 | Kim et al. (2002) |
PO | 60 | 100 | 14 | 21-26 | 43-86 | Choi et al. (1998) | |
PO | 60 | NA | 16 | NA | NA | Insert paper |
IV, intra venous; PO, per os; NA, not applicable.
Table 3 Predicted and observed PK parameters of clarithromycin, amoxicillin, clavulanate, and loxoprofen
Substrates | Route of administration | Dose (mg) | Observed mean ± SD | Predicted mean ± SD | Fold-error (predicted/observed) | |||
---|---|---|---|---|---|---|---|---|
Cmax (μg/mL) | AUC0−t (μg/mL·h) | Cmax (μg/mL) | AUC0−t (μg/mL·h) | Cmax | AUC0−t | |||
Clarithromycin | IV infusion | 250 | 2.78 ± 0.5 | 8.41 ± 2.44 | 3.26 | 8.08 | 1.17 | 0.76 |
PO | 250 | 0.85 | 4.71 | 0.72 | 5.39 | 0.86 | 1.14 | |
Amoxicillin | IV | 250 | NA | 10.4 ± 0.34 | NA | 11.1 | NA | 1.07 |
PO | 250 | 4.35 | 10.6 | 3.27 | 10.4 | 0.75 | 0.98 | |
Clavulanate | PO | 125 | 3.23 | 6.20 | 3.64 | 11.34 | 1.13 | 1.83 |
Loxoprofen | PO | 60 | 5.51 | 7.98 | 4.35 | 4.75 | 0.79 | 0.60 |
IV, intra venous; PO, per os; NA, not applicable.
Table 4 The predicted AUCs and AUCR of clarithromycin, amoxicillin, clavulanate, and loxoprofen after oral doses of each drug either individually or in combination, including clarithromycin 250 mg, amoxicillin 250 mg, clavulanate 125 mg, and loxoprofen 60 mg
Substrates | Dose (mg) | AUC0−t (μg/mL·h) | ||
---|---|---|---|---|
Without co-administration (a) | Co-administration (b) | AUCR (b/a) | ||
Clarithromycin | 250 | 7.14 ± 2.42 | 6.36 ± 2.31 | 0.89 |
Amoxicillin | 250 | 7.87 ± 2.04 | 7.79 ± 2.56 | 0.99 |
Clavulanate | 125 | 11.3 ± 0.71 | 11.6 ± 0.67 | 1.03 |
Loxoprofen | 60 | 4.68 ± 2.55 | 4.72 ± 3.23 | 1.01 |
AUC, area under the curve; AUCR, area under the curve ratio.