Abstract Objectives To calculate the clavulanic acid exposure of oral amoxicillin/clavulanic acid dosing regimens, to investigate variability using a population pharmacokinetic model and to explore target attainment using Monte Carlo simulations. Methods Two groups of healthy male volunteers received amoxicillin/clavulanic acid tablets at the start of a standard meal on two separate days 1 week apart. One group (n = 14) received 875/125 mg q12h and 500/125 mg q8h and the other group (n = 15) received 500/125 mg q12h and 250/125 mg q8h. In total, 1479 blood samples were collected until 8–12 h after administration. Concentrations were analysed using non-compartmental (WinNonLin) and population pharmacokinetic (NONMEM) methods. Results Median Cmax and AUC0–8 were 2.21 mg/L (0.21–4.35) and 4.99 mg·h/L (0.44–8.31), respectively. In 40/58 daily concentration–time profiles, Cmax and AUC0–8 of the morning dose were higher than with later doses. The final population model included a lag time (0.447 h), first-order absorption (3.99 h−1 at 8:00 h, between-subject variability 52.8%, between-occasion variability 48.5%), one distribution compartment (33.0 L, between-subject variability 23.9%) and first-order elimination (24.6 L/h, between-subject variability 26.7%). Bioavailability (fixed at 1 at 8:00 h, between-occasion variability 28.2%) and absorption rate decreased over the day. For 97.5% of the simulated population after 125 mg q12h or q8h, %fT > Ct at 0.5 mg/L was 8.33% (q12h) and 15.2% (q8h), %fT > Ct at 1 mg/L was 0% (q12h + q8h), and fAUC0–24 was 3.61 (q12h) and 5.56 (q8h) mg·h/L. Conclusions Clavulanic acid absorption in healthy volunteers is highly variable. Bioavailability and absorption rate decrease over the day. The model developed here may serve to suggest clavulanic acid dosing regimens to optimize efficacy and prevent underdosing. Introduction Clavulanic acid is a β-lactamase inhibitor that is combined with β-lactam antibiotics such as amoxicillin to target β-lactamase-producing strains. Amoxicillin alone or in combination with clavulanic acid was the most consumed antibacterial agent in primary care in two-thirds of EU/EEA countries in 2012.1 However, despite its widespread use for 30 years, the pharmacokinetics (PK) of oral clavulanic acid have received little attention. Several non-compartmental PK studies showed highly variable PK,2–4 but it still remains unclear how the variable PK influences the exposure and thereby the efficacy of the drug. It is plausible that activity against β-lactamase-producing strains will not be attained if the exposure to clavulanic acid is inadequate. A population PK model can be used to further investigate variability and exposure in the population. Currently, only one population PK model for oral amoxicillin/clavulanic acid suspension is available in a thesis.5 The exposure to an antimicrobial of the microorganism in vivo (dependent on dose and PK) and the potency of a drug in vitro (usually expressed as an MIC) determine the antimicrobial efficacy.6 For antimicrobials, PK/pharmacodynamic (PD) indices, such as area under the concentration–time curve for 0–24 h (AUC0–24)/MIC, maximum concentration (Cmax)/MIC and T>MIC, describe the exposure–response relationships of antimicrobial agents.6 The PD target is the minimal PK/PD index value that ensures a high probability of successful treatment.6 However, clavulanic acid has very weak antimicrobial activity when used alone.7 For β-lactamase inhibitors the time that the free concentration exceeds a threshold concentration (%fT > Ct) and the total daily dose and fAUC have been described to be important for inhibitory activity.8–12 For tazobactam, sulbactam and avibactam, the PK/PD index seems to be %fT > Ct.8–11 In contrast, for relebactam (MK-7655), the total daily dose and fAUC were linked to effect.12 Clavulanic acid is an irreversible suicide inhibitor similar to tazobactam and sulbactam, but has a unique chemical structure and different affinities for β-lactamase enzymes than the other inhibitors.7 Its PD properties cannot therefore be extrapolated. Owing to the lack of preclinical data, the PK/PD index best describing its activity is currently not known. The purposes of this study were therefore twofold. The first was to build a population PK model using NONMEM to determine the clavulanic acid exposure and variability of various oral amoxicillin/clavulanic acid dosing regimens and to investigate whether PK interactions occur between the two agents.13 A population PK model for oral amoxicillin using the same amoxicillin/clavulanic acid data was previously published by the authors.14 The second purpose was to explore the variability of clavulanic acid exposure and its effects on target attainment for standard dosing regimens of clavulanic acid (125 mg q12h and 125 mg q8h). Because the PK/PD index and PD target of clavulanic acid are still unknown, both %fT > Ct and fAUC were calculated. Methods Study design and population The study was designed as an open-label, randomized, two-part, crossover investigation of the PK of oral amoxicillin/clavulanic acid. Male volunteers were enrolled in the study if they were aged between 18 and 50 years and in good general health. Exclusion criteria were >20% deviation from ideal weight for height, use of prescribed medication in the 2 weeks prior to the study (antibiotics: 4 weeks), use of any medication during the study without consent, alcohol intake >3 units/day, participation in a trial within 2 months prior to the start of this study, prior hypersensitivity to the trial drug or to drugs with a similar chemical structure, diseases known to interfere with the drug PK, or blood donation of >1500 mL within the previous year. The study (reference number 25000/360),15 commissioned by SmithKline Beecham Pharmaceuticals (Harlow, UK), was conducted in 1993 at FOCUS Clinical Drug Development GmbH (Neuss, Germany). Ethics The study was conducted in accordance with the Declaration of Helsinki. The study protocol was approved by the Freiburger Ethics Committee (Freiburg, Germany). All volunteers gave written informed consent prior to the study. Study procedures The study consisted of two parts. Each part included a separate group of subjects who each received two dosing regimens over 1 day. The order of the dosing regimens was randomized and treatment days were separated by a washout period of 6 or 7 days. In the first part, 16 subjects were allocated to amoxicillin/clavulanic acid 875/125 mg q12h (2 doses) and 500/125 mg q8h (3 doses). In the second part, 16 other subjects were assigned to 500/125 mg q12h (2 doses) and 250/125 mg q8h (3 doses). Each dose was provided as a single tablet, amoxicillin as trihydrate and clavulanic acid as potassium clavulanate (Augmentin®, SmithKline Beecham Pharmaceuticals, Bristol, TN, USA). Doses were administered with 200 mL water at the start of a standard meal. Each meal (∼800 kcal and 30% fat content) consisted of four slices of pork, slices of cucumber, 250 g of pasta salad and a half slice of coarse wholemeal bread. Dosing times were 8:00, 16:00 and 24:00 h for the q8h regimens and 8:00 and 20:00 h for the q12h regimens. For the q12h regimens, a second standard meal was provided at 12:00 h. The first dose of each day was administered at 8:00 h after having fasted from food and fluids from 22:00 h the night before. Blood samples were collected just before administration and after 0.5, 1, 1.5, 2, 2.5, 3, 4, 6, 8, 10 and 12 h (q8h regimens until 8 h). Samples were frozen at −70 °C within 1 h of sampling and were assayed within 6 weeks of collection. Clavulanic acid plasma concentrations were determined by Hazleton Laboratories (UK) using an ASTED (Automated Sequential Trace Enrichment of Dialysates) system coupled to an HPLC with UV absorbance detection. The lower limit of quantification was 0.05 mg/L.15 PK analysis Non-compartmental PK analysis of the plasma concentration–time data was performed using WinNonLin (version 7.0, Certara, Princeton, NJ, USA). Population PK analysis was performed using non-linear mixed-effects modelling (NONMEM version 7.2, ICON Development Solutions, Ellicott City, MD, USA). The Intel Visual Fortran Compiler XE 14.0 (Santa Clara, CA, USA) was used. The first-order conditional estimation method with interaction was used throughout the model-building process. Tools used to evaluate and visualize the model were RStudio (version 0.98.1028), R (version 3.1.1), XPose (version 4.5.0) and PsN (version 4.6.0), all with the graphical interface Pirana16 (version 2.9.4). First-order, zero-order and Michaelis–Menten absorption models with and without lag time were evaluated in combination with one- and two-compartment distribution models. Between-subject variability (BSV; variability between individual subjects) and between-occasion variability (BOV; variability between the doses in an individual subject) were tested using an exponential variance model. Residual unexplained variability (RUV) was evaluated with a combined (additive and proportional) error model. A stepwise covariate model building was performed with forward addition at P < 0.05 [decrease in the objective function value (OFV) of 3.84 units] followed by backward elimination at P < 0.001 (decrease in the OFV of 10.88 units). Evaluated covariates were body weight, dosing time, amoxicillin dose time and amoxicillin daily dose (dosing time and dose were evaluated as continuous and categorical covariates). Model selection criteria were decrease in OFV, goodness-of-fit plots and visual predictive checks (VPCs). A decrease in the OFV of 3.84 units was considered statistically significant (P < 0.05) in a nested model.17 For each VPC, a set of 1000 simulated datasets was created to compare the observed concentrations with the distribution of the simulated concentrations. The 95% CI of each parameter in the final model was determined from a non-parametric bootstrap analysis, in which the dataset was resampled 1000 times. Monte Carlo simulations Monte Carlo simulations were performed using the final model in NONMEM. Two clavulanic acid dosing regimens were evaluated: 125 mg q12h (at 8:00 and 20:00 h) and 125 mg q8h (at 8:00, 16:00 and 24:00 h). Five thousand subjects were simulated for each dosing regimen. For each simulated concentration–time profile, the fAUC0–24 was calculated as well as the %fT > Ct (the percentage of the dosing period of 24 h that the free clavulanic acid concentration exceeds the threshold concentration Ct) for threshold concentrations of 0.015–64 mg/L. The unbound clavulanic acid concentration was calculated from the total concentration using a fixed value for protein binding of 25%.18,19 Results Study population Thirty-two healthy male volunteers (16 per part) entered the study. After two withdrawals (one because of diarrhoea and one due to personal reasons), clavulanic acid plasma concentrations were determined in 30 volunteers (15 per part) who completed both dosing regimens. PK could not be evaluated in one subject in part one owing to very low plasma clavulanic acid and amoxicillin concentrations. The characteristics of the 29 PK evaluable subjects are shown in Table S1 (available as Supplementary data at JAC Online). The average ± SD values were: age, 33 ± 7 years; height, 179 ± 6 cm; weight, 78 ± 9 kg; and BMI, 24 ± 2 kg/m2. PK analysis One hundred and forty-five clavulanic acid concentration–time profiles (5 profiles per subject) with in total 1479 samples were analysed. The concentrations in the q12h regimens with sampling up to 12 h after administration were detectable for at most 8 h. A summary of the results for the non-compartmental PK analysis is shown in Table 1. Considering the ranges of Cmax and area under the concentration–time curve for 0–8 h (AUC0–8), the ratio between the maximum and minimum values is ∼20 for both parameters (Cmax 4.35/0.21 and AUC0–8 8.31/0.44). The individual concentration–time profiles illustrate that this high variability in Cmax and AUC0–8 not only exists between the individual subjects but also between the different doses in one dosing regimen in a subject. In 40/58 daily concentration–time profiles, the Cmax and AUC0–8 of the first dose were higher than the Cmax and AUC0–8 of the second or third dose. A summary of the individual concentration–time profiles is displayed in Figure 1. Table 1. Results of the non-compartmental PK analysis of clavulanic acid Amoxicillin/clavulanic acid dosing regimen PK parameters of clavulanic acid Cmax (mg/L) Tmax (h) AUC0–8 (mg·h/L) AUC0–24 (mg·h/L) t½ (h) 250/125 mg q8h mean±SD 1.91 ± 0.68 1.30 ± 0.31 4.33 ± 1.54 12.98 ± 3.33 1.07 ± 0.28 median (range) 1.99 (0.21–2.97) 1.50 (1.00–2.00) 4.60 (0.44–6.76) 13.45 (4.63–18.26) 0.98 (0.46–1.69) 500/125 mg q12h mean ± SD 1.99 ± 0.61 1.33 ± 0.40 4.45 ± 1.15 8.90 ± 1.92 1.05 ± 0.19 median (range) 1.91 (0.56–3.00) 1.25 (1.00–2.50) 4.63 (1.08–6.41) 9.26 (4.86–11.55) 0.99 (0.72–1.41) 500/125 mg q8h mean ± SD 2.54 ± 0.76 1.26 ± 0.28 5.39 ± 1.38 16.17 ± 3.89 1.09 ± 0.41 median (range) 2.59 (1.09–4.35) 1.25 (1.00–2.00) 5.39 (2.58–8.31) 17.17 (8.23–21.75) 0.99 (0.66–3.37a) 875/125 mg q12h mean ± SD 2.41 ± 0.90 1.29 ± 0.32 5.25 ± 1.79 10.51 ± 3.07 1.02 ± 0.17 median (range) 2.62 (0.23–4.02) 1.25 (1.00–2.00) 5.63 (0.50–7.78) 10.85 (3.51–13.87) 1.00 (0.73–1.47) All regimens mean ± SD 2.21 ± 0.78 1.29 ± 0.32 4.82 ± 1.53 12.10 ± 4.10 1.06 ± 0.30 median (range) 2.21 (0.21–4.35) 1.50 (1.00–2.50) 4.99 (0.44–8.31) 11.95 (3.51–21.75) 0.99 (0.46–3.37a) Amoxicillin/clavulanic acid dosing regimen PK parameters of clavulanic acid Cmax (mg/L) Tmax (h) AUC0–8 (mg·h/L) AUC0–24 (mg·h/L) t½ (h) 250/125 mg q8h mean±SD 1.91 ± 0.68 1.30 ± 0.31 4.33 ± 1.54 12.98 ± 3.33 1.07 ± 0.28 median (range) 1.99 (0.21–2.97) 1.50 (1.00–2.00) 4.60 (0.44–6.76) 13.45 (4.63–18.26) 0.98 (0.46–1.69) 500/125 mg q12h mean ± SD 1.99 ± 0.61 1.33 ± 0.40 4.45 ± 1.15 8.90 ± 1.92 1.05 ± 0.19 median (range) 1.91 (0.56–3.00) 1.25 (1.00–2.50) 4.63 (1.08–6.41) 9.26 (4.86–11.55) 0.99 (0.72–1.41) 500/125 mg q8h mean ± SD 2.54 ± 0.76 1.26 ± 0.28 5.39 ± 1.38 16.17 ± 3.89 1.09 ± 0.41 median (range) 2.59 (1.09–4.35) 1.25 (1.00–2.00) 5.39 (2.58–8.31) 17.17 (8.23–21.75) 0.99 (0.66–3.37a) 875/125 mg q12h mean ± SD 2.41 ± 0.90 1.29 ± 0.32 5.25 ± 1.79 10.51 ± 3.07 1.02 ± 0.17 median (range) 2.62 (0.23–4.02) 1.25 (1.00–2.00) 5.63 (0.50–7.78) 10.85 (3.51–13.87) 1.00 (0.73–1.47) All regimens mean ± SD 2.21 ± 0.78 1.29 ± 0.32 4.82 ± 1.53 12.10 ± 4.10 1.06 ± 0.30 median (range) 2.21 (0.21–4.35) 1.50 (1.00–2.50) 4.99 (0.44–8.31) 11.95 (3.51–21.75) 0.99 (0.46–3.37a) q12h, every 12 h; q8h, every 8 h; range, minimum–maximum; Cmax, maximum concentration; Tmax, time to maximum concentration; AUC0-8, area under the concentration-time curve for 0-8 h, AUC0-24, area under the concentration-time curve for 0-24 h (q12h: 2 doses, q8h: 3 doses); t1/2, half-life. a Outlier (penultimate t½ was 1.53 for 500/125 mg q8h). Figure 1. View largeDownload slide Clavulanic acid concentration–time curves for four dosing regimens of amoxicillin/clavulanic acid: 875/125 mg q12h (n = 14), 500/125 mg q8h (n = 14), 500/125 mg q12h (n = 15) and 250/125 mg q8h (n = 15). The average concentration and standard deviation at each timepoint is displayed as a filled circle with error bars. Figure 1. View largeDownload slide Clavulanic acid concentration–time curves for four dosing regimens of amoxicillin/clavulanic acid: 875/125 mg q12h (n = 14), 500/125 mg q8h (n = 14), 500/125 mg q12h (n = 15) and 250/125 mg q8h (n = 15). The average concentration and standard deviation at each timepoint is displayed as a filled circle with error bars. Population PK analysis showed that the clavulanic acid data were best described by a model with a lag time and first-order absorption, one distribution compartment and first-order elimination. A second distribution compartment did not further improve the model. Implementation of BSV was significant for volume of distribution (V), clearance (CL) and first-order absorption rate constant (Ka). Implementation of BOV was significant for bioavailability (F) and Ka. BSV for F and lag time and BOV for V, CL and lag time were not significant and were therefore not implemented. The covariate analysis resulted in two significant covariates: dosing time (8:00, 16:00, 20:00, 24:00 h) was proportionally correlated with F and Ka. The effect of dosing time was further studied by implementation of a cosine function.20 Replacing the covariate dosing time with cosine functions on Ka and F did not improve the model and therefore the proportional effects of dosing time on Ka and F were implemented in the model. For example, at dosing time 8:00 h, the population value of Ka was 3.99 h−1 (fixed proportional effect of 1) and at 16:00 h the population value was 3.60 h−1 (estimated proportional effect of 0.903). A model with combined dosing times of 20:00 and 24:00 h was also tested because the proportional effects on Ka and F looked similar at these dosing times, but this combination did not improve the model further and was therefore not implemented. The population PK parameter estimates of the final model are displayed in Table 2. Since no data with intravenous administration were collected in this study, F could not be quantified and was fixed at 1. Consequently, V/F and CL/F values are displayed instead of V and CL values. Table 2. Model-based population PK parameter estimates and values obtained after bootstrap analysis Parameter Final model estimate RSE (%) Bootstrap median Bootstrap 95% CI Fixed effects V/F (L) 33.0 3.8 33.0 30.3–35.6 CL/F (L/h) 24.6 3.8 24.7 22.6–26.6 F (–) 1 (fixed) 1 (fixed) — Ka (h−1) 3.99 14.1 3.95 3.07–5.20 lag time (h) 0.447 1.3 0.447 0.436–0.456 Covariate: proportional effect on Ka dosing time 8:00 h 1 (fixed) — 1 (fixed) — dosing time 16:00 h 0.903 9.9 0.904 0.737–1.08 dosing time 20:00 h 0.610 15.1 0.617 0.442–0.801 dosing time 24:00 h 0.636 14.8 0.638 0.476–0.843 Covariate: proportional effect on F dosing time 8:00 h 1 (fixed) — 1 (fixed) — dosing time 16:00 h 0.873 5.6 0.876 0.765–0.949 dosing time 20:00 h 0.799 7.4 0.802 0.688–0.904 dosing time 24:00 h 0.801 8.8 0.806 0.663–0.944 BSV V (%CV) 23.9 23.2 23.4 12.9–33.3 CL (%CV) 26.7 17.1 26.1 17.8–34.8 Ka (%CV) 52.8 15.8 51.7 34.4–67.8 BOV Ka (%CV) 48.5 12.8 47.3 37.8–60.2 F (%CV) 28.2 21.4 27.2 16.7–38.6 RUV additive (mg/L) 0.0533 8.2 0.0530 0.0450–0.0625 proportional 0.142 8.2 0.141 0.119–0.165 Parameter Final model estimate RSE (%) Bootstrap median Bootstrap 95% CI Fixed effects V/F (L) 33.0 3.8 33.0 30.3–35.6 CL/F (L/h) 24.6 3.8 24.7 22.6–26.6 F (–) 1 (fixed) 1 (fixed) — Ka (h−1) 3.99 14.1 3.95 3.07–5.20 lag time (h) 0.447 1.3 0.447 0.436–0.456 Covariate: proportional effect on Ka dosing time 8:00 h 1 (fixed) — 1 (fixed) — dosing time 16:00 h 0.903 9.9 0.904 0.737–1.08 dosing time 20:00 h 0.610 15.1 0.617 0.442–0.801 dosing time 24:00 h 0.636 14.8 0.638 0.476–0.843 Covariate: proportional effect on F dosing time 8:00 h 1 (fixed) — 1 (fixed) — dosing time 16:00 h 0.873 5.6 0.876 0.765–0.949 dosing time 20:00 h 0.799 7.4 0.802 0.688–0.904 dosing time 24:00 h 0.801 8.8 0.806 0.663–0.944 BSV V (%CV) 23.9 23.2 23.4 12.9–33.3 CL (%CV) 26.7 17.1 26.1 17.8–34.8 Ka (%CV) 52.8 15.8 51.7 34.4–67.8 BOV Ka (%CV) 48.5 12.8 47.3 37.8–60.2 F (%CV) 28.2 21.4 27.2 16.7–38.6 RUV additive (mg/L) 0.0533 8.2 0.0530 0.0450–0.0625 proportional 0.142 8.2 0.141 0.119–0.165 RSE, relative standard error; CV, coefficient of variation. As shown in Table 2, the model-based parameter estimates were similar to the values computed from the bootstrap analysis, indicating the stability of the model. The goodness-of-fit plots in Figure 2 and Figure S1 show that the model adequately described the observed concentrations. The VPC plots, presented in Figure S2, indicate a good predictive performance. Figure 2. View largeDownload slide Observed versus population predicted concentrations (a) and observed versus individual predicted concentrations (b). Figure 2. View largeDownload slide Observed versus population predicted concentrations (a) and observed versus individual predicted concentrations (b). Monte Carlo simulations The results of the simulations with 125 mg q12h and 125 mg q8h are presented in Table 3 (fAUC0–24) and Figure 3 (%fT > Ct). Table 3 shows that 97.5% of the population reached an fAUC0–24 of 3.61 mg·h/L with 125 mg q12h and 5.56 mg·h/L with 125 mg q8h. For 97.5% of the population, the %fT > Ct at 1 mg/L was 0% for both regimens and the %fT > Ct at 0.5 mg/L was 8.33% (125 mg q12h) and 15.2% (125 mg q8h). Half of the population (q12h: 46%, q8h: 53%) attained a concentration of 2 mg/L, but the average %fT > Ct values at 2 mg/L were low: 2.09% with 125 mg q12h and 3.05% with 125 mg q8h. Table 3. fAUC0–24 distribution obtained by 5000 Monte Carlo simulations for two dosing regimens (125 mg q12h and 125 mg q8h) fAUC0–24 (mg·h/L) 125 mg q12h 125 mg q8h Minimum 2.10 3.43 1st percentile 3.17 4.86 2.5th percentile 3.61 5.56 5th percentile 4.07 6.15 50th percentile 6.94 10.4 95th percentile 12.2 17.5 97.5th percentile 13.7 19.2 99th percentile 15.4 21.2 Maximum 28.0 36.1 fAUC0–24 (mg·h/L) 125 mg q12h 125 mg q8h Minimum 2.10 3.43 1st percentile 3.17 4.86 2.5th percentile 3.61 5.56 5th percentile 4.07 6.15 50th percentile 6.94 10.4 95th percentile 12.2 17.5 97.5th percentile 13.7 19.2 99th percentile 15.4 21.2 Maximum 28.0 36.1 Figure 3. View largeDownload slide %fT > Ct displayed as a function of threshold concentration (Ct) for two dosing regimens: (a) 125 mg q12h; and (b) 125 mg q8h. The middle line represents the values for the median of the population and the surrounding lines indicate the 1st, 5th, 95th and 99th percentiles, obtained by 5000 Monte Carlo simulations. Figure 3. View largeDownload slide %fT > Ct displayed as a function of threshold concentration (Ct) for two dosing regimens: (a) 125 mg q12h; and (b) 125 mg q8h. The middle line represents the values for the median of the population and the surrounding lines indicate the 1st, 5th, 95th and 99th percentiles, obtained by 5000 Monte Carlo simulations. Discussion Our non-compartmental PK analysis showed that clavulanic acid Cmax and AUC0–8 in healthy volunteers were highly variable, whereas half-life (t½) had a limited variability. In two-thirds of the subjects, the Cmax and AUC0–8 of the morning dose were higher than later doses. Our population PK model indicated that Ka was the most variable parameter. Ka and F were estimated to be higher in the morning than in the afternoon and evening. Similar to the present study, other non-compartmental PK studies with oral clavulanic acid demonstrated a more variable Cmax and AUC than t½.2–4 A population PK model for oral amoxicillin/clavulanic suspension5 also included a high BSV and BOV of absorption parameters rather than of V and CL. Another study in 10 volunteers who received a single oral and a single intravenous dose of clavulanic acid also found a wide F range (31.4%–98.8%).2 In our population PK model, we could explain part of the observed variability by the effect of dosing time on Ka and F. To our knowledge, time-varying absorption and F of oral clavulanic acid has not previously been described. For two other β-lactams, meropenem and ceftazidime, it has been shown that morning concentrations were higher than afternoon concentrations.21–23 However, since those antibiotics were both given intravenously, the varying concentrations were explained by renal CL variation rather than absorption differences as in our study. Our finding of the inversely proportional effect of dosing time on Ka may be caused by non-linear processes, such as saturation of Ka. However, this seems to be unlikely since implementation of Michaelis-Menten absorption did not improve the model. Unfortunately, it is impossible to predict the results for second and later dosing days, because our data only included dosing regimens of 24 h. Differences in meal composition can be excluded as a reason for variation in Ka and F, because each dose was taken at the start of a standardized meal that was the same for each meal during the day. Administration without food does not eliminate the variable PK, since fasting studies with oral clavulanic acid also showed a high variation in Cmax and AUC.2,4 In a chronopharmacokinetic study with oral midazolam,20 the daily variation in Ka was described by a time-varying covariate and F was parameterized as a cosine function. The Ka and F differences were explained by 24 h variation in gastric emptying, gastrointestinal mobility and splanchnic blood flow,20 which may be the most reliable explanation for the findings in our study as well. Possible clinical implications of time-varying Ka and F for dosing regimens, such as higher afternoon and evening doses than morning doses, should be further studied. The results of our non-compartmental analysis seem to suggest that the amoxicillin dose influences the clavulanic acid PK. However, we tested several covariate types of amoxicillin dose (e.g. dose time, daily dose, categorical covariate, continuous covariate) during the modelling process and none was significant. The literature is not conclusive about the effect of amoxicillin on clavulanic acid PK. It has been reported that the Cmax and AUC of oral clavulanic acid in the presence of amoxicillin were higher than those of clavulanic acid alone and that the AUC ratio of amoxicillin/clavulanic acid was lower (2.55) than expected (500/125 = 4).24 These findings suggest an interaction between the absorption of amoxicillin and clavulanic acid. However, these AUC ratios differ enormously between studies. For oral 500/125 mg, the AUC ratios in our study were 3.4 (500/125 mg q12h) and 3.9 (500/125 mg q8h) whereas other studies found ratios of 2.025 and 4.3.4 We found an AUC ratio of 2.1 for oral 250/125 mg which was the same as the ratio found by another study.4 However, a third study found a ratio of 1.4.26 The AUC ratios for oral 875/125 mg were 5.3 (our study) and 5.7.4 It is possible that the saturable absorption rate of amoxicillin influences the AUC ratio.14 However, for intravenous amoxicillin/clavulanic acid too, the AUC ratios were not as expected: 2.8 (500/100 mg),27 2.7 (1000/200 mg),27 3.2 (500/100 mg),25 7.1 (625/125 mg)2 and 6.5 (2000/200 mg).27 These findings indicate that factors other than absorption also influence the interaction between the two drugs. Although we were not able to find a significant effect of the amoxicillin dose on clavulanic acid PK, the influence of the interaction between both compounds is not yet clear. The EMA guideline on the use of PK/PD in the development of antimicrobials recommends studying the PK interaction of β-lactams and β-lactam inhibitors.13 Future research should elucidate the influence of amoxicillin on clavulanic acid PK. Similar to a study with oral amoxicillin/clavulanic acid suspension in healthy volunteers,5 the BSV and BOV magnitude of the absorption parameter was comparable. We hypothesize that in a patient population the BSV will be higher than the BOV. Owing to highly variable clavulanic acid concentrations, the risk of ineffectiveness (with too-low concentrations) and adverse events (with too-high concentrations) should be considered. When the PD target is known, dosing regimens can be optimized to attain a high probability of successful treatment. However, it is still unknown which PK/PD index and PD target should be taken into account for clavulanic acid. Since clavulanic acid has a structure that is unique among β-lactamase inhibitors,7 it is difficult to extrapolate a PK/PD index from another inhibitor. Clavulanic acid is a clavam isolated from Streptomyces clavuligerus, whereas tazobactam and sulbactam are synthetic penicillinate sulfones.7,28 The different chemical structures possibly explain the differences in enzyme activities of these inhibitors.7 The β-lactamase inhibitors avibactam and relebactam are diazabicyclooctane derivatives that do not have any structural similarity to β-lactams.28 These two compounds have different PD properties, although they are both from the same group. Avibactam activity is primarily dependent on %fT > Ct10,11 and relebactam activity on fAUC.12 Since the PD target for clavulanic acid is unclear, optimization of treatment by ensuring a high probability of attainment is not feasible. However, we do present simulations and attainment for different targets and these will be useful once the PD target of clavulanic acid becomes available. Current EUCAST guidelines use a fixed clavulanic acid concentration of 2 mg/L for susceptibility testing purposes.29 Our Monte Carlo simulations show that with 125 mg q12h or q8h half of the population attains concentrations of 2 mg/L and the average %fT > Ct is only 2%–3%. Similarly, assuming an fAUC0–18 of 36 mg·h/L in vitro (based on the same 2 mg/L concentration and an incubation time of 18 h), the probability of target attainment is 0% with 125 mg q12h or q8h. However, the in vivo effect of clavulanic acid is thereby far underestimated. Ultimately, the susceptibility in vitro has to be correlated to efficacy in vivo and there is at present—in contrast to antimicrobials—no clear consensus for inhibitors. For example, the EUCAST fixed concentration for tazobactam is 4 mg/L,29 which is much higher than the PD target %fT > Ct at 0.5 mg/L.8 PD studies providing data on the clavulanic acid target are clearly required. A limitation of this study is that only a few covariates were available and, unfortunately, creatinine data were lacking. This prohibited an analysis of the impact of renal function on clavulanic acid exposure. However, the participants were healthy volunteers with normal renal function and a clear relationship would likely not have been found in this population. Since oral amoxicillin/clavulanic acid is mostly prescribed to patients with only relatively mild infections and normal renal function, the results of our study can therefore be extrapolated to such patients. Second, it was not possible to evaluate different clavulanic acid doses in the population PK analysis, because only one dose of clavulanic acid (125 mg) was included in this study. However, 125 mg is the dose most generally used for oral dosing. This study included three different tablets and four dosing regimens for amoxicillin/clavulanic acid. It is impossible to extrapolate these results to other oral formulations [e.g. the extended release (XR) tablet or the suspension]. However, because the clavulanic acid formulation and the dosing frequency of the XR tablet are the same as used in our study, we expect that the timing problem may also exist with the XR tablet. A third limitation is that our data only included dosing regimens of 24 h, which makes it impossible to predict the results for second and later dosing days. In conclusion, clavulanic acid concentrations in healthy volunteers are highly variable after oral administration. F and Ka decrease over the course of the day. The consequences of the variable concentrations for underdosing and adverse events should be further studied for multiple-day dosing and dosing regimens should be optimized. Studies on the PK/PD index and PD target are needed. Acknowledgements This study was presented at the Twenty-seventh European Congress of Clinical Microbiology and Infectious Diseases, Vienna, Austria, 2017 (Abstract 3428). We thank Peter de Bruijn for assistance with WinNonLin. Funding This work was supported by the Innovative Medicines Initiative Joint Undertaking under grant agreement no. , resources of which are composed of financial contribution from the European Union’s Seventh Framework Programme (FP7/2007–2013) and EFPIA companies’ in-kind contribution. The research leading to these results was conducted as part of the COMBACTE-NET consortium. For further information please refer to http://www.combacte.com/. Transparency declarations J. W. M. has received research funding from IMI, the EU, ZON-MW, Adenium, AstraZeneca, Basilea, Eumedica, Cubist, Merck & Co., Pfizer, Polyphor, Roche, Shionogi, Thermo-Fisher, Wockhardt, Astellas, Gilead and Pfizer. All other authors: none to declare. Supplementary data Table S1 and Figures S1 and S2 are available as Supplementary data at JAC Online. References 1 ECDC. Surveillance of Antimicrobial Consumption in Europe 2012. Stockholm, Sweden, 2014. http://ecdc.europa.eu/en/publications/Publications/antimicrobial-consumption-europe-esac-net-2012.pdf. 2 Nilsson-Ehle I, Fellner H, Hedstrom SA et al. Pharmacokinetics of clavulanic acid, given in combination with amoxycillin, in volunteers. J Antimicrob Chemother 1985; 16: 491– 8. Google Scholar CrossRef Search ADS PubMed 3 Kaye CM, Allen A, Perry S et al. The clinical pharmacokinetics of a new pharmacokinetically enhanced formulation of amoxicillin/clavulanate. Clin Ther 2001; 23: 578– 84. Google Scholar CrossRef Search ADS PubMed 4 Vree TB, Dammers E, Exler PS. Identical pattern of highly variable absorption of clavulanic acid from four different oral formulations of co-amoxiclav in healthy subjects. J Antimicrob Chemother 2003; 51: 373– 8. Google Scholar CrossRef Search ADS PubMed 5 Bulitta J. Innovative techniques for selecting the dose of antibiotics in empiric therapy—focus on β-lactams and cystic fibrosis patients. Thesis. Julius-Maximilians-Universität, 2006. 6 Mouton JW, Ambrose PG, Canton R et al. Conserving antibiotics for the future: new ways to use old and new drugs from a pharmacokinetic and pharmacodynamic perspective. Drug Resist Updat 2011; 14: 107– 17. Google Scholar CrossRef Search ADS PubMed 7 Drawz SM, Bonomo RA. Three decades of β-lactamase inhibitors. Clin Microbiol Rev 2010; 23: 160– 201. Google Scholar CrossRef Search ADS PubMed 8 Melchers MJ, Mavridou E, van Mil AC et al. Pharmacodynamics of ceftolozane combined with tazobactam against Enterobacteriaceae in a neutropenic mouse thigh model. Antimicrob Agents Chemother 2016; 60: 7272– 9. Google Scholar PubMed 9 Lister PD, Prevan AM, Sanders CC. Importance of β-lactamase inhibitor pharmacokinetics in the pharmacodynamics of inhibitor-drug combinations: studies with piperacillin-tazobactam and piperacillin-sulbactam. Antimicrob Agents Chemother 1997; 41: 721– 7. Google Scholar PubMed 10 Coleman K, Levasseur P, Girard AM et al. Activities of ceftazidime and avibactam against β-lactamase-producing Enterobacteriaceae in a hollow-fiber pharmacodynamic model. Antimicrob Agents Chemother 2014; 58: 3366– 72. Google Scholar CrossRef Search ADS PubMed 11 Berkhout J, Melchers MJ, van Mil AC et al. Pharmacodynamics of ceftazidime and avibactam in neutropenic mice with thigh or lung infection. Antimicrob Agents Chemother 2016; 60: 368– 75. Google Scholar CrossRef Search ADS 12 Mavridou E, Melchers RJ, van Mil AC et al. Pharmacodynamics of imipenem in combination with β-lactamase inhibitor MK7655 in a murine thigh model. Antimicrob Agents Chemother 2015; 59: 790– 5. Google Scholar CrossRef Search ADS PubMed 13 EMA. Guideline on the Use of Pharmacokinetics and Pharmacodynamics in the Development of Antimicrobial Medicinal Products. London, UK, 2016. www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2016/07/WC500210982.pdf. 14 de Velde F, de Winter BC, Koch BC et al. Non-linear absorption pharmacokinetics of amoxicillin: consequences for dosing regimens and clinical breakpoints. J Antimicrob Chemother 2016; 71: 2909– 17. Google Scholar CrossRef Search ADS PubMed 15 SmithKline Beecham. Comparison of the 24-Hour Pharmacokinetic Profile of Oral Augmentin Administered with Food to Healthy Male Volunteers as 1 g 12 Hourly versus 625 mg 8 Hourly, and as 625 mg 12 Hourly versus 375 mg 8 Hourly (Study 25000/360, Final Report). Harlow, UK, 1994. 16 Keizer RJ, Karlsson MO, Hooker A. Modeling and simulation workbench for NONMEM: tutorial on Pirana, PsN, and Xpose. CPT Pharmacometrics Syst Pharmacol 2013; 2: e50. Google Scholar CrossRef Search ADS PubMed 17 Boeckmann A, Sheiner L, Beal S. NONMEM Users Guide—Part V. Ellicott City, MD, USA: ICON Development Solutions, 2011. https://nonmem.iconplc.com/nonmem720/guides/v.pdf. 18 GlaxoSmithKline. Summary of Product Characteristics: AUGMENTIN Amoxicillin/Clavulanic Acid 500mg/125mg Oral Tablets. Zeist, The Netherlands, 2015. 19 de la Pena A, Derendorf H. Pharmacokinetic properties of β-lactamase inhibitors. Int J Clin Pharmacol Ther 1999; 37: 63– 75. Google Scholar PubMed 20 van Rongen A, Kervezee L, Brill M et al. Population pharmacokinetic model characterizing 24-hour variation in the pharmacokinetics of oral and intravenous midazolam in healthy volunteers. CPT Pharmacometrics Syst Pharmacol 2015; 4: 454– 64. Google Scholar CrossRef Search ADS PubMed 21 Vinks AA, Touw DJ, Heijerman HG et al. Pharmacokinetics of ceftazidime in adult cystic fibrosis patients during continuous infusion and ambulatory treatment at home. Ther Drug Monit 1994; 16: 341– 8. Google Scholar CrossRef Search ADS PubMed 22 Mouton JW. Pharmacokinetic and pharmacodynamic studies of β-lactam antibiotics in volunteers and patients with cystic fibrosis. Thesis. Erasmus University, 1993. 23 Mouton JW, Horrevort AM, Michel MF. Circadian rhythm of ceftazidime and meropenem serum levels during continuous and intermittent infusion. In: Conference Proceedings of the International Conference Chemotherapy, Berlin, pp. 1016–7. Futuramed Verlag, Munchen, 1991. 24 Adam D, de Visser I, Koeppe P. Pharmacokinetics of amoxicillin and clavulanic acid administered alone and in combination. Antimicrob Agents Chemother 1982; 22: 353– 7. Google Scholar CrossRef Search ADS PubMed 25 Hampel B, Lode H, Bruckner G et al. Comparative pharmacokinetics of sulbactam/ampicillin and clavulanic acid/amoxycillin in human volunteers. Drugs 1988; 35 Suppl 7: 29– 33. Google Scholar CrossRef Search ADS PubMed 26 Idkaidek NM, Al-Ghazawi A, Najib NM. Bioequivalence evaluation of two brands of amoxicillin/clavulanic acid 250/125 mg combination tablets in healthy human volunteers: use of replicate design approach. Biopharm Drug Dispos 2004; 25: 367– 72. Google Scholar CrossRef Search ADS PubMed 27 Fresenius. Summary of Product Characteristics: Amoxicillin/Clavulanic Acid for Injection/Intravenous Use. Zeist, The Netherlands, 2015. 28 Drawz SM, Papp-Wallace KM, Bonomo RA. New β-lactamase inhibitors: a therapeutic renaissance in an MDR world. Antimicrob Agents Chemother 2014; 58: 1835– 46. Google Scholar CrossRef Search ADS PubMed 29 EUCAST. Breakpoint Tables for Interpretation of MICs and Zone Diameters, Version 7.0, 2017. www.eucast.org. © The Author 2017. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please email: email@example.com.
Journal of Antimicrobial Chemotherapy – Oxford University Press
Published: Feb 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.
All for just $49/month
It’s easy to organize your research with our built-in tools.
All the latest content is available, no embargo periods.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud