Effect of Cranberry Juice on Population Pharmacokinetics of Amoxicillin and Cefaclor
M. Andrew (1), M. Li (2), D. Salinger (1), P. Vicini (1), J. Wang (2), R. Grady (3), B. Phillips (4), D. Shen (1,2,4), G. Anderson (2,5)
(1) Department of Bioengineering, University of Washington, Seattle, Washington, USA; (2) Department of Pharmaceutics, University of Washington, Seattle, Washington, USA; (3) Department of Urology, School of Medicine, University of Washington, Seattle, Washington, USA; (4) Pharmacokinetics Laboratory, Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, Washington, USA; (5) Department of Pharmacy, University of Washington, Seattle, Washington, USA
Objectives: Consumption of cranberry juice with low-dose oral antibiotics is often prescribed as long-term therapy/prophylaxis for recurrent urinary tract infection. β-lactam antibiotic disposition is moderated by multiple carrier-mediated pathways that also may be modified in the presence of cranberry juice. In addition, amoxicillin disposition is known to exhibit strong dose dependence . The aim of this study was to determine whether co-administraton of cranberry juice alters β-lactam antibiotic population pharmacokinetics
Methods: In a crossover design, 18 healthy women received treatments of 500 mg or 2 g amoxicillin p.o. with water or 8 oz cranberry juice cocktail, or 500 mg cefaclor p.o. with water or 12 oz of cranberry juice cocktail. Blood samples were collected pre-dose and 0.25, 0.5, 1, 2, 3, 4, 6 and 8 hr post-dose. Urine collections were made pre-dose and at 2 hr intervals for 8 hr post-dose. Population pharmacokinetic analysis of serum data was performed using SAAM-II, SPK (University of Washington, Seattle, WA, USA) and NONMEM ver. V (Globomax LLC, Hanover, MD, USA). Candidate structural models investigated prior to final model selection included a two compartment model with first order absorption and elimination, a two compartment with delay and first order absorption and elimination, and a two compartment with Michaelis-Menten absorption and first order elimination . The final structural model consisted of one compartment with Weibull nonlinear absorption (profile based on the Weibull distribution) and first-order elimination . Standard-Two-Stage population analysis was used to seed nonlinear mixed effects model runs with full covariance matrices in order to investigate appropriate covariance models. Reduced covariance matrix structures based on block diagonal and diagonal elements were selected. Between Subject Variability (BSV) and Between Occasion Variability (BOV) were modeled as lognormally distributed. Additive and proportional models were investigated for the Residual Unknown Variation (RUV) error model. FO and FOCE estimation methods were explored. The influences of dose, juice consumption and bodyweight on apparent clearance, apparent volume and the Weibull absorption shape (S) and scale (TD) parameters were examined. Several goodness-of-fit measures were used to evaluate model fit.
Results: FO method was selected due to difficulties achieving convergence with FOCE method. True likelihood profiling was used to verify the adequacy of FO method in select cases. Visual inspection revealed that individual predictions fit the data well. Data values were noted to be evenly distributed around the individual predictions, and the weighted residuals were evenly distributed around zero. The population predictions for amoxicillin exhibited a slight bias, overpredicting data values for the highest predicted concentrations. This latter is hypothesized to result from the model failing to capture some of the variation due to approaching the limits of parameterization. Population predictions of cefaclor were well distributed with respect to the data values.
Amoxicillin was best described using an additive error model (RUV of 0.689 mcg/ml), while cefaclor was best described by a proportional error model (RUV of 26.5%). Final amoxicillin and cefaclor models included as covariates a juice consumption effect on the Weibuill scale parameter TD and a bodyweight effect on apparent volume. The amoxicillin final model also included dose-dependence effects on CL/F, VOL/F and the Weibull scale parameter TD.
Amoxicillin CL/F of 22.1 ± 1.04 (standard error) L/hr for a 500 mg dose, with 10.6 ± 1.05 L/hr ΔCL/F due to high dose, correlated well with noncompartmental estimates of 20.8 – 22.0 L/hr for CL/F at 500 mg and 32.3-33.6 L/hr for total CL/F at 2 g dose. BSV and BOV on CL/F were 18.7 and 15.0%, respectively. Estimated dose-dependence covariates on amoxicillin CL/F and VOL/F were found to exhibit relative bioavailability comparable to an independent estimate obtained using urinary excretion data. Population values of amoxicillin VOL/F were 22.5 ± 1.06 L for low dose with 10.6 ± 1.05 L ΔVOL/F due to high dose administration. BSV and BOV were 72.6 and 35.6%, respectively. Weibull population values S and TD were 2.86 ± 0.11 (unitless) and 1.31 ± 0.19 hr, respectively, with BSV of 32.9 and 10.8%, respectively, and BOV of 32.7 and 13.4%, respectively. Bodyweight produced a change in VOL/F of 105 ± 93 ml per kg-bodyweight difference from median. High dose administration increased TD from baseline by 0.375 ± 0.164 hr, or 29%. Juice co-administration increased TD from baseline by 0.533 ± 0.074 hr, or 41%. In simulations using the population parameter values, increased TD produced a modest decrease in Cmax and increase in Tmax. This correlates well with the statistically significant increase in Tmax observed in noncompartmental results for doses co-administered with juice.
Cefaclor CL/F of 24.4 ± 1.24 L/hr correlated well with noncompartmental estimates of 23.8 – 24.5 L/hr. BSV on CL/F was 19.7%; no BOV was indicated. VOL/F was 21.5 ± 1.27 L, with 17.4% BSV and 10.5% BOV. Weibull population parameter values of S and TD were 1.73 ± 0.18 (unitless) and 0.704 ± 0.05 hr, respectively, with BSV of 69.7 and 26.8%, respectively, and BOV of 55.2 and 27.8%, respectively. Bodyweight produced a change in VOL/F of 146 ± 43 ml per kg-bodyweight difference from median. Juice co-administration increased TD from baseline by 0.160 ± 0.060 hr, or 23%, which resulted in a slight decrease in Cmax and increase in Tmax. This change correlates with the statistically significant decrease in Cmax observed in noncompartmental results for co-administration with juice.
Conclusions: The Weibull nonlinear absorption function and model parameterization to accommodate dose-dependent relative bioavailability provide a useful framework for assessing changes in absorption and overall disposition. Cranberry juice cocktail has a statistically significant but modest effect on the population pharmacokinetic profiles of amoxicillin and cefaclor. Changes in the absorption profile of both drugs cause slight to modest increases in Tmax and decreases in Cmax. Amoxicillin also exhibits a strong dose dependence that appears both to affect relative bioavailability in the form of significantly reduced Cmax and AUC, and to cause a slight increase in Tmax at high doses. These findings are consistent with previously published clinical pharmacokinetic results. The modest effects of cranberry juice consumption on disposition suggest that β-lactam antibiotics may be safely co-administered with typical servings of cranberry juice cocktail.
 Sjövall J, Alván G, Westerlund D. Dose-dependent absorption of amoxycillin and bacampicillin. Clin Pharmacol Ther. 1985 Sep;38(3):241-50.
 Reigner BG, Couet W, Guedes JP, Fourtillan JB, Tozer TN. Saturable rate of cefatrizine absorption after oral administration to humans. J Pharmacokinet Biopharm. 1990 Feb;18(1):17-34.
 Andrew MA, Easterling TR, Carr DB, Shen D, Buchanan ML, Rutherford T, Bennett R, Vicini P, Hebert MF. Amoxicillin pharmacokinetics in pregnant women: modeling and simulations of dosage strategies. Clin Pharmacol Ther. 2007 Apr;81(4):547-56.