1. 1.Andersson, T, et al. Clin Pharmacokinet 2001;40:411-26. 2.Hassan-Alin, M, et al. Eur J Clin Pharmacol 2000;56:665-70. Population Pharmacokinetic Modelling of Esomeprazole Nonlinearity Joseph F Standing (1), Marie Sandström (2), Tommy Andersson (2), Kerstin Röhss (2), Mats O Karlsson (1) (1) Department of Pharmaceutical Biosciences, Uppsala University, Sweden; (2) AstraZeneca Ltd., Sweden. Esomeprazole clearance decreases, with increased dose and after repeated administration, probably due to CYP2C19 inhibition [1]. Bioavailability also increases with time and dose [2]. This study aimed to develop a PK model that describes the autoinhibition of esomeprazole clearance. Data from a previously published parallel design study were used [2]. Healthy volunteers received 40mg (n=25) or 20mg (n=16) esomeprazole by 30min intravenous (iv) infusion. After a washout, subjects then received 5 days of once-daily oral esomeprazole 40mg (capsule) or 20mg (solution), followed by a second intravenous dose on day 6. Rich blood sampling was performed after the first and last iv and oral doses. Model building began by including only the iv data; subsequently all data were pooled and analysed. Modelling was undertaken in NONMEM with linear, Michaelis-Menten and saturable turnover models investigated for CL. Background Methods Results Conclusion The model describes nonlinearities in esomeprazole pharmacokinetics. Refinement of the absorption modelling should yield decreased residual variability, and evaluation of the model with larger dose ranges would be useful. A two-compartment model with linear CL fitted the iv data well, but differences in CL with dose were ascribed to inter- individual variability. Michaelis-Menten elimination models collapsed to linear CL. The final model consisted of two disposition compartments, and CL was split into a linear component, and a saturable component described by a turnover model. The turnover model (Figure 2) was described by a target pool by which drug in both the central and depot compartments could be lost. Non-compartmental AUC(0-t) steady state/first dose ratios for 20 & 40 mg iv and oral were: 1.40, 1.96, 1.72 and 2.55 respectively. The same population predicted AUC(0-t) ratios were: 1.69, 1.88, 1.76, and 2.64 respectively. Figure 1. Visual predictive checks for 20mg iv first dose (A), 20mg iv steady- state dose (B), 40mg iv first dose (C), and 40mg iv steady-state dose (D). Shown are raw data (black open circles), median of the raw data (red line), median and 95%CI of simulated data (black line and grey shaded area), and the 10th and 90th percentiles of the simulated data (black dotted lines). Table I. Parameter estimates from the final model. ParameterModel estimate Fixed effect (RSE)Interindividual Between occ variability (RSE) variability(RSE) VD (L/70kg)11.5 (4.26)20% (26.3) CL (L/h/70kg)7.4 (15)34% (34.5) VP (L/70kg)3.97 (15.4)112% (57) Q (L/h/70kg)8.23 (15.6)- Ksyn1.75 (54.5)94% (47.9) Kon (mcmol/h)0.00717 (13.6)- Kdeg (1/h)0.0129 (31)47% (104) Ka solution (1/h)6.09 (19.5)49% (75.2) 46% (66.4) Ka capsule (1/h)4.53 (25.4)98% (47.4) Ktr (1/h)**4.88 (6.66)23% (65.3) 32% (28.6) F*0.84 (2.81)86% (27.6) IV proportional error14% (9.15)- Oral solution prop error30% (10.8)- Capsule prop error39% (11.6)- *Maximum bioavailability when target is saturated. ** Transit rate constant over 4 lag compartments Figure 2. Schematic diagram of the final model with parameters of elimination shown. Ksyn is the zero-order target synthesis rate, Kon is the second-order target-drug association constant, Kdeg is the first-order target degredation constant, and CL is the linear component of clearance. References