CLINICAL PfIARMACOLOGY & THERAPEUTICS P 8 6 American Society for Clinical Pharmacology and Therapeutics FE~RU~Y20O3

PIII-87 AN INTEGRATED PHARMACOKINETIC/PHARMACODY-

NAMIC (PK/PD) MODEL FOR SB-480848 INHIBITION OF PLASMA LIPOPROTEIN-ASSOCIATED PHOSPHOLIPASE A2 (LP-PLA2) ENZYME ACTIVITY IN HUMAN. M. H. Maeg~ PharmD, B. E. Ilson, MD, B. C. Shaddinger, PharmD, M. Hossain, PhD, GlaxoSmithKline, King of Prussia, PA.

Purpose: SB-480848 is a novel, selective Lp-PLA a inhibitor un- der development for the treatment of atherosclerosis. An integrated PK/PD model was developed to characterize the effects of SB- 480848 on plasma Lp-PLA 2 enzyme activity. Methods: SB-480848 plasma concentrations attained following 20 and 40 mg repeat oral dosing to 10 subjects were used to establish the PK model. SB- 480848 concentration and Lp-PLA 2 enzyme activity in plasma fol- lowing 0.5 to 100 mg single oral dosing to 31 subjects were used to characterize the concentration-effect relationship. Modeling and sim- ulations were performed using NONMEM V and Pharsight Trial Simulator. Results: The PK/PD parameter estimates are presented in the following table:

ka CL/F V2/F V3/F Q E o ICs0 h -1 L/h L L L/h nM/min/mL ng/mL Gamma

Pop.Mean 0.467 198 4030 8730 473 22.4 4.37 1.88 (% CV) (36) (6.8) (28) (16) (27) (4.4) (3.7) (3.3)

IIV % N.E. 20 49 N.E. 29 25 7.9 4.0

Residual Variability (PK) proportional (% CV) 14% (26) Residual Variability (PD) proportional (% CV) 20% (22)

additive (nM/mirdmL) 0.17 (184) (% cv)

N.E.: no t e s t i m a t e d

Conclusion: Simulations Ii'om this integrated PK/PD model de- veloped using Phase I data guided dose selection and duration of dosing in designing SB-480848 Phase IIa development program.

PIII-88 A STUDY OF LIVER TRANSAMINASES DURING PLACEBO

TREATMENT IN PHASE I DRUG TRIALS. A. Mathur, PhD, R. Wang, PhD, D. Wallace, MD, S. Patterson, A. Mortimer, L. Brown, MD, W. V. Williams, MD, B. I/son, MD, E. O'Donneli, P. Fernan- dez, MD, GlaxoSmithKline, Philadelphia, PA.

Drug-induced hepatotoxicity (DIH) is of concern during new drug development. A slight to moderate increase of serum transaminases is a recognized phenomenon in clinical trials even during placebo treatment. A major challenge in new drug development is to distin- guish between background transaminase elevations (TE) which are observed sporadically and drug-induced TE which may herald DIH. The present study examined the inherent variation of liver function tests (LFTs) including alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl transferase (GGT), alka- line phosphatase (AP), and total bilirubin (T bifi) in 386 subjects from 8 studies during placebo treatment in Phase I multiple dose trials. Subjects with ALT results that exceeded their first predose (baseline) value by 10 IU or more at least once during placebo administration were arbitrarily classified as 'ALT-susceptible' (ALT-S). Based upon this definition 10% of subjects were ALT-S. The following covariates were considered as possible predictors of ALT-S: gender, age, race, weight, height, body mass index (BMI), screening LFT and choles- terol (CHOL) and triglycerides (TRIG) values. Univariate tests (t-test and Fisher's exact test) comparing ALT-S and ALT-nonsusceptible (ALT-N) groups indicated significant differences in screening values of ALT(p=0.0534), AP(0.0352), T bili(0.0062), GGT(0.0179), age(0.0375), gender(0.00002) and weight(0.0362).

PIII-89 WHEN SHOULD DOSE BE ADJUSTED TO BODY SIZE? A

POPULATION PHARMACOKINETC (PPK) SIMULATION. J= Lu, PhD, S. Gourlay, MD, R. Bruno, PhD, Genentech, Inc., South San Francisco, CA.

Objectives: To understand the impact of weight (WT)-clearance (CL) relationships and inter-patient variability on drug exposure following weight-based vs . flat dosing. Methods: Based on a previ- ously developed PPK model, a total of 15 different scenarios were investigated (500 clinical trials simulated for each scenario, Pharsight Simulator 2.0). In each simulated trial, 400 patients were randomized to one of two treatment arms: the weight-based dosing arm (2 mg/kg), and the flat dosing arm (140 rag). Drug exposure variability, the proportion of patients with steady-state concentration greater than a threshold level (P), and the exposure difference between heavy and light patients (D) were considered as main endpoints. Results: The simulation revealed that even when there is a relationship between WT and CL (CLina=CLsto (WTinJ70) ~') but eL ~< 0.5, flat dosing would do better than weight-based dosing in terms of reducing the drug exposure variability, maximizing P, and minimizing D. Weight- based dosing would only be beneficial if there is a strong relationship between WT and CL (a-->0.75). For intermediate values of a (a =0.5 or 0.75), dose adjustment to a power of WT (close to body surface area) would be better. Conclusions: In summary, the choice of the right dose-adjustment should be based on PK characteristics rather than convention (e.g. . . mg/kg or mg/m a in oncology). Weight-based dosing should only be used if a strong relationship between WT and CL is found in early PK data.

PIII-90 POPULATION PHARMACOKINETICS OF EFALIZUMAB

(HUMANIZED MONOCLONAL ANTI-CD 1 la ANTIBODY) FOL- LOWING WEEKLY DOSING IN PSORIASIS PATIENTS. Y~ Sun, PhD~ J. Lu, PhD, J. Amita, Ptff), P. Compton, PhD, R. A. Bruno, PhD, Genentech, Inc., South San Francisco, CA.

Purpose: To perform population pharrnacokinetic (PK) analysis and to determine covariates affecting efalizumab PK in psoriasis patients. Methods: Efalizumab PK data obtained from 1088 patients who received 1 or 2 mg/kg weekly subcutaneous injections were included in the population analysis. The Base Model (1- compartment, pt -order absorption) was built using a rich dataset from a Phase I study (N=64). The Final Model was determined after the sparse data from three Phase III/IIIb studies with Days 42 or/and 84 trough levels (N= 1024) were incorporated in the analysis (NON- MEM V). Results: The population mean (estimation CV%) values for apparent volume of distribution, absorption rate constant, and apparent clearance (CL/F) were 9.13 L (19.6%), 0.191 Day 1 (35.2%), and 1.29 L/Day (4.1%) for a typical subject receiving 1 mg/kg dose. The inter-individual variability in C L ~ was 48.2% (6.3%). The most important covariate affecting CL/F was baseline body weight (CL/F estimates ranged from -29 .5% to 37.6% for the 2.5 th (57.2 kg) and 97.5 th (139 kg) percentile patient weight). Other covariates in the Final Model (obesity, baseline lymphocyte counts, PASI score, and age) only had modest effects. Patients in the 2 mg/kg dose group had a 24.0% lower CLfF from the population mean CL/F value. This is consistent with the nonlinear PK of efalizumab. Con- clusion: The results support the current dosing strategy, that is, administration of efalizumab based on a body weight-adjusted dose [mg/kg].