OII‐A‐1

BACKGROUND To develop a disease progression model that describes the time course of ACR‐N in RA patients treated with placebo. METHODS Using 536 ACR‐N values from 80 placebo‐treated RA patients enrolled in a 6‐month, randomized, double‐blind study of etanercept, conducted in 13 centers, a disease progression model was developed. Mixture model approach implemented in NONMEM (ver 5.1) was applied with the hybrid estimation method. RESULTS A two normal probability curve model, with each curve multiplied by a coefficient, adequately described the time course of ACR‐N. Based on the multiplication coefficients estimated by mixture model approach, three placebo response groups were identified: no (41.9%), moderate (30.4%) and high (27.7%) responders. Higher baseline value of Health Assessment Questionnaire (BHAQ) significantly decreased the multiplication coefficient of the second curve. Typical patients with BHAQ of 1.7 in the moderate and high responders had early and late peak responses at 4.8 (4.5–5.1) and 19.5 (18.7–20.3) weeks after treatment, respectively. Their ACR‐N values at those peak weeks were 6 (0–12) and 11 (8–15) in the moderate responders, and 25 (2–51) and 65 (54–77) in the high responders, respectively. CONCLUSIONS These results imply that placebo‐treated RA patients show a bi‐modal ACR‐N response pattern with the late peak being higher. Patients with higher BHAQ have lower late ACR‐N. A disease progression model is useful to identify the placebo response pattern in this population. Clinical Pharmacology & Therapeutics (2005) 79 , P3–P3; doi: 10.1016/j.clpt.2005.12.010