Exposure-Response Modeling of the ACR50 Score in Rheumatoid Arthritis Patients Treated with Certolizumab Pegol.
Brigitte D. Lacroix(1,2), Mark Lovern(1), Armel Stockis(1), Maria Laura Sargentini-Maier(1), Mats O. Karlsson(2) and Lena E. Friberg(2)
(1)Pharmacometrics, Global Exploratory Development, UCB Pharma SA, Belgium; (2)Department of Pharmaceutical Biosciences, Uppsala University, Sweden
Objectives: To describe the effect of exposure to certolizumab pegol (CZP) on the ACR20 [1] and ACR50 (American College of Rheumatology 20% and 50% improvement criteria) in rheumatoid arthritis (RA), by developing pharmacodynamic Markov mixed-effects models, and to compare the outcomes from the two models.
Methods: Exposure response analysis of ACR20 and ACR50 data was undertaken as representative measures of response and remission respectively. Data from 1747 patients treated with CZP and 633 patients treated with placebo were used for non‑linear mixed effects modeling using NONMEM VI. Placebo or CZP at doses ranging from 50 to 800 mg was administered subcutaneously every 2 or 4 weeks (Q2W/Q4W), with or without concomitant administration of methotrexate, for 8 to 48 weeks. At each visit, the ACR (20 or 50) outcomes (responder/non-responder) and drop out events were coded in 3 categories. The probabilities of the transitions between these 3 different states were modeled using logit functions and described as function of CZP exposure, disease- and patient-related factors. The models were used to predict the clinical outcome following various treatment regimens in a variety of patient sub-populations.
Results: For both the ACR20 and ACR50 outcomes, the Markov model accurately described the response and dropout events. The best models combined Emax functions on the logit scale to describe the increase in the probability of becoming a responder and the decrease in the probability of becoming a non-responder as a function of the average plasma concentration between successive doses (Cavg). The population value of EC50 was similar for both outcomes (17 and 12 µg/mL, respectively). Simulations from the final models predicted similar response probabilities for the 200 mg Q2W and 400 mg Q4W dosing schedules (response rates of 0.71 vs. 0.69 for ACR20 and 0.40 vs 0.39 for ACR50, at week 22). Loading doses of 400 mg at weeks 0, 2 and 4 were predicted to result in a faster onset of response, with a maximum effect observed at week 8 (median response rate increased by about 10%).
Conclusions: Significant exposure-response relationships for CZP were demonstrated for both the ACR20 and ACR50 outcomes. The exposure-response model supports the proposed dosing regimen of 400 mg at weeks 0, 2 and 4 followed by 200 mg Q2W for CZP in RA. The model also supports 400 mg Q4W as an alternative dosing regimen. A future approach would be to model the two outcomes simultaneously as this would allow simulating consistent ACR20 and ACR50 response at a given visit for a given subject.
References:
[1] Lacroix. Exposure-Response Modeling of the ACR20 Score in Rheumatoid Arthritis Patients Treated with Certolizumab Pegol. PAGE 17 (2008) Abstr 1318 [www.page-meeting.org/?abstract=1318]