Jeffrey Wald, Jagdev Sidhu, David Blum, and Marianne Silver
GlaxoSmithKline, Research Triangle Park, USA and Harlow, UK.
Lamotrigine was studied for pain associated with diabetic peripheral neuropathy in replicate, randomized, double-blind, placebo (PBO) controlled, multicenter studies. Population PK/PD modeling was used to elucidate the exposure (post-hoc estimates of lamotrigine steady-state AUC) versus effect (response on an 11-point ordered categorical scale) as a function of time and baseline pain scores.
Methods: Each study included a 2-4 week screening phase; a 1-week baseline phase; a 7-week dose-escalation phase; and a 19-week treatment period. A total of 720 patients were enrolled in the 2 studies at doses of 0, 200, 300, and 400 mg/day (given BID). A longitudinal PK/PD model for ordinal response data was fit to individual patient diary pain data using NONMEM. The model accounted for the temporal profile of PBO response, individualized exposure to drug, and baseline pain status.Â
Results: The primary outcome, pain intensity change from baseline to week 19, was achieved for only the 400mg/day dose group in one of the two studies. Moreover, the dose-response relationship was not rank ordered in comparison to efficacy for 1 of the 2 trials. The PK/PD model was successful in characterizing the major features of trial results. Moreover, the model successfully accounted for the time profile and distribution of PBO response. Lamotrigine steady-state AUC and baseline were both strongly significant predictors of effect.
Conclusions: An exposure-response relationship was established for lamotrigine in painful diabetic neuropathy. The complexity of trial outcomes required a model-based approach to elucidate the exposure-response relationship and explore performance factors of the current trials, and mechanisms to improve upon the design of future trials.Â
Reference: PAGE 14 (2005) Abstr 740 [www.page-meeting.org/?abstract=740]
Poster: poster