Abstracts

Rationale: Placebo response appears to be rising in randomized epilepsy trials, and recently has threatened the ability to demonstrate a difference between active drug and placebo. In neuropathic pain trials, recent work by Quessy and Rowbotham (2008) show the tendency for placebo cohorts to improve in the later time points of double-blind placebo controlled studies. We have observed this phenomenon in several trials in epilepsy and other disease areas which we have studied. Here we describe this phenomenon in pregabalin epilepsy studies and explore potential factors associated with it. Specifically, we asked whether differential dropout of patients with higher seizure frequency might account for the rise in placebo effect. We also addressed other possible explanatory factors for the placebo effect. Methods: We used data from a large Pfizer database of trials which studied add-on pregabalin treatment (N=1305), an AED approved as adjunctive therapy for adult patients with partial onset seizures. We combined data from four double-blind, randomized, placebo controlled studies. We summarized and modeled percent seizure free patients per day. We examined dropout patterns and associations with seizure outcome. We also looked for subgroups with higher levels of placebo response, including age, gender and baseline seizure frequency using a data-mining tool called Activity Region Finder (ARF). Results: Among the placebo patients the median percent change from baseline in seziure rate was an increase of 21.5% for non-completers versus a decrease of 3.6% for completers (p<0.001 for testing the difference between completers and non-completers). At baseline phase (6-8 weeks), both pregabalin and placebo patients showed a similar per cent seizure free. The proportion of seizure free patients on any given day was similar for placebo and pregabalin patients during baseline and differed significantly post-randomization. However, among the placebo patients a trend toward greater daily seizure freedom was observed for later time periods in the studies. As modeled using a linear approximation the trend for the placebo group was significant (p=0.0002) (see Figure 1). ARF analysis suggested a subgroup of placebo patients with baseline seizure rate between 8.75 and 11.5 have a significantly higher responder rate (26% versus 10%, n=43, p < 0.05). ARF also suggested that elderly (52-82 years old) placebo patients are more likely to complete the study (95% versus 82%, n=56, p<0.05). Conclusions: Placebo response increased over time. Capacity to complete 12 weeks of study was associated with greater placebo response. The longitudinal aspects of this effect are under further investigation.