Gijs Santen(1), Roberto Gomeni(2), Meindert Danhof(1) and Oscar Della Pasqua(1)
(1) Division of Pharmacology, LACDR, Leiden University, Leiden, the Netherlands. (2) Department of Clinical Pharmacokinetics/Modeling & Simulation, GlaxoSmithKline, Verona, Italy
Objectives: Depression trials are known to show large placebo response, which makes it very difficult to separate placebo effect from drug effect before 8 to 12 weeks treatment. The golden standard assessment tool in depression studies is the Hamilton Depression Rating Scale (HAMD), which consists of 17 items which are summed up to provide a total score. Thus far, little attention has been paid to the sensitivity of the HAMD to placebo response and no conclusive evaluation is available about the impact of disease severity on the onset and time course of depression symptoms. Objective of this investigation was to explore the sensitivity of HAMD to treatment effect and characterise the time course of response using retrospective data from clinical studies with SSRIs.
Methods: Data from 8-week to 12-week double blind, placebo-controlled studies were used. The HAMD was analysed to assess the importance of specific sub-scores and their sensitivity to drug efficacy (clinical response defined as >50% sustained decrease from baseline). Data exploration and sensitivity analysis was performed in R. Modelling was performed using NONMEM V.
Results: Exploratory analysis of HDRS revealed that whilst some of the clinical items can clearly distinguish between responders and non-responders, many of the sub-scores do not show any sensitivity to response or variation over time. A sub-scale was developed incorporating only the sub-scores that showed a clear sensitivity to treatment response over time. The profiles and variability of in HDRS as well as the proposed sub-scale could be adequately characterised by a mono-exponential mixture model. According to this evaluation, the 7-item new subscale appeared to be more sensitive to detect drug effect than the 17-item HAMD.
Conclusion: Not all items of the HAMD scale are equally sensitive to drug treatment. This may have consequences in the analysis of depression studies, especially in those cases where a relatively small population is enrolled. A new subscale is presented which may help to overcome these issues.
Reference: PAGE 14 (2005) Abstr 707 [www.page-meeting.org/?abstract=707]
Poster: poster