Welten C C M, Koeter M W J, Wohlfarth T, Storosum J G, van den Brink W, Gispen-de Wied C C, Leufkens H G M, Denys D A J P
Dept. of Psychiatry, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands; Medicines Evaluation Board, Utrecht, The Netherlands.
Dept. of Psychiatry, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands.
Eur Neuropsychopharmacol. 2015 Jul;25(7):1018-26. doi: 10.1016/j.euroneuro.2015.03.010. Epub 2015 Mar 30.
We examined the role of placebo response in acute mania trials. Specifically, whether placebo response: (1) predicts treatment effect, (2) can be predicted by patient and study characteristics, and (3) can be predicted by a parsimonious model. We performed a meta-analysis of individual patient data from 10 registration studies (n=1019) for the indication acute manic episode of bipolar disorder. We assessed the effect of 14 determinants on placebo response. Primary outcome measures were mean symptom change score (MCS) on the Young Mania Rating Scale (YMRS) and response rate (RR), defined as ≥ 50% YMRS symptom improvement from baseline to endpoint. The overall placebo response was 8.5 points improvement on the YMRS (=27.9%) with a RR of 32.8%. Placebo response was significantly associated with the overall treatment response. Five determinants significantly (p<0.05) predicted the placebo response. The multivariate prediction model, which consisted of baseline severity, psychotic features at baseline, number of geographic regions, and region, explained 10.4% and 5.5% of the variance in MSC and RR, respectively. Our findings showed that the placebo response in efficacy trials of antipsychotics for acute mania is substantial and an important determinant of treatment effect. Placebo response is influenced by patient characteristics (illness severity and presence of psychotic features) and by study characteristics (study year, number of geographic regions and region). However, the prediction model could only explain the placebo response to a limited extent. Therefore, limiting trials to certain patients in certain geographic regions seems not a viable strategy to improve assay sensitivity.
我们研究了安慰剂反应在急性躁狂症试验中的作用。具体而言，安慰剂反应是否：（1）预测治疗效果，（2）可由患者和研究特征预测，以及（3）可由一个简约模型预测。我们对10项注册研究（n = 1019）中关于双相情感障碍急性躁狂发作适应症的个体患者数据进行了荟萃分析。我们评估了14个决定因素对安慰剂反应的影响。主要结局指标为杨氏躁狂量表（YMRS）上的平均症状变化评分（MCS）和缓解率（RR），RR定义为从基线到终点YMRS症状改善≥50%。总体安慰剂反应为YMRS改善8.5分（=27.9%），RR为32.8%。安慰剂反应与总体治疗反应显著相关。五个决定因素显著（p<0.05）预测了安慰剂反应。由基线严重程度、基线时的精神病性特征、地理区域数量和区域组成的多变量预测模型分别解释了MCS和RR中10.4%和5.5%的方差。我们的研究结果表明，抗精神病药物治疗急性躁狂症疗效试验中的安慰剂反应相当显著，是治疗效果的一个重要决定因素。安慰剂反应受患者特征（疾病严重程度和精神病性特征的存在）和研究特征（研究年份、地理区域数量和区域）影响。然而，该预测模型只能在有限程度上解释安慰剂反应。因此，将试验限制于特定地理区域的特定患者似乎不是提高分析灵敏度的可行策略。
