 |
| April 15, 2010 Part 3 In this third and final installment, we’ll explore why the virtual study failed and lessons learned. If you would like to review previous installments, click on one or both of the following links: Part 1 or Part 2. The unanswered question is: Why did this virtual study fail to separate the active comparator from placebo? First, enrollment was quite rapid such that approximately 200 patients were recruited in less than 6 months. The recruitment speed was particularly rapid early in the study suggesting the possibility that sites already had a pool of patients to enroll. Second, the entry rating scale score from the investigational sites was notably higher than the score from a central reader. Nevertheless, the investigator’s assessment was the one which determined inclusion into the trial. That decision may have lead to the inclusion of a subset of patients who were not severe enough to provide information of value to this trial. Third, there were over 25 study sites such that there was only approximately 8 patients/site. Fourth, the study permitted liberal use of concomitant anxiolytics to control anxiety/agitation and thus keep patients in the study during the observational period in hospital. That may also have contributed to the failure of this trial. The failed nature of this virtual trial highlights that the adaptive design approaches alone cannot guarantee success. There also has to be an emphasis on getting the fundamentals of the clinical trial right. That means choosing a patient population and the endpoints with a goal to optimize the ability to detect a treatment effect, avoiding confounding factors such as uncontrolled concomitant medication and avoiding the all-too-universal urge to recruit quickly at all cost (including the cost of a failed trial and perhaps a failed development program). These admonitions are particularly relevant in Central Nervous System (CNS) indications where disease and outcome are often assessed using subjective endpoint measures (i.e., patient reported symptom scales). That is the reason why the biomarker endpoints are also gaining ascendancy in CNS drug development research, particularly in early phase proof of concept studies. |
 |