Related Experiment Videos
A model to select regimens for phase III trials for patients with advanced-stage non-small cell lung cancer
Boris Freidlin1, Oscar S Breathnach, Bruce E Johnson
1Biometric Research Branch, National Cancer Institute, Bethesda, MD 20892, USA.
Purpose:
Historical data from pilot, Phase II, and Phase III studies for patients with advanced-stage non-small cell lung cancer (NSCLC) were used to evaluate a statistical model developed to provide assistance in selecting regimens from pilot studies for subsequent use in larger Phase III randomized studies.
Experimental Design:
Information from 33 Phase III trials for patients with advanced-stage NSCLC performed from 1973 and 1994 in the United States and Canada was collected. The data from antecedent pilot or Phase II and subsequent Phase III trials were analyzed using a predictive statistical model. This model uses the number of patients in the pilot/Phase II study, the median survival of patients in the pilot, and the number of deaths observed, to estimate the statistical likelihood that the pilot regimen will be shown superior to standard therapy in a subsequent Phase III trial.
Results:
Ten pilot/Phase II studies were identified that preceded eleven subsequent Phase III studies. The three pilot regimens associated with Phase III trials, revealing statistically significant longer survival, had an expected power of 0.69, 0.85, and 0.94 respectively. The regimens from the seven other pilot studies for which the median power expected was 0.38 (range, 0.07-0.80) showed no difference when compared with standard treatment in a Phase III trial.
Conclusion:
The use of the expected power model provides an important enhancement to the screening of new therapies. Regimens with an expected power of >0.55 may be good candidates for testing in Phase III trials.