Abstract
INTRODUCTION: As currently used, the Alzheimer's Disease Assessment Scale-Cognitive subscale (ADAS-Cog) has low sensitivity for measuring Alzheimer's disease progression in clinical trials. A major reason behind the low sensitivity is its sub-optimal scoring methodology, which can be improved to obtain better sensitivity. METHODS: Using item response theory, we developed a new scoring methodology (ADAS-CogIRT) for the ADAS-Cog, which addresses several major limitations of the current scoring methodology. The sensitivity of the ADAS-CogIRT methodology was evaluated using clinical trial simulations as well as a negative clinical trial, which had shown an evidence of a treatment effect. RESULTS: The ADAS-Cog was found to measure impairment in three cognitive domains of memory, language, and praxis. The ADAS-CogIRT methodology required significantly fewer patients and shorter trial durations as compared to the current scoring methodology when both were evaluated in simulated clinical trials. When validated on data from a real clinical trial, the ADAS-CogIRT methodology had higher sensitivity than the current scoring methodology in detecting the treatment effect. CONCLUSIONS: The proposed scoring methodology significantly improves the sensitivity of the ADAS-Cog in measuring progression of cognitive impairment in clinical trials focused in the mild-to-moderate Alzheimer's disease stage. This provides a boost to the efficiency of clinical trials requiring fewer patients and shorter durations for investigating disease-modifying treatments.