Predicting the time to clinically worsening in mild cognitive impairment patients and its utility in clinical trial design by modeling a longitudinal clinical dementia rating sum of boxes from the ADNI database.

BACKGROUND: Growing interest in treating Alzheimer's disease (AD) patients in the earliest stages requires new clinical endpoints. Currently, there is no established clinical endpoint or treatment duration for mild cognitive impairment (MCI) trials.
OBJECTIVE: This analysis attempts to answer "how long the MCI clinical trial would be necessary" using the Clinical Dementia Rating Sum of Boxes (CDR-SB) as a clinical endpoint, where CDR-SB is an example of a suitable tool to assess both cognition and function as a single primary efficacy outcome.
METHODS: A longitudinal model was developed to predict the CDR-SB time-profile. The CDR-SB is considered ideal to assess both cognition and function as a single primary endpoint in MCI trials. The median time for clinically "worsening", defined using several thresholds for change from baseline, was calculated using individual CDR-SB predictions. Covariates predictive of worsening were also evaluated.
RESULTS: The median time to a 1-point change in CDR-SB was approximately 2 years in MCI patients. Higher baseline severity in disease, lower hippocampal volume, and ApoE4 carrier status were significant covariates predicting shorter times to worsening (faster progress). The results indicate that at least a 2-year trial would be necessary with 30% (or more) disease modifying drug with a sample size of n = 350 to detect the significant difference from placebo (80% power) and to achieve the target mean effect size of 0.5 point change in CDR-SB.
CONCLUSION: Predictions of CDR-SB changes from a longitudinal model are able to inform study design and possible enrichment strategies, based on covariate analyses, for prospective planning of clinical trials in MCI patients.