Automatic identification of MCI progressors significantly reduces the number of subjects in clinical trials

Abstract

AbstractBackgroundAccurately predicting the progression rate in patients with mild cognitive impairment and mild dementia due to AD enables enrichment of trial populations. Here, we evaluate the potential use of our prognostic model (Abstract #54217) as a screening tool for enrichment in clinical trials in early AD.MethodData included 312 patients with mild AD from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and were chosen based on the Clinical Dementia Rating (CDR=0.5) and the Mini‐Mental State Exam (MMSE, range 24‐30) scores. We classified subjects into stable and progressive (defined by at least a 2‐point increase in CDR‐SB) using our automatic classification tool. Required sample sizes to detect a reduction in the mean annual rate of cognitive decline based on CDR‐SB score were estimated. We corrected for the annualized decline due to normal aging so as not to overestimate the benefit of enrichment when computing the treatment effect. We estimated and compared sample sizes for two groups of subjects: Using data from the mild AD subjects in the ADNI dataset that fit the selection criteria above (the unenriched group) and using only the subset of those ADNI MCI subjects identified as pMCI using baseline data and the classifier described above (the enriched group).ResultFigure 1 shows the required sample sizes for different treatment effects for both unenriched and enriched MCI cohorts. Using the unenriched group of MCI subjects, power analysis shows that 1075 subjects are required in a 2‐year (764 subjects for a 3‐year) trial of therapy with a hypothesized 25% effect size (80% power and 5% significance level) to reduce cognitive decline. When using the enriched cohort of MCI subjects, only 279 subjects are required for a 2‐year (373 for 3‐year) trial (solid lines in Fig. 3). These results demonstrate that enrichment using baseline HC, MoCA, ADAS‐13 and MMSE yields a 3.8‐fold decrease in the sample size for a 2‐year study (2.8‐fold decrease for a 3‐year study).ConclusionEnriching clinical trial cohorts of subjects in the early stages of AD with a higher chance of functional and cognitive decline will improve efficiency with smaller, faster, less expensive trials.