PRE-SCREENING FOR EPISODIC MEMORY DEFICITS CAN ENRICH FOR ELIGIBLE SUBJECTS IN EARLY ALZHEIMER’S DISEASE CLINICAL TRIALS

Abstract

Positive biomarkers and a mild cognitive deficit define Prodromal Alzheimer's disease (AD). A key challenge of clinical trials is identifying subjects to meet screening criteria for both biomarker and cognitive deficits. Scores on the CANTAB Paired Associates Learning (PAL) task of episodic memory correlate with AD biomarkers. Pre-screening with PAL is easy to implement and is time as well as cost effective. Here we assess the use of PAL to enhance samples for biomarker positive subjects and deficits in other cognitive screening tasks; RBANS total score, delayed memory index, Clinical Dementia Rating Scale (CDR), and Mini-mental state examination (MMSE). During screening for an early Alzheimer's disease trial, data was collected on cognitive function and CSF biomarkers (AB1–42,Tau) in adults aged over 50 (n= 56). Estimates for improving sample selection were based on the age-specific probability of biomarker positivity in the general population together with the sensitivity and specificity of episodic memory tasks (PAL and RBANS). Ability of PAL to detect a deficit in RBANS, CDR and MMSE was based on demographically adjusted logistic regression, and subject scores relative to quantile bands of normative (PAL) data. The correlation between PAL and the RBANS delayed memory index was -0.81, and PAL predicted a deficit in RBANS (>-1sd) with an accuracy of 89%; the AUC was 0.96. Participants with a predicted deficit on RBANS also showed a lower average MMSE score (24.8 vs 28.6) and a higher CDR-SOB score (1.85 vs 0.48). Performance on PAL (-1sd) worse than age adjusted norms (∼1700) displayed high specificity for identifying deficits in the RBANS and CDR. The number of subjects screened to identify those who are AB biomarker positive is reduced with increasing impairment on PAL, in accordance with the sensitivity and specificity. PAL can validily identify subjects’ with episodic memory deficits, early in the recruitment process, to enrich samples for biomarker positive subjects, and identify those likely to show deficits in other cognitive tasks. As such, PAL has potential for reducing trial costs, screen failure rates, site and patient burden, by targeting subjects with a high probability of meeting screening criteria.