Unpacking cognitive composites: A longitudinal analysis

Abstract

AbstractBackgroundThe development of cognitive endpoints that can accurately assess changes in cognition over short time frames is crucial for clinical trials and research of Alzheimer’s disease (AD). Understanding the changing influence of contributing test scores on composites throughout the disease course provides the opportunity to optimise cognitive composite scores for different stages of AD.MethodAIBL participants with declining cognitive performance were included in this study N=1275 [688 cognitively unimpaired (CU), 277 mild cognitively impaired (MCI), 310 AD; aged 73±9; 718 females]). Two cognitive composite scores (Episodic Memory (EM) and PACC) and their component test scores (California Verbal Learning Test‐II Delayed Recall (CVLT‐II DR), Logical Memory Delayed Recall (LMII), Rey Complex Figure Test 30 minute delayed recall (RCFT‐DR) and CVLT‐II DR, LMII, Digit Symbol Substitution Test (DS), MMSE, respectively) were evaluated. We first examined the relationship between each of component tests score for each composite. We then compared the extent to which longitudinal trajectories of each component test score and each cognitive composite score differed at each disease stage.ResultCVLT‐II DR contributed the most to the EM composite followed by RCFT‐DR and LMII with the influence remaining unchanged across each disease stage. For PACC, CVLT‐II DR contributed the most to the initial decline, with MMSE and LMII contributing similar amounts and DS contributing the least. CVLT‐II DR contributed substantially to changes in PACC earlier in the disease course but MMSE drove the PACC change in later stages of disease. Initially, both composites follow similar longitudinal trajectories. However, the EM composite reaches a floor not observed for the PACC.ConclusionUnderstanding the temporal contribution of component tests scores on cognitive composites could provide improved cognitive endpoints tailored to use. For instance, MMSE is sensitive to change later in the disease trajectory and therefore should be included in a composite endpoint for trials in prodromal or clinical AD, however is unlikely to have value for preclinical AD trials.