The pattern of performance on sensitive tests of four cognitive domains differentiates groups with progressive cognitive decline in Alzheimer's disease

• Published in: Alzheimer's & dementia Vol. 16
• Main Authors: Maljkovic, Vera, Battioui, Chakib
• Format: Journal Article
• Language: English
• Published: 01.12.2020
• ISSN: 1552-5260; 1552-5279
• DOI: 10.1002/alz.047027

Background A screening challenge for Alzheimer’s disease (AD) is a paucity of tests sensitive to the different stages of the disease continuum. Progression of clinical disease suggests that multiple tests, sensitive to individual stages of disease development, may better define the range of cognitive abilities and predict the next cognitive domain to decline. Method 78 healthy controls and 29 clinically diagnosed cognitively impaired participants (23 Mild Cognitive Impairment, MCI, 6 mild Alzheimer’s dementia, mildAD) ages 60‐76, self‐administered at home 9 CANTAB tasks on a provided iPad (Maljkovic et al, AAIC 2019). Four tasks were selected for further analysis: spatial working memory (SWM), paired‐associates learning (PAL), pattern recognition matching‐delayed (PRM‐del) and executive function task ‘One‐touch‐stockings‐of‐Cambridge’ (OTS). We performed two independent cross‐sectional analyses: (1) Top‐down. We manually created good/poor performance regimes for each test, resulting in 16 test‐by‐performance combinations, and sorted subjects according to their performance. (2) Bottom‐up. We performed a K‐means cluster analysis. Result (1) Top‐down. The 5 combinations with the highest number of participants (total of 85%) show progression of cognitive decline from the loss of SWM, through the losses of PAL, PRM‐del, and finally OTS (Table 1, left columns of numbers). (2) Bottom‐up. K‐means cluster analysis produced 5 categories with same cognitive profiles (Table 1, right columns). Both results show the number of healthy controls decreasing and the number of symptomatic participants increasing across groups. The two methods are also highly concordant in determining individual participants’ cognitive profiles: 80% of the participants from the 5 most numerous manually‐generated categories were assigned to the analogous group by the cluster analysis; 19% were assigned to an adjacent category (Table 2). Conclusion The performance pattern on tests of different cognitive domains, relevant to clinical stages of AD, shows separation between groups with progressive levels of cognitive decline. This type of at‐home, self‐administered, testing on a personal device can evaluate current cognitive ability and may predict the next domain of cognitive loss. Results imply that an individual with AD would pass through these stages of cognitive decline over time, a hypothesis that needs to be tested in a longitudinal study.