Underlying clusters of NIA‐AA Research Framework stage 2 criteria and their relationship with AD biomarkers in the DELCODE study

Abstract

AbstractBackgroundAs part of the updated NIA‐AA Research Framework for Alzheimer’s disease (AD), a numeric clinical staging scheme, applicable only to individuals in the Alzheimer’s continuum, has been introduced. “Stage 2” is defined as a transitional stage, characterized by subtle or subjective decline in cognition or neuropsychiatric symptoms (NPS) that might reflect underlying AD pathology (Jack et al., Alzheimers Dement. 2018). Here, we aimed to cluster a sample of cognitively healthy participants based on stage 2 criteria and compare the emerging clusters with regard to differences in CSF biomarker status.MethodFor this analysis, we included baseline data of N = 349 cognitively healthy participants with available CSF biomarker information from the DZNE Longitudinal Cognitive Impairment and Dementia (DELCODE) study. A two‐step cluster analysis (TSCA) was applied based on CSF Aß42/40 status and stage 2 criteria, which included a) subtle objective cognitive impairment (<0.5SD) in any of five factor scores from DELCODE’s neuropsychological test battery, b) subjective cognitive decline (SCD) with concern in any of the five SCD‐Interview domains and c) the presence of agitation, depression, anxiety, disinhibition, irritability or sleep disturbances on the Neuropsychiatric Inventory Questionnaire [NPI‐Q]. The TSCA identifies the optimal number of clusters by grouping cases into preclusters first and then clustering those using a hierarchical algorithm.ResultThe TSCA yielded three clusters. Cluster 1 (n = 114) was characterized by subtle cognitive impairment in memory, language, executive function and working memory as well as SCD in the memory and language domain. In contrast, cluster 2 (n = 104) was predominantly formed by participants that reported SCD and NPS without subtle objective impairment. In cluster 3 (n = 121), none of the stage 2 criteria were prominent. In comparison to clusters 2 and 3, participants in cluster 1 had lower levels of Aß42/40 (H(2) = 11.58, p = .003) and Aß42/ptau‐181 (H(2) = 12.06, p = .002). We observed no significant differences between clusters 2 and 3.ConclusionOur analysis resulted in three distinct clusters of cognitively healthy participants with different distributions of stage 2 criteria. The criteria prominent in participants of cluster 1 seem most promising to identify individuals at risk of progression due to underlying AD pathology.