Neuropsychiatric symptom networks in early- and late-onset Alzheimer's disease

While early- and late-onset Alzheimer’s disease (AD) are defined by the same pathology, presentation of neuropsychiatric symptoms can differ between groups. In this study, we investigate differences in neuropsychiatric symptom co-occurrence between early-onset (EOAD) and late-onset AD (LOAD) using a network-based approach. This cross-sectional study included 655 EOAD (< 65 years, mean age 58.4 ± 4.5 years; 53% female) and 705 LOAD (≥ 65 years, mean age 70.8 ± 4.0 years; 52% female) participants from the Amsterdam Dementia Cohort ( N = 1360). A symptom network for each group was estimated based on the Neuropsychiatric Inventory Questionnaire (NPI-Q) using Ising models for binary data, with dichotomized NPI items as nodes (present or absent) and edge weights (ω) reflecting the strength of the conditional association between two nodes. Networks were compared between EOAD and LOAD by visual inspection, edge strength, node strength centrality (sum of edges), and by comparing the global network strength between both networks (Network Comparison Test). Apathy and depression were most prevalent in both groups (EOAD: 58%,48%; LOAD: 55%,43%). Although global network strength did not differ significantly between groups ( p > .05), subtle differences in symptom connectivity were observed: in EOAD, aggression-disinhibition (ω = 1.22) and aggression-irritability were the strongest edges (ω = 1.22), with aggression and anxiety showing highest node strength within the EOAD network (3.61 and 3.60, respectively), whereas in LOAD, depression-anxiety was the strongest edge (ω = 1.49), with aggression and disinhibition showing highest node strength within the LOAD network (strength = 3.83 and 3.54, respectively). EOAD and LOAD show largely overlapping neuropsychiatric symptom networks, with subtle differences in how symptoms manifest between groups.