Disease trajectories before dementia: Evidence from a large‐scale community‐based prospective study

Abstract Background Dementia patients often have several co‐existing diseases, whereas the specific temporality and development patterns between them remain uncertain. Method Based on the multicenter, community‐based prospective UK Biobank, enriched disease diagnoses were extracted from hospital admission linkage, along with basic demographic information and individual genome data from the baseline visit. We constructed disease trajectory before dementia utilizing the phenome‐wide association analysis to firstly filter potential precursors, the directional test to select temporal disease pairs, and conditional logistic regression to finally quantify association strength. Disease trajectory was also constructed across diverse subpopulations stratified by dementia subtypes (Alzheimer’s disease and vascular dementia), sex, median diagnosis age, and Apolipoprotein E ( ApoE ) ε4 status (carrier or not), respectively. Mendelian randomization (MR) was further adopted to suggest causality for longitudinal associations. Result Our study comprised 434,266 participants without baseline dementia and 4,638 individuals with all‐cause dementia, with a median follow‐up of 10.9 years. Initially, 1,253 diseases were extracted as potential nodes of disease trajectory preceding dementia. We ultimately defined three clusters of disease trajectory preceding all‐cause dementia, initiated by circulatory, metabolic, and respiratory system diseases (Figure 1). The initial diseases in the trajectory were diagnosed approximately 5‐15 years before dementia. Cerebral infarction or chronic renal failure following chronic ischemic heart disease was the specific trajectory before vascular dementia. Angina, progressing to atrial fibrillation and then dementia, is a key trajectory in males, senior dementia patients, and ApoE ε4 non‐carriers. Dementia patients without ApoE ε4 typically developed urinary system disorders after musculoskeletal and connective tissue diseases. Lipid metabolism disorders remained in trajectories across all subgroups. However, MR based on inverse variance weighting indicated a modest causal relationship between lipid disorder and Alzheimer’s disease (Odds ratio = 1.05, P value = 0.234). Conclusion This study offers a comprehensive view of longitudinal disease trajectories before dementia among diverse subpopulations, indicating critical precursor diseases and specific development patterns. The findings emphasize the importance of clinical attention to patients with cardiometabolic dysfunction in midlife for screening and prevention of dementia. Further population studies and mechanism explorations are needed to validate and generalize the trajectories.