Differential expression of spatiotemporal sleep spindle clusters in ageing

Study Objectives Sleep spindles are potential biomarkers for memory decline in aging. However, significant within-person variability in spindle attributes complicates their utility in predicting cognitive deterioration. This study aimed to uncover distinct spindle types and their relevance to memory decline using exploratory, data-driven clustering. Methods Polysomnography was collected from younger (n = 43, ages 20-45 years) and older cognitively healthy adults (n = 34, ages 60-81 years). Clustering analysis was performed using multiple features and spatiotemporal context, irrespective of participant age. Results Resulting clusters were hierarchically defined by the sleep stage, slow oscillation concurrence, and hemisphere. Stage N3 spindles (15%; predominantly coinciding with slow oscillations) formed a distinct group, followed by N2 spindles coinciding with slow oscillations (27%). Remaining N2 spindles were categorized into unilateral (41%) and bilateral clusters (17%). In older adults, there was a lower proportion of N2 bilateral spindles and a higher proportion of N2 spindles concurrent with slow oscillations. Lower proportion of N2 bilateral spindles was associated with better composite memory performance in younger adults, whereas higher spindle power, regardless of cluster belonging, was associated with reduced memory benefit from sleep compared with wakefulness. Conclusions Our results indicate differing expression of spatiotemporal spindle clusters in older age, as well as intertwined dynamics between spindle propagation, SO concurrence, and frequency shifts in ageing. In addition, spindle heterogeneity aligned with global sleep stage dynamics. These results emphasize the interconnectedness of spindle activity with overall sleep patterns, underscoring the importance of spatiotemporal context within and across sleep stages. Statement of significance This study used data-driven clustering to explore sleep spindles as potential markers for age-related memory decline. We identified spindle clusters determined by sleep stage, slow oscillation concurrence, and hemisphere propagation. Notably, older adults showed altered expression of these clusters, indicating age-specific dynamics. Further research should focus on distinguishing spindle deterioration from broader sleep changes in older age. Such insights could pave the way for early detection and intervention strategies in cognitive decline, highlighting sleep's crucial role in maintaining cognitive health and resilience in aging populations. These findings hold promise for developing targeted approaches to enhance mental wellness and quality of life in older adults.