Divergent network trajectories in mild cognitive impairment: Converters versus non-converters to Alzheimer's disease

Background: Mild cognitive impairment (MCI) is a condition between normal aging and Alzheimer's disease (AD), with around 12% of individuals progressing to AD each year. Understanding how cognitive functions evolve over time in people with MCI, especially those who eventually develop AD, is key to improving early detection and intervention. Objective: To investigate the longitudinal changes in the relationships between cognitive functions in patients with MCI, which will or will not convert to AD. Methods: We applied cross-lagged panel network (CLPN) to data from the Alzheimer's Disease Neuroimaging Initiative (ADNI). We analyzed cognitive evaluations over two years in 128 converters to AD (convMCI), 390 non-converters (stabMCI), and 362 healthy participants (HP). Results: HP and stabMCI showed increasing stability in their performance over time (i.e., a performance at one time predicting the same performance at a later point), while convMCI experienced a decline. Additionally, while HP maintained stable relationships between different tests (i.e., a performance at one time predicting a different performance at a later point), individuals with MCI showed more fluctuation, particularly those who progress to AD. We also found that executive functions were the most reliable predictors of future cognition in healthy participants, while memory played a more critical role in predicting outcomes for MCI patients, especially those who convert to dementia. Conclusions: These findings highlight different cognitive trajectories in MCI patients, offering valuable insights into how cognitive decline unfolds in those at risk of developing dementia. Our results shed new light on the longitudinal reorganization of cognitive performances.