Key Takeaways
Extended Takeaways
- In this amnestic MCI/AD cohort, principal components analysis supported a 4-cluster structure after excluding an uninterpretable fifth component, with the final model explaining 62.4% of the variance; the largest cluster was hyperactivity, accounting for 31.4% of the total variance.
- Psychosis symptoms were too infrequent for stable longitudinal imaging analysis, with only 13.9%, 18.2%, and 14.3% of participants scoring >0 at baseline and 1-and 2-year follow-up visits, respectively, so negative psychosis findings should not be overinterpreted clinically.
- Among the longitudinal brain-symptom associations, the largest standardized effects were seen for the neurovegetative cluster with right nucleus accumbens volume (β=−0.55, 95% CI =−0.75 to −0.35, P<.001) and left middle temporal volume (β=−0.50, 95% CI =−0.72 to −0.28, P<.001), suggesting appetite/apathy symptoms may be particularly informative markers of temporolimbic and reward-circuit involvement.
- The affective cluster showed a distributed cortical-subcortical pattern rather than a single focal correlate, with associations spanning left rostral anterior cingulate, right entorhinal, right medial orbitofrontal, right pars opercularis, and left putamen volumes; this may help clinicians conceptualize depression/anxiety/nighttime symptoms in AD as network-level phenomena.
- Functional impairment tracked across symptom dimensions even after adjustment for age, education level, MoCA, time, sex, and cholinesterase inhibitor use: all clusters were associated with iADL outcomes, whereas hyperactivity, neurovegetative, and affective clusters also predicted ADL outcomes.
- Because the minimum detectable standardized effect was approximately β=0.27, associations weaker than this threshold may have been missed; clinicians should view the reported significant cluster-volume relationships as moderate-or-larger effects within a mild-to-moderate AD sample rather than an exhaustive map of all relevant neuropsychiatric circuitry.
