Abstract
BACKGROUND: The structure of psychopathology can be organized hierarchically into a set of transdiagnostic dimensional phenotypes. No studies have examined whether these phenotypes are associated with brain structure or dementia in older adults. METHODS: Data were drawn from a longitudinal study of older adults aged 70-90 years at baseline (N = 1072; 44.8% male). Confirmatory factor models were fit to baseline psychiatric symptoms, with model fit assessed via traditional fit indices, model-based reliability estimates, and evaluation of model parameters. Bayesian plausible values were generated from the best-fitting model for use in subsequent analyses. Linear mixed models examined intraindividual change in global and regional gray matter volume (GMV) and cortical thickness over 6 years. Logistic regression examined whether symptom dimensions predicted incident dementia over 12 years. RESULTS: A higher-order model showed a good fit to the data (BIC = 28,691.85; ssaBIC = 28,396.47; CFI = 0.926; TLI = 0.92; RMSEA = 0.047), including a general factor and lower-order dimensions of internalizing, disinhibited externalizing, and substance use. Baseline symptom dimensions did not predict change over time in total cortical and subcortical GMV or average cortical thickness; regional GMV or cortical thickness in the frontal, parietal, temporal, or occipital lobes; or regional GMV in the hippocampus and cerebellum (all p-values >0.5). Finally, baseline symptom dimensions did not predict incident dementia across follow-ups (all p-values >0.5). CONCLUSIONS: We found no evidence that transdiagnostic dimensions are associated with gray matter structure or dementia in older adults. Future research should examine these relationships using psychiatric indicators capturing past history of chronic mental illness rather than current symptoms.