Cognitive decline and longitudinal atrophy in successful and typical aging
Stefania Pezzoli, Joseph Giorgio, Xi Chen, Theresa M. Harrison, William J. Jagust- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
Greater hippocampal volume (HV) and thicker mid‐cingulate cortex (MCC) are features of successful agers (SA), defined as exceptional cognitive performers, compared to typical agers (TA). This study aimed at examining whether SA experience different longitudinal change in cognition and atrophy from TA, and how baseline ß‐amyloid (Aß) affects these relationships.
Method
190 cognitively normal older adults (70+ years old) underwent 11C‐Pittsburgh compound B (PiB)‐PET to derive baseline Aß positivity, and longitudinal structural MRI and cognitive testing. A machine learning cognitive‐age model was used to predict participants’ cognitive‐age from cognitive testing. Cognitive‐age gap (CAG = cognitive‐age – chronological age) was computed to represent individual variability in cognitive aging, wherein negative values indicate better cognitive performance than expected for age; SA were below the 20th percentile (n = 39). Linear mixed‐effects models were used to investigate the relationship between longitudinal changes in 1) Preclinical Alzheimer’s Cognitive Composite (PACC), 2) HV, and 3) MCC, with baseline CAG or SA‐status, as well as Aß‐status as predictors.
Result
We found a significant main effect on PACC for CAG (p<0.001), but not Aß‐status (p = 0.9), and significant CAG x time (p = 0.002) and Aß‐status x time (p = 0.02) interactions (Figure 1). Results were similar when participants were dichotomized into SA and TA. In addition, baseline CAG showed significant main effects on 1) HV (p<0.001) and 2) MCC (p = 0.02), but the CAG x time interactions were not significant (HV: p = 0.6; MCC: p = 0.7) (Figure 2). Finally, when CAG, Aß‐status, and HV were included in the same model to predict PACC change, we found a significant main effect for CAG (p<0.001), significant CAG x time (p = 0.01) and Aß‐status x time (p = 0.02) interactions, and a marginally significant HV x time interaction (p = 0.06) (Figure 3).
Conclusion
Lower baseline CAG scores, indicating younger cognitive‐age, were related to greater HV and thicker MCC cross‐sectionally, but not longitudinal atrophy differences, suggesting that better cortical integrity in SA may reflect longstanding differences from TA. Lower baseline CAG and Aß negativity were independently associated with slower cognitive decline. These findings suggest that factors underlying SA include separate contributions of both brain maintenance and reserve.