Abstract
Background: Prediction models aimed at detecting risk of progression from Mild Cognitive Impairment (MCI) to Alzheimer’s disease (AD) dementia increase their accuracy when impaired executive functions enter the analysis. This suggests that impaired executive functions in MCI are likely linked to the prodromal stages of AD dementia. Neuroimaging assessment of such patients would allow exploring if they show AD related patterns of brain atrophy. We hypothesized that AD sensitive brain regions would show discrimination between dysexecutive amnestic MCI (maMIC) and healthy controls.
Method: We analysed 32 healthy controls and 23 MCI patients. Patients were divided in single domain amnestic MCI, multidomain amnestic MCI (i.e., with the dysexecutive component), and non-amnestic MCI. Brain volume data entered regression models to analyse which brain regions predict group membership (control vs maMCI). Stepwise lineal regression model was then conducted to identify the brain regions with better prediction power.
Results: Four variables were able to predict group membership in simple lineal regression models: entorhinal cortex, lingual gyrus and parahippocampal gyrus in the left hemisphere and fusiform gyrus in the right hemisphere. The entorhinal cortex provided the most accurate model (F(1, 42) = 14.19, p=0.001, R2=0.24). Linear regression models were run with performance on executive function tasks including tests of switching, planning, verbal fluency and working memory. The most accurate model returned Letters and Numbers and categories fluency (F(2, 44) = 21.35, p=0.000, R2=0.48) suggesting that working memory and category generation are the functions contributing to the dysexecutive profiles observed in maMCI patients.
Conclusion: Dysexecutive profiles in multidomain amnestic MCI together with neuroimaging volumetric analysis increase the probability of identifying the prodromal stages of AD dementia.
Method: We analysed 32 healthy controls and 23 MCI patients. Patients were divided in single domain amnestic MCI, multidomain amnestic MCI (i.e., with the dysexecutive component), and non-amnestic MCI. Brain volume data entered regression models to analyse which brain regions predict group membership (control vs maMCI). Stepwise lineal regression model was then conducted to identify the brain regions with better prediction power.
Results: Four variables were able to predict group membership in simple lineal regression models: entorhinal cortex, lingual gyrus and parahippocampal gyrus in the left hemisphere and fusiform gyrus in the right hemisphere. The entorhinal cortex provided the most accurate model (F(1, 42) = 14.19, p=0.001, R2=0.24). Linear regression models were run with performance on executive function tasks including tests of switching, planning, verbal fluency and working memory. The most accurate model returned Letters and Numbers and categories fluency (F(2, 44) = 21.35, p=0.000, R2=0.48) suggesting that working memory and category generation are the functions contributing to the dysexecutive profiles observed in maMCI patients.
Conclusion: Dysexecutive profiles in multidomain amnestic MCI together with neuroimaging volumetric analysis increase the probability of identifying the prodromal stages of AD dementia.
| Original language | English |
|---|---|
| Number of pages | 1 |
| Publication status | Published - 29 Jul 2020 |
| Event | Alzheimer’s Association International Conference 2020 - Virtual Event Duration: 27 Jul 2020 → 31 Jul 2020 https://aaic2020.vfairs.com/en |
Conference
| Conference | Alzheimer’s Association International Conference 2020 |
|---|---|
| Period | 27/07/20 → 31/07/20 |
| Internet address |
Keywords
- cognitive performance
- Alzheimer disease
- dementia