Premium
Potential contributions of MRI and blood‐based biomarkers for cognitive status classification based on cognitive performance examinations
Author(s) -
Farina Mateo,
Saenz Joseph,
Crimmins Eileen M.
Publication year - 2021
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.053326
Subject(s) - dementia , cognition , neuroimaging , medicine , effects of sleep deprivation on cognitive performance , logistic regression , population , neuropsychology , gerontology , disease , clinical psychology , psychology , pathology , psychiatry , environmental health
Background The global burden of dementia is growing from 50 million people in 2020 to 150 million in 2050 (World Health Organization). Health researchers often use national survey studies to understand the social and population dynamics of dementia. Many of these studies have relied on classification algorithms based on cognitive performance exams such as the MMSE and MOCA. However, researchers are concerned about misclassification leading to incorrect conclusions about dementia in the population. This study assesses how the addition of MRI markers, along with blood‐based biomarkers, would improve cognitive status classification based on cognitive performance exams. Method We use the data from the Alzheimer’s Disease and Neuroimaging Initiative (ADNI) to assess whether MRI measures (whole brain volume, hippocampal volume, and ventricle volume) along with plasma Aß and P‐Tau improve cognitive status classification accuracy, net of MMSE scores. To assess improvements, nested multinomial logistic regression equations adjusting for sex, age, and APOE alleles are used. In total, we have total of 1,442 respondents. Results We found marginal improvements in cognitive status classification when structural MRI measures and blood‐based biomarkers were included compared to only MMSE scores. The R 2 for model with only MMSE (0.401) increased when both structural MRIs, and plasma Aß and P‐Tau were added (0.445, an approximately 10% improvement). Furthermore, while structural MRI measures and blood‐based biomarkers alone did explain a modest amout of variability in dementia status classification (R 2 of .182 and 142, respectively), models with MMSE scores alone had by far the largest explanatory power at .401. Conclusion While MRI markers and blood‐based biomarkers may lead to novel insight in understanding how brain pathology leads to cognitive impairment, their application to improving cognitive status classification in survey research may be more limited. Our findings do not show large improvements in cognitive status classification when including them in our predictive models of cognitive status classification. Additionally, the marginal improvement may not justify the costs associated with data collection in large nationally representative datasets.