Risk prediction models identify Alzheimer’s disease before symptoms

The study Covered in this abstract is published on medRxiv.org as a preliminary publication and has not yet been peer-reviewed.

Takeaway key

Why is this important

study design

  • Participants were older adults from two ongoing combined cohort studies of aging and dementia: the Rapid Memory and Aging Project (MAP) (n = 1179), and the Religious Orders Study (ROS) (n = 1103).

  • At enrollment, participants were without known dementia. 1742 had no cognitive impairment (NCI), 540 had it Mild cognitive impairment (MCI).

  • Follow-up duration ranged from 2 to 26 years (median 8 years; standard deviation 5.42).

  • Risk prediction models used five sets of clinical predictors of Alzheimer’s dementia: (1) common risk factors (age, gender, education, mini mental status test). [MMSE] scores, status of APOE E4 allele; (ii) Health measures (blood pressure, depression, cardiovascular disease); (3) the use of medicines; (4) other variables (complex cognition scores, physical activity, social network size); (5) Motion and sleep measures.

  • Cognitive assessments used 17 tests that evaluated five domains of cognitive ability: episodic memory, semantic memory, working memory, visual-spatial abilities, and perceptual speed.

  • Participants’ cognitive status was classified as NCI, MCI, Alzheimer’s dementia, or other types of dementia, according to criteria set by the National Institute on Aging and the Alzheimer’s Association.

  • Associations between expected risk scores for Alzheimer’s dementia and brain disease were evaluated by autopsy.

  • The models examined expected transitions from non-dementia (NCI or MCI) to Alzheimer’s dementia, NCI to Alzheimer’s dementia, or NCI to MCI or Alzheimer’s dementia.

  • The accuracy of the prediction model was determined: Model A (non-cognitive covariates alone), Model B (MMSE + non-cognitive covariates), Model C (MMSE + complex cognition and non-cognitive variables), and Model D (combined cognition covariates alone).

Main results

  • The performance of the general model using only non-cognitive covariates showed a good model predictor of Alzheimer’s dementia.

  • The incremental addition of the cognitive covariates resulted in an improvement in model performance using only the noncognitive covariates.

  • Combined models of non-cognitive and cognitive measures provide better predictors of cognitive impairment than using cognitive covariates alone.

determinants

  • The study used a minimally diverse population sample.

  • Brain imaging and fluid biomarkers have not been evaluated.

  • Diagnosis of dementia with Alzheimer’s disease can be associated with mixed brain diseases.

Disclosures

  • The study was funded by the National Institute of Health, the Illinois Department of Public Health, and the Robert C. Burwell Endowment.

  • The authors have disclosed no relevant financial relationships.

This is a pre-print summary research study, “Risk models based on non-cognitive measures may identify Alzheimer’s disease before symptoms,” conducted by Jingjing Yang of the Computational and Quantitative Genetics Center, Department of Human Genetics, Emory University School of Medicine, Atlanta, Georgia, and colleagues, published in medRxiv.org, and brought to you From Medscape. This study has not yet been peer-reviewed. The full text of the study can be found on medRxiv.org.

For more news, follow Medscape at FacebookAnd the TwitterAnd the InstagramAnd the Youtube.