Noncognitive Measures Can Help Identify Presymptomatic Alzheimer's Disease

Susan Kreimer for Medscape

April 07, 2022

The study covered in this summary was published in as a preprint and has not yet been peer reviewed.

Key Takeaways

  • To enable early intervention, this study explored whether noncognitive metrics alone can be applied to design risk models for identifying adults at risk for Alzheimer's dementia and cognitive impairment.

  • Risk models using noncognitive metrics predict both Alzheimer's dementia and cognitive impairment.

  • Noncognitive covariates do not enable incremental predictivity for models that include cognitive metrics in predicting Alzheimer's dementia, but they do in models predicting cognitive impairment.

  • Risk models of Alzheimer's dementia demonstrate more reliable predictions than the models of cognitive impairment.

  • Deriving improved risk prediction models, particularly for cognitive impairment, will be essential to ensure early interventions.

Why This Matters

  • There is an asymptomatic stage during which Alzheimer's disease (AD) pathology accumulates. Following that stage is mild cognitive impairment (MCI). Dementia occurs later.

  • AD varies widely, with most patients progressing through all three clinical stages — from no cognitive impairment (NCI) to MCI followed by dementia.

  • Some individuals make a direct transition from NCI to dementia, whereas some never progress beyond MCI, and one third with NCI prior to death also exhibit pathologic AD at death.

  • AD is a progressive and complex disorder that develops over the years.

  • It negatively affects cognition and also has an adverse impact on other important noncognitive aging phenotypes.

  • For instance, accumulation of AD pathology is associated with the decline of body mass index or impaired motor function that may predate and predict incident cognitive impairment and Alzheimer's dementia.

  • Devising a risk profile based on noncognitive covariates may make it easier to identify at-risk adults during the earliest stages of AD.

Study Design

  • The researchers extracted clinical data from older adults who did not have dementia from two harmonized cohort studies, the Memory and Aging Project (MAP, n = 1179) and the Religious Orders Study (ROS, n = 1103).

  • They performed an analysis using Cox proportional hazard models with backward variable selection to establish risk prediction models for Alzheimer's dementia and cognitive impairment.

  • The investigators compared models utilizing only noncognitive covariates of physical function, psychosocial, health conditions, and medications to models that added cognitive covariates of Mini–Mental Status Examination (MMSE) and composite cognition score summarizing 17 cognitive tests.

  • All models were trained in MAP and were evaluated in ROS.

  • The researchers assessed model performance by the area under the curve (AUC) of the receiver operating characteristic (ROC) curve.

Key Results

  • Models based on noncognitive covariates alone attained AUCs of 0.800 and 0.785 for predicting Alzheimer's dementia 3 and 5 years from baseline.

  • With the inclusion of additional cognitive covariates, AUCs improved to 0.916 and 0.881.

  • A model with a single covariate of composite cognition score reached AUCs of 0.905 and 0.863.

  • Models based on noncognitive covariates alone achieved AUCs of 0.717 and 0.714s for the prediction of cognitive impairment 3 and 5 years from baseline.

  • The inclusion of additional cognitive covariates resulted in improvement of AUCs to 0.783 and 0.770.

  • A model with a single covariate of composite cognition score attained AUCs of 0.754 and 0.730.


  • Participants were mainly Americans of European descent and possessed higher than average levels of education, so the findings will need to be reproduced in more diverse populations.

  • The investigators did not use brain imaging and fluid biomarkers.

  • This specific cognitive battery is more comprehensive than what may be accessible other than in a research setting, which underlines the significance of the findings employing the MMSE.

Study Disclosures

  • The authors have disclosed no relevant financial relationships.

This is a summary of a preprint research study, "Risk Models Based on Non-cognitive Measures May Identify Presymptomatic Alzheimer’s Disease," written by Jingjing Yang from Emory University School of Medicine in Atlanta, published on medRxiv, and provided to you by Medscape. This study has not yet been peer reviewed. The full text of the study can be found on

For more news, follow Medscape on Facebook, Twitter, Instagram, and YouTube.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.