Predicting cognitive decline in amyloid-negative individuals with amnestic mild cognitive impairment
(2025) In Alzheimer's & dementia : the journal of the Alzheimer's Association 21(S8).- Abstract
BACKGROUND: A considerable portion of patients with amnestic mild cognitive impairment (aMCI) have negative amyloid-β (Aβ) biomarkers and are therefore unlikely to have Alzheimer's disease (AD). Potential causes of cognitive decline in this heterogeneous group include limbic-predominant age-related TDP-43 encephalopathy (LATE), cardio/cerebrovascular diseases, primary age-related tauopathy (PART), and subthreshold Aβ. The prognosis of Aβ-negative (Aβ-) aMCI patients, putatively more benign, is unclear. We aim to investigate which predictors - including demographics, baseline cognition, fluid and imaging biomarkers - can best predict cognitive decline and progression to dementia in Aβ- aMCI. METHOD: We included 140 Aβ- aMCI patients (Aβ... (More)
BACKGROUND: A considerable portion of patients with amnestic mild cognitive impairment (aMCI) have negative amyloid-β (Aβ) biomarkers and are therefore unlikely to have Alzheimer's disease (AD). Potential causes of cognitive decline in this heterogeneous group include limbic-predominant age-related TDP-43 encephalopathy (LATE), cardio/cerebrovascular diseases, primary age-related tauopathy (PART), and subthreshold Aβ. The prognosis of Aβ-negative (Aβ-) aMCI patients, putatively more benign, is unclear. We aim to investigate which predictors - including demographics, baseline cognition, fluid and imaging biomarkers - can best predict cognitive decline and progression to dementia in Aβ- aMCI. METHOD: We included 140 Aβ- aMCI patients (Aβ status based on cerebrospinal fluid (CSF) and positron emission tomography, when available; 'amnestic' based on norm scores for AD Assessment Scale delayed word recall) from BioFINDER-1/2 with longitudinal Mini-Mental State Examination (MMSE), and subsets with longitudinal data on Clinical Dementia Rating Sum of Boxes (CDR-SB; n = 67) and progression to dementia (n = 134, 43% progressors; Table 1). Predictors included global and regional atrophy measures, specific for LATE and PART, CSF Aβ42/40 and p-tau181, hypertension, white matter hyperintensities and global cognition. Individual MMSE and CDR-SB slopes were estimated with linear mixed-effects models. Associations of predictors with MMSE/CDR-SB slopes and progression to dementia were tested. Significant predictors and demographic variables were included in the model selection process using R package MuMIn, which tests linear combinations of variables and ranks models by the Akaike information criterion (AIC). RESULTS: Figure 1 shows individual associations for the identification of significant predictors for the model selection process. For MMSE (AIC: 350.57; Figure 2a), the selected most parsimonious model included baseline MMSE and whole-brain cortical thickness. For CDR-SB (AIC: 176.73; Figure 2b), baseline CDR-SB, amygdala volume, and middle frontal gyrus cortical thickness were included. For progression to dementia (AIC: 97.65; Figure 2c), the selected model included MMSE, lateral ventricles volume, entorhinal and whole-brain cortical thickness, sex and CSF Aβ42/40. CONCLUSIONS: Baseline cognition, global and regional atrophy measures are valuable predictors of cognitive decline in Aβ- aMCI, with regional brain measures hinting at specific pathologies. These prediction models are relevant for the new LATE clinical criteria. We aim to validate our findings in ADNI.
(Less)
- author
- organization
-
- Diagnostic Radiology, (Lund)
- MultiPark: Multidisciplinary research on neurodegenerative diseases
- LU Profile Area: Proactive Ageing
- Clinical Memory Research (research group)
- eSSENCE: The e-Science Collaboration
- Brain Injury After Cardiac Arrest (research group)
- WCMM-Wallenberg Centre for Molecular Medicine
- MR Physics (research group)
- publishing date
- 2025-12
- type
- Contribution to journal
- publication status
- published
- subject
- in
- Alzheimer's & dementia : the journal of the Alzheimer's Association
- volume
- 21
- issue
- S8
- article number
- e110861
- publisher
- Wiley
- external identifiers
-
- pmid:41433399
- scopus:105025740966
- ISSN
- 1552-5279
- DOI
- 10.1002/alz70862_110861
- language
- English
- LU publication?
- yes
- id
- 2d9b28b3-68cc-427f-bb99-8d96a14df5bc
- date added to LUP
- 2026-02-11 14:07:04
- date last changed
- 2026-02-12 03:45:22
@misc{2d9b28b3-68cc-427f-bb99-8d96a14df5bc,
abstract = {{<p>BACKGROUND: A considerable portion of patients with amnestic mild cognitive impairment (aMCI) have negative amyloid-β (Aβ) biomarkers and are therefore unlikely to have Alzheimer's disease (AD). Potential causes of cognitive decline in this heterogeneous group include limbic-predominant age-related TDP-43 encephalopathy (LATE), cardio/cerebrovascular diseases, primary age-related tauopathy (PART), and subthreshold Aβ. The prognosis of Aβ-negative (Aβ-) aMCI patients, putatively more benign, is unclear. We aim to investigate which predictors - including demographics, baseline cognition, fluid and imaging biomarkers - can best predict cognitive decline and progression to dementia in Aβ- aMCI. METHOD: We included 140 Aβ- aMCI patients (Aβ status based on cerebrospinal fluid (CSF) and positron emission tomography, when available; 'amnestic' based on norm scores for AD Assessment Scale delayed word recall) from BioFINDER-1/2 with longitudinal Mini-Mental State Examination (MMSE), and subsets with longitudinal data on Clinical Dementia Rating Sum of Boxes (CDR-SB; n = 67) and progression to dementia (n = 134, 43% progressors; Table 1). Predictors included global and regional atrophy measures, specific for LATE and PART, CSF Aβ42/40 and p-tau181, hypertension, white matter hyperintensities and global cognition. Individual MMSE and CDR-SB slopes were estimated with linear mixed-effects models. Associations of predictors with MMSE/CDR-SB slopes and progression to dementia were tested. Significant predictors and demographic variables were included in the model selection process using R package MuMIn, which tests linear combinations of variables and ranks models by the Akaike information criterion (AIC). RESULTS: Figure 1 shows individual associations for the identification of significant predictors for the model selection process. For MMSE (AIC: 350.57; Figure 2a), the selected most parsimonious model included baseline MMSE and whole-brain cortical thickness. For CDR-SB (AIC: 176.73; Figure 2b), baseline CDR-SB, amygdala volume, and middle frontal gyrus cortical thickness were included. For progression to dementia (AIC: 97.65; Figure 2c), the selected model included MMSE, lateral ventricles volume, entorhinal and whole-brain cortical thickness, sex and CSF Aβ42/40. CONCLUSIONS: Baseline cognition, global and regional atrophy measures are valuable predictors of cognitive decline in Aβ- aMCI, with regional brain measures hinting at specific pathologies. These prediction models are relevant for the new LATE clinical criteria. We aim to validate our findings in ADNI.</p>}},
author = {{Annettesdotter, Amanda and Spotorno, Nicola and Wuestefeld, Anika and Mattsson-Carlgren, Niklas and Strandberg, Olof and Ossenkoppele, Rik and Stomrud, Erik and Palmqvist, Sebastian and Wolk, David A. and Hansson, Oskar and Wisse, Laura E.M.}},
issn = {{1552-5279}},
language = {{eng}},
note = {{Conference Abstract}},
number = {{S8}},
publisher = {{Wiley}},
series = {{Alzheimer's & dementia : the journal of the Alzheimer's Association}},
title = {{Predicting cognitive decline in amyloid-negative individuals with amnestic mild cognitive impairment}},
url = {{http://dx.doi.org/10.1002/alz70862_110861}},
doi = {{10.1002/alz70862_110861}},
volume = {{21}},
year = {{2025}},
}
