Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Biological classification of memory clinic patients

Mastenbroek, Sophie E. LU ; Collij, Lyduine E. LU ; Anijärv, Toomas Erik LU orcid ; Rittmo, Jonathan LU orcid ; Young, Alexandra L. LU ; Strandberg, Olof LU ; Smith, Ruben LU ; Spotorno, Nicola LU ; Palmqvist, Sebastian LU orcid and Mattsson-Carlgren, Niklas LU orcid , et al. (2026) In Brain 149(4). p.1239-1253
Abstract

Neurodegenerative diseases have traditionally been defined in vivo based on clinical symptoms. However, the development of biomarkers has enabled a shift towards in vivo biological definitions. There is now a need to characterize memory clinic populations using multi-dimensional biomarker information. Here, we employed a data-driven approach to develop a biological framework for categorizing individuals in a heterogeneous memory clinic cohort based on the presence, extent and sequence of several common pathologies. We studied 1677 individuals, including subjective cognitive decline (SCD, n = 255), mild cognitive impairment (MCI, n = 400), all cause dementia (n = 393) and cognitively normal controls (n = 625) from the BioFINDER-2 cohort... (More)

Neurodegenerative diseases have traditionally been defined in vivo based on clinical symptoms. However, the development of biomarkers has enabled a shift towards in vivo biological definitions. There is now a need to characterize memory clinic populations using multi-dimensional biomarker information. Here, we employed a data-driven approach to develop a biological framework for categorizing individuals in a heterogeneous memory clinic cohort based on the presence, extent and sequence of several common pathologies. We studied 1677 individuals, including subjective cognitive decline (SCD, n = 255), mild cognitive impairment (MCI, n = 400), all cause dementia (n = 393) and cognitively normal controls (n = 625) from the BioFINDER-2 cohort [median age (interquartile range) = 72.0 (16.2) years; 50.3% female]. The Subtype and Stage Inference (SuStaIn) model was applied to biomarkers of amyloid-β (Aβ) (CSF Aβ42/Aβ40), tau (temporal meta-region of interest PET), neuronal α-synuclein [CSF seed amplification assay (SAA)], vascular pathology [MRI-based white matter hyperintensities (WMHs)] and regional atrophy (MRI-based cortical thickness) to identify biomarker-based clusters across the entire dataset. We then applied this framework to cognitively symptomatic individuals (n = 788) to compare clinical symptoms, disease progression rate and brain changes (atrophy and functional connectivity) across profiles. We identified five biomarker clusters reflecting established clinico-pathological entities, closely corresponding to (i) Alzheimer’s disease [AD, n = 317 (40.2%)]; (ii) α-synuclein disease [αSyn, n = 123 (15.6%)]; (iii) vascular disease [Vascular, n = 67 (8.5%)]; (iv) mixed AD and Vascular [Mixed, n = 207 (26.3%)]; and (v) a heterogenous group of individuals characterized by atrophy without any of the major brain pathologies, here termed ‘non-vascular, Alzheimer's, and synuclein pathology' [NOVAS, n = 74 (9.4%)]. The AD profile was characterized by global cognitive impairment and cortical atrophy in AD-associated regions. The αSyn profile was associated with visuospatial and executive dysfunction, motor impairment, hallucinations and functional connectivity disruptions throughout the brain, despite less overall atrophy compared to all others. The Vascular profile showed language and motor impairments, and both the Vascular and Mixed profiles demonstrated atrophy in cingulate and subcortical regions, alongside reduced periventricular white matter integrity. The NOVAS profile was older, demonstrated pronounced hippocampal and amygdala atrophy and baseline memory deficits, possibly reflecting neurodegenerative diseases for which currently no robust biomarkers are available, such as primary tauopathies and TDP-43 proteinopathies [e.g. limbic-predominant age-related TDP-43 encephalopathy (LATE)]. In longitudinal analyses, the AD profile showed the fastest global cognitive decline, while αSyn demonstrated an accelerated decline in language and executive and visuospatial functioning. To conclude, classifying individuals using a multimodal biomarker approach can provide valuable diagnostic and prognostic insights, with potential implications for clinical trials.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; ; ; ; and , et al. (More)
; ; ; ; ; ; ; ; ; ; ; ; ; ; and (Less)
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
amyloid-β, biological framework, data-driven, tau, vascular, α-synuclein
in
Brain
volume
149
issue
4
pages
15 pages
publisher
Oxford University Press
external identifiers
  • pmid:41157967
  • scopus:105035242452
ISSN
0006-8950
DOI
10.1093/brain/awaf411
language
English
LU publication?
yes
additional info
Publisher Copyright: © The Author(s) 2025. Published by Oxford University Press on behalf of the Guarantors of Brain.
id
837b90b3-0bc1-428a-878e-060bad314e63
date added to LUP
2026-06-22 11:04:54
date last changed
2026-07-07 17:41:16
@article{837b90b3-0bc1-428a-878e-060bad314e63,
  abstract     = {{<p>Neurodegenerative diseases have traditionally been defined in vivo based on clinical symptoms. However, the development of biomarkers has enabled a shift towards in vivo biological definitions. There is now a need to characterize memory clinic populations using multi-dimensional biomarker information. Here, we employed a data-driven approach to develop a biological framework for categorizing individuals in a heterogeneous memory clinic cohort based on the presence, extent and sequence of several common pathologies. We studied 1677 individuals, including subjective cognitive decline (SCD, n = 255), mild cognitive impairment (MCI, n = 400), all cause dementia (n = 393) and cognitively normal controls (n = 625) from the BioFINDER-2 cohort [median age (interquartile range) = 72.0 (16.2) years; 50.3% female]. The Subtype and Stage Inference (SuStaIn) model was applied to biomarkers of amyloid-β (Aβ) (CSF Aβ42/Aβ40), tau (temporal meta-region of interest PET), neuronal α-synuclein [CSF seed amplification assay (SAA)], vascular pathology [MRI-based white matter hyperintensities (WMHs)] and regional atrophy (MRI-based cortical thickness) to identify biomarker-based clusters across the entire dataset. We then applied this framework to cognitively symptomatic individuals (n = 788) to compare clinical symptoms, disease progression rate and brain changes (atrophy and functional connectivity) across profiles. We identified five biomarker clusters reflecting established clinico-pathological entities, closely corresponding to (i) Alzheimer’s disease [AD, n = 317 (40.2%)]; (ii) α-synuclein disease [αSyn, n = 123 (15.6%)]; (iii) vascular disease [Vascular, n = 67 (8.5%)]; (iv) mixed AD and Vascular [Mixed, n = 207 (26.3%)]; and (v) a heterogenous group of individuals characterized by atrophy without any of the major brain pathologies, here termed ‘non-vascular, Alzheimer's, and synuclein pathology' [NOVAS, n = 74 (9.4%)]. The AD profile was characterized by global cognitive impairment and cortical atrophy in AD-associated regions. The αSyn profile was associated with visuospatial and executive dysfunction, motor impairment, hallucinations and functional connectivity disruptions throughout the brain, despite less overall atrophy compared to all others. The Vascular profile showed language and motor impairments, and both the Vascular and Mixed profiles demonstrated atrophy in cingulate and subcortical regions, alongside reduced periventricular white matter integrity. The NOVAS profile was older, demonstrated pronounced hippocampal and amygdala atrophy and baseline memory deficits, possibly reflecting neurodegenerative diseases for which currently no robust biomarkers are available, such as primary tauopathies and TDP-43 proteinopathies [e.g. limbic-predominant age-related TDP-43 encephalopathy (LATE)]. In longitudinal analyses, the AD profile showed the fastest global cognitive decline, while αSyn demonstrated an accelerated decline in language and executive and visuospatial functioning. To conclude, classifying individuals using a multimodal biomarker approach can provide valuable diagnostic and prognostic insights, with potential implications for clinical trials.</p>}},
  author       = {{Mastenbroek, Sophie E. and Collij, Lyduine E. and Anijärv, Toomas Erik and Rittmo, Jonathan and Young, Alexandra L. and Strandberg, Olof and Smith, Ruben and Spotorno, Nicola and Palmqvist, Sebastian and Mattsson-Carlgren, Niklas and Janelidze, Shorena and Parchi, Piero and Vogel, Jacob W. and Barkhof, Frederik and Ossenkoppele, Rik and Hansson, Oskar}},
  issn         = {{0006-8950}},
  keywords     = {{amyloid-β; biological framework; data-driven; tau; vascular; α-synuclein}},
  language     = {{eng}},
  number       = {{4}},
  pages        = {{1239--1253}},
  publisher    = {{Oxford University Press}},
  series       = {{Brain}},
  title        = {{Biological classification of memory clinic patients}},
  url          = {{http://dx.doi.org/10.1093/brain/awaf411}},
  doi          = {{10.1093/brain/awaf411}},
  volume       = {{149}},
  year         = {{2026}},
}