Predicting Limbic‐predominant Age‐related TDP‐43 Encephalopathy (LATE) based on In‐vivo MRI Features
Alzheimer's & Dementia · 2025-12 · 21 (S2)
DOI: 10.1002/alz70856_096121 ↗Kurz gefasst
Fachartikel in der Zeitschrift Alzheimer's & Dementia, online erschienen am 24.12.2025. Verfasst von Mahir Tazwar und 5 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref). Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background
Limbic‐predominant age‐related TDP‐43 encephalopathy neuropathologic change (LATE‐NC) is a common pathologic finding in aged brain, however, definitive diagnosis of this disease is only possible at autopsy. This work aimed to develop a marker of LATE‐NC based on in‐vivo MRI features from a large group of community‐based older adults.
Method
This study included ex‐vivo MRI, in‐vivo MRI, and pathology data from four longitudinal clinicopathological cohort studies of aging conducted at the Rush Alzheimer's Disease Center (ROS, Rush MAP, MARS, LATC). LATE‐NC was evaluated based on pathologic TDP‐43 inclusions in 8 brain regions and categorized into 4 stages (Figure 1). MRI data was processed to obtain fractional anisotropy (FA), deformation‐based morphometry (DBM) measurements, and lobar volumes. To develop a marker of LATE‐NC, we first trained a classifier to distinguish between advanced (stages 2‐3) and early LATE‐NC stages (stages 0‐1) based on ex‐vivo MRI features ( N = 863). Our classifier used a two‐level stacking model for training and cross‐validation, where level‐1 estimators generated risk scores based on single‐modality features, and level‐2 estimator (logistic regression) provided the final LATE‐NC prediction score based on the combined risk scores. We then translated the classifier to in‐vivo and validated it on a separate set of participants ( N = 60) with in‐vivo MRI and pathology data. The entire pipeline was packaged into an automated software container named MARBLE (MARker of Brain LatE).
Result
In the training group, the ex‐vivo classifier demonstrated excellent performance, achieving an average AUC of 0.85±0.05 (sensitivity=78%, specificity=76%, balanced accuracy=77%) based on FA, DBM, and lobar volume features (Figure 2). In‐vivo validation of MARBLE scores yielded an overall AUC of 0.76 in the test group. Additionally, ordinary least‐squares regression revealed higher MARBLE scores with greater LATE‐NC stages ( p
Zitieren
APA
Tazwar, M., Evia, A. M., Ridwan, A. R., Bennett, D. A. A., Schneider, J. A., & Arfanakis, K. (2025). Predicting Limbic‐predominant Age‐related TDP‐43 Encephalopathy (LATE) based on In‐vivo MRI Features. Alzheimer's & Dementia, 21(S2). https://doi.org/10.1002/alz70856_096121
Vancouver
Tazwar M, Evia AM, Ridwan AR, Bennett DAA, Schneider JA, Arfanakis K. Predicting Limbic‐predominant Age‐related TDP‐43 Encephalopathy (LATE) based on In‐vivo MRI Features. Alzheimer's & Dementia. 2025;21(S2). doi:10.1002/alz70856_096121
BibTeX
@article{Tazwar2025,
author = {Tazwar, Mahir and Evia, Arnold M and Ridwan, Abdur Raquib and Bennett, David A. A. and Schneider, Julie A and Arfanakis, Konstantinos},
title = {Predicting Limbic‐predominant Age‐related TDP‐43 Encephalopathy (LATE) based on In‐vivo MRI Features},
journal = {Alzheimer's & Dementia},
year = {2025},
volume = {21},
number = {S2},
doi = {10.1002/alz70856_096121}
}
EUVIMED