Brain MRI signatures across sex and CSF Alzheimer’s disease biomarkers

You Cheng, Yingnan He, Karthik Gopinath, Benjamin Billot, Juan Eugenio Iglesias, Chao‐Yi Wu, Hiroko H. Dodge, Anne‐Marie Wills, Becky C. Carlyle, Pia Kivisäkk, Bradley T. Hyman, Steven E. Arnold, Sudeshna Das, Frank Tennigkeit, Andrea Pfeifer, Christine Vincenzetti, Fabien Monin, G. C. Maitra, Joseph A. O’Sullivan, Alex Teligladas, Ana Diaz, Christophe Bintener, Cindy Birck, Dianne Gove, Jean Georges, Katherine Ellis, Stefanie Peulen, Eva Danasi, G. Vlachos, Nicolaos Scarmeas, Alexandre Mencik, Chi‐Fai Ng, Hernàn Picard, Mats Ericson, Michael E. Gaunt, Nicolas Pourbaix, Rob Lenz, Rodolphe Eyraud, Ross McGarrity, Véronique Chauvet, Ahmad Allouche, Béatrice Le Balanger, Bruno Dubois, Edith Benmansour, Francis Nyasse, Hicham Filali, Ihsen Youssef, Leslie Ameil, Marion Dubois, Nabila Lajaail, Nabila Nyasse, Nadja Younsi, Patrick Feukam Talla, Souza Benadjaoud, Stéphane Lehéricy, Tatiana Akake Beshelemu, Beatriz Molinuevo, Beatriz Díaz‐Molina, Ian Sherriff, Pedro Pesini, Pilar Martínez Tirado, Vinicio de Jesús Pérez, Alan McNeill, A. A. M. Gibson, Andrew Fox, Bert Overduin, Christine G. Lee, David Allan, David Taylor, Emma‐Jane McDonald, Harry Peaker, Karen O'Hanlon, Kara J Thornton, Lorna Harper, Pamela Brankin, Rhonda Bryce, Rodrigo Barnes, Ana Belén Callado, Andreea Rădoi, Annabella Beteta, Arcadi Navarro, Beatriz Blasco, Carolina Herrero, Carolina Minguillón, Eider Arenaza, E. Nebot, Gema Huesa, Gerard Blasco, G Roessler Oliver, Gonzalo Sánchez Benavides, Grégory Operto, Iva Knezevic, Jordi Camı́, José Luís Molinuevo, Juan Domingo Gispert, J. Pereda, Karine Fauria, Laia Tenas, Mahnaz Shekari, Maríateresa BuongiornoBrain 2025 · DOI

TL;DR

This study uses advanced MRI and CSF biomarker analysis to reveal distinct patterns of brain atrophy and connectivity associated with Alzheimer's disease (AD) pathology, differentiating between amyloid and tau markers, and importantly, highlighting significant sex differences in these brain changes. The findings underscore the value of integrating multiple data modalities and large, diverse datasets to better understand the complex neurobiological underpinnings of AD.

What Problem Does This Paper Address?

The relationship between cerebrospinal fluid (CSF) biomarkers for Alzheimer's disease (AD) – specifically amyloid and tau proteins – and the resulting neurodegenerative effects in the brain is not fully understood. While CSF biomarkers are increasingly used for diagnosis, their precise impact on brain structure and connectivity, especially when considering the interplay between amyloid and tau, and how these effects differ between males and females, remains an area needing deeper exploration. This paper aims to clarify these relationships by investigating neurodegeneration patterns across different CSF biomarker groups and examining sex-specific differences in brain volume and connectivity.

Methods

The researchers employed a multi-faceted approach combining advanced neuroimaging and biomarker analysis. Key methods include:

  • Brain Segmentation and Volume Calculation: Using the SynthSeg+ pipeline, a deep learning algorithm, they segmented brain images into various regions and calculated their volumes. Quality control measures were implemented to ensure data reliability.
  • Structural Covariance Network (SCN) Analysis: This technique examines how the volumes of different brain regions co-vary across individuals. It helps understand the coordinated growth or atrophy patterns, providing insights into the "connectome" of brain volumes. Metrics like global efficiency, path length, nodal degree, and small-worldness were calculated.
  • High-Dimensional Analysis (UMAP): Uniform Manifold Approximation and Projection (UMAP) was used to visualize and explore complex, non-linear relationships among brain volumes, allowing for the identification of distinct clustering patterns related to biomarker groups and sex.
  • Logistic Regression and Machine Learning: Various classifiers (LASSO, ridge, Firth logistic regression, random forest) were used to predict amyloid and tau status based on brain volumes, assessing model performance with AUROC, sensitivity, and specificity.
  • Meta-Analyses: To ensure robustness and generalizability, findings from the primary dataset were validated and synthesized with data from three independent, larger datasets using random-effects meta-analyses.
  • Functional Network Overlap: Dice coefficients and spin tests were used to quantify the overlap between identified brain regions and established functional brain networks, linking structural changes to functional systems.

Datasets

The study utilized several datasets:

  • Mass General Brigham (MGB) Healthcare System Dataset: This served as the primary discovery dataset, comprising 306 participants aged 50+ with clinical MRI scans and CSF data collected within one year of each other.
  • Validation Datasets:
    • NACC (National Alzheimer's Coordinating Center): A large database of AD research.
  • ADNI (Alzheimer's Disease Neuroimaging Initiative): A well-established longitudinal study of AD.
  • EPAD (European Prevention of Alzheimer's Dementia): A European consortium for AD research.

Combined, these datasets totaled 3137 participants for validation and meta-analyses, significantly enhancing the generalizability of the findings.

Key Insights and Ideas

  • Distinct Neurodegeneration Patterns: The study revealed that different CSF biomarker profiles (control, amyloid-only, amyloid+tau, other dementia) are associated with unique patterns of brain atrophy and connectivity disruption.
  • Control Group's Robust Connectivity: The cognitively unimpaired control group exhibited the most interconnected structural network, suggesting preserved brain organization.
  • Amyloid's Impact on Specific Regions: Amyloid pathology was consistently linked to atrophy in subcortical regions (thalamus, amygdala), cerebellum, and brainstem across multiple datasets.
  • Tau's Global Effect (with Amyloid): Tau pathology, particularly in the presence of amyloid, was associated with more generalized brain shrinkage, characterized by enlargement of extracerebral CSF and surprisingly, ventricular shrinkage.
  • Sex Differences Emerge: Significant sex differences were observed in brain volume patterns across most biomarker groups, suggesting that AD pathology may manifest differently in males and females.
  • A+T+ Group Shows Less Sex Difference: Interestingly, the A+T+ (amyloid-positive, tau-positive) group showed less pronounced sex differences in connectivity patterns compared to other groups.
  • Importance of Global Connectivity: The study highlighted the value of examining global structural connectivity, as whole brain volumes showed greater differences between groups than subregions.

Key Results

  • Amyloid Prediction: A ridge logistic regression model predicted amyloid status with an AUROC of 0.795 (95% CI: [0.788, 0.802]) in the MGB dataset.
  • Tau Prediction: A random forest model predicted tau status in the presence of amyloid with an AUROC of 0.694 (95% CI: [0.686, 0.702]).
  • Consistent Amyloid Associations: The amygdala and thalamus showed consistent significant associations with amyloid across all four datasets.
  • Functional Network Overlap: Amyloid-associated brain volumes showed significant overlap with the dorsal attention, control C, and visual association networks. Tau-associated volumes overlapped with somatomotor (increased volume) and language/auditory networks (decreased volume).
  • Sex Differences in Connectivity: While overall SCN analysis showed no significant sex difference, visualization revealed distinct connectivity patterns between sexes in certain groups, particularly in the A+T- and A+T+ groups.
  • Sex-Based Clustering: UMAP analysis revealed distinct sex-based clustering in subcortical gray matter for control, A+T-, and A-T-&CI groups.

Limitations and Future Work

The authors acknowledge several limitations:

  • Slice Thickness: The slice thickness of clinical images may have posed challenges for precise cortical thickness measurements, potentially impacting analyses related to tau pathology.
  • Cross-Sectional Design: The study used a cross-sectional design, limiting the ability to track how brain volume changes over time. Longitudinal studies are needed to fully understand these dynamics.
  • Generalizability of Education Level: The average education level of participants was higher than the general population, which might limit the generalizability of findings to less educated groups.
  • Hormone Therapy: Information on hormone therapy, which can affect brain volumes, was not consistently recorded across all datasets.

Future work could focus on longitudinal studies to track changes over time, investigate the role of cortical thickness in relation to tau pathology, and explore the impact of factors like hormone therapy. The study also suggests that understanding the interplay of CSF biomarkers, sex, and brain structure is crucial for personalized diagnostic and treatment approaches for AD.

Summary updated

Abstract

The relationship between cerebrospinal fluid (CSF) biomarkers of Alzheimer's disease and neurodegenerative effects is not fully understood. This study investigates neurodegeneration patterns across CSF Alzheimer's disease biomarker groups, the association of brain volumes with CSF amyloid and tau status and sex differences in these relationships in a clinical neurology sample. MRI and CSF Alzheimer's disease biomarkers data were analysed in 306 patients of the Mass General Brigham healthcare system aged 50+ (mean age = 68.4 ± 8.8 years; 43.1% female), who had lumbar punctures within 1 year of clinical MRI scans. We first analysed neurodegeneration patterns across four biomarker groups: 60 controls (A-T-&CU; amyloid negative, tau negative, cognitively unimpaired), 25 A+T- (amyloid positive, tau negative), 121 A+T+ (amyloid positive, tau positive) and 100 other dementia (A-T-&CI; amyloid negative, tau negative, cognitively impaired). Second, we examined volumetric associations with amyloid (amyloid positive, tau negative versus control) and tau in the presence of amyloid (amyloid positive, tau positive versus amyloid positive, tau negative) across 52 brain areas. Third, we examined sex differences in these relationships. Finally, we validated core analyses across three independent datasets-NACC (National Alzheimer's Coordinating Center), ADNI (Alzheimer's Disease Neuroimaging Initiative) and EPAD (European Prevention of Alzheimer's Dementia)-totalling 3137 participants, and performed meta-analyses to obtain more robust estimates. We observed distinct neurodegeneration patterns across biomarker groups, with disrupted connectivity (brain volume covariance networks) in amyloid positive and other dementia groups, while amyloid and tau negative, cognitively unimpaired controls exhibited the most connected network. Amyloid was associated with subcortical, cerebellar and brainstem atrophy, with consistent association observations in the thalamus and amygdala across all four datasets. Tau in the presence of amyloid demonstrated general brain shrinkage through enlargement of extracerebral CSF, alongside unexpected ventricle shrinkages. Sex-based analyses revealed that A+T+ (amyloid positive, tau positive) had lower sex differences in connectivity patterns compared with other groups. Sex differences were also noted in amyloid-related ventricular volume changes. This study reveals how amyloid and tau affect brain connectivity and volume across sex and CSF biomarker groups, emphasizing global brain changes and sex differences. By leveraging automated pipelines and advanced MRI and biomarker analyses, we extracted meaningful and replicable findings from heterogeneous clinical samples from real-world data. The meta-analyses across four datasets enhance the generalizability of our results.
BibTeX
@inproceedings{cheng2025brain,
  author = {You Cheng and Yingnan He and Karthik Gopinath and Benjamin Billot and Juan Eugenio Iglesias and Chao‐Yi Wu and Hiroko H. Dodge and Anne‐Marie Wills and Becky C. Carlyle and Pia Kivisäkk and Bradley T. Hyman and Steven E. Arnold and Sudeshna Das and Frank Tennigkeit and Andrea Pfeifer and Christine Vincenzetti and Fabien Monin and G. C. Maitra and Joseph A. O’Sullivan and Alex Teligladas and Ana Diaz and Christophe Bintener and Cindy Birck and Dianne Gove and Jean Georges and Katherine Ellis and Stefanie Peulen and Eva Danasi and G. Vlachos and Nicolaos Scarmeas and Alexandre Mencik and Chi‐Fai Ng and Hernàn Picard and Mats Ericson and Michael E. Gaunt and Nicolas Pourbaix and Rob Lenz and Rodolphe Eyraud and Ross McGarrity and Véronique Chauvet and Ahmad Allouche and Béatrice Le Balanger and Bruno Dubois and Edith Benmansour and Francis Nyasse and Hicham Filali and Ihsen Youssef and Leslie Ameil and Marion Dubois and Nabila Lajaail and Nabila Nyasse and Nadja Younsi and Patrick Feukam Talla and Souza Benadjaoud and Stéphane Lehéricy and Tatiana Akake Beshelemu and Beatriz Molinuevo and Beatriz Díaz‐Molina and Ian Sherriff and Pedro Pesini and Pilar Martínez Tirado and Vinicio de Jesús Pérez and Alan McNeill and A. A. M. Gibson and Andrew Fox and Bert Overduin and Christine G. Lee and David Allan and David Taylor and Emma‐Jane McDonald and Harry Peaker and Karen O'Hanlon and Kara J Thornton and Lorna Harper and Pamela Brankin and Rhonda Bryce and Rodrigo Barnes and Ana Belén Callado and Andreea Rădoi and Annabella Beteta and Arcadi Navarro and Beatriz Blasco and Carolina Herrero and Carolina Minguillón and Eider Arenaza and E. Nebot and Gema Huesa and Gerard Blasco and G Roessler Oliver and Gonzalo Sánchez Benavides and Grégory Operto and Iva Knezevic and Jordi Camı́ and José Luís Molinuevo and Juan Domingo Gispert and J. Pereda and Karine Fauria and Laia Tenas and Mahnaz Shekari and Maríateresa Buongiorno},
  title = {{Brain MRI signatures across sex and CSF Alzheimer’s disease biomarkers}},
  booktitle = {Brain},
  year = {2025},
  doi = {10.1093/braincomms/fcaf210},
}
Figure