Unraveling Microstructural and Macrostructural Brain Age Dynamics in Multiple Sclerosis

Xinjie Chen^{1

2

3}, Po-Jui Lu^{1

2

3}, Mario Ocampo-Pineda^{1

2

3}, Alessandro Cagol^{1

2

3

4}, Sabine Anna Schaedelin^{1

2

3}, Esther Ruberte^{1

2

3}, Matthias Weigel^{1

2

3

5}, Federico Spagnolo^{1

2

3

6}, Pascal Benkert⁷, Johanna Maria Lieb⁸, David Leppert^{2

3}, Özgür Yaldizli^{2

3}, Johanna Oechtering^{2

3}, Marcus D'Souza^{2

3}, Bettina Fischer-Barnicol², Tobias J Derfuss^{2

3}, Clara Ekerdt⁹, Willeke M Menks⁹, Kwok-Shing Chan^{10

11}, Marcel Zwiers⁹, Andrew Chan¹², Robert Hoepner¹², Franca Wagner¹³, Caroline Pot¹⁴, Renaud A Du Pasquier¹⁴, Sebastian Finkener¹⁵, Michael Diepers¹⁶, Claire Bridel¹⁷, Patrice H Lalive¹⁷, Marjolaine Uginet¹⁷, Claudio Gobbi^{18

19}, Chiara Zecca^{18

19}, Emanuele Pravatà^{19

20}, Giulio Disanto¹⁸, Patrick Roth²¹, Jochen Vehoff²², Stefanie Mueller²³, Olaf Chan-Hi Kim²⁴, Ludwig Kappos^{1

2

3}, Jens Kuhle^{2

3}, Lester Melie-Garcia^{1

2

3}, José P Marques⁹, Cristina Granziera^{1

2

3}

Affiliations

¹ Department of Biomedical Engineering, Translational Imaging in Neurology (ThINk) Basel, Faculty of Medicine, University Hospital Basel and University of Basel, Switzerland.
² Department of Neurology, University Hospital Basel, Switzerland.
³ Research Center for Clinical Neuroimmunology and Neuroscience Basel (RC2NB), University Hospital Basel and University of Basel, Switzerland.
⁴ Department of Health Sciences, University of Genoa, Italy.
⁵ Division of Radiological Physics, Department of Radiology, University Hospital Basel, Switzerland.
⁶ MedGIFT, Institute of Informatics, School of Management, HES-SO Valais-Wallis University of Applied Sciences and Arts Western Switzerland, Sierre, Switzerland.
⁷ Clinical Trial Unit, Department of Clinical Research, University Hospital Basel, University of Basel, Switzerland.
⁸ Division of Diagnostic and Interventional Neuroradiology, Clinic for Radiology and Nuclear Medicine, University Hospital Basel and University of Basel, Switzerland.
⁹ Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, the Netherlands.
¹⁰ Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA.
¹¹ Department of Radiology, Harvard Medical School, Boston, MA.
¹² Department of Neurology, Inselspital, Bern University Hospital and University of Bern, Switzerland.
¹³ Department of Diagnostic and Interventional Neuroradiology, Inselspital, Bern University Hospital and University of Bern, Switzerland.
¹⁴ Department of Clinical Neurosciences, Lausanne University Hospital (CHUV) and University of Lausanne, Switzerland.
¹⁵ Department of Neurology, Cantonal Hospital Aarau, Switzerland.
¹⁶ Department of Neuroradiology, Cantonal Hospital Aarau, Switzerland.
¹⁷ Department of Clinical Neurosciences, Division of Neurology, Geneva University Hospitals and Faculty of Medicine, Switzerland.
¹⁸ Neurology Department, Neurocenter of Southern Switzerland, EOC, Lugano, Switzerland.
¹⁹ Faculty of Biomedical Sciences, Università della Svizzera Italiana, Lugano, Switzerland.
²⁰ Department of Neuroradiology, Neurocenter of Southern Switzerland, Lugano, Switzerland.
²¹ Department of Neurology, University Hospital of Zurich, University of Zurich, Switzerland.
²² Department of Neurology, Cantonal Hospital St. Gallen, Switzerland.
²³ Department of Neurology, University Teaching and Research Hospital, St. Gallen, Switzerland; and.
²⁴ Clinic for Radiology, Cantonal Hospital St. Gallen, Switzerland.

PMID: 40829091
PMCID: PMC12366035
DOI: 10.1212/NXI.0000000000200459

Unraveling Microstructural and Macrostructural Brain Age Dynamics in Multiple Sclerosis

Xinjie Chen et al. Neurol Neuroimmunol Neuroinflamm. 2025 Sep.

. 2025 Sep;12(5):e200459.

doi: 10.1212/NXI.0000000000200459. Epub 2025 Aug 19.

Authors

Affiliations

¹ Department of Biomedical Engineering, Translational Imaging in Neurology (ThINk) Basel, Faculty of Medicine, University Hospital Basel and University of Basel, Switzerland.
² Department of Neurology, University Hospital Basel, Switzerland.
³ Research Center for Clinical Neuroimmunology and Neuroscience Basel (RC2NB), University Hospital Basel and University of Basel, Switzerland.
⁴ Department of Health Sciences, University of Genoa, Italy.
⁵ Division of Radiological Physics, Department of Radiology, University Hospital Basel, Switzerland.
⁶ MedGIFT, Institute of Informatics, School of Management, HES-SO Valais-Wallis University of Applied Sciences and Arts Western Switzerland, Sierre, Switzerland.
⁷ Clinical Trial Unit, Department of Clinical Research, University Hospital Basel, University of Basel, Switzerland.
⁸ Division of Diagnostic and Interventional Neuroradiology, Clinic for Radiology and Nuclear Medicine, University Hospital Basel and University of Basel, Switzerland.
⁹ Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, the Netherlands.
¹⁰ Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA.
¹¹ Department of Radiology, Harvard Medical School, Boston, MA.
¹² Department of Neurology, Inselspital, Bern University Hospital and University of Bern, Switzerland.
¹³ Department of Diagnostic and Interventional Neuroradiology, Inselspital, Bern University Hospital and University of Bern, Switzerland.
¹⁴ Department of Clinical Neurosciences, Lausanne University Hospital (CHUV) and University of Lausanne, Switzerland.
¹⁵ Department of Neurology, Cantonal Hospital Aarau, Switzerland.
¹⁶ Department of Neuroradiology, Cantonal Hospital Aarau, Switzerland.
¹⁷ Department of Clinical Neurosciences, Division of Neurology, Geneva University Hospitals and Faculty of Medicine, Switzerland.
¹⁸ Neurology Department, Neurocenter of Southern Switzerland, EOC, Lugano, Switzerland.
¹⁹ Faculty of Biomedical Sciences, Università della Svizzera Italiana, Lugano, Switzerland.
²⁰ Department of Neuroradiology, Neurocenter of Southern Switzerland, Lugano, Switzerland.
²¹ Department of Neurology, University Hospital of Zurich, University of Zurich, Switzerland.
²² Department of Neurology, Cantonal Hospital St. Gallen, Switzerland.
²³ Department of Neurology, University Teaching and Research Hospital, St. Gallen, Switzerland; and.
²⁴ Clinic for Radiology, Cantonal Hospital St. Gallen, Switzerland.

PMID: 40829091
PMCID: PMC12366035
DOI: 10.1212/NXI.0000000000200459

Abstract

Background and objectives: In multiple sclerosis (MS), neurodegeneration results from the interplay between disease-specific pathology and normal aging. Conventional MRI captures morphologic changes in neurodegeneration, while quantitative MRI (qMRI) provides biophysical measures of microstructural alterations. Combining these modalities may reveal how aging and pathology interact and contribute to disability progression in people with MS.

Methods: We analyzed cross-sectional and longitudinal morphometry data from 1,353 patients with MS and 3,462 healthy controls (HCs). In addition, cross-sectional qMRI data, available for 378 HCs and 169 patients with MS, were analyzed separately. Morphometric measures and quantitative metrics were used to estimate brain-predicted age differences (brain-PADs) with machine learning. We assessed the added value of quantitative metrics over a model based exclusively on morphometric measures in brain age prediction. We also investigated the associations of brain-PADs derived from conventional and qMRI-based predictive models with clinical disability, serum inflammatory biomarkers of neuroaxonal and astrocytic injury, and lesion burden.

Results: Models combining morphometry and qMRI data achieved the best performance (mean absolute error: 5.73), outperforming those based on qMRI (6.62) or morphometry alone (8.00). Cross-sectional and longitudinal morphometry-based brain-PAD correlated with clinical disability, serum neurofilament light chain, and serum glial fibrillary acidic protein levels (all p < 0.01), with significant longitudinal interactions with time (all p < 0.05). Cross-sectional qMRI-based brain-PAD correlated with white matter lesion count (p = 0.042, R² = 0.028) and paramagnetic rim lesion volume (p = 0.028, R² = 0.020).

Discussion: Integrating qMRI improves brain age predictions. Brain-PAD serves as an imaging biomarker to quantify MS-associated aging and track disability and neuroinflammation progression.

PubMed Disclaimer

Conflict of interest statement

A. Cagol is supported by the Horizon 2020 Eurostar program (grant E!113682) and received speaker honoraria from Novartis and Roche. M. Weigel received research money from Biogen in the past. Ö. Yaldizli received grants from ECTRIMS/MAGNIMS, University of Basel, Pro Patient Stiftung University Hospital Basel, Free Academy Basel, Swiss Multiple Sclerosis Society, Swiss National Science Foundation and advisory board/lecture and consultancy fees from Roche, Sanofi Genzyme, Allmirall, Biogen and Novartis. R. Hoepner received speaker/advisor honorary from Merck, Novartis, Roche, Biogen, Alexion, Sanofi, Janssen, Bristol-Myers Squibb, Teva/Mepha and Almirall. He received research support within the last 5 years from Roche, Merck, Sanofi, Biogen, Chiesi, and Bristol-Myers Squibb. He also received research grants from the Swiss MS Society, the SITEM Insel Support Fund and is a member of the Advisory Board of the Swiss and International MS Society. He also serves as deputy editor in chief for Journal of CNS Disease and is part of the ECTRIMS Young Investigator Committee. P. Roth has received honoraria for lectures or advisory board participation from Alexion, Bristol-Myers Squibb, Boehringer Ingelheim, Debiopharm, Galapagos, Merck Sharp and Dohme, Laminar, Midatech Pharma, Novocure, QED, Roche, Sanofi and Servier and research support from Merck Sharp and Dohme and TME Pharma. C. Zecca: Ente Ospedaliero Cantonale (employer) received compensation for C.Z.’s speaking activities, consulting fees, or grants from AbbVie, Alexion, Almirall, Biogen, Bristol Meyer Squibb, Eisai, Lilly, Lundbeck, Merck, Merz, Novartis, Organon, Pfizer, Sandoz, Sanofi, Teva Pharma, Roche. C. Zecca is recipient of a grant for senior researchers provided by AFRI (Area Formazione accademica, Ricerca e Innovazione), EOC. S. Müller received honoraria for travel, honoraria for lectures/consulting and/or grants for studies from Almirall, Alexion, Bayer, Biogen, Bristol-Myers Squibb SA/Celgene, Genzyme, Merck-Serono, Teva, Novartis and Roche. L. Kappos' institutions (University Hospital Basel and RC2NB) have received compensation that was used exclusively to support research for the following activities: consultancy fees from Bayer HealthCare, Biogen, Bristol Myers Squibb, Celltrion Inc., Eli Lilly SA, EMD Serono Research and Development, GlaxoSmithKline, Galapagos NV, Janssen, Japan Tobacco Inc., Kiniksa Pharmaceuticals, Merck Healthcare AG, Minoryx, Neurostatus UHB AG, Novartis, Roche, Santhera Pharmaceuticals, Shionogi BV, Wellmera AG, and Zai Lab; contracted research fees from the European Union, InnoSuisse, Merck Healthcare AG, Novartis, Neurostatus UHB AG, Sanofi, and Roche; speaker fees and support of educational activities from Bristol Myers Squibb, Janssen, Roche, Sanofi, Merck, and Novartis; serving on the Steering Committee or Advisory Board or Data Safety Monitoring Board for Biogen, Clene Nanomedicine Inc., EMD Serono Research and Development, Genentech, Galapagos NV, Immunic AG, Janssen, Minoryx Therapeutics S.L., Novartis, Roche, Santhera Pharmaceuticals, and Sanofi. J. Kuhle received speaker fees, research support, travel support, and/or served on advisory boards by Swiss MS Society, Swiss National Research Foundation (320030_212534/1), University of Basel, Progressive MS Alliance, Alnylam, Bayer, Biogen, Bristol Myers Squibb, Celgene, Immunic, Merck, Neurogenesis, Novartis, Octave Bioscience, Quanterix, Roche, Sanofi, Stata DX. C. Granziera as the employer of the University Hospital Basel has received the following fees which were used exclusively for research support: (1) advisory boards, and consultancy fees from Actelion, Novartis, Genzyme-Sanofi, GeNeuro, Hoffmann La Roche and Siemens; (2) speaker fees from Biogen, Hoffmann La Roche, Teva, Novartis, Merck, Jannsen Pharmaceuticals and Genzyme-Sanofi; (3) research grants: Biogen, Genzyme Sanofi, Hoffmann La Roche, GeNeuro. X. Chen, P-J. Lu, M. Ocampo-Pineda, S. Schädelin, E. Ruberte, F. Spagnolo, P. Benkert, J.M. Lieb, D. Leppert, J. Oechtering, M. D'Souza, B. Fischer-Barnicol, T. Derfuss, C. Ekerdt, W.M. Menks, K.-S. Chan, M. Zwiers, A. Chan, F. Wagner, C. Pot, R. Du Pasquier, S. Finkener, M. Diepers, C. Bridel, P.H. Lalive, M. Uginet, C. Gobbi, E. Pravatà, G. Disanto, J. Vehoff, O.C-H. Kim, L. Melie-Garcia, and J.P. Marques have nothing to disclose. Go to Neurology.org/NN for full disclosures.

Figures

**Figure 1. Research Framework Overview**
Panel (A): For predictive model training, 2 healthy combined datasets were included: a larger dataset with morphometric data and a smaller dataset with both morphometric and qMRI data. (B): The morphometry-based predictive model was trained on the larger combined cohort, while the qMRI-based predictive model (Mq), supplementary morphometry-based predictive model (Mm), and a combined modalities predictive model (Mm-q) were trained on the smaller dataset. (C): In the brain-PAD analysis, the morphometry-based brain-PAD was calculated for each patient in the SMSC cross-sectional cohort and at each follow-up for longitudinal analysis. The qMRI-based brain-PAD (Mq-based), supplementary morphometry-based brain-PADs (Mm-based), and combined-based brain-PADs were calculated for each patient in the INsIDER cohort to assess the intercorrelation between them and their association with clinical data. INsIDER, ABRIM, and SMSC: Names of Cohorts. qMRI = Quantitative MRI; Brain-PAD = Brain-Predicted Age Difference; EDSS = Expanded Disability Status Scale; sNfL = serum Neurofilament Light Chain; sGFAP = serum Glial Fibrillary Acidic Protein.

**Figure 2. Brain Age Prediction and Correction**
(A) shows corrected brain-PAD (colored points) and uncorrected brain-PAD (gray points) across healthy controls for the morphometry-based predictive model. The table above the plot presents the average root mean square error (RMSE) and mean absolute error (MAE) from five-fold cross-validation. Colored points represent different cohorts, and the red dashed lines indicate the 10th, 25th, 50th, 75th, and 90th percentiles of the corrected brain-PAD distribution. Following the format of (A),(B) presents corrected and uncorrected brain-PAD for participants using the qMRI-based predictive model. (C and D) show the relationship between chronologic age and predicted age in the morphometry-based and qMRI-based predictive models, respectively. The black dashed line represents the relationship before bias correction, while the red dashed line represents the relationship after correction. Colored points indicate corrected predicted ages by cohort, with the distribution and corresponding color for each cohort shown in the bar charts above panels (A and B). RMSE = Root Mean Square Error; MAE = Mean Absolute Error; brain-PAD = Brain-Predicted Age Difference; and qMRI = Quantitative MRI.

**Figure 3. Brain-PAD Distribution and Correlation**
(A and D) Scatterplots showing chronologic age vs predicted brain age for patients (red) and controls (blue) from morphometry-based (A) and qMRI-based (D) predictive models. Regression lines indicate group trends, with patients exhibiting higher predicted brain ages. (B and E) Histograms of the predicted age difference for patients (red) and controls (blue) using morphometry-based (B) and qMRI-based (E) predictive models. (C) Split violin plots displaying chronologic (blue) and predicted (red) ages in healthy controls and patients' phenotype groups, with medians indicated by bold lines. (F) Scatterplot with linear fit and marginal histograms illustrating the correlation between qMRI-based brain-PAD and supplementary morphometry-based brain-PAD. Spearman correlation (ρ) and p value are shown. MAE = Mean Absolute Error; RMSE = Root Mean Squared Error; Brain-PAD = Brain-predicted age difference.

**Figure 4. Cross-Sectional Clinical Correlation Results**
(A–C) For the morphometry-based predictive model, partial residual plots show the relationship between brain-PAD and EDSS (log-transfer), sNfL (Z score), and sGFAP (Z score) (blue). Linear (dashed black) and locally estimated scatterplot smoothing (blue) fits are displayed, with marginal histograms illustrating the brain-PAD distribution. Separate fits for RRMS (dot-dash, blue) and PMS (dot-dash, red) phenotypes are displayed in (C), showing their significant difference. (D–F) showing the results for the qMRI-based predictive model on patients in red. (D) A linear regression fitting shows the relationships between brain-PAD and log-transformed EDSS in progressive MS (PMS) patients. (E) presents a linear regression model between brain-PAD and log-transformed PRLs accounting for WMLs and CLs volumes and disease duration. (F) presents a linear regression model between brain-PAD and WMLs count accounting for CLs and PRLs counts and disease duration. Significant correlations are annotated with adjusted R² and p values corresponding to the predictor variable on the top right corner in each plot. EDSS = Expanded Disability Status Scale; PRL = Persistent Rim Lesion; WML = White Matter Lesion; PMS = Progressive Multiple Sclerosis; sNfL = Serum Neurofilament Light Chain; sGFAP = Serum Glial Fibrillary Acidic Protein.

**Figure 5. Multiple Time Points and Longitudinal Clinical Correlation Results**
Partial residual plots illustrate the relationships between morphometry-based brain-PAD and (A) EDSS (log-transformed), (B) sNfL (Z score), and (C) sGFAP (Z score), accounting for multiple time points in each subject through random effects. Linear (dashed black) and locally estimated scatterplot smoothing (dashed red) fits are displayed, with marginal histograms depicting the brain-PAD distribution. Significant correlations are annotated with p values and conditional R² values. (D–F) present a longitudinal analysis that illustrates the interaction effects of brain-PAD and time on (D) EDSS, (E) sNfL, and (F) sGFAP. These figures depict the predicted clinical outcomes over time for the different brain-PAD percentiles (25th, 50th, and 75th), with each panel including annotations that indicate the significance of the interaction effects and the conditional R² values. EDSS = Expanded Disability Status Scale; PMS = Progressive Multiple Sclerosis; sGFAP = Serum Glial Fibrillary Acidic Protein; sNfL = Serum Neurofilament Light Chain. brain-PAD = brain-predicted age difference.

See this image and copyright information in PMC

References

1. Graves JS, Krysko KM, Hua LH, Absinta M, Franklin RJM, Segal BM. Ageing and multiple sclerosis. Lancet Neurol. 2023;22(1):66-77. doi: 10.1016/S1474-4422(22)00184-3 - DOI - PubMed
1. Sanai SA, Saini V, Benedict RH, et al. Aging and multiple sclerosis. Mult Scler. 2016;22(6):717-725. doi: 10.1177/1352458516634871 - DOI - PubMed
1. Cagol A, Schaedelin S, Barakovic M, et al. Association of brain atrophy with disease progression independent of relapse activity in patients with relapsing multiple sclerosis. JAMA Neurol. 2022;79(7):682-692. doi: 10.1001/jamaneurol.2022.1025 - DOI - PMC - PubMed
1. Bakshi R, Thompson AJ, Rocca MA, et al. MRI in multiple sclerosis: current status and future prospects. Lancet Neurol. 2008;7(7):615-625. doi: 10.1016/S1474-4422(08)70137-6 - DOI - PMC - PubMed
1. Cole JH, Raffel J, Friede T, et al. Longitudinal assessment of multiple sclerosis with the brain‐age paradigm. Ann Neurol. 2020;88(1):93-105. doi: 10.1002/ana.25746 - DOI - PubMed

MeSH terms

Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions

LinkOut - more resources

Full Text Sources
- Atypon
- PubMed Central
Medical
- MedlinePlus Health Information

Save citation to file

Email citation

Add to Collections

Add to My Bibliography

Your saved search

Create a file for external citation management software

Your RSS Feed

Unraveling Microstructural and Macrostructural Brain Age Dynamics in Multiple Sclerosis

Affiliations

Unraveling Microstructural and Macrostructural Brain Age Dynamics in Multiple Sclerosis

Authors

Affiliations

Abstract

Conflict of interest statement

Figures

References

MeSH terms

LinkOut - more resources

Full Text Sources

Medical