Multimodal Biomarker Characterization of the ALS/FTD Spectrum: A Real-World Clinical Dataset Analysis.

Area of Science:

• Neuroscience
• Biomarker Discovery
• Clinical Diagnostics

Background:

• Amyotrophic lateral sclerosis and frontotemporal dementia (ALS/FTD) diagnosis relies heavily on clinical assessment due to a lack of definitive fluid biomarkers.
• Existing biomarkers like neurofilaments and CSF pTau:tTau ratio have limitations in specificity and clinical implementation.

Purpose of the Study:

• To characterize the biological features of the ALS/FTD spectrum by integrating routinely available clinical, neuroimaging, and biofluid data.
• To develop and evaluate a multimodal model for improved diagnostic performance in the ALS/FTD spectrum.

Main Methods:

• Retrospective analysis of 229 samples (ALS, FTD, other neurodegenerative diseases, healthy controls).
• Utilized propensity score weighting, optimal cut-point determination for CSF pTau:tTau ratio, and XGBoost-based multimodal modeling.
• Performance evaluated using area under the precision-recall curve (AUC-PR).

Main Results:

• The CSF phosphorylated-tau/total-tau (pTau:tTau) ratio showed modest classification performance (AUC-PR 0.32) for ALS/FTD.
• A multimodal model integrating clinical, biofluid, and neuroimaging data achieved significantly better performance (AUC-PR 0.75).
• Key features identified included global cognition, age, Aβ42/pTau ratio, and immunoglobulin levels (CSF IgG:IgA, serum IgG).

Conclusions:

• Multimodal data integration offers a more robust biological characterization of the ALS/FTD spectrum than single biomarkers.
• The identified signature beyond pTau:tTau ratio highlights tau, amyloid, and immunological pathways.
• This approach provides a foundation for biomarker-supported diagnosis and hypothesis generation in ALS/FTD.