Recent proteomic advances in neurology

Publication highlights July 2026

Neurological diseases are driven by complex molecular changes that can begin long before symptoms appear, involving diverse biological pathways that warrant a systems-level understanding of disease mechanisms. Advances in highly specific, multiplex proteomics technologies such as the Olink® PEA platform have made it possible for researchers to identify protein signatures in blood and cerebrospinal fluid linked to neurodegeneration, neuroinflammation, and disease progression. By uncovering novel biomarkers and potential therapeutic targets, proteomics is helping to enable the next generation of precision neurology. Here we highlight some recent outstanding studies that illustrate the key role of protein biomarkers in neurology research.

A high-performance protein signature for Parkinson’s disease research

A research study from the National Institutes of Health used a combination of high-multiplex proteomics with Olink® Explore 3072 and advanced machine learning (ML) analysis to identify a highly accurate, multi-protein model to discriminate samples from individuals with Parkinson’s Disease (PD) from those taken from healthy controls and individuals with other neurological conditions. The model displayed high specificity and reliability across multiple independent cohorts, as well as identifying important proteins and pathways involved in PD pathophysiology.

• 482 proteins associated with PD vs healthy controls after false discovery rate (FDR) adjustment – enrichment analysis highlighted pathways including NOD-like receptor signaling, toll-like receptor signaling, apoptosis, & interaction of chemokine receptors with their ligands.
• Machine learning produced an 11-protein model that could discriminate PD from both healthy controls and other neurological conditions with extremely high accuracy (AUC = 0.939).
• The model retained high performance in independent PD disease (PDBP) and population-based (UKB-PPP) cohorts.
• Pathway analysis of the 11 proteins highlighted enrichment for pathways closely associated with PD pathobiology.

A comparison of protein and RNA-based models for Parkinson’s disease research

Parkinson’s disease (PD) is a progressive neurodegenerative disorder whose global burden is increasing, with diagnosis primarily based on clinical symptoms that arise only after significant neuronal loss.

The circulating secretome has emerged as a promising biomarker source for PD research, but the relative discriminatory values of proteomic versus transcriptomic signals—and the benefits of multimodal integration across cohorts—remain unclear. Researchers at George Mason University have addressed this question by training PD classifiers using plasma proteomics (Olink® Explore 1536) and whole-blood RNA sequencing data from the Parkinson’s Progression Markers Initiative (PPMI).

• A proteomics-only model achieved perfect discrimination (AUC = 1.000) in the internal hold-out dataset. The model also showed high performance in an external dataset (AUC = 0.8724).
• The RNA-only model classified PD only modestly above chance in the external comparison dataset (AUC = 0.5978) and multi-modal models did not significantly improve on the proteomics-only model.
• A 32-protein “Proteomic Severity Index” (PSI) was derived by modeling baseline protein expression against symptom burden – linear PSI explained 28.2% of the variance observed in PD severity.

CSF proteomics help enable future prediction of disease trajectories in Alzheimer’s Disease

Scientists from Saint Louis University carried out a CSF proteomics research study in a large longitudinal cohort of individuals with AD and controls. Using the Olink® Explore HT platform, they measured CSF SDC4 along with >5,400 other proteins to look for associations with the trajectories of amyloid and tau pathologies and cognitive impairment in AD. Baseline and longitudinal measurements were used to develop “pseudo-time models” to estimate CSF biomarker trajectories across the AD progression period.

• Elevation of SDC4 was observed to begin very early in AD pathogenesis, coincident with amyloid positivity, accelerating further by the time of tau positivity.
• Higher baseline levels of SDC4 also predicted more rapid progression of brain amyloid and tau, and faster decline in global cognition.
• Machine learning applied to the complete 5,400+ protein dataset identified SDC4 among the top 10 proteins with the highest importance for predicting the pseudo-time models of AD progression.
• The top 10 proteins included those with known AD associations, as well as more novel proteins in this context (e.g., SHC3, DTX3, ITGAM, ACHE, FABP3, TXNRD1).

A protein-based model for predicted brain age discordance

Predicted age deviation (PAD) – the difference between MRI-predicted brain age and chronological age) is a potentially valuable biomarker for identifying structural brain changes that precede the clinical manifestation of neurological disease. However, MRI-based brain age assessment depends on costly imaging infrastructure and specialized technical expertise.

To explore the potential of circulating proteins as a simple, minimally invasive approach to support brain aging research, scientists at Johns Hopkins School of Nursing measured the levels of >5,400 proteins in plasma samples taken from 137 cognitively normal individuals from the LIMBIC-CENC cohort. The study participants were divided into two groups, one with MRI-based estimated brain ages ≥5 years older than chronological age, and the other with brain age equal to or younger than chronological age.

• 418 dysregulated proteins were identified that showed at least a 2-fold change between groups – pathway analysis showed enrichment in NF-κB, heat-shock protein, and Wnt signaling.
• Machine learning and ROC curve analysis identified multiple 6- or 7-protein models that could discriminate samples from the two brain age groups with AUCs above 0.9
• The top model (FGFBP3, CRTAC1, CHUK, DSG3, IMSL3, SH3GL1) discriminated samples from the two groups with an AUC of 0.9035 and false-negative rate of just 0.086.
• Several of the key proteins identified have known biological associations with neurobiology (e.g., FGFBP3 with neuronal development and anxiety-related behavior, CRTAC1 with synaptic development and function).

For Research Use Only. Not for use in diagnostic procedures