Proteomic-based approaches to complex neuroimmunology: A window into mechanistic insight

Insights from Dr. Abdelhak

The biological onset of multiple sclerosis (MS) remains poorly defined. A recent Nature Medicine study by Abdelhak et al. used high-throughput discovery proteomics with Olink Explore HT to analyze serum from presymptomatic individuals who later developed MS alongside matched healthy controls. The goal: to define the earliest biological changes preceding neurological symptoms and better understand disease mechanisms. The findings suggest that MS begins years before clinical manifestation. Evidence of myelin injury was detected approximately seven years before symptom onset and preceded signs of axonal injury by about one year. Astrocyte involvement, by contrast, became evident only around the time of diagnosis. The study also revealed widespread proteomic changes during the presymptomatic phase, including signals implicating interleukin-3 and NF-κB–related pathways. Individuals with a previously identified MS-associated autoantibody signature showed increased immune cell activity, reinforcing the role of early immune dysregulation. Together, these results support the development of protein biomarker panels capable of distinguishing presymptomatic individuals from healthy controls—pending validation in future studies—and provide new insight into the cascade of central nervous system injury that precedes MS.

How do you determine when to start monitoring patients in the preclinical
stage? Should testing focus on individuals with family history or other risk factors?

This is an important question. In my opinion, a universal screening approach for the general public is unlikely to be useful. Instead, we need a strategy to identify individuals at high risk. Family history is one factor, but it is only part of the picture. We now know that Epstein–Barr virus infection, among other factors, plays a significant role in MS risk. Ongoing work aims to define the optimal at-risk populations and the most appropriate timing for monitoring.

How were pre-symptomatic MS patients defined in this study?

Our lab is one of the first labs that developed different models to look at the Sjögren’s syndrome disease process. We use a spontaneous mouse model that very much mimics what happens in patients. With the model that we developed in the lab, we can identify the precise mechanism from the initiation to the full-blown clinical disease. The mice go through different stages. At an early stage, we learn about the immunological processes leading to the disease, how the immune system is evolving, and lastly, what impact it has on the clinical stage. Having a mouse model allows us to learn about different disease stages which we cannot see in humans. In patients, we have access to the very end stage of the disease, and you never know what actually initiated the whole process. With the mouse model, we can understand the disease process really well. Because of that, we can design targets that can prevent the early stage, the immunological stage or the clinical stage.

Which murine sample type do you predominantly use in your studies?

We leveraged the U.S. Department of Defense Serum Repository, which collects annual blood samples from service members. Individuals who later developed MS and received care through the VA system could be identified retrospectively. We then retrieved their stored samples from years before symptom onset. Similar approaches are possible in large population cohorts, particularly in countries with comprehensive health registries, such as those in scandinavian countries. However, determining how to screen and monitor at-risk individuals remains an open challenge.

We won’t be collecting cerebrospinal fluid (CSF) samples years before clinical onset, but what might your results have looked like in CSF?

We would likely observe CNS injury markers in CSF, but not necessarily the same immune signatures seen in blood. In these early stages of MS, much of the immune dysregulation appears detectable in peripheral blood. As the disease progresses, inflammation becomes more compartmentalized within the CNS. Presymptomatic CSF might therefore reflect early brain and spinal cord injury, while systemic immune changes could remain more evident in blood. However, this remains speculative.

Given that myelin injury appears years before axonal damage, do you interpret this as a primary oligodendrocytedriven pathology, or as early immunemediated demyelination that later triggers axonal degeneration?

This remains a central question in MS research. Our data cannot definitively resolve whether the initial event is immune-driven or oligodendrocyte-intrinsic. The earliest pathological changes may be subtle or transient, potentially occurring between annual sampling points. Animal models may ultimately be necessary to address this question. My current view is that immune-mediated injury likely plays a primary role, but this remains debated.

Do you see interleukin-3 (IL-3) activation as an early driver of disease pathogenesis or as secondary responses to subclinical CNS injury?

IL-3 is closely associated with T-cell expansion. Elevated levels may reflect early immune activation that either contributes to CNS injury or precedes immune cell migration into the CNS. It could represent both a driver and a consequence of early disease activity. Examining IL-3 trajectories alongside markers of myelin and axonal injury will be an important next step.

Does IL-3 kinetics correlate with clinical response or toxicity outcomes like CRS?

Most acute toxicities, including ICANS and CRS, occur during the phase of rapid T-cell expansion. In the cases presented here, we did not focus on studying these complications, so we did not directly assess correlations with IL-3. However, in a separate study examining delayed toxicities, we are seeing an association between IL-3 elevation—reflecting T-cell proliferation—and the development of later certain pattern of neurotoxicities. This suggests IL-3 kinetics may have value as a marker of prolonged immune activation rather than acute toxicity.

Given IL-3’s low baseline abundance, what was your strategy for ensuring Olink’s proximity extension assay sensitivity was sufficient to reliably capture this signal without false positives?

We used a combination of statistical and analytical approaches, including outlier exclusion and consideration of limits of detection. Importantly, we do not automatically exclude proteins near the detection limit. Even if only a subset of cases shows measurable levels, that signal may be biologically meaningful. Ultimately, results must align with biological plausibility.

From your presentation, it seems like these biomarkers may relate to other neurodegenerative diseases as well. How do you differentiate the biomarkers specific to MS and not other diseases?

Relying on any single biomarker to provide specificity for differentiating neurodegenerative diseases from MS would be very challenging, if not even impossible. MS, compared to other neurodegenerative diseases, is not a proteinopathy. As we know in Alzheimer’s, we have tau accumulation, as well as we have amyloid beta, meaning there are specific proteins related to this neurodegenerative condition. In MS we know that there is inflammation and widespread injury to the brain and spinal cord. Of course, the myelin is affected more than others, but it’s not the single component. Myelin is also affected in many other conditions as well. So while I doubt that we will ever have that one single highly sensitive and specific marker for MS, I believe we might not even need that. This is the beauty of having protein biomarker panels, which reflect different aspects of the pathology, including CNS injury characteristic for MS, as well as the immune changes. Specificity will therefore come from protein biomarker panels, rather than individual markers.

Want to watch the full webinar? Click here!