Using blood protein tests for earlier detection of MND

Amyotrophic lateral sclerosis (ALS) is by far the commonest motor neuron disease (MND). It is a devastating condition where the nerve cells that control movement gradually stop working and die. This leads to muscle weakness, worsening over time and can eventually affect speaking, swallowing and breathing.

Diagnosing ALS is difficult (especially challenging in the early stages of the condition) because its symptoms can overlap with other neurological conditions. There is currently no single test that can confirm ALS. As a result, diagnosis often relies on ruling out other diseases, which can lead to long days before people receive the right specialist care and support.

Many scientists are searching for biomarkers, which are measurable biological signs of disease in the body, that can help identify ALS earlier. Blood tests are particularly useful in biomarker research because they are widely accessible, relatively simple to perform, and can be repeated over time. Proteins in blood may be informative because they can reflect changes happening inside the body even before symptoms appear. At present, there are no reliable blood tests available that could be used routinely to diagnose ALS.

Why is the study important?

Earlier studies looking at blood biomarkers in ALS often included a very small number of people living with MND or examined only a small set of proteins, which made it very hard to draw clear conclusions that can be applied to a larger population. This specific study, however, used blood samples from large groups of people (over 500 different individuals), including those with ALS, healthy individuals, and people with other neurological conditions. By comparing these groups carefully, the researchers were able to look for patterns that were specific to ALS, rather than changes linked to neurological conditions in general.

The study also examined blood samples taken years before people developed ALS symptoms. This allowed the researchers to explore whether changes linked to the disease appear long before diagnosis, giving a clear picture of how ALS may begin and develop over time.

What did the authors do and how did they do it?

Almost 3,000 proteins were measured using a laboratory test designed to examine many proteins at once. They then used a machine learning approach to look for patterns in the data “freely”, meaning it searched for patterns without being told in advance what the correct answer should look like. Then, the researchers identified which proteins were uniquely present at different levels in people with ALS.

Using this information, the authors applied the machine learning model to new sets of data from several sources to validate how well it performs. Importantly, they were also able to use data from long-term health studies to see whether the same protein changes were present in the blood years before ALS symptoms began.

The study identified 33 blood plasma proteins that were consistently different in people with ALS vs people without. When these proteins were analysed together, the computer model was able to identify ALS with very high accuracy of up to 98.3%. Excitingly, the results were similar when the tested in different groups, showing that the findings were not limited to the set of data the machine learning model was trained on.

The researchers also found that some of these proteins could be detected up to ten years before ALS symptoms appeared. Providing strong evidence for what has long been suspected, that biological changes linked to ALS may begin long before diagnosis is possible using currently available methods.

What do the findings mean going forward for people with the disease?

These findings suggest that blood tests could one day help diagnose ALS earlier and more reliably. This earlier detection would undoubtedly help people access care sooner, which, once more effective treatments become available in the future, could increase the chances of recovery or slow the speed of disease progression. It may also help researchers understand how ALS develops over time, even before symptoms appear, and which biological processes are involved at each stage of the ALS.

A particularly exiting results from this research is how well the protein panel performed when applied to very large independent data. For example, when tested in the UK Biobank dataset (over 23,000 people), the authors report that the biomarker panel achieved over 99% accuracy in predicting ALS or no ALS status. Results like this increase confidence that the approach is not only promising in small research studies but may also translate well into large populations.

However, this is not yet a test that can be rolled out in clinics. What makes this study exciting is that it moves the idea of a blood-based diagnosis from “promising biology” towards something that feels achievable. The authors also spell out what needs to happen to get there: the protein signature must be shown to hold up across broader and more diverse populations, be refined beyond the limits of the current protein panel and be tracked over longer periods to confirm how early, and how reliably, it flags ALS in real-world settings.

Overall, this study provides strong evidence that blood plasma proteins can act as useful biomarkers for ALS. It may not yet be a clinical test, but it does mark a real turning point. The results suggest that ALS leaves a detectable trace in the blood, and that this trace can be recognised with outstanding accuracy, sometimes even years before symptoms are obvious. That is the kind of progress that changes what feels possible, not just diagnosing ALS faster, but opening the door to studying the earliest stages of the condition and, in the future, giving people the best chance of benefiting from treatments that work effectively when started early.

This study can be found at https://www.nature.com/articles/s41591-025-03890-6#Sec14

Paper title
A plasma proteomics-based candidate biomarker panel predictive of amyotrophic lateral sclerosis

Lead author
Ruth Chia, Bryan J. Traynor

Publication details including date of publication
Published 19 August 2025 in Nature Medicine.