A new study from University College London offers a more precise way to understand and potentially treat multiple sclerosis (MS), a neurological disease affecting over 1 million people in the United States alone. researchers have identified two distinct biological subtypes of MS using artificial intelligence, blood tests, and MRI scans, challenging the traditional classification of the disease based solely on symptom presentation. The findings, published in the journal Brain, could lead to more targeted therapies and improved outcomes for patients with this often-debilitating condition.
A new study suggests that artificial intelligence (AI) combined with readily available tests could help doctors better understand multiple sclerosis (MS) and personalize treatment plans for patients, a significant step toward more effective care for this complex neurological disease. Currently, treatments for MS are often chosen based on a patient’s symptoms, but the disease progresses differently in each individual.
Researchers at University College London (UCL) and Queen Square Analytics have identified two distinct biological subtypes of MS using AI, a simple blood test, and magnetic resonance imaging (MRI). The findings, published recently in the medical journal Brain, could pave the way for more targeted therapies and improved outcomes for the millions worldwide affected by MS.
The study analyzed data from 600 patients, tracking levels of light chain neurofilament protein (NfL) in their blood. NfL is a biomarker that indicates nerve cell damage and can signal how active the disease is. Researchers then used a machine learning model called SuStaIn to interpret the blood test data alongside MRI scans of the patients’ brains.
The analysis revealed two distinct patterns, which the authors termed “early NfL” and “late NfL.” Patients in the “early NfL” subtype showed elevated NfL levels from the onset of the disease, coupled with lesions visible in a brain region called the corpus callosum. These patients also experienced a rapid development of brain lesions, leading researchers to describe this subtype as more aggressive and active.
In contrast, the “late NfL” subtype initially presented with shrinkage of brain volume (atrophy) in areas like the limbic cortex and deep gray matter, before NfL levels began to rise. Disease progression appeared slower in this group, with noticeable damage emerging later.
According to Dr. Arman Eshaghi, the study’s lead author from UCL, MS “is not a single disease,” and current subtypes don’t adequately reflect changes at the tissue level—information crucial for treatment selection. He explained, as quoted by The Guardian, that combining an AI model with a readily available blood marker and routine MRI scans allowed researchers to identify these two clear biological patterns for the first time.
This new understanding could help doctors determine which patients are at higher risk of complications and tailor their care accordingly. Researchers suggest that patients identified as “early NfL” could be eligible for more effective treatments and closer monitoring. For those in the “late NfL” subtype, alternative therapies, including those focused on protecting brain cells or neurons, might be considered.
“This is an important evolution in our understanding of multiple sclerosis,” said Dr. Eshaghi. He envisions a future where AI algorithms support clinical and neurological assessments, allowing for treatment decisions to be adjusted based on each patient’s unique biological profile.
The study focused on individuals with relapsing-remitting MS (RRMS) and secondary progressive MS (SPMS). While approximately 20 treatment options exist for relapsing forms of the disease, and some are emerging for progressive forms, many patients still lack effective solutions. The researchers emphasize that classifying MS based on biological patterns, rather than clinical labels like “relapsing” or “progressive,” could help identify those at higher risk of rapid progression and guide more personalized treatment strategies.
As scientists gain a deeper understanding of MS, the potential for treatments that halt disease progression increases. These findings support the idea that a more nuanced classification of MS, based on underlying biological processes, is essential for improving patient care.
