Brazilian and French researchers have developed an innovative tool that combines machine learning with high-content cell imaging (HCI) to predict how patients with multiple sclerosis (MS) will respond to natalizumab, one of the most widely used therapies for the disease. The findings mark a significant step toward precision medicine in neurology.
Natalizumab is an effective monoclonal antibody therapy that reduces relapses and slows MS progression. However, up to 35% of patients experience a return of symptoms within two years. The drug also carries risks, such as serious infections and other side effects, making it essential to identify in advance who is most likely to benefit. With a monthly cost of around BRL 10,000 per patient in Brazil’s public health system (SUS), predicting treatment response is also crucial for controlling healthcare expenditure.
Imaging immune cell behaviour
The therapy works by preventing immune cells from entering the brain and triggering inflammation. After treatment, CD8+ T cells, which are part of the immune system, typically become more rounded due to actin remodelling, a process that shapes how cells move and interact. Using HCI technology, the researchers observed that non-responders show distinct actin remodelling: their T cells remained elongated and retained higher migration capacity. This cellular behaviour suggested why treatment fails in some patients.
By extracting more than 400 morphological features from patient blood samples and feeding them into a machine learning model, the team created over one million combinations. The system achieved 92% accuracy in the discovery cohort and 88% in the validation cohort, demonstrating high predictive power.
From images to numbers
“The breakthrough was turning images into quantifiable data and applying machine learning,” explains Helder Nakaya, senior researcher at Albert Einstein Hospital and professor at the University of São Paulo. “This approach could be extended to other therapies, including cancer immunotherapies such as CAR-T.”
Lead author Beatriz Chaves, now based at INFINITy in France, emphasizes the broader impact: “These results can improve quality of life by avoiding ineffective treatments and unnecessary side effects, while reducing costs for health systems like the SUS.” The findings were published in Nature Communications.
Implications for multiple sclerosis care
MS is an autoimmune disease affecting 2.8 million people worldwide, with symptoms ranging from muscle weakness and mobility problems to cognitive decline. Most diagnoses occur in young adults between 20 and 50, with women disproportionately affected. By tailoring treatment choices earlier, clinicians can minimize disease progression and optimize outcomes.
The team now aims to validate the tool in larger and more diverse patient populations, while exploring ways to make the technique more accessible through simpler and less costly equipment. Expanding this methodology beyond MS could also accelerate personalized medicine across other chronic and complex diseases.
MS therapy developments
Earlier this year, researchers at the Dutch Brain Institute (NIN) identified new therapeutic targets for MS, offering hope for regenerative treatments that move beyond inflammation control toward actual tissue repair. Using a unique donor cohort from the Dutch Brain Bank, researcher Alida Chen analysed brain tissue from deceased MS patients, comparing well-, and poorly-healed lesions. She identified genes and signalling pathways linked to natural myelin repair, as well as the pivotal role of microglia, immune cells in the central nervous system, whose shape appears to influence repair success.
The findings mark an important step toward gene therapies capable of repairing damaged brain tissue, supported by viral vector technology now in preclinical development. Currently, MS therapies only limit damage and delay progression, but these insights point to treatments that could restore neurological function. The publicly available transcriptomic dataset may accelerate global research into regenerative MS therapies, offering new prospects for millions of patients worldwide.
