Cantex, Headlamp team up on AI-driven MS drug trial

Cantex Pharmaceuticals and Headlamp Health have agreed to collaborate on azeliragon in multiple sclerosis, using Headlamp's Lumos AI platform to identify patients most likely to respond to the drug.

The effort will focus on severe depression and fatigue in people with multiple sclerosis, two symptoms that remain difficult to treat with existing medicines. Headlamp will analyse data from azeliragon's development programme to classify patients as responders or non-responders.

Azeliragon is an oral inhibitor of RAGE, the receptor for advanced glycation endproducts. Cantex is developing the drug to target neuroinflammatory pathways linked to depression and fatigue in multiple sclerosis, rather than the pathways typically addressed by SSRI and SNRI medicines.

These symptoms affect a large share of patients. Around 40% of those with progressive multiple sclerosis have clinically significant depression, while fatigue affects 60% to 80% of people with the disease over time, according to the companies.

The collaboration reflects a broader push in central nervous system drug development to improve patient selection in clinical trials. The companies plan to combine biological, behavioural, biomarker and clinical information to identify patterns associated with treatment response.

Headlamp describes Lumos AI as a neuro-symbolic, multi-agent platform for neuropsychiatric drug development. It says the system draws on more than 100 million longitudinal, multi-modal data points and is designed to model patient trajectories beyond what episodic clinical data can capture.

Focus on stratification

Patient stratification has become a central issue in neurological and psychiatric drug development, where broad study populations can obscure benefits in a subset of patients. In multiple sclerosis, depression and fatigue reflect overlapping inflammatory, clinical and behavioural drivers, making that approach especially relevant, the companies said.

Cantex is a clinical-stage biopharmaceutical company developing azeliragon for major depressive disorder and fatigue in patients with multiple sclerosis. It licensed worldwide rights to the drug from vTv Therapeutics, which had originally advanced it in Alzheimer's disease.

Earlier studies generated clinical safety data from more than 2,000 individuals dosed for up to 18 months, according to Cantex. Those trials indicated that azeliragon was well tolerated.

Stephen Marcus outlined the rationale for the programme in multiple sclerosis.

"Severe depression and fatigue are among the most debilitating aspects of MS, affecting approximately 40% of patients with progressive MS and in need of pharmacologic treatment that is more effective and safer than medications currently available. Having led the development of Betaseron, the first disease-modifying therapy approved for MS, which has been followed by several other effective medicines for relapsing MS, we still need to address the symptoms that severely impact daily functioning, especially for progressive MS, for which current treatments need improvement. Azeliragon's mechanism of action, inhibition of "RAGE" (the receptor for advanced glycation endproducts), directly addresses the neuroinflammatory pathways linked to depression and fatigue that SSRI and SNRI medications do not address," said Stephen Marcus, M.D., chief executive officer of Cantex Pharmaceuticals.

The biology behind that approach centres on RAGE expression in the brain on microglia, astrocytes, endothelial cells and infiltrating immune cells. Activation by ligands including HMGB1 and S100 proteins can amplify neuroinflammatory signalling, according to the companies.

AI in trial design

For Headlamp, the deal is another example of drug developers using artificial intelligence tools to sift through large, varied datasets in search of more precise trial populations. Rather than treating neuropsychiatric conditions as uniform across a study cohort, the aim is to identify narrower patient phenotypes with a stronger likelihood of response.

Headlamp said its system creates an intelligence layer that integrates multi-modal signatures to define distinct responder and non-responder phenotypes. That information can then be used to refine trial design and support development decisions.

Erwin Estigarribia, Headlamp's chief executive officer, said the collaboration fits the platform's original purpose.

"Lumos AI was built to reason across complex biological, clinical and longitudinal data so that drug development can move beyond broad populations and toward more precise patient phenotypes," said Erwin Estigarribia, CEO of Headlamp Health. "This collaboration highlights how agentic AI can unlock insights within complex data and is exactly the use case we built for."