Inflammation and progressive neurodegeneration within the central nervous system (CNS) are key pathological mechanisms underlying multiple sclerosis (MS) [1]. Disease progression can start early, even though it mainly becomes clinically evident over time as disability accumulates [2]. Current disease-modifying therapies (DMT) substantially reduce relapse rates, but they have a limited success in controlling progression [2]. To date, the clinical impact of treatments approved for progressive MS remains relatively modest and is most evident in individuals with signs of disease activity [3]. There is an urgent need for treatments that can halt or slow progression and restore function [4].
At present, a burning field of research focuses on neuroprotection and neuroregeneration [4]. Neuroprotection aims at avoiding CNS damage and preventing the worsening of disability [2]. Neuroregeneration mostly consists in the promotion of remyelination, which refers to the regeneration and repair of injured myelin sheaths. In pilot trials, remyelination has been primarily investigated in cases of chronic optic neuritis, a common symptom of MS, for which we have tools to measure axonal damage quantitatively [5, 6].
Assessing regeneration in clinical practice
Professor Stankoff, President of ECTRIMS, tells us, “Today the primary tool in clinical trials for evaluating neuroregeneration is the functional analysis of electrical conduction along the visual pathways with the visual evoked potentials (VEP). It is not perfect but has the highest sensitiveness [7, 8]. Alongside VEP, we can also use advanced structural measures, such as the advanced magnetic resonance imaging (MRI) with magnetization transfer (MT) or myelin water fraction, or the myelin positron emission tomography (PET) [9, 10]. The problem is that these imaging techniques are limited to pilot studies. None of them are available in clinical practice, due to their cost and complexity. Therefore, we are working on developing advanced yet affordable and clinically usable tools using artificial intelligence. By leveraging the most classical MRI sequences, like T1 FLAIR, our goal is to reproduce specific myelin maps in multiple modalities.”
At ECTRIMS 2024, Dr Théodore Soulier showed the possibility of using a deep learning algorithm to convert routine clinical MRI scans into myelin maps derived not only from PET but also from other sophisticated techniques, like quantitative T1 and MT. All these modalities give a unique perspective on the same myelinic information [11]. When tested on large datasets in Paris (France), San Paulo (Brasil) and Genova (Italy), the algorithm could generate demyelination and remyelination indices – measures that indicate the extent of myelin damage and repair.
The concentration of serum biomarkers, such as neurofilament light chain (NfL), may represent another possible outcome in clinical trials. The ReBUILD study showed that treatment with clemastine, an agent with remyelinating effect, was associated with a 9.6% reduction of plasma NfL levels [12]. However, as Professor Stankoff, from the Sorbonne University in Paris, notes, “there is currently no super biomarker for regeneration. We need to intensify our research efforts to discover a specific biomarker.”
Tackling negative results in clinical trials: key takeaways
Over the past thirty years, many clinical trials addressing neuroprotection have yielded modest results [13]. There are several factors behind this.
The current approach to neuroprotection is predominantly “neuron-centered”, focusing on the mechanisms that lead to neuronal death, as Professor Stankoff highlighted at ECTRIMS 2024. Various mechanisms, including neuroinflammation, repair failure, hypoxia, oxidative stress, mitochondrial damage, ion channel dysfunction and maldistribution, gradually contribute to neuroaxonal damage and represent promising targets for drug development [13, 14].
Professor Stankoff, from the Sorbonne University in Paris, emphasises, “Many studies pinpoint a single, specific molecular mechanism. However, relying only on one pathway may not fully capture the complexity of neurodegeneration across all disease stages. We must avoid targeting only endpoint processes – resulting from multiple preceding mechanisms. The idea might be to adopt a multitarget approach and investigate a variety of candidate molecules.”
Multi-arm clinical trials can help testing multiple treatment options concurrently to accelerate drug discovery. One study simultaneously assessed three drugs – amiloride, fluoxetine, and riluzole – against a placebo to investigate key molecular processes within axons. These compounds were tested in individuals with secondary progressive MS, with the primary outcome measure being the rate of brain atrophy. Results suggest that solely targeting axonal pathobiology is insufficient to achieve neuroprotection. In fact, none of the three compounds were effective in slowing disease progression [13].
Beyond neurons, glial cells play a key role in regulating neurodegeneration [15]. At ECTRIMS 2024, Professor Francisco Quintana presented methods to examine the dialogue between microglia and astrocytes in vivo [16]. One technique is known as RABID-seq (rabies barcode interaction detection followed by sequencing) and uses the Rabies virus to track cell interactions within the CNS. A transcriptional analysis allows then to reconstruct the mechanisms and biological effects of these communications [16]. Using this method, the researchers identified a miscommunication between astrocytes and microglia involving the EphB3 protein, which appears to promote inflammation and neurodegeneration in the CNS of mice with experimental autoimmune encephalomyelitis (EAE) [16]. Treatment with an EphB3 inhibitor effectively suppressed the proinflammatory response, leading to disease improvement [16].
Professor Stankoff says, “We should maintain a focus on pharmacology. Robust pharmacokinetic trials are crucial to ensure localised action, an effective target, optimal concentration, and efficient penetration into the CNS. How we measure a compound’s effect can make the difference. We must include intermediate outcomes beyond late-stage measures, such as brain atrophy or the EDSS (Expanded Disability Status Scale). We need to add target engagement measures to assess whether and how effectively a compound binds to the intended target. For instance, if targeting neuroinflammation, we must measure markers of neuroinflammation; without these, it is unclear whether the compound had its intended effect. This approach will enable more accurate interpretation of results in early clinical trials.”
A critical window for early repair?
Another challenge lies in the individual’s capacity for repair. About half of individuals with MS experience spontaneous repair in most demyelinated areas, regardless of their age or disease duration [17]. Nonetheless, cortical myelin loss tends to worsen as MS progresses [17].
Professor Stankoff tells us, “One year ago, at the ECTRIMS symposium in Nice, we agreed that clinical trials targeting neuroregeneration should include individuals with MS at the earliest stages of the disease. Focusing on younger participants can ensure sufficient sensitivity in detecting therapeutic effects. This is a strategic choice, given that the capacity for regeneration diminishes with age and in the progressive forms of MS, when remyelination has less impact due to extensive neuronal loss. Additionally, we may have a narrow, early window of time to act on myelin repair. Another solution could involve stratifying participants into groups based on disease stage. It is essential to intervene early, prevent comorbidities and work on neurorehabilitation and physical activity.”
Neurorehabilitation and physical activity
Myelin is not a passive sheath of axons; rather, it actively changes throughout life, supporting learning [18]. This means that when we learn a new activity – like playing piano or speaking a new language – the myelin may strengthen the neural circuits underlying that skill [18]. Rehabilitation programs that build practical, everyday skills could be a valuable addition to clinical trials for remyelinating compounds. Professor Michelle Ploughman, from the Memorial University of Newfoundland, tells us, “To optimise the effectiveness of remyelinating compounds, individuals must actively participate in both physical and mental training tailored to enhance real-life skills. For example, if the goal is to improve balance, physical rehabilitation should involve exercises that simulate real-life scenarios, while cognitive training can target attention skills. A comprehensive plan is essential to challenge balance in realistic ways. Simply walking on a treadmill does not necessarily generalise into walking into the supermarket or crossing a parking lot. In those situations, you walk while carrying bags, staying aware of people and cars, looking around, orienting yourself spatially. In real-life you need to keep your balance while being distracted, so you require more attention. We identify the specific challenges associated to each task and design exercises to address them. And it is fundamental to practice these exercises every day.” Physical activity has neuroprotective and neurorestorative effects [19]. It may be important to test the combination of remyelination compounds with neurorehabilitation and exercise [20].
***
Written by Stefania de Vito
Special thanks to Professor Bruno Stankoff (President of ECTRIMS, Université Sorbonne, Hôpital Universitaire Pitié-Salpêtrière) and to Professor Michelle Ploughman (Memorial University of Newfoundland) for their insights.
References
[1] Oh J et al. J. Cent. Nerv. Syst. Dis. 2024; 16: 1-7.
[2] Bittner S et al. Nat. Rev. Neurol. 2023; 19(8): 477-488.
[3] Bittner S and Zipp F Curr. Opin. Neurol. 2022; 35(3): 293-298.
[4] Villoslada P & Steinman L Expert Opin. Investig. Drugs 2020; 29(5): 443-459.
[5] Andorra M et al. JAMA Neurol. 2020; 77(2): 234-244.
[6] Lubetzki C et al. Lancet Neurol. 2020; 19(8): 678-688.
[7] Green AJ et al. Lancet 2017; 390(10111): 2481-2489.
[8] Brown JWL et al. Lancet Neurol. 2021; 20(9): 709-720.
[9] Bodini B et al. Ann. Neurol. 2016; 79(5): 726-738.
[10] Mancini M et al. elife 2020; 9: e61523.
[11] Soulier T et al. Medical Imaging 2024: Image Processing. SPIE, 2024; 12926.
[12] Abdelhak A et al. JNNP 2022; 93(9): 972-977.
[13] Chataway J et al. Lancet Neurol. 2020; 19(3): 214-225.
[14] Friese MA, Schattling B, and Fugger L Nat. Rev. Neurol. 2014; 10(4): 225-238.
[15] Charabati M et al. 2023; Cell 186(7): 1309-1327.
[16] Clark IC et al. Science 2021; 372(6540): eabf1230.
[17] Lazzarotto A et al. 2024; Brain 147(4): 1331-1343.
[18] Bonetto G, Belin D, and Káradóttir RT Science 2021; 374(6569): eaba6905.
[19] Lozinski B and Yong VW Mult. Scler. J. 2022; 28(8): 1167-1172.
[20] Ploughman M et al. Mult. Scler. Relat. Disord. 2022; 58: 103539.