Beyond One Drug: What the Tolebrutinib Decision Teaches Us About Progressive MS Trials.

Reflections on risk, biology, and the trials progressive MS now demands

In my last post, I tried to offer a balanced reading of the safety concerns surrounding tolebrutinib by comparing its clinical trial data with that of other FDA-approved BTK inhibitors, by looking back at past MS therapies that faced significant regulatory hurdles, and by clarifying what REMS programs do and do not actually accomplish.

My hope was that, by sharing my interpretation of the available data, I could help put this complex and emotionally charged information into context.

But what has unsettled so many of us after the FDA’s letter is not just the safety signal itself, but the deeper question it forces us to confront:

💡 When a drug carries real risk, should it still be considered for approval, and what degree of evidence do patients deserve in return?

This isn’t just a regulatory question. It’s a moral one. A clinical one.

Progressive MS remains one of the greatest unmet needs in Neurology. For many patients, progression may not be dramatic, but it is devastating. It is slow, grinding, and relentless. Moving becomes harder. Thinking takes more effort. Memory slips away. Fatigue reshapes daily life. Independence narrows quietly, year after year. In that context, the idea that we might only approve therapies with pristine safety profiles begins to feel disconnected from reality. If the standard for approval were zero serious adverse events, many of the therapies we now consider transformative in Neurology, and certainly in Oncology, would never have reached patients.

So yes, risky drugs must remain on the table. For some patients, the alternative is not safety. It is certainty of decline.

We often talk about benefit–risk as if it were a calculation: benefit on one side, risk on the other, weigh them carefully and decide. But benefit–risk is not an exact ratio. Benefit in MS, particularly in progressive MS, is often heterogeneous. It may be modest on average, but meaningful for a subset of patients. Risk, on the other hand, may be rare but catastrophic, falling on a small number of individuals with potentially devastating consequences. Those two realities don’t cancel each other out neatly. The ethical challenge is not deciding whether benefit outweighs risk in the abstract. It’s deciding whether we understand who benefits, who is at risk, and how we can protect patients while offering choice.

Risky therapies may be necessary. They may even be transformative. But the riskier the drug, the higher the bar for clarity on benefit. This is not about demanding certainty. Medicine has never offered that. It is about proportionality. When a drug’s risks are minimal, we can tolerate ambiguity in benefit. A modest effect, a trend in the right direction, or population-level averages may be enough. But as risk increases, especially when that risk includes rare, severe, or irreversible harm, the ethical threshold shifts.

People living with MS deserve more than protection from harm, and more than access to innovation. They deserve to know:

• What kind of benefit is being offered

• How meaningful that benefit may be in daily life

• How likely they personally are to experience it

If this moment pushes our field toward smarter trial designs and greater collaboration to provide our patients with the answers they need, then the delay will not have been entirely wasted. Sometimes the hardest pauses are the ones that move us forward more responsibly.

Tolebrutinib trials: FDA’s comments on design and population selection

Below are my thoughts on some of the comments the FDA made regarding the design of the tolebrutinib studies, particularly HERCULES, and the effectiveness data that was presented.

The FDA’s Complete Response Letter to Sanofi regarding tolebrutinib is publicly available here:

https://download.open.fda.gov/crl/CRL_NDA219624_20251223.pdf

“Your enrollment criteria defined the targeted population for the proposed indication of nrSPMS, which does not align with the current multiple sclerosis (MS) course descriptor paradigm. Enrolled subjects could have either non-active SPMS (i.e., no clinical relapses and no inflammatory MRI activity) for which there are no approved therapies, or active SPMS (i.e., no clinical relapses but evidence of inflammatory MRI activity) which is considered a relapsing form of MS (RMS) and for which there are approved therapies.”

The FDA’s first argument is that the HERCULES trial population itself is not sufficiently well defined.

Sanofi, like other manufacturers conducting BTK inhibitor trials in MS, chose to define its study population as non-relapsing rather than non-active secondary progressive MS. In other words, participants were required to have no recent clinical relapses, but not necessarily an absence of MRI activity.

I agree with the FDA that the target population is not well defined.

Tolebrutinib is a drug with a dual theoretical mechanism: it acts on the peripheral immune system, through modulation of B cells, and it is also thought to exert effects within the central nervous system, through modulation of microglia. Its promise in progressive MS rests largely on the latter, its potential to address compartmentalized, smoldering inflammation that drives progression independent of relapses.

By including patients with active disease, the trial blurs an important line. It becomes difficult to determine whether the observed benefit reflects an effect on progression biology within the CNS, or whether it is driven, at least in part, by suppression of peripheral inflammatory activity.

That distinction matters.

If the drug’s apparent efficacy is mediated through its effect on disease activity, then we already have therapies that are more effective at controlling peripheral inflammation and carry less risk than tolebrutinib. In that context, accepting a higher safety burden becomes much harder to justify.

“Subgroup analyses of Study EFC16645 indicate that the observed treatment effect was greater in subjects who had T1 gadolinium-enhancing (GdE) lesions at baseline (i.e., active SPMS), which comprised 13% of the enrolled population. For subjects with GdE lesions at baseline, the hazard ratio (HR) (95% confidence interval [CI]) for the primary endpoint was 0.346 (0.183, 0.656), and without GdE lesions at baseline was 0.777 (0.601, 1.006)”

I gather from this statement that the FDA requested additional subgroup data from Sanofi. And I know many of us have been asking for such analyses. As the full extent of the subgroup findings have not yet been published, we are limited to what is reported in the FDA’s letter.

According to that summary, when the FDA examined how tolebrutinib performed across different subgroups, the treatment effect appeared strongest in participants who had evidence of active inflammation on MRI at baseline, specifically, those with T1 gadolinium-enhancing lesions. In this subgroup, which represented about 13% of the study population, tolebrutinib was associated with a clear reduction in the risk of progression. In contrast, among participants without gadolinium-enhancing lesions at baseline, the estimated treatment effect was smaller. The confidence interval for this group crossed 1, meaning the result did not quite meet conventional thresholds for statistical significance.

These subgroup analyses are not sufficient to conclude that tolebrutinib does not slow progression in people with non-active SPMS. Subgroup analyses are inherently exploratory, and absence of statistical significance is not the same as absence of effect. Still, the pattern is notable. This does not close the door on tolebrutinib’s potential role in non-active SPMS, but it does give me pause and it reinforces why clearer population definition and more precise benefit signals matter so much when we are asked to accept higher risk.

“We also note that patients with SPMS in the United States would typically have been treated with at least one approved MS therapy for RMS prior to reaching the secondary progressive phase of MS.”

This is entirely correct in my opinion.

In my own practice, the patients with whom I had discussed the possibility of switching to tolebrutinib prior to this FDA decision were individuals experiencing gradual disability progression despite being on high-efficacy disease-modifying therapies, with no evidence of ongoing disease activity on MRI or serum biomarkers.

That is the population for whom a drug like tolebrutinib would, in theory, offer something distinct.

Is there a way forward for tolebrutinib?

Yes, I believe there could be a way forward for tolebrutinib.

From my interactions with the folks at Sanofi, I believe they care deeply about meeting the needs of people living with progressive MS. They now have a 12-month window to put forward their strongest case. In my view, for tolebrutinib to have a meaningful chance of approval, two parallel strategies will be essential: refining the target population and stratifying risk.

Better defining the target population and window of opportunity

When I look back at what has been the most transformative shift in MS therapeutics, the widespread adoption of high-efficacy treatments stands out. That transformation did not occur because of a single breakthrough, but through two sequential developments. First, therapies with strong efficacy and acceptable safety profiles, most notably B-cell–directed treatments, became available. But just as importantly, the MS community collectively defined the right window of opportunity for their use: early in the disease course.

That insight fundamentally changed clinical practice. Instead of waiting for treatment failure and escalating stepwise, we moved toward earlier, proactive intervention. The result was twofold: better control of relapses and disability accumulation, and fewer adverse events, as we treated younger patients with fewer comorbidities and lower infection risk. Had we continued to reserve these therapies for later stages of disease, their full potential would never have been realized.

The same principle must apply to therapies developed for progressive MS. It is not enough to develop drugs that might target progression, we must also develop a clear understanding of how to use them. That challenge can be distilled into two questions: who stands to benefit most, and when is the optimal time to intervene?

This is where uncertainty persists with tolebrutinib.

The population with the greatest unmet need includes those who continue to experience disability progression despite optimal control of inflammatory disease activity. I deliberately avoid traditional labels such as RRMS, SPMS, or PPMS. MS is one disease, and progression independent of overt inflammatory activity occurs across labels. I see these patients every week in my clinic. Their steady decline is devastating, not just clinically, but personally.

Many do everything asked of them. They exercise, eat well, take recommended supplements, educate themselves, and remain deeply engaged in their care. And yet, despite their best efforts, progression continues. These are the individuals most willing to accept risk to slow the relentless course of their disease. I believe Sanofi needs to rise to this challenge and examine this population with particular care.

One starting point would be a deeper exploration of existing data from the HERCULES trial and its long-term extension. At the ACTRIMS 2025 meeting, Sanofi presented intriguing subgroup analyses from GEMINI I and II and HERCULES, suggesting greater delays in progression among individuals with paramagnetic rim lesions (PRLs) at baseline. Because PRLs are thought to reflect chronic active lesions marked by ongoing microglial activation, these findings raise the possibility that identifying patients with microglial-driven pathology could help refine the target population for tolebrutinib.

The challenge, of course, is that PRL detection is not yet routine and remains difficult to operationalize in everyday practice. For that reason, attention may need to shift toward biomarkers that are more readily deployable, such as serum glial fibrillary acidic protein (GFAP), to determine whether they can help identify patients most likely to benefit.

Finally, tolebrutinib’s approval in the United Arab Emirates since August 2025 presents a unique opportunity. Real-world use, if paired with a thoughtful and rigorous Phase 4 program, could help answer questions that clinical trials alone cannot, particularly around effectiveness, safety, and patient selection. Close partnership with prescribers in the UAE may allow Sanofi to generate the kind of pragmatic data needed to reduce uncertainty and clarify where tolebrutinib fits in the treatment of MS progression.

Predicting risk

This may be an unpopular opinion, but I do not think the situation we are facing with tolebrutinib is directly comparable—at least not yet—to what happened with natalizumab (Tysabri). I have written more about this previously, but the key distinction is this: once cases of PML began to emerge after natalizumab’s approval, it quickly became clear that the risk was not entirely idiosyncratic. Prior treatment exposure and duration of therapy were identified early as major risk factors, and the addition of JC virus serostatus ultimately allowed us to personalize treatment decisions in a meaningful way.

With tolebrutinib-associated drug-induced liver injury, the risk currently appears less predictable, and that is the central challenge. If liver injury is treated as purely idiosyncratic, it becomes difficult to justify exposing patients to that risk. History suggests, however, that many serious cases of drug-induced liver injury initially appear unpredictable, until patterns are carefully sought.

The path forward is not to dismiss the risk, but to understand it. That means systematically comparing patients who develop liver injury with those who do not, and looking beyond routine liver tests to identify clinical, biological, genetic, and pharmacokinetic factors that confer susceptibility. Rare toxicity becomes more acceptable when it is no longer mysterious.

What would truly change the conversation is evidence that liver injury risk can be anticipated, detected early, or meaningfully reduced through patient selection and monitoring. Mechanism-informed biomarkers, pharmacogenomic signals, exposure–response analyses, and real-world validation of safety strategies could transform a diffuse safety concern into a stratified risk. That is the difference between asking patients to take a leap of faith and offering them an informed choice.

In diseases with serious unmet need, like progressive MS, risk IS acceptable, but only when it is understood, contextualized, and paired with clear evidence of benefit.

Why MS progression trials need to change—and how

Beyond the specifics of tolebrutinib, the recent letter from the U.S. Food and Drug Administration to Sanofi offers lessons for how future trials in progressive MS should be designed, especially if the goal is to demonstrate benefit in patients with non-active disease.

At the heart of the FDA’s message is a simple but increasingly unavoidable truth: population definition matters more than ever.

For decades, clinical trials in multiple sclerosis followed a familiar template. Define a diagnostic category, enroll a broad population, randomize, and wait. That approach served us well in an era when relapses were frequent, inflammatory activity was easy to capture, and treatment effects were large enough to measure over relatively short periods of time.

But progressive MS has exposed the limits of that model.

Progression is slow, heterogeneous, and biologically complex. Patients who share the same diagnostic label may be driven by very different underlying mechanisms. As therapies move beyond suppressing inflammation toward targeting CNS-intrinsic processes, the cost of imprecision becomes harder to ignore.

The FDA’s letter did not just raise concerns about one drug. It highlighted a deeper tension: our traditional trial designs are struggling to keep up with the biology we are now trying to treat.

That tension is why I believe the field needs to move toward a different way of thinking about trials: one that is more biology-driven, patient-centered, and ultimately platform-based.

Biology-driven trials: aligning mechanism with population

Being biology-driven means asking, from the outset, whether the mechanism of the drug, the biology of the enrolled population, and the outcomes being measured actually align.

One of the FDA’s core concerns was interpretability. When trials enroll patients with different underlying drivers of disease, it becomes difficult to know why a drug appears to work or whether it is offering something meaningfully distinct from existing therapies.

Biology-driven trials are designed with this problem in mind. Their goal is not to eliminate heterogeneity altogether, but to reduce the kind of heterogeneity that matters mechanistically. When biology, mechanism, and population are misaligned, subgroup analyses proliferate after the fact. When they are aligned, trials become easier to interpret, and regulators, clinicians, and patients can have greater confidence in what the results actually mean.

Biomarkers offer a way to move past heterogeneous diagnostic categories and ask more precise questions:

• Is progression accelerating or stable?

• Is there evidence of inadequately controlled inflammatory activity?

• Are there features suggesting ongoing compartmentalized CNS inflammation?

Blood-based markers, imaging features, and digital measures do not replace clinical judgment but they add critical context. They allow trials to enrich for patients who are biologically aligned with a drug’s proposed mechanism, rather than hoping that a broad population will contain enough responders to generate a signal.

It is time for the MS research community to better leverage these existing tools in research and broaden their implementation and access in clinical practice, so patients can be matched to the right trials based on their individual biology.

Patient-centered: starting with people, not protocols

A patient-centered trial model begins with a simple but profound shift in perspective.

Instead of asking, “How do we recruit patients into this trial?” it asks,
“How do we build a system that allows patients to participate in research over time, as their disease evolves?”

In traditional clinical trials, participation is episodic. Patients are screened once, enrolled once, and then exit the research ecosystem when the trial ends or when they no longer meet eligibility criteria. For a disease like MS, where biology, disability, treatment exposure, and life circumstances change over years, this episodic approach is poorly aligned with reality. It treats patients as static snapshots rather than as individuals on a trajectory, and in doing so, it misses valuable information about how disease evolution, prior therapies, and accumulated burden shape both response and risk.

A patient-centered model would treat research participation as longitudinal rather than transactional. Patients consent once to share data over time, to remain visible to the research ecosystem even when they are not actively enrolled in a trial, and to be recontacted as new opportunities emerge. Their clinical history, imaging, biomarkers, and outcomes do not disappear when a protocol ends, they continue to inform future questions. Additionally, by following patients over time, we reduce the burden on individuals to constantly search for trials and instead place that responsibility on systems designed to match the right patient to the right study at the right moment.

Platform trials: building infrastructure that learns over time

Platform trials have the potential to bring these ideas together.

A platform trial is a different way of running investigational studies, one designed to learn continuously rather than starting from scratch each time. Instead of testing one drug in one trial and then shutting everything down, a platform trial creates a standing research framework with shared patients, shared comparison groups, and shared measurements. New treatments can be added as they become available, and ineffective ones can be dropped, without restarting the entire trial process.

In plain terms, it is like building a permanent testing ground rather than a series of pop-up experiments. Patients do not have to keep re-enrolling, fewer people receive placebo, and researchers can more quickly learn who benefits from which intervention.

For progressive MS, where change is slow and biology varies from person to person, this approach offers a path to clearer answers while making better use of patients’ time and trust.

At the core of such a platform would be a standing, longitudinal patient cohort: deeply phenotyped, consented once, and followed over time. Trials are then layered onto that infrastructure, matching patients to studies based not just on diagnosis, but on biology, trajectory, and timing.

Key features would include:

• clinical, imaging, and biomarker data collected longitudinally

• Continuous reassessment of trial eligibility as disease evolves

• Enrichment strategies based on mechanism, not just diagnostic labels

• The ability to test multiple interventions within a shared framework

While platform trials have not yet been fully embraced in progressive MS, similar models have emerged in other neurologic conditions. Initiatives such as ALS MyMatch in amyotrophic lateral sclerosis, DIAN-TU in genetically determined Alzheimer’s disease, and GBM AGILE in glioblastoma all reflect a shared philosophy: building patient-centered, biology-driven research ecosystems capable of answering complex questions in high-need diseases.

Building this kind of infrastructure for progressive MS nationally, and perhaps internationally, will require teamwork, a shared vision, agreement on target populations, standardized biomarker and imaging pipelines, harmonized outcome measures, and a safety infrastructure that matches the potential risks of future therapies. It will also require rethinking governance and funding models to ensure that the ecosystem is durable. It is not easy, but it is exciting!

Conclusion

Taken together, the FDA’s letter should not be read as a rejection of innovation in progressive MS, but as a signal that the next phase of progress will demand greater precision. Therapies like tolebrutinib are forcing us to confront a new reality: when benefits are likely to be heterogeneous and biology-dependent and when risks are real, we need to go beyond broad labels, one-off trials, or post hoc explanations. Patient-centered, biomarker-driven trials offer a way to align mechanism, population, and outcome from the start, while honoring the urgency felt by people living with MS progression. If we are going to ask patients to accept greater uncertainty and greater risk, we owe them trials that are clearer, smarter, and built for the complexity of the disease we are trying to treat.