How to Pick an MS Disease-Modifying Therapy in 2026

Bottom Line: More than a dozen disease-modifying therapies now span five mechanistic classes, and few have been compared head-to-head. A rational choice is best made along two axes — efficacy tier and mechanistic class — and then filtered through the individual patient: disease activity and prognosis, pregnancy plans, age, JC virus and comorbidity status, and the route and monitoring the patient can sustain. The most consequential decision is whether to begin with a high-efficacy agent or to escalate after breakthrough activity, because disability accrued during inadequate control is largely irreversible. No agent has been shown to reverse established disability, and no disease-modifying therapy alters the underlying neurodegenerative component once it is established.

The Clinical Problem: An Expanding, Largely Un-Compared Field

The number of disease-modifying therapies (DMTs) for multiple sclerosis has grown faster than the framework most clinicians use to choose among them. The decision is no longer simply which drug, but which efficacy tier, then which agent within that tier, for a given patient at a given point in the disease. The evidence base complicates this: nearly every pivotal trial compared a drug with placebo or with an older active agent, and the comparators changed over time. Ranking therapies by the relapse-rate reduction quoted in their pivotal papers is therefore unreliable, because those reductions were measured against different standards.

Efficacy Tiers, and the Limits of Cross-Trial Comparison

Grouping DMTs into modest, intermediate, and high-efficacy tiers is a useful organizing principle and underlies most contemporary algorithms. The percentages that populate those tiers, however, are not directly comparable, and treating them as if they were leads to clinical error.

The comparator column explains the apparent paradox. Natalizumab’s 68% and fingolimod’s 54–60% reductions were measured against placebo (AFFIRM, NEJM 2006; FREEDOMS, NEJM 2010). Ofatumumab’s 51–58% and ublituximab’s 49–59% reductions were measured against active teriflunomide (ASCLEPIOS, NEJM 2020; ULTIMATE, NEJM 2022), a drug that itself suppresses roughly a third of relapses. The anti-CD20 agents and natalizumab cluster at the top of the efficacy hierarchy not because of any single percentage but because of converging evidence: head-to-head superiority over platform agents, near-elimination of gadolinium-enhancing lesions, and the largest reductions in serum neurofilament light chain. Tier, rather than the headline percentage, is the unit that survives cross-trial scrutiny.

Drug Classes and Within-Class Differences

Efficacy tier indicates how strongly a drug suppresses inflammatory activity. Class determines nearly everything else: route and frequency of administration, monitoring burden, the patients it excludes, and its behavior on discontinuation. Within several classes, the differences between members are large enough to drive the choice.

Anti-CD20 monoclonal antibodies

All anti-CD20 agents deplete CD20-positive B cells and produce profound suppression of relapses and MRI activity; efficacy is broadly similar, and the choice among them is largely logistical.

The shared liabilities are more clinically important than the differences: progressive hypogammaglobulinemia, blunted vaccine responses, and a gradual accumulation of infection risk, which together make pre-treatment vaccination and immunoglobulin monitoring essential. Ublituximab illustrates a separate point: marked suppression of relapses and lesions did not translate into a measurable disability benefit within the trial window (12-week confirmed worsening, 5.2% vs 5.9%; P=0.51), a reminder that relapse control and disability protection are related but distinct endpoints.

Sphingosine-1-phosphate receptor modulators

The members of this class differ enough in receptor selectivity, monitoring requirements, and half-life to suit different patients.

All share class characteristics: confirmation of varicella immunity before initiation, attention to first-dose bradycardia, and a genuine risk of rebound disease activity on discontinuation, occasionally exceeding baseline severity. That rebound liability is a practical reason to be cautious about starting an S1P modulator in a patient who may wish to conceive in the near term.

Fumarates, teriflunomide, and the immune-reconstitution agents

The fumarates differ chiefly in tolerability: dimethyl fumarate (DEFINE/CONFIRM, NEJM 2012) and diroximel fumarate share the same active metabolite, but diroximel was developed to reduce the flushing and gastrointestinal effects that drive early discontinuation. Both require lymphocyte monitoring for the rare risk of progressive multifocal leukoencephalopathy. Teriflunomide (TEMSO, NEJM 2011; TOWER, Lancet Neurology 2014) has become less a first choice than the active comparator against which newer high-efficacy agents are measured; its teratogenicity (requiring accelerated elimination), hepatotoxicity, and hair thinning constrain its use. The immune-reconstitution therapies operate on a different principle — a short treatment course followed by prolonged benefit without continuous immunosuppression. Two years of cladribine tablets followed by two years of placebo matched four years of continuous treatment (CLARITY Extension, Mult Scler J 2018). Alemtuzumab is more potent still, with most patients requiring no further courses through five years (CARE-MS I 5-Year, Neurology 2017), but its risk of secondary autoimmunity (thyroid disease, immune thrombocytopenia, anti-glomerular basement membrane nephritis) and four years of mandatory monthly monitoring have moved it to a late line for all but the most active disease.

Natalizumab

Natalizumab remains among the most effective single agents in relapsing MS, reducing the annualized relapse rate by 68% and the risk of sustained disability progression by 42% versus placebo (AFFIRM, NEJM 2006). Its use is governed almost entirely by a measure unrelated to efficacy: the JC virus antibody index. In a JC virus–negative patient it is a highly effective, rapidly acting therapy; as the index rises and exposure lengthens, the risk of progressive multifocal leukoencephalopathy increases. Extended-interval dosing reduces that risk, and discontinuation carries a risk of rebound. Natalizumab is thus a therapy whose efficacy is settled and whose use is a risk-stratification exercise.

Bruton tyrosine kinase inhibitors

The newest class warrants precise interpretation. Brain-penetrant BTK inhibitors act on both B cells and CNS-compartmentalized microglia, and were developed to address the smoldering, progression-independent biology that anti-inflammatory DMTs largely do not affect. The evidence to date is mixed by design. Tolebrutinib significantly slowed disability progression in nonrelapsing secondary progressive MS, the first agent to do so in that population (HERCULES, NEJM 2025). In relapsing MS, however, tolebrutinib was not superior to teriflunomide on annualized relapse rate, and the primary endpoint was not met (GEMINI 1 & 2, NEJM 2025). The class therefore holds promise in progressive disease rather than as a more effective therapy for relapses, and it carries a hepatotoxicity signal that requires liver monitoring. It should be regarded as emerging rather than established.

Treatment Strategy: Escalation versus Early Highly-Effective Therapy

The choice of strategy is as consequential as the choice of drug. Escalation begins with a modest, well-tolerated agent and steps up only after breakthrough activity. Early highly-effective treatment begins with a high-efficacy agent and de-escalates later if warranted. A third approach, immune reconstitution, front-loads a short high-efficacy course followed by treatment-free benefit.

The argument for early high-efficacy treatment is that disability accrued during inadequate control is largely irreversible, and large observational registries consistently associate earlier high-efficacy treatment with better long-term disability outcomes. The argument for escalation is that registry data are confounded by indication, that many patients do well on platform agents, and that high-efficacy agents carry cumulative risk. The first randomized strategy trials designed to address this question, DELIVER-MS and TREAT-MS, have begun to report, and the early data are nuanced rather than definitive; the randomized evidence is still consolidating. A parallel shift is toward treating earlier in the disease spectrum: teriflunomide reduced the risk of a first clinical event in radiologically isolated syndrome by 63–72% (TERIS, JAMA Neurology 2023), and high-dose vitamin D reduced disease activity in clinically isolated syndrome by 34% (D-Lay MS, JAMA 2025).

A Framework for the Individual Patient

The two axes can be combined into a practical sequence:

1. Assess disease activity and prognosis. Favor high-efficacy or early high-efficacy therapy when prognosis is poor: frequent or severe relapses, incomplete recovery, high T2 lesion burden, infratentorial or spinal-cord or gadolinium-enhancing lesions, early disability, high serum neurofilament light, older age at onset, male sex, and African-ancestry or Hispanic background.
2. Address pregnancy and family planning. Glatiramer acetate and interferon-β are the lowest-risk to continue; natalizumab and the anti-CD20 agents have workable peri-conception strategies; teriflunomide (teratogenic, requiring washout) and the S1P modulators (teratogenic, with rebound risk) are best avoided in patients planning conception soon.
3. Weigh age and immunosenescence. Beyond approximately 55 years, the relapse-prevention benefit of high-efficacy therapy narrows while infection risk rises, favoring consideration of de-escalation.
4. Screen risk modifiers. JC virus index (natalizumab), cardiac and macular history (S1P modulators), liver disease (teriflunomide, BTK inhibitors), vaccination status before any depleting or S1P agent, and malignancy history.
5. Match the patient’s circumstances. Oral, subcutaneous home, or infusion administration; monitoring access; adherence; and insurance coverage.
6. Match the phenotype. Relapsing MS opens the full range; active secondary progressive MS indicates siponimod; primary progressive MS with inflammatory activity indicates ocrelizumab; nonrelapsing secondary progressive disease is the emerging niche for tolebrutinib.

Monitoring and Safety

Disease-modifying therapy in multiple sclerosis is as much a monitoring program as a prescription. Baseline screening and ongoing surveillance differ substantially by class, and omitting them accounts for a large share of preventable harm.

Managing Adverse Events

A small number of class-defining adverse events account for most of the serious risk, and each has an established mitigation pathway.

Assessing Treatment Response

Choosing a therapy is only the first step; recognizing inadequate control determines when to switch. The pragmatic composite is NEDA-3 — no relapses, no confirmed disability progression, and no new or enlarging T2 or gadolinium-enhancing lesions on MRI. In practice, a surveillance MRI is repeated 3–6 months after starting a therapy to establish a new on-treatment baseline, then annually, with closer intervals for higher-risk agents (for example, JC virus–positive patients on natalizumab). Serum neurofilament light chain is an emerging marker of subclinical activity and treatment response, but it is not yet a stand-alone decision tool. Genuine breakthrough activity — a clinical relapse or new MRI lesions despite adequate therapy and confirmed adherence — should prompt escalation rather than continued observation.

Vaccination

Immunization status should be reviewed and completed before starting any depleting or S1P-modulating therapy, ideally at least 4–6 weeks beforehand. Live-attenuated vaccines must be given before B-cell depletion or S1P initiation and avoided during treatment; confirming varicella immunity, and vaccinating seronegative patients, is part of pre-treatment screening. Inactivated vaccines — including annual influenza, COVID-19, and pneumococcal — are safe but produce a blunted antibody response under anti-CD20 therapy, so they are best timed before a depletion course or just before the next cycle when B cells have partially reconstituted.

Switching and Sequencing Therapies

Most harm in switching comes from two sources: leaving a treatment gap that permits rebound, and stacking immunosuppression that compounds infection or PML risk. Escalating from a lower- to a higher-efficacy agent is generally straightforward and tolerates a short washout. The higher-risk transitions are off natalizumab and off the S1P modulators, both of which can produce rebound disease activity — sometimes worse than baseline — if discontinued without a prompt successor. Discontinuing natalizumab in a JC virus–positive patient adds the competing concern of carry-over PML risk, so the transition window (commonly 4–8 weeks) should be neither too long, which invites rebound, nor too short, which compounds risk. After an immune-reconstitution therapy (cladribine or alemtuzumab), the approach is to monitor and re-treat or switch only for genuine breakthrough activity rather than to bridge continuously. When a patient on an anti-CD20 agent plans pregnancy, the usual strategy is to time conception after B-cell depletion rather than to continue dosing through conception.

Distinguishing Multiple Sclerosis from NMOSD and MOGAD

Before committing a patient to MS disease-modifying therapy, aquaporin-4–antibody neuromyelitis optica spectrum disorder (NMOSD) and myelin oligodendrocyte glycoprotein antibody–associated disease (MOGAD) must be excluded, because they mimic MS but require different treatment — and because several MS therapies can worsen NMOSD. Interferon-β, natalizumab, and fingolimod have each been associated with NMOSD exacerbation, so an antibody-positive patient misclassified as having MS can be harmed by an otherwise reasonable MS drug. NMOSD has its own evidence-based therapies: the complement inhibitors eculizumab (PREVENT, NEJM 2019) and ravulizumab (CHAMPION-NMOSD), the anti-CD19 agent inebilizumab (N-MOmentum, Lancet 2019), and the interleukin-6 receptor antagonist satralizumab (the SAkura program), alongside off-label rituximab. MOGAD, by contrast, is frequently steroid-responsive and monophasic, with immunoglobulin or rituximab reserved for a relapsing course. Sending aquaporin-4 and MOG antibodies before starting therapy is therefore a prerequisite, not an afterthought.

Pregnancy and Lactation

Family planning frequently overrides efficacy in therapy selection, and the management differs sharply by class.

Special Situations

Progressive MS. The agents with progressive-phase evidence — ocrelizumab in primary progressive MS (ORATORIO, NEJM 2016), siponimod in active secondary progressive MS (EXPAND, Lancet 2018), and tolebrutinib in nonrelapsing secondary progressive MS (HERCULES) — share an important qualification: they are most effective where inflammatory activity persists. The “active” descriptor in their labeling identifies the patients in whom the drug has a target.

De-escalation and discontinuation. Stopping is not uniform across classes. Anti-CD20 agents and the immune-reconstitution therapies can generally be spaced out or stopped with manageable risk; natalizumab and the S1P modulators carry a meaningful risk of rebound disease activity that can exceed baseline severity. The exit strategy should be considered at the time of initiation, not only later.

Symptomatic Management

Disease-modifying therapy does not address the symptoms that most affect daily function, which require their own targeted treatment.

Supplementary Table: Disease-Modifying Therapies at a Glance

A consolidated reference to the agents discussed above, organized from platform therapies through the high-efficacy classes to the BTK frontier. Indications follow FDA labeling conventions, where “relapsing forms” encompasses clinically isolated syndrome (CIS), relapsing-remitting MS (RRMS), and active secondary progressive MS (active SPMS).

Conclusion

The choice of a disease-modifying therapy in multiple sclerosis can be made systematically: identify the efficacy tier the patient’s disease requires, select an agent within that tier on the basis of mechanism, monitoring, pregnancy plans, and phenotype, and decide deliberately between early high-efficacy treatment and escalation. The remaining uncertainty is not in the drugs themselves but in long-term strategy — whether beginning with a high-efficacy agent measurably changes the disease course decades later compared with escalation. Observational data favor early high-efficacy treatment; the randomized trials designed to settle the question are only now reporting. Until they do, the decision rests on individualized risk assessment as much as on trial data.