Remibrutinib is real. The question is how big

On September 1st Novartis announced that remibrutinib, its experimental pill for relapsing multiple sclerosis (MS), had passed the biggest test on its road to market. Two Phase III trials in roughly 2,000 patients beat the comparator on the annualized relapse rate (the number of attacks a patient suffers per year), and the company reported a favourable safety profile. The market had been told to expect something grand. Four months earlier Vas Narasimhan, the chief executive, had called the impending results "the single largest potential driver of upside" to the company's guided 5–6% growth to 2030, and described a result superior to the injected standards of care as "massive". This morning he ran the race he had summoned. Whether he won it is a more delicate question.

MS therapy has long been a study in trade-offs. The drugs with the strongest relapse control — Ocrevus, Roche's twice-yearly infusion, and Kesimpta, Novartis's monthly injection — work by depleting the B cells that sustain the autoimmune attack. The convenient pills, such as teriflunomide, offer only moderate efficacy. The industry's prize has been the thing neither side has supplied: an oral medicine with top-grade relapse control and a credible claim on the smouldering inflammation that drives disability. Inhibitors of Bruton's tyrosine kinase, an enzyme that matters to B cells and to brain-resident microglia, were the bet that this trade-off could be broken.

The record of that bet is a graveyard with one surviving contender. Merck KGaA dropped evobrutinib in 2024 after it failed two Phase III trials against the same teriflunomide. Sanofi's tolebrutinib was refused by the FDA in December 2025 despite a positive pivotal trial for progressive MS — a rebuke that said as much about the class's liver problem as about that drug. Only Roche's fenebrutinib has converted promise into evidence: its two relapse studies cut the annualized relapse rate by roughly half (51% and 59% against teriflunomide), and in progressive MS it came out at least as good as Ocrevus. But it carried baggage — liver-enzyme elevations that needed monitoring, several times more common than on Ocrevus in one study, and an imbalance in deaths across the relapse trials that investigators judged unrelated but regulators will keep weighing.

It is against that backdrop that "favourable safety profile", the blandest phrase in Novartis's announcement, turns into its most consequential one. Remibrutinib beat teriflunomide on the primary endpoint and on every key secondary endpoint, including brain lesions seen on MRI; in a pre-planned combined analysis of the two studies it showed a positive trend on three-month confirmed disability progression and a nominally significant effect at six months. And there was no liver-safety signal: no cases meeting the Hy's Law pattern that presages severe drug-induced liver injury, across a programme now spanning more than 4,500 participants. Novartis says it will file for approval globally.

Now notice what was not said: the size of the relapse reduction. The figure is being kept for a late-breaking presentation at the MS congress in Toronto and an investor call. The magnitude matters because teriflunomide is a small gate. Clearing it is an entry ticket every credible entrant has paid; the B-cell standard itself was proven by beating the same comparator — Kesimpta's pivotal trials ran against teriflunomide. Being "significant" against that drug does not by itself certify "high efficacy" against Ocrevus. Whether remibrutinib lands near fenebrutinib's halving of relapses, above it or below it is the one number that lets investors test the claim management has been asking them to capitalise. It is either topline discipline or an unflattering decimal; the market will know at Toronto.

For Novartis's shareholders the context is sharp. The company is navigating what management calls the largest patent expiry in its history, has guided 2026 core operating income down and sales up by only low single digits, and has staked its 5–6% compound growth to 2030 on launches of exactly this kind. There is also a subtlety: Novartis already owns the second-biggest B-cell medicine, its own Kesimpta, a franchise doing more than $1.1bn a quarter and growing by roughly a quarter a year. Selling a pill that competes with your own injection looks like self-sabotage. It is better read as defence: if an oral high-efficacy segment is coming anyway, and Roche is marching fenebrutinib towards regulators, the rational move is to own both needle and pill and keep the patient in the family. The readout also de-risks more than MS. The same molecule is already marketed as Rhapsido for chronic hives, an indication approved in America last year and Europe this year; its Phase III work in hidradenitis suppurativa was accelerated, and the drug is being studied in food allergy and progressive MS. That is the "pipeline-in-a-pill" logic Narasimhan has been selling.

None of this answers the durable question, which is whether oral high-efficacy medicine takes share at all. The market for MS drugs is roughly $22bn across the seven biggest health systems and is projected to reach only about $26bn by 2034 — a slow-growing pond in which a newcomer grows mainly by dispossessing incumbents. The B-cell therapies carry a decade of outcomes data and installed infrastructure; patients settled on six-monthly infusions do not rush to a twice-daily pill; and payers price orals against injections, not beside them. Two oral rivals arriving within months of each other in the same class is historically a recipe for price competition rather than land-grab.

So what did this morning actually change? It retired the class's standing fear, at least for this molecule: the drug works, and its liver is clean. What it did not do is hand investors the number that would justify "massive". That number is still in the company's filing cabinet, to be produced at Toronto and then tested in the first years of prescriptions. Aggregate rating signals from AInvest label the shares a Hold — a market declining to pay for an option whose payoff is not yet visible. That reluctance is defensible, and it will resolve one way or the other when the withheld decimal appears.