It treats fewer patients than its rival, but each one costs a third more. And the company that is behind has a move to make.
Medicare is the public health insurance of the United States, mostly for people over 65, and it works in two ways. In traditional Medicare, the government pays the site that treats the patient directly. In Medicare Advantage, chosen by more than half of enrollees, it pays a private insurer. The public data by drug covers only traditional Medicare.
Kisunla, from the American company Lilly, came out a year after Leqembi, from the Japanese company Eisai, and already costs traditional Medicare more even though it treats fewer patients: each patient costs a third more. Eisai has a move Lilly does not have today: an injectable version of Leqembi that, after the first doses, the patient can give themselves, without going back to the infusion site. I would use it first where there is a waiting list.
In the first quarter of 2026, traditional Medicare spent USD 60.4 million on Kisunla and 54.3 million on Leqembi, the two antibodies it pays for in early Alzheimer's. Leqembi still treats more people: 12,876 against 10,782.
Think about it: if the site charges a percentage of the price of the drug, which one suits it better?
Medicare pays differently depending on where the drug is given: if a professional gives it in an office or a hospital, as with these intravenous infusions, that is Part B; if it is picked up at the pharmacy, Part D.
Under Part B, the site buys the vial, gives it, and charges the average sales price plus 6%, and on top of that a fixed amount for administering it.
Think about how many steps come before the first dose.
I divided each one's spending by its patients.
In the first quarter of 2026 it was USD 5,605 per Kisunla patient and 4,219 per Leqembi patient: a third more. The difference is in the price: each milligram of Kisunla is charged 55% higher.
And what suits the site? It depends on how many patients it can treat. In my calculation, each Kisunla infusion leaves the site between 1.6 and 1.9 times more than a Leqembi one, but Leqembi is given twice as often, so over a year, per patient, Leqembi leaves more. In my hypothesis, with free chairs Leqembi suits the site; with more patients than chairs, Kisunla.
Kisunla went from 41.6% of the patients and 39.2% of the spending in 2025 to 45.6% and 52.7% in the first quarter of 2026.
Eisai has already moved. Leqembi comes in two forms, and the doctor chooses with the patient which one to use and can switch: the infusion, or an injection, Leqembi Iqlik, picked up at a pharmacy and, after the first doses given under a professional's guidance, administered by the patient or their care partner. According to Eisai, since October 2025 it served to continue treatment at 18 months, and since 25 August 2026, to start it.
Whoever uses the injection does not show up in Part B spending, where the figures above come from: in the first quarter of 2026 there were already 619 patients like that in Part D, which does include Medicare Advantage. Even adding them in, in my calculation Kisunla would still be ahead in spending.
Eisai's goal is for the new patient to start on Leqembi and stay. Leqembi sold USD 97 million in the United States in the second quarter of 2026, according to Biogen, which markets it with Eisai.
The one who chooses the drug is the doctor. But in 2024, 6 out of 10 Leqembi patients were treated in an office, where the same practice that treats them buys the drug and charges for each infusion: in my calculation, between USD 3,081 and 4,765 per patient in the first year. With the injection, the patient saves the trip and the hour in the chair, and the practice stops charging for those visits. That is where the risk is: in my hypothesis, if the injection empties its slots, that practice may be better off prescribing Kisunla, which is only given in the chair.
Where there are more patients than chairs, in my hypothesis the injection does not take money away: the slot it frees is taken by another patient who was waiting. There I would push it from the start. Where chairs are to spare, every patient who moves to the injection is an empty slot: there I would offer it without pushing it.
The first step is listing the sites that infuse Leqembi and marking where there is a wait, something that is not public and that I assume Eisai has from its sales.
How I would know it works, with cutoffs set before seeing the data (a new patient is one who received neither of the two in the previous 12 months):
I would not lower the price of the infusion: since the site charges a percentage of the price, it would charge even less per slot. What I accept in exchange is that, where chairs are to spare, Eisai keeps depending on the infusion. And the clock is running: Roche expects results for its antibody, trontinemab, by the middle of 2028.
If you were on the Eisai team, would you push the injection at every site even though some would stop charging for infusions, or would you go site by site as I propose?
Independent analysis with public data. I have no commercial relationship with Eisai, Biogen, Lilly or Roche, and never have had one.
If you read this far, something about the problem caught your interest. I post every new case on LinkedIn and discuss these decisions there with people who live them. To take it private, write me what you would do differently or pick a time to talk.
See all projects