Stelara's First Year Off Patent: The Copies Kept One Patient in Six
When the brand starts shedding patients, how many stay on some version of the same drug? And does a cheap version win back patients you were already losing?
The Setup
Stelara lost US exclusivity in January 2025. Interchangeable biosimilars launched through the year. On January 1, 2026, Medicare's negotiated price took effect: $4,695 for a 30-day supply against a $13,836 list. Within twelve months, dermatologists had cheap copies — and a much cheaper Stelara.
We looked only at plaque psoriasis: self-administered subcutaneous fills at the pharmacy — the only setting where interchangeable substitution operates at the counter. Crohn's and colitis load IV under the medical benefit; mixing routes would obscure the answer. Most of Stelara's revenue was in IBD, so every figure here describes a minority of the franchise. National biosimilar-share numbers cover the whole molecule — they are not measuring the same thing this is.
- Substitution was fast. Copies went from a rounding error to nearly 30% of ustekinumab patients in a year.
- Retention was not. Roughly one in six who left Stelara was on a copy eighteen months later.
- They didn't go to a competitor. No rival took more than about one in ten. Under 7% moved to anything cheaper.
- Nobody new chose it. Low-single-digit new-start share fell steadily through both events — and dermatologists still start it first-line when they use it at all.
Where the Patients Went
Everyone established on Stelara during 2024 — before any copy existed — tracked to June 2026. Base below: patients actively on the brand in January 2025.
| Channel | On Stelara, Jan 2025 | On Stelara, Jun 2026 | On a copy | Kept on ustekinumab |
|---|---|---|---|---|
| Commercial | 3,858 | 1,122 | 409 | 15.0% |
| Medicare | 739 | 274 | 95 | 20.4% |
Blended: 504 of 3,201 patients — roughly one in six. Brand→copy conversion ~8%/quarter compounds above that endpoint; copies also lost some. The number is who were still on ustekinumab 18 months later.
Of patients on copies, 67.3% commercial and 70.1% Medicare arrived from the brand — about a third of copy volume was never Stelara's to lose. Of the established commercial cohort, IL-17s took 7.3%, Skyrizi 5.6%, Tremfya 3.8%. Methotrexate, acitretin and phototherapy absorbed 2.5%; Otezla and Sotyktu another 2.7%. Under 7% went anywhere less expensive. About one in ten stopped advanced therapy over six months. Patients weren't trading down. They were leaving. Ninety-seven patients switched into a copy from a competitor across 2025 — twenty from expensive injectables.
The finding: a cheap version of your drug does not recover patients from the drugs that took your market. It catches a minority of the ones already walking out.
The Coverage Mandate Moved Retention, Not Acquisition
Medicare kept one in five against commercial's one in seven — in the channel required to keep covering the brand, while commercial formularies dropped it. Express Scripts excluded the reference product from its 2026 formulary. MMIT (May 2025): payers covering 83% of commercial lives expected to drop it. All three large PBMs launched private-label vehicles: Cordavis, Quallent, Nuvaila.
Patients aging into Medicare while still labeled commercial push commercial retention up — the real gap is wider. On a tighter group the gap flips sign across eras: Medicare retained the brand less before copies, more during the mandate year. A fixed demographic difference can't flip sign. The mandate held existing patients on the brand and did nothing where new patients were being started.
The Share Figure and the Shrinking Base
Copies reached 29.0% in fifteen months — quick. Total ustekinumab patients fell 24.2%; the brand alone fell 46.0%. When the total is falling, copy share can rise while copy volume goes nowhere. Three constructions (diagnosis required / derm+rheum / neither) gave 27.9%, 27.2–27.5%, and 29.0%.
Nobody New Chose It — Before or After
Class volume rises every Q1 when benefits reset. The clean comparison is same quarter, year over year. The slide begins three quarters before any copy and continues through them. Cheap copies and a 66% cut on the brand recovered nothing at the first-prescription decision.
Dermatologists Still Use It First
The comfortable reading is that ustekinumab became a fallback. It isn't. First-line rates sit inside the Skyrizi/Tremfya range; doctors reach for it at the same decision point — about one-seventh as often.
| Drug | New starts | First-line | Avg. prior classes |
|---|---|---|---|
| Skyrizi, H1 2025 | 10,781 | 71.5% | 0.32 |
| Skyrizi, H2 2025 | 8,287 | 70.2% | 0.33 |
| Tremfya, H1 2025 | 5,023 | 68.2% | 0.36 |
| Tremfya, H2 2025 | 4,281 | 63.7% | 0.42 |
| Ustekinumab, H1 2025 | 1,536 | 73.2% | 0.31 |
| Ustekinumab, H2 2025 | 984 | 67.1% | 0.40 |
The Cheap Version Didn't Change the Decision Either
If lower price reached the moment a doctor picks a first drug, patients whose first-ever ustekinumab was a copy should sit earlier in treatment history than brand starters. Same drug, same everything except cost. Second half of 2025:
| First ustekinumab Rx | Patients | First-line | Avg. prior classes |
|---|---|---|---|
| A copy | 239 | 66.1% | 0.38 |
| The brand | 745 | 67.4% | 0.41 |
No difference worth the name — if anything copy patients sit marginally later. Cheap ustekinumab gets prescribed like ordinary ustekinumab.
What We Checked
| Question | Finding |
|---|---|
| Pushed later in treatment? | No. 67–73% first-line, inside Skyrizi/Tremfya range. |
| Definition-dependent share? | No. Three definitions: 27.2%–29.0%. |
| Why below national copy share? | National = all indications + routes. This = PsO SC only. |
| Did patients leave over cost? | No. Under 7% went anywhere less expensive. |
| One competitor took the base? | No. Nothing above ~1 in 10; destinations scattered. |
| Copy-first cheaper entry? | No. Same first-line rate as brand-first (H2 2025). |
What the Data Covers
Methodology
- Open claims, plaque psoriasis, ~32 months through July 2026
- SC ustekinumab via pharmacy benefit; derm/rheum specialty
- Patients counted, not prescriptions (dosing intervals differ)
- Indication by specialty (Dx-code window drops long-tenured pts)
- Sequence = distinct classes in prior year (data from Nov 2023)
- 2024-established cohort — before any copy existed
Limitations
- No reason for choice, severity, or post-rebate paid amount
- Interchangeable switches may occur without prescriber knowledge
- Payer channel in open claims is imprecise
- Stop at 2026 Q1 — later quarters show late-claims artifact
- Minority of franchise (PsO SC; most revenue was IBD)
- Copy-first test doesn't fully separate price from plan mix
What This Means for the Next One
The only biologic among three drugs leaving — pushed out by copy competition itself. Whether remaining brand patients convert faster once Part D is no longer required to cover it will say how much that requirement was holding them in place.
This is independent research. No manufacturer funded this analysis.
We can analyze biosimilar conversion, retention, and new-start share for your molecule.
Book a Consultation →