Ipsen closes the Kartos acquisition — and inherits one recruiting Phase 3 with a 31 December 2026 primary completion date

Ipsen said on 21 August 2026 that it has completed — not signed — its acquisition of Kartos Therapeutics, and with it takes ownership of navtemadlin, an oral MDM2 inhibitor the company describes as in Phase III clinical development in myelofibrosis ^[1]. What Ipsen owns, in public-record terms, is one recruiting trial: NCT06479135, 600 patients planned, navtemadlin or placebo added 2:1 to ongoing ruxolitinib in patients with a suboptimal response, co-primary endpoints of spleen volume reduction ≥35% and total symptom score reduction ≥50%, primary completion date 31 December 2026 ^[4]. Four earlier Kartos-sponsored studies, including the Phase 2/3 comparison against best available therapy, sit on ClinicalTrials.gov with an unverified status and no results posted ^[5][6][8][9].

What happened

The announcement is Ipsen's own, issued in English and in French under the same release number on the same day — a bilingual corporate communication rather than a journalist's pickup ^[1][2]. It names navtemadlin as the acquired late-stage asset and frames the transaction as strengthening Ipsen's late-stage oncology pipeline ^[1].

The price is where the record thins. A news headline dated 21 August 2026 puts the deal at USD 450m ^[3]. That figure rests on an aggregator headline and not on the text of the Ipsen release available to Prognyx; no upfront-versus-milestone split, no consideration structure and no financing detail is established in the primary sources reviewed, and readers should treat the number as secondary reporting until Ipsen's own regulated disclosure carries it ^[3][1].

Navtemadlin's identity is unambiguous. ChEMBL registers it as CHEMBL3125702, mechanism "Tumour suppressor p53/oncoprotein Mdm2 inhibitor", maximum phase 3 — an oral small molecule that blocks the MDM2–p53 interaction rather than the JAK pathway ^[7]. The registry record for the older Kartos study states the positioning plainly: "Inhibition of MDM2 is a novel mechanism of action in MF" ^[5].

Why it matters

The interesting thing about this deal is not the mechanism. It is the pivot the mechanism is now being tested through.

Kartos originally ran navtemadlin as a replacement for a failing JAK inhibitor: NCT03662126 randomized patients relapsed or refractory to JAK-inhibitor treatment 2:1 to KRT-232 versus best available therapy chosen by the treating physician, with crossover permitted after six months or on disease worsening, 385 patients, primary outcome spleen volume reduction ^[5]. That study started on 15 January 2019 with a primary completion date of 31 December 2023 ^[5].

The asset Ipsen bought is running the opposite experiment. NCT06479135 puts patients on ruxolitinib alone in a run-in period, identifies those with a suboptimal response, and then randomizes them 2:1 to navtemadlin or matching placebo on top of continuing ruxolitinib — double-blind across subjects, physicians, central endpoint assessors and the sponsor ^[4]. The design was flagged publicly in March 2025 ^[14] and is referred to in trade coverage as POIESIS, a name that does not appear in the registry fields Prognyx retrieved ^[13]. Its dose-finding basis is NCT04485260, a 36-patient Phase 1b/2 whose stated objectives were the recommended Phase 2 dose of KRT-232 with ruxolitinib and spleen volume reduction at week 24 ^[8].

Two design facts decide what this trial can and cannot prove. First, the comparator arm is continuing ruxolitinib plus placebo — so the readout is a clean add-on effect, not a head-to-head against another agent ^[4]. Second, the co-primaries are SVR35 and TSS50, the ruxolitinib-era endpoint pair; there is no survival, no transformation and no marrow-fibrosis endpoint among the primaries ^[4]. In Prognyx's reading, the run-in enrichment cuts both ways: it selects a population that has already demonstrated it cannot reach an optimal spleen response on ruxolitinib alone, which lowers the bar the placebo arm has to clear and simultaneously narrows the commercial claim to a defined salvage-by-addition population. Myelofibrosis itself supplies the pressure — progressive marrow fibrosis, symptomatic splenomegaly from extramedullary haematopoiesis, constitutional symptoms, progressive cytopenias and leukaemic transformation in a subset of patients, in a field organised around JAK1/2 inhibition ^[11].

Who is exposed

One name is certain: Ipsen. It has closed on an asset whose entire public efficacy record, as far as the sources reviewed go, is a congress presentation. Data on navtemadlin in relapsed or refractory myelofibrosis versus best available therapy after JAK-inhibitor treatment were presented by John O. Mascarenhas, MD, per an item dated 10 December 2024 — a headline with no numbers attached, and nothing posted to the registry to check it against ^[12][5].

Beyond that, Prognyx did not assemble the competitive set for this briefing. No comparator asset in JAK-inhibitor-treated myelofibrosis is supported by a primary source in the material reviewed, so this briefing names none rather than describing a field it cannot cite. The one adjacent claim the sources do support is that the class thesis is broader than the indication: a May 2026 review in Leukemia argues that pharmacologic MDM2 targeting is "one of the most compelling strategies for therapeutic reactivation of wild-type p53 in hematologic malignancies" and is relevant across myeloid neoplasms including AML ^[10]. Anyone working that biology now has a large-cap owner of a Phase 3 MDM2 asset to plan around.

What Prognyx could not verify

Four Kartos-sponsored records — the Phase 2/3 versus best available therapy ^[5], the KRT-232 plus TL-895 combination and JAK-inhibitor-intolerant study ^[6], the ruxolitinib dose-finding study ^[8] and the JAK-inhibitor-naïve Phase 2 ^[9] — all carry an unverified overall status on ClinicalTrials.gov, none has results posted, and none records a reason for stopping. That is genuinely ambiguous: a programme that was discontinued and a sponsor that simply stopped updating its records look identical from the outside, and nothing in the sources reviewed distinguishes them. Any statement about navtemadlin's efficacy versus best available therapy — response rates, statistical significance, whether the study met its primary endpoint — is unsourced here and is not asserted. Nor could Prognyx confirm any FDA or EMA designation for navtemadlin, the full indication wording in the Ipsen release (the retrieved summary truncates it), or a regulated-information filing carrying the deal terms; Ipsen trades on Euronext as IPN with an ADR as IPSEY, so no SEC 8-K is expected in this chain ^[1].

What to watch next

31 December 2026 — the primary completion date recorded for NCT06479135 ^[4]. That is the date for final data collection on SVR35 and TSS50, not a date on which anything becomes public. Prognyx's working window for a topline disclosure is therefore the first half of 2027 at the earliest, and only if enrollment is complete: the registry snapshot reviewed is dated 25 September 2025 and still shows the trial recruiting toward 600 patients against a 3 June 2024 start ^[4]. Treat that range as ours, derived from the registry's own dates, not as a company guidance figure.

Sponsor transfer on NCT06479135. The record still names Kartos Therapeutics, Inc. ^[4]. The change to Ipsen is administrative — and it is the cheapest public confirmation available that Ipsen is operating the trial rather than parking it.

The four quiet records. Whether Ipsen updates, posts results for, or withdraws NCT03662126, NCT04640532, NCT04485260 and NCT04878003 is the single clearest signal of how much of the legacy programme it intends to stand behind ^[5][6][8][9].

The POIESIS name. If it appears in the NCT06479135 registry fields, the press-level name-to-trial link becomes a documented one ^[13][4].

The now-what

If you run an add-on in suboptimal ruxolitinib responders: re-baseline against this design, not against the old replacement paradigm. A run-in-enriched, placebo-controlled, sponsor-blinded 2:1 trial with SVR35 and TSS50 co-primaries is now the shape a large-cap sponsor has bought into ^[4]. Your control-arm assumptions and your enrichment criteria should be defensible against it.

If you need the missing efficacy picture before December 2026: assign someone to retrieve the December 2024 congress presentation of the relapsed/refractory comparison ^[12]. The registry will not give it to you — the record has no results posted ^[5] — and waiting for Ipsen to publish it is a decision to be last.

If your interest is the class rather than the indication: the MDM2/p53 rationale extends across myeloid neoplasms including AML ^[10]. What changed on 21 August 2026 is not the biology but the ownership: partnering or in-licensing conversations in this class now run through a company that has just paid to own a Phase 3 ^[1].

Verdict — threat level low today, binary from 2027. Ipsen bought a trial, not a dataset. The window to act is the four months to the 31 December 2026 primary completion date and the disclosure interval that follows it.