Command Palette

Search for a command to run...

Rocatinlimab in prurigo nodularis: an OX40 program stopped by three Kaposi sarcoma cases

OtherSafetyJuly 28th, 2026·6 min read·10.5281/zenodo.20479005

Amgen terminated a 469-patient Phase 3 of the anti-OX40 antibody rocatinlimab in prurigo nodularis after the trial had already reached primary completion. The posted reason is a sponsor decision, but the program-level record shows a malignancy signal with a mechanistic link to OX40 blockade.

Mechanism Risk Score

ComponentPoints
Phase-weighted failure burden15.7 / 40
Archetype severity8.5 / 25
Temporal recency4.3 / 15
Genetic evidence deficit11.1 / 15
Programmatic saturation1.0 / 5

For TNFRSF4 in Prurigo nodularis, the Mechanism Risk Score is 41/100 (yellow band). The score is a failure-burden index derived from Claidex post-mortems on this target–disease pair, not a probability of approval.

Single Phase 3 failure driven by a virus-associated malignancy signal. No genetic association evidence supports the target in this disease and five clinical programs target OX40.

This score does not predict whether the next trial will succeed. It flags how heavy the documented mechanistic failure record is before a new program is justified.

Primary figure supporting this claim (Rocatinlimab / TNFRSF4 / Prurigo nodularis): Rocatinlimab in prurigo nodularis: an OX40 program stopped by three Kaposi sarcoma cases

What was tried

Amgen ran NCT06527404, a 52-week, multicentre, randomised, double-blind, placebo-controlled Phase 3 study of rocatinlimab in adults with prurigo nodularis inadequately controlled on topical therapy or not eligible for it. The design was parallel-group with double masking across four arms: two blinded rocatinlimab dose levels, matching placebo, and an open-label arm in the second treatment period. Actual enrollment was 469 participants at 192 locations. The study started on 18 July 2024, reached an actual primary completion date of 17 April 2026, and closed on 25 June 2026.

Eligibility required disease of at least three months, at least 20 bilaterally distributed nodules, and inadequate response to topicals of medium or higher potency. Patients with active atopic dermatitis were excluded. The global coprimary endpoint was the number of participants achieving a reduction from baseline in the weekly average of the daily itch score at week 24, with a United States coprimary that also required improvement on the Prurigo Nodularis Stage Assessment. ClinicalTrials.gov records the study as terminated for "Sponsor Decision", last updated 27 July 2026, with no results posted.

Rocatinlimab is a monoclonal antibody registered in ChEMBL as CHEMBL4594567 at maximum phase 3, synonyms AMG 451 and KHK4083.

The biological hypothesis

OX40, encoded by TNFRSF4, is a costimulatory receptor on activated T cells. Rocatinlimab was developed as a T cell rebalancing therapy that inhibits and reduces the number of pathogenic T cells by targeting that receptor. Prurigo nodularis is a neuroimmune disease in which type 2 inflammation and sensory nerve signalling sustain the itch-scratch cycle, so an agent that shrinks the activated T cell pool should reduce the cytokine drive feeding it, including the interleukin 31 axis that dominates the itch phenotype.

The evidence base for that step is thin at the target level. Open Targets scores the TNFRSF4 association with prurigo nodularis at 0.259, and the entire score comes from clinical evidence at 0.426, meaning trials of the drug itself. There is no genetic association evidence for the pair. The strongest scoring mechanisms in the same disease are IL31RA at 0.494 and IL4R at 0.354, which also draw mainly on clinical evidence but rest on a longer chain of mechanistic work. The figure shows this decomposition.

What actually happened

Two events preceded the termination. On 30 January 2026, Amgen and Kyowa Kirin ended their rocatinlimab collaboration, described by the companies as the result of a strategic portfolio prioritisation by Amgen, with Kyowa Kirin assuming global control. The Phase 3 ROCKET program then comprised eight pivotal studies with more than 3,300 enrolled atopic dermatitis patients, and a regulatory submission was planned for the first half of 2026.

On 3 March 2026, Kyowa Kirin announced discontinuation of the rocatinlimab clinical trials in atopic dermatitis, uncontrolled asthma and prurigo nodularis. The stated basis was a planned safety update that identified emerging concerns of malignancies with possible viral or immune-related links: one newly confirmed case of Kaposi sarcoma, one suspected case, and one previously confirmed case. The company noted that the overall number of malignancy cases across the program remained below expected background rates, and concluded that the potential risks may outweigh the benefits in the studied populations.

The efficacy side of that judgement is public for atopic dermatitis. In ROCKET-IGNITE and ROCKET-HORIZON, reported in the Lancet in January 2026, rocatinlimab 300 mg reached EASI-75 at week 24 in 138 of 326 patients (42%) versus 28 of 219 on placebo (13%), and in 178 of 543 (33%) versus 25 of 183 (14%). Both trials met their coprimary endpoints.

Failure mechanism, best guess

The archetype is a safety signal, not the sponsor decision recorded on the registry. Kaposi sarcoma is driven by human herpesvirus 8 and emerges when T cell surveillance is impaired. An antibody whose stated action is to reduce activated T cells produces that state, so the tumour type carries more information than the count does. Comparison against background malignancy rates was the weaker test, because a pooled rate dilutes a signal concentrated in one virus-driven histology.

The benefit side made that risk harder to absorb. The atopic dermatitis effect sizes above are placebo-adjusted differences of roughly 19 to 30 percentage points on EASI-75, achieved in a market with entrenched competitors. In prurigo nodularis the comparison is sharper, because nemolizumab and dupilumab are approved and the primary endpoint was itch reduction, a measure on which interleukin 31 receptor blockade acts quickly. A slower upstream mechanism was asked to win on the endpoint where the incumbent mechanism is strongest.

How to prevent this next time

No endpoint-level prurigo nodularis data were released, so a quantitative reweighting of the efficacy prior is not supportable. The remaining levers are design levers.

First, adverse events of special interest should be anchored to the mechanism rather than to a generic oncology list. For a therapy that reduces activated T cells, that means prospective adjudication of virus-associated malignancy, including Kaposi sarcoma and Epstein-Barr virus lymphoproliferation, with a stopping rule that triggers on histology rather than on a pooled rate against background.

Second, indication sequencing should follow the benefit ceiling. An itch endpoint placed a slow upstream mechanism against a fast one, and the onset kinetics were knowable before the study opened.

Third, the absence of a genetic anchor for TNFRSF4 in this disease argues for responder enrichment. Lesional or tape-strip transcriptomics identifying OX40-high patients would have turned an unselected 469-patient trial into a test of the mechanism rather than of the average patient.

The single highest leverage change would have been a prespecified, mechanism-anchored virus-associated malignancy stopping rule, adjudicated on histology rather than benchmarked against a pooled background malignancy rate, since the three Kaposi sarcoma cases were only interpretable as a class of one.

What this means for similar programs

The OX40 axis is now carrying two negative reads in inflammatory disease within weeks of each other. Sanofi announced on 24 July 2026 that it was discontinuing amlitelimab, an OX40 ligand antibody, in atopic dermatitis, citing efficacy and safety data insufficient to improve on current standard of care. The Claidex graph already holds an amlitelimab claim in alopecia areata recorded as a sponsor decision. The two antibodies differ mechanistically, one blocking the receptor and the other the ligand, and only the receptor-directed antibody has reported a virus-associated malignancy cluster. That distinction is the testable question for the remaining programs.

Open Targets lists five clinical candidates against TNFRSF4, three of them oncology agonists, including GSK-3174998, ivuxolimab and MEDI-6469. Those programs push the receptor in the opposite direction, which makes the Kaposi sarcoma observation a coherence check on the whole axis rather than a fact about one antibody. The Claidex mechanism risk score for TNFRSF4 is 41, in the yellow band, driven by phase burden and by the absence of genetic support.

Open questions

  • Will the week-24 prurigo nodularis dataset from NCT06527404 be published, given that primary completion was reached before termination?
  • Were the three Kaposi sarcoma cases concentrated in a population with higher human herpesvirus 8 seroprevalence, and were they dose or duration related?
  • Does receptor blockade differ from ligand blockade in its effect on antiviral T cell surveillance, and can that be tested in stored samples?

Sources