JNJ-95475939, a bispecific IL-4Rα x IL-31 antibody, stopped for futility in atopic dermatitis at the DUPLEX-AD interim

What was tried

Janssen Research and Development ran DUPLEX-AD (NCT06881251, study 95475939ADM2001), a Phase 2b, randomized, triple-masked, parallel-group study of JNJ-95475939 in adults with moderate-to-severe atopic dermatitis (ClinicalTrials.gov, NCT06881251). The protocol used five arms: dupilumab as an active comparator, three dose regimens of JNJ-95475939, and a placebo arm that crossed over to active drug after Week 10 (ClinicalTrials.gov, NCT06881251). Eligibility required an Eczema Area and Severity Index (EASI) score of at least 16, a validated Investigator Global Assessment of at least 3, at least 10 percent body surface area involvement, and a baseline Peak Pruritus Numeric Rating Scale of at least 4 (ClinicalTrials.gov, NCT06881251). The primary endpoint was the percentage of participants achieving EASI-75 at Week 12 (ClinicalTrials.gov, NCT06881251). The study opened on 2025-02-26 and reached an actual enrollment of 256 participants (ClinicalTrials.gov, NCT06881251). It was terminated with the posted reason "Terminated: futility criteria met at interim efficacy analysis," and the primary completion date was recorded as 2026-03-26 (ClinicalTrials.gov, NCT06881251).

The biological hypothesis

JNJ-95475939, also called JNJ-5939, is a bispecific antibody that engages two type 2 cytokine axes at once, IL-4Rα and IL-31 (BioSpace, 2026-01-05). Johnson and Johnson acquired the asset from Numab Therapeutics for 1.25 billion dollars upfront in 2024 on the thesis that combining two validated targets in a single molecule could raise the efficacy ceiling of either alone (BioSpace, 2026-01-05). The IL-4Rα arm overlaps with dupilumab, the approved IL-4 and IL-13 pathway blocker. The differentiating arm is IL-31, the cytokine that John Reed of J&J described as an attempt to break through the efficacy ceilings of the monotherapies and that David Lee called a big driver of itch (BioSpace, 2026-01-05). IL-31 is a well-described pruritogen that acts on sensory neurons and keratinocytes (Allergy, Datsi et al., 2021), and the IL-31 network in atopic dermatitis links macrophages, basophils, and thymic stromal lymphopoietin (JACI, 2023). The single-target IL-31 receptor antibody nemolizumab was approved for atopic dermatitis in December 2024 and reduces itch with modest lesion responses (NEJM, Kabashima et al., 2020, doi:10.1056/NEJMoa1917006; Lancet, 2024, doi:10.1016/S0140-6736(24)01203-0).

What actually happened

A planned interim analysis met prespecified futility criteria, and J&J halted the study (Johnson & Johnson statement, 2025-12-26). The company reported that JNJ-5939 did not meet the high-bar efficacy it set for atopic dermatitis programs and that the molecule was well tolerated with no new safety signal (Johnson & Johnson statement, 2025-12-26). No efficacy data have been released beyond the futility outcome. The openFDA FAERS database held no reports for JNJ-95475939 at the time of writing, consistent with an investigational bispecific (openFDA FAERS, queried 2026-06-05). In Open Targets, IL31 carried an overall association with atopic eczema of 0.1862, built from a literature score of 0.395 and a genetic association score of 0.2865, with no clinical or somatic datatype contributing (Open Targets Platform, IL31 ENSG00000204671, EFO_0000274). The combined IL-31 axis is lightly populated, with two mechanism-annotated molecules against IL31 or its receptor in ChEMBL plus the JNJ bispecific (ChEMBL v34).

Failure mechanism, best guess

The most likely explanation is a translational mismatch between the endpoint and the added biology. EASI-75 measures lesion area and severity, not itch. IL-31 governs pruritus far more than it governs lesion clearance, and nemolizumab monotherapy produces strong itch relief with only modest EASI gains (NEJM, 2020). Stacking an IL-31 arm onto an IL-4Rα arm therefore added an axis that is weakly tied to the primary readout, while the IL-4Rα arm duplicated biology that dupilumab already saturates. The bet that two validated targets would multiply on a lesion endpoint did not survive contact with the interim data. The weak genetic anchoring supports this read. IL31 had no clinical datatype and an overall Open Targets association below 0.19, so the program rested on pharmacology and a mechanistic itch story rather than human genetic causality for eczema severity (Open Targets Platform, IL31 ENSG00000204671, EFO_0000274).

How to prevent this next time

Two quantitative tools would have sharpened the go decision before a 256-patient Phase 2b.

First, a historical base-rate adjustment anchored to selection biomarkers. In the 2011 to 2020 record, the likelihood of approval from Phase 1 was 7.9 percent across all programs, but it rose to 25.9 percent for programs that used selection biomarkers versus 8.4 percent without (Clinical Development Success Rates 2011-2020, BIO, Informa Pharma Intelligence, QLS Advisors). A bispecific whose differentiating arm targets an itch cytokine should have entered with a biomarker that identifies the IL-31-high, itch-dominant subgroup most likely to separate from dupilumab, rather than an all-comers EASI-75 design.

Second, a Bayesian predictive-probability framework with an explicit interim gate. The predictive probability of trial success integrates the probability of a positive final result over the posterior for the treatment effect given interim data.

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

Pre-specifying this quantity, with a prior centered on the modest incremental effect that IL-31 blockade adds over IL-4Rα blockade, would have set a quantitative stop boundary tied to the realistic ceiling rather than to an aspirational one. The same machinery supports a biomarker-enriched interim that tests the dual-mechanism claim where it is biologically strongest.

The single highest leverage change would have been to enrich enrollment for an itch-dominant, IL-31-high subgroup and gate continuation on a pre-specified Bayesian predictive probability rather than testing the dual-mechanism claim in an all-comers EASI-75 design.

What this means for similar programs

Bispecific antibodies that fuse two validated targets inherit the burden of proving that the combination beats the better single agent on the registrational endpoint, not on a mechanistic surrogate. When one arm duplicates a saturated pathway and the other governs a symptom that the primary endpoint does not measure, the math favors the incumbent. Atopic dermatitis programs that add an itch axis should power and enrich for itch, and should treat EASI-75 superiority over dupilumab as a separate and harder claim than itch superiority. For targets with no clinical or genetic datatype, the absence of human causal evidence is a signal to demand a biomarker before committing to a large parallel design.

Open questions

What were the EASI-75 separation curves between JNJ-95475939, dupilumab, and placebo at the interim, and will J&J release them? Did the IL-31 arm deliver itch benefit even as lesion endpoints failed? Would an IL-31-high or pruritus-dominant subgroup have separated from dupilumab? Does any bispecific that includes an IL-4Rα arm have room to beat dupilumab on lesion clearance, or is that ceiling already reached?

Sources

1. • ClinicalTrials.gov, NCT06881251 (DUPLEX-AD design, five arms, enrollment 256, EASI-75 Week 12 primary endpoint, termination reason, start 2025-02-26, primary completion 2026-03-26). - Johnson & Johnson Statement on the Phase 2b DUPLEX-AD Study, 2025-12-26 (interim futility, well tolerated, no new safety signal). - BioSpace, 2026-01-05, J&J Stops Mid-stage Eczema Trial as 1.25B Asset Misses Efficacy Mark (IL-4 and IL-31 bispecific, Numab 1.25 billion dollar deal, dual-mechanism rationale quotes). - Open Targets Platform: IL31 (ENSG00000204671) association with atopic eczema (EFO_0000274) overall 0.1862, literature 0.395, genetic association 0.2865. - openFDA FAERS, queried 2026-06-05 (no reports for JNJ-95475939). - ChEMBL v34: two mechanism-annotated molecules against IL31 or IL31RA. - Kabashima et al., Trial of Nemolizumab and Topical Agents for Atopic Dermatitis with Pruritus, NEJM, 2020,- Nemolizumab with concomitant topical therapy in adolescents and adults with moderate-to-severe atopic dermatitis, Lancet, 2024,- Datsi et al., Interleukin-31: the itchy cytokine in inflammation and therapy, Allergy, 2021,- IL-31-generating network in atopic dermatitis, JACI, 2023,- Clinical Development Success Rates and Contributing Factors 2011-2020, BIO, Informa Pharma Intelligence, QLS Advisors (Phase 1 likelihood of approval 7.9 percent; 25.9 percent with selection biomarkers versus 8.4 percent without). Available from: https://clinicaltrials.gov/study/NCT06881251.