GEN1047 and the B7-H4 T-cell engager problem: 179 patients, no posted result

What was tried

Genmab ran NCT05180474, a first-in-human, open-label, sequential dose-escalation trial with expansion cohorts, testing GEN1047 in participants with advanced solid tumors. GEN1047, also registered as DuoBody-CD3-H101GxB7H4, was a bispecific antibody with one arm against B7-H4 on tumor cells and one arm against CD3 on T cells. It was given by intravenous infusion. The design used sequential assignment, no randomization, and no masking, which is conventional for dose finding. Primary endpoints were dose-limiting toxicities and treatment-emergent adverse events.

Eligibility was restricted to histologically or cytologically confirmed breast cancer, endometrial cancer, ovarian cancer, or squamous non-small cell lung cancer with no remaining standard therapy expected to confer benefit. Ovarian cancer participants had to show documented progression within 60 days of screening. The trial started on 2021-12-13 and reached an actual enrollment of 179 participants. Primary completion and study completion were both recorded as 2026-03-25. The registry gives one reason for stopping: "The Sponsor has made a strategic decision to stop the development of GEN1047." No results were posted to ClinicalTrials.gov. Genmab announced the discontinuation of GEN1047 alongside GEN3017 and GEN1056 in November 2024, with chief executive Jan van de Winkel saying the programs "simply didn't meet the high bar we have set internally."

The biological hypothesis

B7-H4, encoded by VTCN1, is a B7-family surface glycoprotein that inhibits T cell proliferation and cytokine production. Its appeal as a target rests on a contrast in expression: broad, high-level presence on carcinomas of the breast, ovary, and endometrium, with restricted expression in normal adult tissue. Work published in Immunity in 2018 also placed B7-H4 on tumor-infiltrating myeloid cells, where it induced T cell dysfunction, which framed the protein as both a tumor antigen and an immune brake.

A CD3 bispecific converts that expression pattern into a killing signal. The logic does not require B7-H4 to drive the tumor. It requires only that B7-H4 sit densely enough on the cancer cell surface, and sparsely enough on normal cells, for redirected T cells to form productive synapses at tolerable doses. Genmab's DuoBody platform supplied the format. Open Targets scores the VTCN1 association with ovarian carcinoma at 0.0638, and the evidence behind that number is literature co-occurrence (0.511 datatype score) and RNA expression (0.054), with no genetic association evidence contributing. Open Targets also records no small-molecule tractability buckets for VTCN1 and no known drugs, leaving antibody-based modalities as the only route.

What actually happened

The registry record is thin. There is no posted efficacy result, no arm-level safety table, and no disclosed recommended phase 2 dose for GEN1047. What can be established from the record is scale and duration: 179 participants dosed across roughly four years and three months, ending in a sponsor decision rather than a protocol-specified futility or safety rule.

The informative data point sits one program over. Pfizer's PF-07260437, a B7-H4 x CD3 bispecific T-cell engager, published its phase 1 dose escalation in Investigational New Drugs in 2026. Thirty patients with breast, ovarian, and endometrial cancer received 100 to 1600 micrograms subcutaneously every two weeks, with and without priming. There were zero confirmed objective responses. Stable disease occurred in 8 of 30 patients (26.7%), giving a disease control rate of 33.3%. Grade 3 treatment-related adverse events occurred in 21 of 30 patients (70.0%), led by ALT increase (46.7%), AST increase (30.0%), and alkaline phosphatase increase (23.3%). Four patients had dose-limiting toxicities including transaminase elevation and cytokine release syndrome. That study, too, was terminated by its sponsor on an overall assessment of safety, preliminary efficacy, and pharmacology.

Failure mechanism, best guess

The archetype recorded here is strategic_reprioritization, because that is what the sponsor stated and because no GEN1047 efficacy data exist to support a stronger claim. The mechanism underneath it looks like a modality mismatch rather than a target mismatch.

B7-H4-directed antibody-drug conjugates have kept moving. Preclinical and early clinical programs for SGN-B7H4V, AZD8205, and XMT-1660 all reported antitumor activity in B7-H4-expressing models, and B7-H4 ADCs remain in active development. The two B7-H4 CD3 engagers both stopped. A payload-delivery agent needs the antigen to be present and internalizing. A T-cell engager needs antigen density above a synapse-formation threshold, on the tumor cell rather than on stromal or myeloid compartments, plus a T cell infiltrate capable of responding. In breast, ovarian, and endometrial carcinoma, those additional conditions are not guaranteed by the immunohistochemistry that qualifies a tumor as B7-H4-positive. The transaminase signal seen with PF-07260437 further suggests that dose escalation ran into off-tumor T cell activation before it reached the exposure a low-density antigen would require.

How to prevent this next time

Endpoint-level data for GEN1047 were never released, so a quantitative reanalysis is not possible here. The available levers are qualitative and were available before the trial opened.

First, biomarker enrichment. Neither B7-H4 engager pre-selected participants by antigen level. For a T-cell engager, the relevant quantity is surface antigen density with a defined cutoff, not a positive or negative call by immunohistochemistry.

Second, base-rate adjustment. When GEN1047 entered escalation in 2021, the tumor-antigen CD3 engager class already had a visible pattern in solid tumors: manageable toxicity, low single-agent response rate, and long dose-escalation phases that ended without a clear recommended dose. That prior belonged in the go/no-go decision.

Third, red-team the modality choice against the ADC arm of the same target. Two modalities were being tested against B7-H4 in the same indications at the same time, and the ADC arm was the one accumulating responses.

The single highest leverage change would have been to require a prospectively defined B7-H4 surface-density threshold for enrollment, so that the absence of responses could be attributed to the mechanism rather than to the population.

What this means for similar programs

Any CD3 bispecific aimed at a B7-family ligand in carcinoma should treat the GEN1047 and PF-07260437 pair as a class signal rather than two independent business decisions. The Claidex graph already carries adjacent entries: PRO1107 against PTK7 in an overlapping ovarian, endometrial, and triple-negative breast population, and TAK-280 against CD276, another B7-family member, both stopped in phase 1 or 1/2. The pattern is consistent. Broad B7-family surface expression has repeatedly qualified a target for T cell redirection and then failed to deliver single-agent responses.

Open questions

Did GEN1047 reach a recommended phase 2 dose, and what were the response and cytokine release syndrome rates in the expansion cohorts? Was B7-H4 expression measured at baseline in the 179 participants, and did any signal track with it? Was hepatotoxicity dose-limiting for GEN1047 as it was for PF-07260437? Genmab has not published the dataset, and until it does, 179 patients of exposure produce no public evidence.

Sources

1. • ClinicalTrials.gov v2 record, NCT05180474, GEN1047 first-in-human trial. https://clinicaltrials.gov/study/NCT05180474 - Phase 1 study of PF-07260437, a B7-H4 x CD3 bispecific T-cell engager. Investigational New Drugs, 2026. https://- PF-07260437 preclinical characterization. Molecular Cancer Therapeutics, 2025. https://- B7-H4: a multifaceted immune checkpoint and oncoprotein in cancer biology and immunotherapy. Frontiers in Immunology, 2026. https://- Co-inhibitory molecule B7-S1 on tumor-infiltrating myeloid cells induces T cell dysfunction. Immunity, 2018. https://- SGN-B7H4V, a vedotin antibody-drug conjugate directed to B7-H4. Journal for ImmunoTherapy of Cancer, 2023. https://- Design and preclinical evaluation of AZD8205, a B7-H4-directed antibody-drug conjugate. Clinical Cancer Research, 2023. https://- Discovery and preclinical characterization of XMT-1660, a B7-H4-targeted antibody-drug conjugate. Molecular Cancer Therapeutics, 2023. https://- Open Targets Platform, target VTCN1 (ENSG00000134258), association with ovarian carcinoma (MONDO_0005140), accessed 2026-07-26. https://platform.opentargets.org/target/ENSG00000134258 - Genmab axes 3 programs to focus on plumped-up pivotal pipeline. Fierce Biotech, November 2024. https://www.fiercebiotech.com/biotech/genmab-axes-3-programs-focus-plumped-pivotal-pipeline.