Command Palette

Search for a command to run...

BND-35 and the myeloid checkpoint that was tested in the wrong tumors

OncologySponsorJune 23rd, 2026·5 min read·10.5281/zenodo.20479005

Biond's anti-ILT3 antibody targets a myeloid checkpoint whose strongest human evidence sits in acute myeloid leukemia, but the Phase 1 ran in solid tumors and stopped at 12 patients on a strategic call.

Mechanism Risk Score

ComponentPoints
Phase-weighted failure burden4.7 / 40
Archetype severity3.8 / 25
Temporal recency4.3 / 15
Genetic evidence deficit13.3 / 15
Programmatic saturation2.5 / 5

For LILRB4 in Advanced solid tumor, the Mechanism Risk Score is 29/100 (yellow band). The score is a failure-burden index derived from Claidex post-mortems on this target–disease pair, not a probability of approval.

MRS 29/100 (YELLOW). 1 programs across LILRB4 have been documented for LILRB4 in Advanced solid tumor: 0 Phase 3, 0 Phase 2, 1 Phase 1 — of which 0 were efficacy failures, 0 safety, 0 biomarker, and 0 operational (enrollment, sponsor, or funding). The most informative failure on file is BND-35 and the myeloid checkpoint that was tested in the wrong tumors. This score quantifies the documented failure burden; the Open Targets association score of 0.11 reflects weak genetic anchoring, compounding the documented failure record. The MRS is not a prediction of future trial outcomes — it is a structured summary of the empirical record, recomputed live from the Claidex claims table, and intended to flag mechanisms where any new program must explicitly resolve each prior failure mode before pursuit is justified.

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 (BND-35 / LILRB4 / Advanced solid tumor): BND-35 and the myeloid checkpoint that was tested in the wrong tumors

What was tried

BND-35 is a humanized IgG4 antibody that antagonizes ILT3, the receptor encoded by LILRB4. Biond Biologics ran it in NCT06274437, a Phase 1 dose-escalation and dose-optimization study in advanced solid tumors, dosing BND-35 alone and in combination with the anti-PD-1 antibody nivolumab or the anti-EGFR antibody cetuximab. The study used a non-randomized sequential design with no masking, opened in June 2024, and enrolled 12 patients before it was terminated. The registry reason was "Study terminated due to strategic considerations," with an actual primary completion date of 29 April 2026 and no posted results (NCT06274437). The cetuximab combination arm was supported by a clinical supply collaboration with Merck KGaA announced in October 2024 (Biond Biologics, press release, 2024).

The biological hypothesis

ILT3, or LILRB4, is an inhibitory receptor on myeloid cells that helps build an immunosuppressive tumor microenvironment. BND-35 binds LILRB4 and blocks its interaction with ligands in that environment, including apolipoprotein E and fibronectin, which is meant to reprogram suppressive myeloid cells toward a pro-inflammatory state and restore T-cell and NK-cell activity. The design also aimed to relieve LILRB4-driven suppression of antibody-dependent cellular cytotoxicity and phagocytosis, which is the rationale for pairing it with cetuximab (Biond Biologics, AACR 2024 presentation). The clinical bet was that taking the brakes off myeloid cells would convert cold solid tumors into responsive ones, alone and on top of established antibodies.

What actually happened

The trial reached 12 patients and stopped on a strategic call before any efficacy or dose-optimization readout was posted (NCT06274437, hasResults = false). A 12-patient escalation is built to find a tolerable dose, not to estimate activity. If such a cohort recorded zero responses, the Wilson 95 percent upper bound on the true response rate would still be 24.3 percent (computed from n = 12). The study could not have separated an inactive antibody from one with a real signal near a quarter of patients. The program halted at the point where the activity question is still wide open, and well before the data needed to close it existed.

Failure mechanism, best guess

The registry calls this a strategic decision, and the data support reading it that way rather than as a biology readout. The deeper issue is where the human evidence for LILRB4 actually points. In Open Targets, LILRB4 associates with the broad neoplasm class at 0.111 and with acute myeloid leukemia at 0.111, while its solid-tumor associations are weaker, and the receptor's best-characterized role is in myeloid malignancy (Open Targets Platform, LILRB4, ENSG00000186818). The foundational work on LILRB4 describes a leukemia-cell receptor that suppresses T cells and drives tissue infiltration, and the receptor remains primarily an AML biomarker and target (Chinese Medical Journal, 2024; Biomedicine and Pharmacotherapy, 2025). BND-35 took a target whose strongest human signal is in a blood cancer and tested it first in solid tumors, where the myeloid-suppression rationale is real but diffuse and harder to convert into single-agent or combination benefit. When early data are ambiguous and the genetic anchoring for the chosen indication is thin, a program becomes an easy candidate to cut in a portfolio review.

How to prevent this next time

Two quantitative tools would have made the stop a decision rather than an accident of timing. The first is an explicit power and precision plan. A 12-patient escalation gives a 95 percent upper response bound of 24.3 percent at zero events, so a credible activity claim needs a pre-specified, biomarker-defined expansion of roughly 25 to 40 evaluable patients to bring that bound near a 10 percent threshold. The second is a Bayesian predictive probability gate fixed before dosing. The predictive probability of eventual success is

with a prior anchored to the historical oncology Phase 1 base rate near 5.3 percent for the path to approval (Wong, Siah, and Lo, Biostatistics, 2019), then updated by interim pharmacodynamics. A third lever, biomarker enrichment, would have selected patients by LILRB4 and ILT3 expression in the tumor microenvironment, since the mechanism only operates where suppressive myeloid cells are present, and it would have prioritized indications such as myeloid-rich tumors where the human evidence is strongest. The single highest leverage change would have been choosing the lead indication by where LILRB4 evidence is strongest, then enriching enrollment for tumors with documented myeloid suppression, so the first cohort tested the mechanism in its best setting rather than a generic solid-tumor population.

What this means for similar programs

Myeloid checkpoints are a crowded and largely unproven space, and BND-35 is a useful data point about sequencing rather than about ILT3 itself. Targets whose genetics and biology cluster in one disease, here acute myeloid leukemia, carry hidden risk when a program launches in a different setting to chase a broader market. The discipline is to let the evidence pick the first indication and to size the first activity cohort to clear a meaningful bound, because a 12-patient solid-tumor escalation will be cut in any tightening market before it can teach anyone anything.

Open questions

Did BND-35 produce measurable myeloid reprogramming or T-cell activation in the 12 treated patients, and will those translational data be released. Would an AML-focused or myeloid-enriched cohort have changed the trajectory of the program. And is single-agent or cetuximab-combination ILT3 blockade ever going to move solid tumors, or is the myeloid-checkpoint thesis better tested in the blood cancers where LILRB4 evidence is concentrated.

Sources

Related failure claims

Linked claims sharing target, indication, or failure mechanism.