OX40 Finally Worked. Inhibrx Cracked a 10-Year Class Failure

Inhibrx Biosciences released interim Phase 2 data on May 11 for INBRX-106 plus pembrolizumab in first-line PD-L1 positive head and neck squamous cell carcinoma. The combination delivered a 44.0% confirmed objective response rate (n=25) against 21.4% for pembrolizumab alone (n=14) in the HexAgon-HN study.

A 22.6 percentage point absolute improvement over pembrolizumab in 1L head and neck cancer is the kind of magnitude that changes standard of care if it holds. The control arm at 21.4% is dead on the KEYNOTE-048 benchmark, so the trial mechanics are clean. Combination patients also showed a 15-fold systemic T-cell expansion, confirming the drug is doing what the biology says it should.

Why This Matters Beyond One Drug

OX40 is a costimulatory receptor on activated T cells. Engage it in the presence of TCR signaling and T cells proliferate and attack tumors. On paper, the perfect partner for checkpoint inhibitors. Take the brakes off with anti-PD-1, hit the gas with OX40 agonism.

The class has been a graveyard. Roche, Pfizer, Genentech, MedImmune, and Bristol Myers Squibb all ran bivalent OX40 antibody programs. Almost every one failed.

The leading theory for the failures was valency. OX40 signaling requires higher-order receptor clustering. Bivalent antibodies bind without triggering the cluster. INBRX-106 is a hexavalent multispecific with six binding sites engineered to force the cluster.

The HexAgon-HN interim data is the first randomized evidence that the valency hypothesis was right. Hexavalent engineering can activate OX40 where bivalent could not.

The Caveats

Interim numbers. 25 vs 14. Statistical fragility is real at that scale. Final HexAgon-HN data is what supports registration, not the interim. Safety in the larger combo population matters as much as efficacy when the full readout comes.

PD-L1 positive patients are the easier patients to show benefit in because the underlying checkpoint response is enriched. PD-L1 negative HNSCC is the harder commercial unlock and still needs to be tested.

The Implications

If the final dataset holds, this is bigger than one drug. It validates hexavalent multispecific engineering as a class. The same approach can be applied to 4-1BB, GITR, CD27, ICOS, and other costimulatory receptors that have failed in bivalent formats.

Large pharma has been hunting for a working OX40 asset for a decade. Merck, BMS, AstraZeneca, and Roche all ran programs that died. A clinically validated OX40 agonist that plugs directly into a checkpoint franchise is exactly the kind of asset that draws acquisition interest.

The Read

INBRX-106 just delivered the first credible randomized signal for an OX40 agonist in 10 years of class failures. The 22.6 point ORR delta, the matched control arm, and the 15-fold T-cell expansion all argue the hexavalent engineering hypothesis was right.

Numbers are small. Phase 3 still has to confirm. But the class is finally alive.