In a first-human trial of a cancer drug, something violated the usual rules: the lower dose worked better than the higher one. Fifty-five percent of first-line lung cancer patients responded at 20 milligrams per kilogram. At 30 mg/kg, the response rate fell to 44%. Nobody has a clean explanation for why.
The finding comes from a 112-patient Phase 1/2 trial of MK-2010, a bispecific antibody that attacks both PD-1 and VEGF, two of the most heavily exploited molecular pathways in oncology, according to The Globe and Mail. The abstract does not account for the inversion, and the dose subgroups may not be directly comparable: a standard caveat in early-phase dose-exploration studies. But it is a real signal in a real trial, and it is the most interesting thing about a drug that entered the world quietly, licensed globally to Merck for $588 million upfront in November 2024, with first human data presented at a major cancer research conference and only now becoming public. The partial data showed six of eleven lung cancer patients responded at the low dose, according to Endpoints News.
That same morning, Biogen paid $100 million upfront and up to $750 million more in milestones to TJ Biopharma for Greater China commercial rights to felzartamab, a CD38 antibody already in late-stage trials for two kidney diseases. The deal is not about bispecifics. It is about geographic optionality: a manufacturing and distribution foothold in one of the world's largest drug markets if the ongoing trials succeed.
The through-line connecting both announcements is the bispecific story. The combination-therapy era, where oncologists ran two drugs in sequence or simultaneously to attack cancer from multiple angles, is giving way to a single-molecule approach. The logic is straightforward: one drug, one trial, one regulatory pathway. The FDA has shown willingness to approve bispecifics on single-arm or smaller datasets when the mechanism is compelling. And the money has noticed.
Chinese biotech companies signed more than $30 billion in oncology licensing deals in the 2024–2025 cycle, the majority of them bispecific or multi-targeted assets. Bristol Myers Squibb paid BioNTech $11.1 billion for a bispecific program. Akeso sold global rights to its ivonescimab bispecific to Summit Therapeutics for up to $5 billion. Pfizer spent $1.25 billion on a 3SBio asset. The pipeline logic, cheaper, faster, and cleaner, has become an industry consensus.
Ivonescimab became the first PD-1/VEGF bispecific approved anywhere when Chinese regulators cleared it in April 2025. No PD-1-based bispecific has yet cleared the FDA. Sixteen pivotal trials of PD-1 or PD-L1 paired with VEGF are now registered in first-line lung cancer globally, a list that includes programs from Sinocelltech, Huabo, BioNTech, Bristol Myers Squibb, and Pfizer. The space is not a niche anymore. It is the new standard.
MK-2010 enters that crowded field with competitive-looking early data. The 55% overall response rate at 20 mg/kg in first-line non-small cell lung cancer tracks within range of what ivonescimab posted at comparable stages. Whether the inverted dose response is a statistical artifact, a tolerability signal, or something biology does not yet have language for is a question the ongoing trial will answer. Merck has the capital and regulatory experience to run the Phase 3 it will need.
Biogen's deal is a different kind of bet. Felzartamab is a CD38 antibody with a longer development history, previously acquired through Biogen's purchase of Human Immunology Biosciences in 2024. The Greater China rights give Biogen a commercial hedge in one of the world's largest pharmaceutical markets, assuming the ongoing Phase 3 trials in IgA nephropathy and primary membranous nephropathy deliver. The $100 million upfront is modest by bispecific standards, but the milestone structure and royalty terms suggest both parties see a real product if the science holds.
The velocity of the shift is the real story. Big pharma has decided, at scale, that combination therapy is a transitional technology. The single-molecule approach is the destination. That conclusion was visible in the licensing data before Monday. Two announcements in one day just made it harder to miss.
