A team in Shanghai built a small double-looped peptide that carries a dose of radioactive gallium, injected it into tumor-bearing mice, and watched it glow on a PET scan only where the tumor carried a lot of one specific cancer protein. The brighter the protein, the brighter the signal. That is the whole pitch, and it is a useful one.
The protein is B7-H3, also called CD276. It sits on the surface of many solid tumors, from lung to prostate to brain, and stays mostly quiet on healthy tissue, which is exactly the profile drugmakers want in a target. Several antibody-drug conjugates and engineered T-cell therapies are now in trials trying to kill cells that display it. The catch is that these treatments only help patients whose tumors actually carry the target, and there is no easy way to know that from the outside. A biopsy samples one spot with a needle. A scan reads the whole body at once.
That is the gap the new tracer is meant to fill. In the Journal of Medicinal Chemistry ↗, published online June 9, Fengsheng Zhang and colleagues at the Fudan University Shanghai Cancer Center describe three versions of a B7-H3-binding bicyclic peptide, each tagged with gallium-68, a short-lived radioactive isotope that PET scanners are built to detect. The versions differ only in the length of a small chemical spacer. The best of them, called FZ1, bound B7-H3 with a strength of 83 nanomolar, which is decent for a peptide but far looser than an antibody's grip, and all three came out more than 96 percent pure with fast clearance from the blood.
The imaging is where the idea earns its keep. Across several mouse tumor models, how much tracer collected in a tumor tracked how much B7-H3 that tumor displayed. In one lung-cancer line, engineering the cells to overexpress B7-H3 pushed tracer uptake from 1.09 to 3.50 percent of the injected dose per gram of tumor at half an hour, a bit more than a threefold jump, with no obvious toxicity along the way. In plain terms, the scan could tell a target-rich tumor from a target-poor one.
The choice of a bicyclic peptide rather than an antibody is the technical heart of this. Antibodies bind their targets far more tightly, but they linger in the bloodstream for days, so an antibody-based scan means waiting days for the background to clear before the picture is readable. A small peptide washes out in minutes to hours, which is why the team could image at 30 minutes and pair it with a short-lived isotope. The price of that speed is the weaker binding, and 83 nanomolar leaves room to do better.
This is preclinical work in mice, and the cleanest specificity result leans on tumor cells deliberately engineered to overexpress the target, not on the messier reality of human tumors. No one has imaged a person with it. But the logic is the companion-diagnostic logic that peptide chemistry keeps circling back to: see which patients carry the target, then treat the ones who do. The double-looped peptide scaffolds that make this possible, small enough to clear the blood in an afternoon, are the same class of anticancer ↗ molecule peptidemodel hosts, and imaging is the half of the theranostic pairing that decides who the treatment is even for.