An evidence-gated strategy for translating targeted oncologic PET radiopharmaceuticals from molecular design to clinical use

Targeted oncologic positron emission tomography (PET) radiopharmaceutical development is often presented as a linear progression from target discovery to clinical imaging. Translation instead depends on interdependent biological, chemical, pharmacokinetic, manufacturing, and regulatory decisions. This Review examines those decisions as an integrated design problem. We consider how target accessibility and heterogeneity, vector format, linker and chelator chemistry, radionuclide properties, and whole-body pharmacokinetics jointly determine image contrast and normal-organ exposure. We then assess how radiosynthesis, analytical validation, molar activity, automation, and batch release determine whether a promising construct can become a reproducible clinical product. Prostate-specific membrane antigen (PSMA) and somatostatin receptor platforms show how molecular imaging can support patient selection for radionuclide therapy when the diagnostic agent, therapeutic construct, trial population, and regulatory indication are aligned. By contrast, fibroblast activation protein (FAP)-targeted and other emerging platforms show why biodistribution or early human feasibility cannot establish clinical benefit. Across these examples, affinity alone is an insufficient design objective. The relevant endpoint is a clinically usable exposure profile that preserves target uptake while limiting background retention and dose to critical organs. Major unresolved issues include inconsistent head-to-head evaluation, incomplete dosimetry, product-specific manufacturing constraints, heterogeneous evidence standards, and region-dependent regulatory indications. We propose an evidence-gated framework that separates target suitability, molecular exposure, manufacturability, human performance, and implementation. This framework distinguishes established practice from early investigation and identifies studies most likely to reduce translational uncertainty.