Why promising cancer PET tracers fail: a new evidence-gated translation roadmap

A new review argues that the development of targeted PET radiopharmaceuticals for cancer has been badly misdescribed as a straight line from target discovery to clinical imaging. Writing in Frontiers in Nuclear Medicine, Qiheng Xiong, Wen He, Deshun Fu and Yu Su of Jiangxi Cancer Hospital in Nanchang, China, reframe translation as an integrated design problem in which biology, chemistry, pharmacokinetics, manufacturing and regulation are interdependent decisions. A construct that binds its target beautifully in the laboratory can still fail if it cannot be produced reproducibly, if it irradiates the wrong organs, or if the regulatory evidence never matches the population in which it was actually tested.

The authors’ central claim is that affinity alone is an insufficient design objective. The endpoint that matters, they argue, is a clinically usable exposure profile: enough radioactivity delivered to the intended biological target on the right time scale, with background retention and dose to critical organs kept low. Molecular changes that improve binding can simultaneously alter plasma protein binding, metabolism, clearance, or retention in kidneys, liver, salivary glands and marrow. For therapeutic constructs, the classical tracer assumption is strained further, because the goal is not merely to observe a biological process but to deliver cytotoxic radiation while preserving an adequate therapeutic window.

To organize this argument, the review proposes six author-derived decision gates: target suitability, molecular architecture, pharmacokinetics and dosimetry, manufacturing and quality control, human performance, and regulation and implementation. Each gate asks a different question and is tied to a minimum evidence type and a stop-or-redesign trigger. The authors are explicit that this framework is a decision aid derived by mapping recurring translational decisions across the literature, not a validated score, regulatory acceptance criterion, or universal numerical threshold.

Target suitability, the first gate, extends well beyond expression level. A candidate target is useful only if it is accessible to the selected vector and produces a clinically interpretable difference between disease and background tissues. High expression in an excised tumour sample is not enough. The review distinguishes between-patient heterogeneity, which determines how many patients qualify for a target-directed strategy, from within-patient heterogeneity, which determines whether an imaging-positive subset of disease adequately represents the total tumour burden. A single patient-level label such as target positive can conceal lesion-level discordance, so a programme must specify whether its intended use is lesion detection, whole-patient selection, response assessment, or treatment delivery.

Vector format sets the biological time scale. Small molecules, peptides, engineered proteins and intact antibodies differ in extravasation, target binding, blood clearance, metabolism and excretion. Rapidly clearing small molecules and peptides often permit same-day imaging, whereas antibodies usually require delayed acquisition to let blood-pool activity decline. Human comparisons of gallium-68- and copper-64-labelled somatostatin ligands show that radionuclide and construct choices change quantification and lesion detection rather than acting as interchangeable labels.

Linkers and chelators are pharmacokinetic variables, not passive connectors. Adding a chelator changes mass, charge, hydrophilicity and sometimes protein binding; a linker can introduce a metabolic soft spot or redirect renal and hepatobiliary clearance. The review highlights paired-comparison studies as the strongest design evidence. In an early human evaluation, gallium-68-labelled FAPI-04 showed longer tumour retention than FAPI-02, proving that a structural modification can change kinetics while target class and radionuclide stay constant. But the authors caution that such observations are informative only for those constructs and are not universal rules for all FAP ligands. Cleavable linkers illustrate the same bounded inference: a cleavable bond is a design hypothesis, not evidence that organ dose is lower, because premature cleavage in blood or tumour can reduce target delivery.

Manufacturability imposes a separate translational filter. A compound with favourable preclinical uptake may still fail if its precursor is hard to obtain, labelling is sensitive to reagent quality, purification cannot be automated, molar activity varies between batches, or analytical methods cannot establish identity, purity and stability within a clinically useful release window. The review stresses that molar activity links measured radioactivity to the amount of radiolabelled and non-radioactive chemical species, and that in saturable receptor systems, differences in precursor-derived carrier or administered mass can change binding, biodistribution and image quality. Reporting activity without chemical amount is therefore insufficient for interpreting a receptor-targeted agent. Automation controls execution but does not establish compliance: a module or cassette is not intrinsically compliant with good manufacturing practice, and the same hardware can produce a nonconforming batch if materials, calibration or documentation are inadequate.

The clinical evidence is strikingly uneven, and the review insists that evidence categories answer different questions. PSMA-targeted imaging in prostate cancer and somatostatin receptor imaging in neuroendocrine tumours are the two established platforms, defined as having a reproducible selection method, randomized therapeutic evidence, and at least one region-specific authorization. The VISION trial showed that adding lutetium-177-labelled PSMA-617 to standard care improved imaging-based progression-free survival from 3.4 to 8.7 months and overall survival from 11.3 to 15.3 months in men with progressive PSMA-positive metastatic castration-resistant prostate cancer, though grade 3 or higher adverse events were more frequent. PSMAfore extended the evidence to taxane-naive patients, with radiographic progression-free survival of 9.3 versus 5.6 months against an alternative androgen-receptor pathway inhibitor, although overall survival was not statistically different at the FDA analysis, complicated by 57 percent of control-arm patients crossing over to the radioligand. The US indication was expanded on 28 March 2025, while the European Medicines Agency description remained framed differently at the review’s cut-off.

On the somatostatin receptor side, NETTER-1 established a marked progression-free-survival benefit for lutetium-177 DOTATATE in advanced midgut neuroendocrine tumours, but the prespecified long-term analysis found median overall survival of 48.0 versus 36.3 months that did not reach statistical significance. NETTER-2 extended the evidence to first-line treatment of grade 2 to 3 gastroenteropancreatic tumours, with progression-free survival of 22.8 versus 8.5 months. The review also notes that the FDA’s 2024 paediatric approval for patients aged 12 and older rested on pharmacokinetic, dosimetry and safety data from nine paediatric patients plus extrapolation of adult efficacy, with no paediatric randomized efficacy trial performed. These distinctions, the authors argue, must be preserved rather than collapsed into a generic story of approval.

Fibroblast activation protein, or FAP, exemplifies the opposite risk. FAP-targeted agents interrogate activated fibroblasts and stromal remodelling, producing uptake across diverse malignancies, but uptake is not synonymous with a cancer-cell-specific target and can occur in inflammation, fibrosis and wound repair. A retrospective description of gallium-68 FAPI uptake in 80 patients across 28 cancer types found substantial variation between and within tumour categories and was not designed to establish sensitivity, specificity or patient benefit. Therapeutic studies remain early: a first-in-human report of lutetium-177 FAP-2286 covered 11 patients with advanced adenocarcinomas, and a dose-escalation study of lutetium-177 EB-FAPI in radioiodine-refractory thyroid cancer included 12 patients without a randomized comparator. No FDA or EMA authorization for a FAP-targeted radiopharmaceutical was identified in the official pages searched as of 9 August 2026, a deliberately narrower statement than claiming it is unapproved worldwide.

The review closes with a warning about the theranostic loop. Patient-specific dosimetry is a measurement framework, not proof of benefit: an EANM position paper explicitly noted that the superiority of dosimetry-based prescription had not been established. The recently proposed somatostatin receptor PET response framework was developed by expert consensus and had not been validated against survival outcomes at publication, so it should be treated as a harmonization tool awaiting prospective validation, not a proven surrogate. The authors’ prescription is concrete: paired chemistry comparisons with multiple sampling times instead of isolated high-uptake examples; early human studies reporting administered mass, metabolites and organ kinetics; diagnostic trials testing lesion-level accuracy and management impact; and therapeutic development progressing through reproducible manufacturing and dosimetry to controlled, platform-specific trials. Each study, they argue, should address the largest unresolved translational risk with a comparator and endpoint matched to the claim, so that molecular innovation supports clinically relevant exposure and reproducible production without overstating maturity.

Subject of Research: An evidence-gated framework for translating targeted oncologic PET radiopharmaceuticals from molecular design to clinical use

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

Article References: Xiong, Q., He, W., Fu, D., & Su, Y. (2026). An evidence-gated strategy for translating targeted oncologic PET radiopharmaceuticals from molecular design to clinical use. Frontiers in Nuclear Medicine, 6, 1976156. https://doi.org/10.3389/fnume.2026.1976156

Image Credits: AI Generated

DOI: 10.3389/fnume.2026.1976156

Keywords: PET radiopharmaceuticals, theranostics, PSMA, somatostatin receptor, FAP, radiopharmaceutical manufacturing, molar activity, dosimetry, radionuclide therapy, molecular design, regulatory authorization, radiochemistry