What Makes a Good Radioligand Therapy Target? A 6-Criteria Framework

Radioligand therapy (RLT) has gone from niche to one of the fastest-moving modalities in oncology. Two franchises — ¹⁷⁷Lu-DOTATATE (Lutathera) for neuroendocrine tumors and ¹⁷⁷Lu-PSMA-617 (Pluvicto) for prostate cancer — proved the model, and the field is now racing to find the next validated target. As of mid-2025 there were more than 400 registered RLT trials exploring new molecular targets.

But most proteins make poor radioligand targets, and the reasons are specific to how RLT works: you are attaching a radioactive payload to a ligand and asking it to deliver a lethal dose to tumor cells while sparing everything else. That constraint rules out the majority of the proteome before biology even enters the conversation.

This is the framework we use at Nuclens to score every target in our catalog. It is six criteria. A strong target clears all six; a fatal weakness in any one is usually disqualifying, no matter how good the others look.

First-pass triage, not a verdict. This framework compresses weeks of database screening into a defensible shortlist. It does not replace wet-lab internalization assays, clinical dosimetry, or freedom-to-operate analysis. Treat it as the filter that decides what's worth validating — not the validation itself.

1. Cell-surface accessibility

The question: Can the ligand physically reach the target?

A radioligand circulates in the blood and extracellular space. It cannot cross an intact plasma membrane on its own. So the target has to present an epitope on the outside of the cell — a cell-surface receptor, a transmembrane protein's ectodomain, or a membrane-anchored enzyme. Intracellular targets (transcription factors, cytoplasmic kinases) are effectively unreachable for RLT, and purely secreted proteins drift away from the tumor rather than concentrating the dose in it.

This is why localization is the first gate, not a nice-to-have. PSMA (FOLH1), the field's canonical success, is a type II transmembrane glutamate carboxypeptidase with its catalytic domain facing the extracellular space — a textbook cell-surface handle.

What to look for: authoritative localization evidence. UniProt subcellular-location annotations with experimental evidence codes (ECO:0000269) are the standard. "Single-pass type I/II membrane protein" or "Cell membrane" with a defined topology is what you want to see; "Cytoplasm" or "Nucleus" is disqualifying.

Related target profiles: PSMA/FOLH1, SSTR2, DLL3.

2. Normal-tissue expression and dosimetry risk

The question: What healthy organs will also take up the dose?

RLT toxicity is on-target, off-tumor: wherever the target is expressed in normal tissue, the radioligand delivers radiation. A handful of organs dominate the risk because they are either radiosensitive or naturally accumulate radioligands:

• Kidney — the classic dose-limiting organ for peptide radioligands (renal clearance concentrates them in the proximal tubules).
• Bone marrow — radiosensitive; drives hematologic toxicity, and the primary constraint for potent alpha emitters like ²²⁵Ac.
• Salivary glands — the source of the xerostomia that limits ¹⁷⁷Lu-PSMA dosing.
• Liver — high perfusion and clearance role.

A target with high expression in any of these starts with a serious handicap. The ideal profile is low, near-background expression across all dose-limiting organs and expression concentrated in the tumor.

What to look for: normal-tissue RNA and, better, protein expression. We use Human Protein Atlas consensus nTPM with practical bands — under 5 nTPM is excellent, 5–20 is moderate, over 20 is concerning — for kidney, liver, marrow, and salivary gland.

The honest limit: expression predicts risk, not absorbed dose. Renal dose in particular is driven heavily by ligand pharmacokinetics and clearance route, which no expression database captures. Low kidney expression is a good sign; it is not a dosimetry result. We cover this in depth in Reading Normal-Tissue Expression for Radioligand Dosimetry Risk.

3. Tumor specificity and tumor-to-background ratio

The question: Is the target high enough in tumor, and in enough patients, to deliver a therapeutic dose?

Criterion 2 is about the denominator (normal tissue); this is about the numerator (tumor) and their ratio. Two things matter:

• Tumor-to-background ratio (T:B) — how much higher expression is in tumor than in the relevant normal tissue, especially the dose-limiting organs. Treat this as a continuous signal, not a pass/fail threshold: a target with a 2,000× tumor-to-kidney ratio is meaningfully better than one at 30×, even though both clear any reasonable bar.
• Tumor positivity — the fraction of patients (and of cells within a tumor) that actually express the target. A spectacular ratio in a target expressed by only 10% of patients is a narrow, high-risk opportunity. Broad positivity (>70%) means a more predictable response across a population.

A favorable ratio in a rarely-expressed target is a trap. Both numbers have to be good.

What to look for: tumor expression relative to normal (HPA pathology / cancer datasets), per-patient positivity, and ideally single-cell data confirming the target is on cancer cells rather than surrounding stroma or immune cells.

Related target profiles: FAP (stromal — a cautionary example), GRPR.

4. Internalization

The question: Does the target pull the radionuclide inside the cell — and does that matter for your isotope?

When a ligand binds and the receptor internalizes, it drags the radionuclide into the cell, where residualizing radiometals like ¹⁷⁷Lu and ²²⁵Ac get trapped and keep irradiating. This intracellular retention is a major driver of delivered tumor dose. PSMA's rapid internalization on ligand binding is part of why it works so well.

The nuance: internalization is not universally required. A radiolabeled antibody bound to a surface antigen can still deliver a cytotoxic dose from the membrane, particularly with longer-range beta or alpha emissions. But for small-molecule and peptide radioligands with residualizing metals, internalization is a strong positive and its absence is a real concern.

What to look for: direct evidence of receptor-mediated internalization on ligand binding. There is no clean structured database for this — it lives in the primary literature — so it is one of the harder criteria to assess at scale. (Nuclens extracts an internalization flag from PubMed and labels it AI-derived; verify before relying on it.)

Related target profiles: PSMA/FOLH1, Nectin-4.

5. Clinical maturity and precedent

The question: How much has someone already de-risked this target for you?

Precedent cuts development risk dramatically. A target with an approved radioligand, or one in active clinical trials, comes with validated ligand chemistry, known biodistribution, and clinical safety signals. A discovery-stage target with only association evidence may be a bigger opportunity — less competition — but carries far more unknowns.

This criterion isn't about preferring mature targets; it's about pricing the risk honestly. Approved > clinical-stage with several active trials > clinical-stage with one > discovery with strong genetic/association evidence > discovery with weak evidence.

What to look for: clinical stage and active-trial counts (ClinicalTrials.gov), approved analogs, and the strength of the underlying target-disease association (e.g., Open Targets association scores).

6. Competitive landscape

The question: Is this space open, or is everyone already in it?

The best target scientifically can still be the wrong bet commercially if a dozen programs are ahead of you. PSMA and SSTR2 are validated and crowded. The strategic value increasingly sits in targets that clear criteria 1–5 and remain relatively open — DLL3, FAP, GRPR, CXCR4, Nectin-4, STEAP2 and others are where much of the field's current energy is going.

This is a judgment call, not a pure data point: a crowded space can still be worth entering with a differentiated ligand or isotope, and an empty space can be empty for a good reason.

What to look for: number of active programs and trials against the target, recency of entrants, and whether existing efforts are diagnostic, therapeutic, or both.

See our full landscape: Beyond PSMA: Emerging Radioligand Therapy Targets in Solid Tumors.

Putting the six together

No single criterion selects a target; the interaction does. A target can have a perfect normal-tissue profile and still fail because it's intracellular. It can be beautifully cell-surface and clinically mature and still be a poor bet because a dozen programs got there first. Good target selection is finding the candidates that clear all six — and doing it before a faster competitor does.

Doing this by hand means pulling localization, four organs of expression, tumor ratios, internalization literature, trial counts, and competitive data — one database tab at a time, for every candidate. That's the weeks-long grind Nuclens was built to compress: describe your target profile in plain English, and get a ranked, fully-sourced shortlist scored on exactly these six dimensions.

Run a free analysis → — no credit card; every data point traces back to its public source.

Nuclens is a first-pass target-triage platform for radioligand therapy. It does not replace experimental validation, clinical dosimetry, or regulatory assessment. AI-extracted literature signals are flagged for verification.