How this index is built

What gets included, what does not, and what an empty field means. Read this before relying on anything here.

The one rule everything else follows

This is why the index looks incomplete. It is incomplete, and visibly so on purpose. A blank cell means the literature did not say, not that the value is zero or unimportant.

What counts as an agent

An entry is created for a compound that is administered, or intended to be administered, and that has a name of its own.

Included

• Diagnostic and therapeutic radiopharmaceuticals, whether approved, in trials or preclinical
• Radiometal starting materials such as gallium chloride, flagged as such — they are not tracers, but the choice between generator and cyclotron production is a real one
• Development codes that are genuine compound identifiers and appear in more than one publication

Not included as agents

• Prosthetic groups and synthons — fluoroethyl tosylate, SFB, F-Py-TFP. They are steps in a synthesis, not what goes into a patient. They are kept in the data, marked as synthons, and excluded from the tracer lists
• Reagents, solvents, phase-transfer catalysts and intermediates
• Metabolites of an existing agent, such as FDG-6-phosphate
• Radiochemical methodology reported without a named product
• Compounds identified in a paper only by a table number, with no name of their own. The synthesis is still recorded against the publication, but no entry is created for a compound called "17"

Where a name looks like a code, the test is whether it identifies the compound outside its own paper. A code used across several publications earns an entry; a label that exists only inside one table does not.

Same compound, different isotope

These are separate entries. The gallium-68 and lutetium-177 forms of the same ligand differ in what they are for, how they are made and how they are released, so they are not merged. Where they form a theranostic pair, they are linked to each other instead.

Members of a compound family are also kept apart: PSMA-11, PSMA-617, PSMA-1007 and PSMA-I&T are four agents, not four spellings. Where two records turned out to be the same substance under different names, they were merged and the discarded spellings kept in the alternative names field, so a search for either still finds it.

One row per synthesis, not per paper

A paper describing six tracers on one platform produces six rows; a paper comparing one tracer across three modules produces three. They share a publication reference. This is why the number of syntheses is larger than the number of papers, and why a count of syntheses is not a count of studies.

Why yields must not be compared across rows

Reported yields depend on starting activity, scale, whether the figure is corrected for decay, and the practice of the group reporting it. Two numbers in the same column can differ twofold for reasons that have nothing to do with the equipment. Where a paper does not state whether the yield is decay-corrected, that is recorded as unknown rather than assumed.

The honest comparison between platforms is how many independent groups published a working synthesis on each, not whose number is highest.

Quality control entries

Where a pharmacopoeial monograph exists, it is cited by number, title and edition. Monograph texts are copyrighted and are not reproduced here, and no numerical specification is taken from them. Where no monograph exists, the applicable tests are derived from the general monograph on radiopharmaceutical preparations together with what the published synthesis says about solvents, catalysts and purification. Those derived entries are marked as such, and they are a starting point for scoping equipment — not a release specification.

Regulatory status

Approvals are checked against the regulator's own register, not against what a paper says. A centralised European approval and a national marketing authorisation are recorded separately, because an agent can be absent from one and routine in the other. Absence of an approval is recorded as absence of an approval, not as absence of the drug.

Indications

The indication vocabulary is ours. It is not a clinical classification and does not map onto any standard. It exists so that a disease area can be used as a way in, and it will be wrong at the edges. Tags record what has been reported in the literature, never what should be used in a patient.

Vendor and supplier profiles

Every module and precursor supplier page has two halves. The literature half is compiled the same way as everything else on this site and nobody edits it by hand. The manufacturer half is empty until the manufacturer claims that profile and fills it in themselves — it is marked with who submitted it and when, and it is never blended into the literature half.

Coverage, stated plainly

Coverage is uneven and will stay that way. A tracer with forty publications gets forty rows; a research compound with one paper gets one; many entries have none yet. Abstracts alone yield roughly a third of the fields that full texts do, so agents whose literature sits behind paywalls will stay thinner regardless of effort.

Where it is likely to be wrong

Automated extraction misreads things. Compound names get truncated, a number occasionally lands in the wrong column, and a supplier named only in a figure legend is missed. Records are checked mechanically for broken references, impossible values and duplicate substances, but mechanical checks do not catch a plausible wrong number.

If you know a tracer well and something here is wrong about it, that is the single most useful thing you can tell us.