Two numbers follow every batch of radiolabeled peptide out of the hot cell, and they are routinely read as if one implied the other. Molar activity says how much radioactivity you have per mole of peptide. Radiochemical purity says what fraction of the radioactivity in the vial is the compound you meant to make. They have different denominators, neither converts into the other, and a preparation can be excellent on one and poor on the other. Most of the trouble in reading a radiolabeling method section starts there.

The vocabulary is the first failure mode

The field has said this about itself. An international working group on nomenclature in radiopharmaceutical chemistry was established in 2015 in response to what its authors described as an increased incidence of incorrect usage of established scientific terms and conventions, and even the emergence of self-invented terms; its consensus guidelines followed in 2017 [1]. A later comment reconsidered one term in particular: radiochemical conversion should define reaction efficiency, measured as the transformation of components in a crude reaction mixture at a given time, whereas radiochemical yield is better suited to defining the efficiency of an entire reaction process including separation, isolation, filtration and formulation [2].

That distinction matters at the bench because the two quantities are measured on different material. Conversion is what an analytical injection of the crude mixture tells you. Yield is what survived the cartridge, the filter and the transfer into the final vial. Neither is radiochemical purity, which is a property of the finished formulation and gets measured again later, and again at expiry. A method reporting "95%" without saying which of the three it means has told you very little.

Matching the chelator to the radiometal

A bifunctional chelator has two ends. One is a coordination cage sized, charged and shaped for a particular metal ion. The other is a handle — an activated ester, an isothiocyanate, a click partner — that attaches to the peptide. The cage is chosen against the ion, not against the peptide, and that choice propagates directly into what the peptide has to survive during labeling.

The cleanest demonstration is a same-peptide comparison. Persson and colleagues prepared DOTA and NODAGA conjugates of the same uPAR-binding sequence, AE105-NH2, and labeled both with gallium-68: the DOTA conjugate at 95 °C, the NODAGA conjugate at room temperature [3]. Same peptide, same nuclide, roughly seventy degrees of difference in thermal stress; the macrocycle is the only thing that differs between the two conjugates, though the authors report the conditions each was labeled under rather than testing either chelator at the other's temperature. The same work reports that fractionated elution of the generator gave the highest radiochemical yield and purity, and that acetone had a negative effect on labeling efficiency for the NODAGA conjugate [3] — a solvent left over from one processing route interfering with the next step.

Denticity is the other axis, and it is set by the ion. H4picoopa was designed as a high-denticity ligand for actinium-225 and characterized alongside lanthanide and indium complexes; the reported pM values were 17.6 for La3+, 21.3 for Lu3+ and 31.2 for In3+, and the ligand gave quantitative radiochemical conversions with both actinium-225 and indium-111 under mild conditions — room temperature, ten minutes — with more than 97% radiochemical purity retained over six days in serum [4]. Worth reading what the same paper reports next. A bifunctional analog conjugated to a melanocortin-targeting peptide and labeled with indium-111 showed good radiotracer stability in mice bearing B16-F10 tumors, but the authors state that improved tumor targeting could not be achieved for imaging purposes [4]. Holding the metal is a necessary condition for a usable tracer, not a sufficient one.

The competition you cannot see

Radiometal arrives in vanishingly small amounts. Almost everything else in the vial that binds a trivalent cation is present in far larger amounts, and some of it arrives with the radionuclide. Cusnir and colleagues quantified trace metals in eluate from an Eckert & Ziegler 68Ge/68Ga generator by ICP-MS and found Al, Fe, natGa, Pb, Ti and natZn at significantly higher concentrations than in the starting eluent solution: Fe and natGa in the range 0.01–0.1 µM, Al, Zn and Pb in the range 0.1–1 µM, and Ti in the range 0.9–1.5 µM [5].

What those metals then did was selective, and the scope of the finding is narrow enough to be worth stating precisely. Working with a tris(3-hydroxy-1,6-dimethylpyridin-4-one) chelator, the authors found that Al3+, Fe3+, natGa3+ and Ti4+ reduced radiochemical yield at concentrations equimolar with the chelator and higher but not at lower concentrations, while Pb2+, Zn2+, Ni2+ and Cr3+ had no effect on yield even when each was present at a hundredfold molar excess over the chelator [5]. That is a statement about one chelator. It is not a general ranking of which contaminating metals matter — a different cage with different donor preferences will be poisoned by a different subset. What does carry across is the shape of the problem: the interference threshold is a ratio of competing ion to chelator, so cutting the amount of chelator in order to raise molar activity moves you toward it. The authors draw that conclusion for their own system, noting that increasing specific activities by using still lower chelator concentrations will likely require purifying the gallium-68 from trace metal ions first [5].

Molar activity, and what sits in its denominator

Molar activity is activity divided by moles of compound, and the moles include every copy of the peptide whether or not it is carrying a metal. That is a convention rather than a result, but everything downstream depends on it. Because labeling reactions are usually run with a large excess of chelator-conjugate over radiometal, most of the peptide in a finished vial has no metal in it at all. Zeng and colleagues put the point plainly in describing a microreactor that achieved over 90% radiometal incorporation at 1:1 stoichiometry of bifunctional-chelate biomolecule conjugate to radiometal, which they note potentially obviates the extensive chromatographic purification typically required to remove the large excess of unlabeled biomolecule in radioligands prepared by conventional methods [6].

Two examples show what the number looks like and how little it travels. A gastrin-releasing peptide receptor antagonist conjugated to the chelator AAZTA5 through a piperidine spacer, labeled within five minutes at room temperature, gave apparent molar activities of 50–60 GBq/µmol for the gallium-68 product and 10–20 GBq/µmol for the indium-111 and lutetium-177 products [7]. Separately, a small-scale 44Ti/44Sc generator of 370 kBq, evaluated by radiolabeling, gave an apparent molar activity of 2 MBq/nmol with DOTA [8]. The adjective is doing work in both: "apparent" signals that the mole figure is the precursor charged into the reaction rather than an independently measured total. That is convention in this literature, not something either measurement establishes.

Now set those against routine production. Gallium-68 labeling of a PSMA inhibitor bearing the HBED-CC chelator, optimized around cation-exchange and ethanol post-processing, is reported at specific activities of 326 ± 20 MBq/nmol obtained reproducibly for clinical-routine patient batches, across more than 200 applications, alongside radiochemical purity greater than 98% [9]. That count is of applications, not of independent preparations. One GBq per micromole is one MBq per nanomole, so the units are directly comparable. Taking the AAZTA5 gallium-68 range as the denominator, the PSMA-HBED figure is about five and a half to six and a half times higher. That is not a ranking of chelators. The two preparations differ in starting activity, in whether the generator eluate was post-processed and in precursor amount, and neither paper states the reference time its activity is decay-corrected to — and molar activity is sensitive to all four. It describes a preparation under one site's conditions, not a molecule.

Radiochemical purity is a different quantity

Radiochemical purity is the fraction of total radioactivity present in the stated chemical form. Its denominator is radioactivity; molar activity's denominator is moles of substance. Several consequences follow, and they are worth writing out because they are so often assumed the other way round. None of them is an experimental finding; each follows from the definitions.

  • Adding unlabeled peptide to a finished vial lowers molar activity and leaves radiochemical purity untouched. The added material carries no radioactivity, so it cannot appear in a radiodetector trace at all.
  • Removing unincorporated radiometal raises radiochemical purity. It does not remove unlabeled peptide, so it does not raise molar activity, and whatever activity is lost in the step can lower it.
  • Whether labeled and unlabeled peptide can be separated at all depends on whether metallation shifts retention enough to resolve them. Where it does not, no amount of purification raises molar activity.
  • A radiodetector is blind to anything without a nuclide in it. Chemical purity and precursor identity are separate release tests on separate instruments.

Which is why a batch described as 99% radiochemically pure tells you nothing whatever about how much peptide mass is in the vial, and a high molar activity tells you nothing about whether the activity is where you think it is.

Radiolysis: the batch degrades while you watch

A therapeutic radionuclide irradiates its own carrier. Radiolysis of water generates reactive species, those attack the peptide, and radiochemical purity falls with time although nothing was added and nothing removed. Where this has been dissected, the governing variable appears to be absorbed dose rather than elapsed time as such. In a study of [177Lu]Lu-PSMA-617 that evaluated dependence on time, radionuclide activity, buffer agent concentration, precursor amount and preparation volume, absorbed dose was the dominant factor influencing radiochemical purity, which the authors report as inversely proportional to absorbed dose with an exponential dependence [10]. The same work demonstrates an influence of sodium acetate buffer concentration on radiolytic degradation and states it should be considered both when developing a dosage form and when comparing the results of independent studies [10]. That caveat is the practically useful part: two labs quoting different stability windows for the same conjugate may simply be buffering differently.

Stabilizers are the standard countermeasure, and they are not free. For [177Lu]Lu-PSMA-I&T formulated in saline, stability was reduced by radiolysis, with radiochemical purity falling below 95% within 24 hours; a formulation containing 20 mg/mL ascorbic acid in saline prevented radiolysis and ensured stability over 30 hours [11]. The same paper reports that labeling PSMA-I&T with lutetium-177 needs a considerably longer time than labeling PSMA-617 [11], a reminder that the precursor, not only the nuclide, sets the conditions.

The cost side shows up in sequences with oxidizable residues. Studying a 177Lu-labeled DOTA-minigastrin analog containing methionine, with an optimized HPLC system that distinguished 22 species over time, Trindade and Balter report that gentisic acid promoted selective oxidation of the methionine residue, and that with gentisic acid as the only stabilizer present, across pH 3.5 to 6.0, raising the pH decreased peptide impurities while increasing that selective oxidation; they conclude that adding methionine becomes necessary to decrease the selective oxidation in this methionine-containing peptide [12]. A quencher that protects the molecule broadly can attack one residue specifically. Which residues a sequence contains therefore constrains which stabilizer is appropriate, and a formulation validated on one conjugate does not transfer to another by default.

How much of this survives into routine production is now measurable at scale. A multifactorial analysis of 355 clinical batches — 101 [177Lu]Lu-DOTA-TATE, 169 [177Lu]Lu-PSMA-617 and 85 [177Lu]Lu-PSMA-I&T — reported yields of at least 98% for all three regardless of starting materials, with DOTA-TATE and PSMA-617 maintaining radiochemical purity above 90% through 24 hours while PSMA-I&T, under that center's standardized protocol, was stable for 8 hours and fell below specifications at 24 hours [13]. Negative correlations between bulk activity or concentration and radiochemical purity appeared across all preparations, and the authors report that carrier-added formulations exhibited greater radiostability than non-carrier-added ones at elevated concentrations, with notably distinct quality profiles between lutetium-177 suppliers [13]. The direction is the one radiolysis predicts: more activity in less volume, less purity.

The QC a labeled batch receives

Release testing is layered because no single method sees everything.

  • Radiochemical purity by two orthogonal systems. Radio-HPLC resolves the labeled peptide from radiolysis products and other labeled species; radio-TLC or ITLC catches free metal and colloid, which may never come off an HPLC column at all. Running systems with different retention behavior is convention, precisely so that a species invisible to one is caught by the other. Eppard and colleagues identified four mobile phases with suitable separation properties and complementary Rf values on silica gel 60 plates, validated the results against HPLC, and found two systems equivalent on ITLC silica gel plates, with the ITLC analysis finished within 5 minutes against 20 minutes for the TLC system [9].
  • Radionuclidic purity. A separate measurement, usually by gamma spectrometry, asking which nuclides are present rather than which chemical forms. It is orthogonal to everything above: a chemically perfect preparation of the wrong nuclide passes every chromatographic test.
  • Chemical purity and precursor identity. UV or mass spectrometric detection on the same separation, because the radiodetector cannot see the unlabeled majority of the peptide.
  • Formulation checks. Appearance, pH, residual solvent, sterility and bacterial endotoxin. These are routine practice rather than findings, and their limits come from pharmacopoeial monographs rather than from any single paper.
  • Stability to the stated expiry. The number that matters is radiochemical purity at the end of the assigned window, not at end of synthesis — exactly the distinction the 8-hour and 24-hour results above draw [11, 13].

What these numbers do not settle

Quite a lot remains convention rather than evidence. There is no agreed reference time at which molar activity must be quoted, so two figures for the same conjugate may differ mostly by decay correction. There is no standard set of chromatographic systems for a given conjugate, so an impurity resolved in one lab can co-elute in another. Stabilizer selection is still largely empirical: the field has a long list of candidates — one systematic study evaluated DMSA, cysteine, gentisic acid, vanillin, methionine, adenine, dobesilic acid, thymine, uracil, nicotinamide, meglumine and mannitol against radiolysis of a lutetium-177 preparation [10] — but no predictive rule that maps a sequence to the right quencher, and at least one documented case where a quencher damages a specific residue [12].

The most reliable habit is also the least glamorous: when a paper or a certificate quotes a number, ask which quantity it is, what the denominator was, what time it refers to, and on which system it was measured. In this field, more than most, the four answers are not implied by the number.