The generator checks, the thin layer chromatography that takes fifteen minutes, and what a failed kit looks like on the image. How a hot lab turns radiopharmaceutical quality control into a routine instead of an investigation after a bad scan.
A bone scan comes back with a stomach, a thyroid and a set of salivary glands on it, and the department spends an afternoon deciding whether the camera has a problem. It does not. The kit did not label, the patient was injected with a mixture of bone agent and free pertechnetate, and nobody would have known before the injection because nobody tested it.
Radiopharmaceutical quality control is fifteen minutes of work per preparation, done at the bench with strips, solvent and a counter. It is the cheapest quality assurance in the department and the one most often left out.
A cold kit is a sterile vial holding a freeze dried ligand and a reducing agent, usually a stannous salt, under an inert headspace. Adding technetium pertechnetate eluate reduces the technetium so it can bind to the ligand. Everything that goes wrong is a failure of that chemistry, and the failures come in three recognisable shapes.
The bench habits that prevent most of it are small. Use the eluate while it is fresh. Respect the maximum activity and maximum volume printed on the insert rather than the volume that happens to be in the syringe. Do not vent the vial to open air. Swirl, do not shake. Start the clock at the moment of addition and write it on the vial label, not on a scrap of paper.
Before the kit is blamed, the eluate has to be cleared. Two tests do that, and both are done on the first elution.
Molybdenum 99 breakthrough. Molybdenum leaking from the column follows the technetium into the vial, and because it has a much longer half-life the ratio gets worse through the day rather than better. The test uses a lead assay canister that absorbs the 140 keV technetium photons while passing the far more energetic molybdenum emissions, so the calibrator reads the molybdenum alone. The result is expressed as molybdenum activity per unit of technetium activity at the time of administration, and the acceptance limit is set by the pharmacopoeia and repeated on the generator insert. Take the number from the current insert rather than from memory, and record the elution time as well as the result, because the limit applies at injection and not at elution.
Aluminium ion. Alumina from the column can break through as aluminium ion, and it does two things: it interferes with labelling, and in a sulphur colloid preparation it causes flocculation which shows as lung uptake on a liver study. The test is a colorimetric spot comparison against a standard supplied with the generator, done on indicator paper in under a minute. The pass limit is on the insert.
Add a visual check to both. The eluate should be clear and colourless. Anything cloudy, coloured or containing particles stops the morning, whatever the tests say.
Radiochemical purity is measured by thin layer chromatography. A drop of the preparation is spotted on a strip of silica gel impregnated glass fibre at a marked origin, the strip is stood in a small depth of solvent in a closed chamber, and the solvent front carries the different chemical species different distances up the strip. Cutting the strip and counting the pieces gives the fraction of the activity in each form.
One strip is not enough for a full answer, because a single solvent cannot separate all three species. Two systems run in parallel do it.
Four details decide whether the numbers are worth anything. Keep the spot small, so the origin is a point rather than a smear. Do not let the solvent level touch the origin when the strip is stood in the chamber. Develop in a closed vessel so the atmosphere is saturated. And keep the strips dry, which in Dhaka is not a trivial instruction: silica gel takes up moisture from humid air, and a damp strip gives a diffuse front and a result that is quietly wrong. Dry the strips in an oven and store them in a desiccator with fresh silica, and treat the desiccant as a consumable that gets replaced rather than as furniture.
Counting is done in a well counter. Departments without one sometimes count the strip pieces in the dose calibrator, which is geometry dependent and not what the instrument is for. It can be defensible as a consistent relative measurement, but only if the method is written down, the same every time, and its limitation is stated in the procedure rather than discovered by an inspector.
Reporting staff should be able to recognise a labelling failure from the pictures, because the images are the last line of defence when the quality control was skipped.
Two further habits are worth building. When a labelling failure is suspected from an image, keep the vial, and test it: an assumption is not a record. And be suspicious of a department that has never recorded a failed preparation. It is far more likely to be a department that is not testing than a department with perfect chemistry.
Quality control that happens when somebody remembers is quality control that happens after the complaint. Turning it into a routine takes five decisions, all of them cheap.
The records join the rest of the department's quality assurance file, and an inspector reads them together with the camera results. How the two fit into one programme is set out in building a gamma camera quality assurance programme, and the layout that makes the bench work is in designing and equipping a radiopharmacy hot lab.