Cleaning validation residue limits: where the number comes from

A residue limit belongs to a product pair, not to a substance

Equipment shared between products carries a limit that neither product owns on its own. It is set by the compound left behind, by the batch that will be made next, and by the surfaces the two share. Change the recipient batch size, add a third product to the train, or revise the toxicology of the donor, and the number moves while the cleaning procedure stays exactly as it was. Programmes that treat the limit as a fixed property of the residue end up revalidating against a figure that no longer describes the plant.

The arithmetic runs in two stages. The first gives a maximum allowable carryover, MACO: a mass of donor residue, in milligrams, permitted across the whole recipient batch. The second divides that mass by the shared product-contact area, producing a surface limit in micrograms per square centimetre, which is the only form an analyst can actually test.

MACO = (daily limit for the donor x minimum batch size of the recipient) / maximum daily dose of the recipient

A daily limit in milligrams divided by a daily dose in milligrams leaves a dimensionless fraction, and the fraction multiplies a batch mass. The formula carries no safety judgement of its own; all of the judgement sits in the choice of the donor limit.

Three routes, and the lowest one governs

Most programmes compute every route they can and adopt the smallest result. That is not conservatism for its own sake: the routes fail in different directions, so the smallest figure is the only one that survives an argument about the other two.

The visual criterion deserves its own paragraph because it is the one most often misused. It is a necessary condition and never a sufficient one. Detection thresholds for residues on stainless steel sit in the low micrograms per square centimetre, and they shift with the residue and the lighting. Where the calculated limit falls below what an eye can resolve, a shiny surface proves that the cleaning step did something, not that it did enough.

The health-based route, worked through

The health-based route borrows its arithmetic from the residual solvent methodology of ICH Q3C, which derives a permitted daily exposure from a no-observed-effect level:

PDE = (NOEL x body weight) / (F1 x F2 x F3 x F4 x F5)

The body weight term is 50 kg by convention, deliberately below the 60 and 70 kg figures used elsewhere so that the result errs low. The five factors cover extrapolation between species, variability between individuals, the duration of the study, the severity of the effect, and whether only a lowest-observed-effect level was available.

A chronic rat study with a no-effect level of 5 mg/kg/day, no severe findings and a proper no-effect level gives 5 multiplied by 50, divided by 5 multiplied by 10, which is 5 mg/day. That figure becomes the donor side of the MACO equation. Run the same equation on an elemental impurity and the answer arrives in micrograms, which is the subject of the elemental impurity piece; the logic is identical.

Two failures recur at this step. The first is a PDE quoted without its report: a figure on a spreadsheet with no assessment behind it cannot be defended once the input data are questioned. The second is a compound with no threshold at all. A mutagenic impurity carrying a structural alert has no no-effect level to divide, so its acceptable intake is derived another way and lands in the microgram-per-day region; there the health-based route governs by two or three orders of magnitude. The reasoning is set out in the ICH M7 walkthrough.

From a mass in the batch to a number on the swab

Take one donor and one recipient on a shared train: minimum daily dose of the donor 1.0 mg, PDE of the donor 0.15 mg/day, maximum daily dose of the recipient 400 mg, minimum batch of the recipient 50 kg, shared product-contact area 25,000 square centimetres.

The spread matters more than any single value. Five, twenty and seven hundred and fifty micrograms per square centimetre differ by more than two orders of magnitude. A site that computes only the health-based route, finds 750 and validates to it has validated nothing for a low-dose drug. A site that computes only 10 ppm has no answer when the auditor asks about the sensitising potential of the donor. A site that computes all three and takes the smallest can show the working.

The same exercise proves the opposite point as well: for an ordinary oral product with solid dose data, the health-based route is often the least restrictive of the three. Adopting it everywhere without the arithmetic is as indefensible as ignoring it.

The quantitation limit has to fit under the limit you chose

A swab limit describes a surface, so the method has to be able to see what the limit permits. Staying with 5 micrograms per square centimetre and a swab covering 25 square centimetres, the mass on the swab at the limit is 125 micrograms. Extracted into 10 mL, that is 12.5 micrograms per millilitre.

Recovery changes the figure the instrument has to produce, and it changes it in the direction people get wrong. At 80 per cent recovery, a sample truly carrying the limit reads 10 micrograms per millilitre, not 12.5. The calculation divides by recovery, so the reported surface residue returns to the right value, but the method now has to quantify at 10 and preferably at a third of that, near 3.3 micrograms per millilitre, to leave any margin. A method whose lower limit of quantification sits at 15 micrograms per millilitre cannot demonstrate compliance with a 5 microgram per square centimetre limit, however clean the equipment is; it can report only that the result fell below quantification, which is a different statement.

Recovery also has to be established on the right material. A coupon study on polished stainless steel says nothing about a PTFE gasket, a sight glass or a roughened weld. The figure belongs to a combination of surface, residue and swab, and at low levels it is concentration dependent in a way a single high-level spike will conceal.

What the reference standard contributes to the result

Every microgram per square centimetre reported is a mass obtained from an area through a calibration curve, and the curve rests on one number: the assigned content of a standard. Three properties of that standard move the answer.

• Content and its basis. A standard assigned at 92.3 per cent as-is, weighed as though it were 100 per cent, makes the calibrator 7.7 per cent weak, and every residue then reads about 8 per cent high. That is conservative, and it is still enough to turn a compliant swab into an investigation. The basis matters as much as the value: a content stated on the dried basis and applied to an undried standard shifts the error the other way. Reading the certificate properly is the cheapest control in the whole method.
• Form. If the residue is a salt and the standard is the free base, the mass ratio belongs in the calculation exactly as it does in assay work, and the conversion is worked through in the salt and hydrate piece. Salt factors of ten to fifteen per cent are common, and larger than most of the differences under discussion.
• Traceability and validity. A working standard prepared in-house needs a link to a characterised material and a documented potency assignment. An expired standard needs no interpretation: it invalidates the run, and hygroscopic material weighed straight from the fridge gains water while the analyst is still taring it.

A fourth case is worth naming. Where the parent compound is not available as a standard, the method targets a marker. Choose one shown to be present in the residue at a known and stable ratio to the parent, and state the ratio in the method.

Where these programmes fail

• A dose fraction applied to an intermediate that is never administered, where the input dose does not exist.
• The limit recalculated after a batch size change, with no revalidation and no change control record.
• All three routes computed, then the highest adopted because it was the easiest to pass.
• Recovery measured once, on one material, at one concentration, several years ago.
• Cleaning agent residue left out of the calculation, although the detergent has its own acceptance criterion.

Checklist before the next campaign

• Donor limit: dose-based and health-based both calculated, reports on file, governing route named.
• Recipient inputs taken from current registered documents, not from memory.
• Shared area measured or drawn from the equipment file, and updated when a line changes.
• Surface limit printed with its unit, since 5 micrograms and 5 milligrams per square centimetre both read as a bare 5.
• Method: quantitation limit at or below a third of the limit after recovery correction.
• Standard: content, basis, form, storage and retest date checked before anything is weighed.