Reporting, identification and qualification thresholds by dose

Three thresholds, three different obligations

A reporting threshold is the level above which a result has to appear in the application, numerically and per batch. An identification threshold is the level above which the structure has to be established. A qualification threshold is the level above which the safety of that level has to be supported by data. They are three separate obligations that happen to be printed next to each other, and the most common error in a batch review is to treat the first number in the row as the limit for all three.

All of these thresholds scale with dose because the concern is intake, not concentration. Ten micrograms a day from a 10 mg tablet and ten micrograms a day from a 1 g tablet are the same exposure in very different percentages, which is why a table keyed only to percentage would be wrong at both ends.

Drug substance: one table with two limbs

The substance thresholds sit in a single table with a step at 2 g per day.

The phrase whichever is lower turns each cell into two rules, and only one of them can bind at any given dose. Both convert into the same currency: at a percentage p and a dose D in grams, the intake in milligrams is p multiplied by D and by ten. So a 1.0 mg cap corresponds to a percentage of 0.1 divided by D.

Two crossovers fall out of that arithmetic. The qualification cap takes over above about 667 mg per day, because 1.0 mg is 0.15 per cent of 667 mg. The identification cap takes over above 1.0 g, because 1.0 mg is 0.10 per cent of 1.0 g. Between 667 mg and 1.0 g the qualification limit is a mass while the identification limit is still a percentage, so the two numbers on the same row are in different units and cannot be compared until one is converted.

The step at 2 g is not smooth. Reporting falls from 0.05 to 0.03 per cent, which is the point where the analytical requirement moves: the quantitation limit of the method must be no more than the reporting threshold, so a 3 g per day product needs a method that quantifies 0.03 per cent of a concentrated sample solution. That is a method development task, not a reporting formality.

Drug product: three tables banded three different ways

The degradation product thresholds are split across three tables, and the bands do not share boundaries. Reporting changes at 1 g. Identification changes at 1 mg, 10 mg and 2 g. Qualification changes at 10 mg, 100 mg and 2 g. A single dose therefore sits in three different bands at once.

Here the qualification limit is usually a mass, and it is usually the tighter of the two. Work it out at three doses that a laboratory actually meets.

At 5 mg per day the identification threshold is 0.4 per cent, because 20 micrograms is the binding limb, and the qualification threshold is a full 1.0 per cent. Between those two figures a peak must be identified and does not need safety data. At 50 mg, the figures are 0.2 and 0.4 per cent, so an impurity at 0.35 per cent is identified and left alone. At 1.5 g the order inverts: identification is 0.133 per cent while qualification is 0.2 per cent, so any peak that has to be identified is already below the qualification limit and that limit never decides anything. A reviewer who assumes qualification is always the stricter number will misread the last row in particular.

The comparison rule decides the borderline case

Thresholds are exceeded on a strict greater-than basis, and the comparison is made after rounding. When the threshold is a percentage, the reported result is rounded to the same number of decimal places as the threshold and compared directly. When the threshold is a mass, the result is converted to a total daily intake, rounded to the same number of decimal places, and compared.

The published example is worth memorising. At a 0.8 g dose the qualification threshold is 1.0 mg per day, which is 0.125 per cent. A result of 0.12 per cent converts to 0.96 mg, rounds to 1.0 mg, and does not exceed the threshold. The same result at a 0.5 g dose, where the qualification threshold is 0.15 per cent, is above it. Nothing about the chromatogram changed; only the dose did.

Reporting conventions compound this. Below 1.0 per cent results are reported to two decimal places and at or above 1.0 per cent to one, so an early result of 0.0963 per cent becomes 0.10 per cent on the form. If the laboratory has not fixed that convention before the batch record is written, the same numerical result can be argued either side of the threshold.

What each threshold forces you to buy or build

The reporting row carries an obligation that is easy to miss and expensive to fix late. If the quantitation limit sits above the reporting threshold, the method cannot support the threshold at all, and no amount of extra batch data repairs it. The lower range limit is treated as its own validation subject for exactly this reason.

The identification row is where a relative retention time is often offered instead of a standard. That substitution works for tracking a peak between runs and does not work for establishing identity, and the difference becomes visible the moment a second method gives the peak a different relative retention. For the qualification row, the same standard is used differently: not to locate a peak but to prepare samples with a known added amount, which is what turns a safety argument into a demonstration that the method measures what the argument is about.

Where a standard for the impurity does not exist, the practical sequence is isolation from an enriched mother liquor, structure elucidation, then assignment of a response factor. The impurity profile comparison work that follows depends on that response factor, and an assumed value of 1.0 is only defensible when the chromophore is close to the parent.

Where the tables stop

Three exclusions decide more cases than the table itself. DNA reactive impurities do not follow a percentage at all; they are controlled against a threshold of toxicological concern of 1.5 micrograms per day, which is orders of magnitude below the percentage thresholds and is the subject of the M7 route from classification to control strategy. Enantiomeric impurities are outside the substance and product guidelines, and are handled through chiral methods and their own acceptance criteria. Residual solvents and elemental impurities have their own limits, keyed to permitted daily exposure rather than to percentage of the drug.

Two further points keep the table honest. A degradation product seen only at the recommended storage conditions drives identification, while one produced only under forced degradation belongs to the specificity argument for the method. And the percentage is a percentage of the substance, not of the peak area, so the impurity standard used for the calculation has to carry an assigned value with a stated uncertainty, not merely a certificate that says the material is the right molecule.

A review checklist

• Confirm the dose used is the maximum recommended daily dose, not the strength on the label.
• Write down which limb binds for each of the three thresholds at that dose, in per cent and in mass.
• Check that the substance table was used for process impurities and the product table for degradation products.
• Confirm the quantitation limit is at or below the reporting threshold, at the sample concentration actually injected.
• List every peak above the identification threshold and state how each structure was assigned.
• For every peak above the qualification threshold, state the basis: study, literature or metabolic argument.
• Record the rounding convention used, and re-check any result within one decimal place of a threshold.
• Confirm that no DNA reactive structure, enantiomer, solvent or elemental impurity was routed through these tables.