Area percent purity (revision 35)

Area percent purity is the integrated area of the principal peak in a chromatogram expressed as a percentage of the total integrated area of all peaks. It is the figure almost always meant when a research-peptide certificate reports "purity", and it is reported without further qualification often enough that its limitations are worth stating plainly.^[1]

The measure has one defining property from which everything else follows: it is relative to what was detected, not to what was weighed. A species that does not elute, does not absorb at the detection wavelength, or co-elutes with the principal peak contributes nothing to the denominator and therefore cannot reduce the figure. Water, counterion and inorganic residue are invisible to it in the ordinary case, which is why a vial can honestly carry a certificate reading 99.2% and contain substantially less than 99.2% peptide by mass.^[1]

Area percent is nevertheless the right measure for the question it answers, which is how much of the material that eluted was the intended species relative to everything else that eluted. Read as a purity ratio it is informative; read as a mass fraction it is simply a different quantity, and that substitution is the commonest analytical error in this field.

A sample is dissolved, injected onto a reverse-phase column, and eluted under a gradient of increasing organic modifier. A detector — most often ultraviolet absorbance at 214 nm, where the peptide bond absorbs — produces a signal against time. Peaks are integrated, and the principal peak's area is divided by the sum of all integrated areas.

Four choices in that description change the answer, and none of them is usually stated on a certificate:

• The gradient. A shallow gradient resolves species that a steep one runs together. Two laboratories can obtain two honest and different figures for one vial by choosing differently here.
• The detection wavelength. At 214 nm the peptide bond dominates and most peptide-related impurities are seen. At 280 nm only aromatic residues respond, and a related substance without one is invisible.
• The integration rule. Valley-to-valley and drop-line integration assign the area under an unresolved shoulder differently, and the difference lands entirely on the principal peak.
• The integration threshold. Peaks below the threshold are not counted at all, so raising it raises the reported purity without any change to the material.^[1]

Four categories of material are systematically under-represented or absent from an area percent determination. Each of them has been the subject of a documented quality failure in this sector, which is why they are enumerated rather than left implicit.^[2]

Non-absorbing species

Salts, most sugars and many small-molecule excipients give little or no response at 214 nm. Charged-aerosol or refractive-index detection sees them; ultraviolet detection does not.

Non-eluting material

Anything that does not come off the column within the run — strongly retained aggregates, particulate — contributes to neither numerator nor denominator.

Co-eluting species

A related substance that shares the principal peak's retention time is counted as principal peak. This is what a peak-purity assessment by diode array or by mass spectrometry is for.

Everything that was never dissolved

The determination is performed on the injected solution. Material that did not dissolve is not in it, so an appearance observation on reconstitution is a separate finding from a purity figure and has to be reported separately.

The last point generalises: a purity assay measures what went into the injector. Any statement about the vial requires a separate determination of how much of the vial the injected material represented.

A certificate reports 99.0% area percent for a 5 mg vial of a peptide. Suppose, as is ordinary for a lyophilised trifluoroacetate salt, that the vial also holds 5% water by mass and that trifluoroacetate accounts for 12% of the mass of the remaining solid.

The peptide mass is then 5 mg × 0.95 × 0.88 = 4.18 mg, or 83.6% of the nominal fill. The purity figure of 99.0% is unaffected by either the water or the counterion, because neither is integrated in the chromatogram. Both figures are correct and they describe different things.

Reading 99.0% as "4.95 mg of peptide" overstates the content by roughly 18%. This arithmetic is the whole reason Peptide content exists as a separate determination, and it is why a purity figure and a content figure appearing together on one certificate — 99% and 83% — are consistent rather than contradictory.^[1]

Three conventions make area percent figures comparable, and their absence is the ordinary reason two figures for one lot differ.

First, report the method with the figure: column chemistry, gradient, flow rate, detection wavelength, injection volume. Second, itemise related substances individually rather than reporting a single aggregate impurity figure — an aggregate conceals whether one impurity is at 0.9% or nine are at 0.1%, which are different situations. Third, state the integration threshold, because it determines what was counted at all.

Where a purchaser is comparing an independently obtained figure with a certificate figure, a difference smaller than the repeatability of the method is not a finding. Repeatability for a well-behaved peptide on a validated method is typically a few tenths of a percentage point across replicate injections; differences of that size between two laboratories are expected.^[3][4] Repeatability, intermediate precision and reproducibility are separate parameters, and only the first is measured within one laboratory on one day. Larger differences have ordinary methodological explanations before they have any other kind, and the article on Third-party testing sets out how to distinguish them.

• Peptide content
• Certificate of analysis
• Karl Fischer titration
• Reference standard
• Third-party testing
• Purity claim inflation

• Project:No original research — why the worked example above is routine calculation and an estimated plasma concentration would not be.