How to read a certificate of analysis

Certificates of analysis turn up constantly in this field — attached to regulatory filings, quoted in enforcement letters, reproduced in litigation exhibits, and posted online by people who assume the document settles an argument. It usually doesn't. A COA is a summary of specific tests run on a specific material at a specific time, and it is silent on everything it did not test.

This piece is about reading the document. It is not a guide to obtaining material, commissioning testing, or handling anything; we do not publish that and will not help with it. What follows applies equally to a COA in a court exhibit and a COA in a scientific supplement.

What a purity figure is a figure about

The number most people look at first is a purity percentage from high-performance liquid chromatography — typically reversed-phase HPLC for a peptide. The method separates components of a mixture by how strongly they interact with a stationary phase, and a detector records what comes off the column and when. The purity figure is almost always an area percentage: the area under the main peak divided by the total area of all detected peaks.

That framing carries three limits worth naming. First, it is relative, not absolute. It tells you the proportion of what the detector saw, not the proportion of what was in the vial. Second, detection is usually by UV absorbance at a wavelength chosen for the target; a contaminant that absorbs weakly at that wavelength contributes little area and shrinks toward invisibility. Third, anything that does not elute from the column — material that stays bound, or aggregates that never make it on — is not in the denominator at all.

None of that makes the number worthless. It makes it a number about chromatographic behaviour under one set of conditions. Method conditions, column, gradient, and detection wavelength should be stated on the certificate. If they aren't, the purity figure is uninterpretable, whatever it says.

Identity is a different question from purity

Mass spectrometry usually appears next, and it answers a different question: is this the molecule the label claims? An observed mass matching the theoretical monoisotopic or average mass of the intended sequence is good evidence for identity. Tandem MS, which fragments the molecule and reads the pieces, goes further and can support the sequence itself.

But a mass match is a statement about mass. Isomers — same formula, different arrangement — can share a mass. Stereochemistry, including racemisation at individual residues, is largely invisible to a simple mass measurement. And a correct identity says nothing whatsoever about whether the material is sterile, whether it contains bacterial endotoxin, or what the remaining percentage consists of. Identity, purity, and safety are three separate axes. A certificate can be excellent on one and silent on the others.

Lot-specific, or merely representative

This is the distinction that does the most work and gets the least attention. A lot-specific certificate reports tests run on the identified batch: it carries a lot or batch number, a manufacture or test date, and results traceable to that material. A representative or typical certificate reports results from some earlier batch, or a composite, and is offered as an illustration of what the product generally looks like.

The two documents can be laid out identically. The words that distinguish them — "representative," "typical," "for illustration" — are usually in small type near a header, and a certificate with no lot number and no date should be read as representative by default. A representative COA tells you about a batch that is not the batch in front of you, which for most purposes means it tells you nothing about the batch in front of you.

Accreditation, sterility, and what is simply absent

"Third-party tested" is a marketing phrase, not a technical one. The technical claim is accreditation: a laboratory assessed by a recognised accreditation body against ISO/IEC 17025 for competence in named methods within a defined scope. That scope matters. A laboratory accredited for chromatographic assay is not thereby accredited for microbiological testing. Accreditation also certifies competence and quality systems — it does not certify that any particular result is correct, and it says nothing about chain of custody or whether the sample tested was drawn from the batch in question.

Sterility and endotoxin are separate again, and separate from each other. A sterility test asks whether viable organisms can be recovered under specified culture conditions. An endotoxin test asks whether lipopolysaccharide fragments — which are not alive and survive processes that kill the bacteria that produced them — are present above a limit. A material can pass sterility and fail endotoxin. Both are microbiological tests, run under their own pharmacopeial chapters, and neither is implied by a chromatographic purity figure.

The most useful habit is to read a certificate for its absences. What was not tested? What was tested against a specification so wide it could not fail? Which lot does this actually describe? A document that answers those questions well is doing its job. A document that looks authoritative while leaving them open is doing something else — and no reader, however careful, can close that gap by reading harder.