What a certificate of analysis is, and what it is not
A certificate of analysis is a report on a specific batch. It lists a set of tests, the procedure used for each, an acceptance criterion, and the result obtained on a sample drawn from that batch. In the vocabulary the ICH Q6A guideline established, a specification is the combination of tests, procedures and acceptance criteria; the certificate is the record that one batch was measured against that specification and passed. It is a factual document about a sample at a point in time.
It is not a licence, a marketing authorisation, a safety evaluation, or a warranty that anything is suitable for use in a person. It carries no pharmacology, no toxicology, and no regulatory status. In the research-peptide market it is often the only technical artefact anyone produces, so it ends up shouldering the evidentiary weight of an approval it was never designed to carry.
The productive way to read one is line by line, asking four questions of each entry. What physical property was measured? By what method? On which sample? And what would have failed to show up if it were wrong? Answered honestly, that turns most certificates into a short list of things measured and a much longer list of things not measured. The absences are where the reading actually happens.
The header block: lot, dates, and who ran the tests
Before any analytical result comes an identification block: product name, stated sequence, batch or lot number, quantity, dates of manufacture and analysis, sometimes a retest date, and the issuing laboratory. This block is checkable against itself without any instrument at all. Does the stated sequence correspond to the named compound? Does the molecular formula match the sequence, and the theoretical mass match the formula? Internal contradictions here are informative, because they are errors a genuine analysis would not have produced.
Dates matter more than they look. Analysis long after manufacture describes a batch that has already aged, and a retest date is a stability claim only as good as the programme behind it, which a research certificate rarely has.
The issuing party is the most consequential line and the least standardised. A certificate from the manufacturer reports the manufacturer's own testing. One on independent laboratory letterhead means a different party ran the assay, which closes exactly one gap. It does not mean an independent party selected the sample, collected it, or observed what happened to the batch afterwards. Third-party testing, as the phrase is normally used, is independent analysis of a submitted specimen, not independent sampling.
Identity: what mass spectrometry confirms and what it cannot
The identity line on almost every peptide certificate is a mass spectrometry result: electrospray ionisation produces multiply charged ions that are deconvoluted to a single mass and compared against the theoretical mass calculated from the stated sequence. Close agreement is a useful screen, ruling out a wholly different compound, a grossly truncated chain, or an inert filler substituted for the peptide.
It does not confirm sequence. Intact mass is order-blind, so any rearrangement of the same residues gives an identical mass and a scrambled chain passes. Leucine and isoleucine are exactly isobaric and can never be separated by mass. Glutamine and lysine differ by roughly 0.036 daltons, which a high-resolution instrument resolves and a nominal-mass instrument does not. Deamidation shifts the mass by about one dalton and methionine oxidation by sixteen, so the tolerance window a certificate accepts decides which of these it can flag at all.
Establishing the actual order of residues takes tandem mass spectrometry with fragment ion assignment, or classical sequencing chemistry, and those results are uncommon on research certificates. The point sharpens for modified peptides: semaglutide carries an aminoisobutyric acid substitution and a fatty diacid chain attached through a linker, and a correct intact mass is compatible with that chain sitting at the wrong position. Mass answers what, at best, not where or in what order.
Purity: how to read an HPLC area percent
The purity figure is nearly always reversed-phase high-performance liquid chromatography with ultraviolet detection, usually near 214 nanometres, where the peptide bond itself absorbs. The instrument integrates every peak, and the reported purity is the main peak area divided by the total integrated area. The phrase for this is area percent, and the two words do a great deal of work.
Area percent assumes every impurity responds to the detector in proportion to the main peak, which is never exactly true. A contaminant with few chromophores is undercounted; one rich in tryptophan or tyrosine is overcounted. More importantly, the denominator contains only material that both absorbed and eluted. Inorganic salt, residual water and many solvents pass through effectively invisible or leave in the void volume, and a species that never comes off the column under the gradient used does not exist as far as the number is concerned.
So a certificate reading 99.4 percent means that 99.4 percent of the ultraviolet-absorbing material eluting under this gradient, on this column, at this wavelength, appeared in one peak. Without the chromatogram, column chemistry, gradient, wavelength and run time, the figure is not checkable by anyone, and a bare percentage with no trace attached is a claim rather than a result. The difference between a shallow gradient over forty minutes and a fast generic one over eight is the difference between resolving closely related impurities and averaging them into the main peak.
The impurities a single chromatographic run will not resolve
Solid-phase synthesis fails in characteristic ways, and its characteristic failures are the hardest impurities to see. A deletion sequence, missing one internal residue because a coupling step did not go to completion, differs from the target by a single amino acid and often elutes very close to it. A truncated sequence, where chain growth stopped early, may be further away or may not be. Whether either appears as a separate peak is a property of the method, not of the material.
Stereochemical impurities are worse. Racemisation during activation and coupling, most readily at cysteine and histidine, produces diastereomers with the same formula, the same mass, and retention behaviour that can be nearly indistinguishable on a standard reversed-phase column. Detecting them requires hydrolysing the peptide and analysing the liberated amino acids by a chiral method, a separate assay research certificates essentially never include. Aggregates are a third category, invisible under reversed-phase conditions that dissociate them and detectable instead by size exclusion chromatography.
The general principle is orthogonality. Two methods that separate on different physical grounds tell you far more than one method reported to an extra decimal place. A certificate showing a single reversed-phase run and a single mass result has established two things about the material, not an impurity profile.
Purity is not how much peptide is in the vial
Purity is a relative measure: what fraction of the peptide-like material present is the intended species. Net peptide content is absolute: what fraction of the powder in the container, by weight, is peptide at all. These answer different questions, and a certificate can report the first while saying nothing about the second.
Net peptide content comes from an orthogonal quantitative method, most commonly amino acid analysis, in which the peptide is hydrolysed and the released residues are quantified against standards. Elemental nitrogen determination and quantitative nuclear magnetic resonance serve the same purpose. All give a weight-per-weight figure that no chromatographic area percent can supply or substitute for.
The consequence is arithmetic rather than subtle. Lyophilised synthetic peptides, especially basic sequences purified as trifluoroacetate salts, commonly fall well short of being pure peptide by mass once counterion and water are counted. A container labelled with a nominal milligram quantity of material reported at high chromatographic purity can hold appreciably less peptide than the label suggests, and the purity figure will not have moved. Most research certificates omit net peptide content, so the number most relevant to how much active substance is present is the one missing.
Water and counterion: the mass that is not peptide
Residual water in a lyophilised cake is measured by Karl Fischer titration, the compendial method for water in solids. Freeze-dried peptides retain some moisture, and hygroscopic material takes up more once a container is opened. Water is invisible at 214 nanometres and contributes nothing to an area percent, but it contributes directly to the weight on the label.
Counterion is the larger term. Reversed-phase purification normally runs in trifluoroacetic acid, so the peptide leaves that process as a trifluoroacetate salt with counterions associated with each basic site. Trifluoroacetate has a formula weight near 114, so on a peptide with several basic residues the counterion contribution is not a rounding error. Quantifying it takes ion chromatography or fluorine nuclear magnetic resonance, and converting to another salt form is a deliberate extra step. A certificate reporting purity but not salt form leaves the composition of the powder partly undefined.
Salt form is not a technicality to regulators either. When compounded semaglutide products proliferated, the FDA stated plainly that semaglutide sodium and semaglutide acetate are different active ingredients from the semaglutide base its approvals cover. That is the regulatory expression of the same chemical point: a document naming a salt describes a different substance from one naming the free base, and the two are not interchangeable because the peptide chain matches.
Sterility, endotoxin and particulates: absent by design
The tests that determine whether something can be injected are almost never on a synthesis certificate, and that is not an oversight. Compendial sterility testing, the bacterial endotoxins test using limulus amebocyte lysate, and particulate limits for injections are attributes of a finished sterile product, set by the manufacturing environment, the filling process and the container closure. They are not properties of a synthesis, and no identity or purity result implies anything about them.
Endotoxin deserves separate emphasis because the intuition about it is usually wrong. It is a heat-stable bacterial cell wall component that survives autoclaving and passes through sterilising-grade filters. Material can therefore be sterile, in the sense of containing no viable organisms, and still be pyrogenic. It can equally be 99 percent pure by chromatography and carry a substantial endotoxin burden, because purity and endotoxin load are unrelated quantities measured by unrelated assays.
Lyophilisation compounds the misunderstanding: freeze-drying removes water, but it does not sterilise and does not destroy endotoxin. A certificate reporting identity, purity and water content has said nothing about microbiological quality, and the honest conclusion from that silence is that the question was not asked.
The question the document cannot answer: does it describe this vial?
Every result on a certificate applies to a sample that reached a laboratory. The step from there to any particular container in the world is not an analytical step, and no instrumentation inside the report can supply it. The only bridge is a batch number on the paper matching one on a label, and in the research market both are printed by the same party.
The document is also trivially forgeable. A certificate is a PDF; text in a PDF is editable, letterhead is copyable, and a genuine certificate for one good batch can be reissued alongside every later batch without anything on its face changing. Some laboratories publish verification portals precisely because the document cannot authenticate itself, and one naming no verifiable issuer offers no route to checking.
The counterfeit semaglutide episodes made the stakes concrete. The FDA seized counterfeit Ozempic that had entered the legitimate supply chain in late 2023, and the World Health Organization issued a medical product alert on falsified semaglutide identified across multiple regions in 2024. Some counterfeit units contained insulin rather than semaglutide. In each case the falsified product travelled with convincing packaging and paperwork, which is the point: documentation is the easiest part of a product to fake, because faking it requires no chemistry. A certificate is evidence about a sample, and its connection to a specific vial is an assertion the analytical chemistry on the page is silent about.
What a regulated release package looks like by comparison
For an approved product, batch release under good manufacturing practice generates a file that a single certificate merely summarises: identity by orthogonal methods, a quantitative assay, related substances measured against thresholds qualified during development, water content, sterility, endotoxin, particulates, container closure integrity, and stability data supporting the expiry. Behind that sit a quality system, validated methods, retained samples, and a regulator empowered to inspect the facility and the records. The certificate is meaningful because of the system it emerged from.
The FDA's guidance on abbreviated applications for certain highly purified synthetic peptides referencing recombinant listed drugs shows the characterisation regulators actually expect. The question it poses is not whether a single purity number clears a threshold, but whether the complete impurity profile of the synthetic material is comparable to that of the reference product, with impurities identified and their relevance argued. That is a materially different exercise from reporting one area percent.
For compounds with no approval anywhere, the comparison has no other side. BPC-157 holds no marketing authorisation in any major jurisdiction; the FDA placed it in the category of its 503A bulk drug substances evaluation reserved for substances raising significant safety concerns, and the World Anti-Doping Agency has named it on the Prohibited List since 2022. TB-500, a synthetic fragment marketed in reference to thymosin beta-4, is likewise unapproved and falls in the prohibited growth factor class in sport. For these there is no compendial monograph, no established specification and no qualified impurity threshold, so a certificate is written against acceptance criteria the issuer selected. That does not make the measurements false; it means passing them shows consistency with a self-chosen standard, and reading the page as equivalent to a regulated release document mistakes the form of the evidence for its weight.