What testing actually finds
When regulators and independent laboratories analyse peptide vials and pens seized from outside the legitimate supply chain, three failure modes dominate. The product contains a different active ingredient than the label claims. It contains the right molecule at the wrong amount. Or it contains roughly the right molecule at roughly the right amount, but was never made, filled or stored under conditions that make injecting it safe.
The first two are chemistry problems, and chemistry can find them. The third is a manufacturing and handling problem, and no chemical assay can rule it out. That asymmetry is the most useful thing to understand here: the analyses that are cheap to run and easy to display on a certificate address the failures least likely to send someone to hospital, while the assays that address the dangerous failures are destructive, expensive and almost never performed on grey-market material.
The pattern repeats across compounds and countries. Falsified semaglutide pens have been confirmed by regulators in South America, Europe and North America. Compounded glucagon-like peptide-1 preparations have been flagged for dosing errors and for salt forms that are not the approved active ingredient. Unapproved research peptides such as BPC-157 circulate with certificates reporting a single purity number and nothing else.
The falsified semaglutide alerts, in order
In June 2024 the World Health Organization issued its first global medical product alert covering falsified semaglutide, after batches presented as Ozempic were confirmed in Brazil and the United Kingdom in late 2023 and in the United States that December. WHO advises removing falsified product from circulation because its contents are unverifiable, not merely suboptimal.
The United States case is worth studying because the counterfeit did not stay on the internet. In December 2023 the FDA and Novo Nordisk warned that counterfeit units bearing a copied lot number had entered the legitimate American supply chain and reached patients through retail pharmacies. The agency seized units and confirmed that the pen devices and needles were not authentic, meaning the amount actually injected could not be assumed to match the label even where some active ingredient was present.
European regulators dealt with a parallel problem the same year: falsified pens were traced back through wholesalers in Germany and Austria, and the UK medicines regulator seized falsified GLP-1 injection pens. The common thread is scarcity. Counterfeiting concentrates where demand outruns legitimate supply and price supports the effort, which is exactly what the incretin shortages of 2022 to 2024 created.
Wrong compound entirely, not a weaker version
The most instructive finding of the whole episode came from Austria in late 2023, where people were hospitalised after injecting pens labelled as Ozempic that had been obtained outside pharmacy channels. The clinical picture was severe hypoglycaemia, in some cases with seizures, and analysis pointed to insulin rather than semaglutide in the cartridges.
That substitution makes commercial sense and clinical nonsense. Insulin is cheap, widely available, already supplied in pen cartridges, and indistinguishable by eye as a clear solution. But semaglutide stimulates insulin release in a glucose-dependent way and rarely drives blood glucose to dangerous lows on its own, whereas injected insulin does exactly that in someone who is not diabetic and has not eaten accordingly.
This case breaks the intuition that a bad batch is a diluted version of the real product. When identity is wrong, the harm profile has nothing to do with the drug on the label, and nothing about the packaging, the solution or the injection signals the difference until the pharmacology arrives.
Compounded is not counterfeit, and neither is automatically clean
Three categories get flattened into one word, and they behave differently. There are FDA-approved semaglutide and tirzepatide products from the licence holders. There are compounded preparations from 503A pharmacies and 503B outsourcing facilities, legal as copies only while the approved products sat on the drug shortage list. And there are unapproved products sold as research chemicals, with no lawful route to human use at all.
Within the compounded category the FDA has raised two specific quality problems. The first is chemical identity: the agency has stated that semaglutide sodium and semaglutide acetate are salt forms that are not the same active moiety as the semaglutide in approved products, so preparations built on them are not compounded copies of an approved drug. The second is presentation. In July 2024 the FDA warned about dosing errors with compounded semaglutide supplied in multi-dose vials with syringes rather than metered pens, including reports of people injecting many times the intended amount and requiring hospital care.
That second problem is not a manufacturing defect at all. The molecule can be correct and the vial sterile, and the harm still arrives through a change in how the product is measured.
The legal basis for routine compounding of copies closed when the shortages resolved, tirzepatide in December 2024 and semaglutide in February 2025. Demand did not close with it, and the predictable consequence is migration toward suppliers that are not pharmacies.
Underdosing is the quieter failure
Counterfeits containing the wrong drug are dramatic and rare. Products containing less peptide than the label states are undramatic and common, and the reason is partly technical rather than fraudulent.
A purity figure from reversed-phase chromatography is an area percentage: what fraction of the material that eluted and absorbed light at the detection wavelength was the target peak. That is a statement about composition, not quantity. A vial can hold material that is genuinely 99 percent pure and still contain substantially less peptide than the number printed on it.
The reason is that lyophilised peptide is not only peptide. The dried cake includes water, the counterion left over from purification, and residual salts, and trifluoroacetate from preparative chromatography can account for a large share of the dry mass of a peptide carrying several basic residues. The measurement that captures all of this is net peptide content, determined by amino acid analysis or nitrogen determination alongside a Karl Fischer water determination. It appears on serious pharmaceutical documentation far more often than on grey-market certificates.
The practical consequence is a product that is simultaneously genuine, high purity and underdosed, with no false statement anywhere on the paperwork. Underdelivery also produces no acute signal. It presents as the compound not working, which is indistinguishable from the compound genuinely not working, and that is why anecdotal reports from unregulated supply carry almost no evidentiary weight.
What HPLC and mass spectrometry can and cannot resolve
Reversed-phase HPLC separates by hydrophobicity. A single sharp peak at the expected retention time is consistent with the labelled peptide and is good evidence that the preparation is not a crude mixture, but it is not proof of identity. Closely related species, including deletion and truncated sequences missing a residue, can co-elute with the target under a routine gradient, and anything without a peptide bond, such as residual solvents and inorganic salts, does not appear in the trace at all.
Mass spectrometry answers identity far better, confirming molecular mass to within a few daltons. Semaglutide sits near 4,100 daltons, tirzepatide near 4,800, and BPC-157, a fifteen-residue sequence, near 1,400. A matching mass is meaningful, but mass has blind spots that matter for peptides specifically. Leucine and isoleucine are isobaric, so a substitution between them is invisible. A D-amino acid substituted for its L counterpart has exactly the same mass while potentially changing receptor binding and protease resistance. Separating these requires tandem mass spectrometry sequencing or chiral analysis, a different and more expensive experiment.
The assay closest to the question people actually care about is a potency assay: a cell-based measurement of whether the material activates the intended receptor, and at what concentration. It measures function rather than mass, catching misfolding, racemisation and quantity error in one step. It is standard for approved biologics and effectively absent from grey-market documentation.
The tests nobody runs: sterility, endotoxin and particulates
Everything above is chemistry, and chemistry says nothing about whether a preparation is safe to put under the skin. The two tests that predict acute harm are the compendial sterility test and the bacterial endotoxin test, the latter usually run as a Limulus amebocyte lysate assay against a defined endotoxin limit.
Endotoxin deserves attention because it defeats the intuitive fix. It is a heat-stable fragment of Gram-negative bacterial cell wall, and it survives both sterile filtration of a contaminated bulk solution and lyophilisation. A vial can pass a sterility test, contain no living organisms, and still be pyrogenic enough to cause fever, rigors and an inflammatory response. Filtering does not undo contamination that already happened.
Sterility testing has limits that certificates rarely make clear. It is destructive and statistical: units are sampled and the lot is inferred, so the result never applies to the specific vial in someone's hand. It also expires the moment the material is handled again. Repackaging, relabelling, transport without cold chain and hand-filling outside a controlled environment all happen after the certificate was issued and are invisible to it.
Visual particulate inspection is the only check most people can perform, and it detects only gross failures: visible particles, cloudiness, discolouration, a cake that will not dissolve. It is worth doing and it resolves almost nothing.
Certificates of analysis, and the forgery problem
A certificate of analysis is a document, not a measurement. Its value depends entirely on who ran which method, on which lot, with instrument records that can be traced back. Detached from that chain it is a picture of a number.
The recurring failure patterns are consistent enough to list. The lot number does not match the vial, or is absent. The chromatogram is unlabelled, lacks axis scales, or is the same image used on other products. The testing laboratory is unnamed or cannot be shown to exist. Purity is quoted to a precision the stated method cannot support. Most commonly the certificate was issued by the seller rather than an independent laboratory, which makes it a marketing claim in the format of a lab report.
Outright forgery of certificates is documented in enforcement actions against suppliers of falsified medical products, and it is cheap to do. A certificate that cannot be tied to a named laboratory and a matching batch number is unverifiable rather than reassuring, and unverifiable evidence should not raise confidence.
Third-party testing changes this only under specific conditions: the buyer selects the laboratory, the sample chain is controlled, and the result covers the lot actually held. Testing arranged by the seller examines whatever the seller chose to send, a different experiment with a predictable answer.
Unapproved research peptides sit outside all of this
BPC-157 shows what happens when there is no approved reference product. It is a synthetic fragment related to a sequence found in gastric juice, with rodent evidence covering tendon and ligament healing, gastrointestinal injury and vascular repair, and no adequate controlled human trials establishing efficacy for any indication. It is not approved as a drug in the United States, the European Union or the United Kingdom.
In 2023 the FDA placed BPC-157 in the category of its bulk drug substances evaluation reserved for substances raising significant safety risks, which in practice removes it from lawful compounding for human use under the 503A pathway. The stated concerns included limited safety characterisation and uncertainty about the material actually being supplied.
The research use only label carried by most of this material is a statement about intended use, not a quality grade. It imposes no sterility requirement, no endotoxin limit, no cold chain and no good manufacturing practice obligation, and it is fully compatible with a product that is exactly what it claims to be and still unsuitable for injection. There is also no pharmacopoeial reference standard for these compounds, so even a well-run laboratory has no authoritative definition of a correct result.
What a clean-looking vial proves
It proves the packaging worked. Holograms, tamper-evident seals, batch codes and printed cartons are the cheapest components of a medicine to reproduce convincingly, and counterfeiters optimise them first because they are the only ones a buyer inspects. The counterfeit semaglutide that reached American retail pharmacies passed enough visual scrutiny to be dispensed by professionals who handled the genuine article daily.
A clear, colourless solution proves that the solid dissolved. A sterile solution and one carrying a clinically significant endotoxin load look identical, because endotoxin is a soluble molecular fragment rather than visible material. Bacterial growth dense enough to see is a late-stage failure, not the threshold of harm.
The signals that carry real information are documentary and institutional rather than physical: an approved product moving through an unbroken licensed supply chain, or a preparation from a named facility that is registered and subject to inspection. Everything else reaching a buyer, including appearance, price, review volume and a seller-supplied certificate, is downstream of what somebody chose to present. That is the honest summary of what testing finds, and the reason testing after the fact is a poor substitute for provenance before it.