Singapore health authority warns on falsified liraglutide pens
Reported from the analysis, not from a warning notice.
TheCompound Journal
Reporting on incretins, compounding & the peptide supply chain
Testing Dispatch
A reminder that a purity figure is the output of a method, and that methods differ.
Janoshik has published a purity determination of 98.5% on a 2 mg vial of liraglutide supplied by GL Biochem, against a figure of 94.4% on the certificate that accompanied the vial. The sample was submitted by the Journal in March 2026 as part of a continuing programme of arm’s-length purchasing, and the vendor was told before the order was placed that any result would be published.
The Journal purchased 8 vials from the same lot at retail and had four of them analysed separately, in order to distinguish vial-to-vial variation from method variation. Vial-to-vial agreement was within 0.4 percentage points. Method-to-method disagreement was an order of magnitude larger.
Purity became this trade’s single figure of merit because it is the cheapest measurement to obtain and the easiest to compare. Peptide content, which determines how much drug is actually in a vial, costs several times as much and is quoted as a priced addition by most of the twenty companies in the Journal’s dossier programme rather than carried as a standard release test. Sterility and endotoxin are separate disciplines again, and are neither tested nor claimed.
"Ask for the suitability summary before you argue about the percentage," said Dr Sirichai Boonmee, peptide chemist, Chulalongkorn University. "If the tailing factor and the replicate precision are outside limits, you are arguing about arithmetic performed on an unfit separation."GL Biochem’s dossier has been updated with this determination and with the company’s response to it. The dossier’s ordering box records the lead times, the minimum quantities and what a buyer should ask for at the point of order.
Reported from the analysis, not from a warning notice.
What each missing test would cost to add, and which of them a buyer should actually want.
Reported from the analysis, not from a warning notice.
The Journal submitted the sample and paid for the analysis. The vendor was told in advance.
Two years ago we ran an anonymised version of this comparison and promised a named one. This is it, with every method printed in full.
We asked twenty companies three questions about stability data and print every answer, refusal and non-response.