Independent analysis puts a TFC liraglutide lot at 97.8%, against 94.8% on the certificate

Two laboratories analysing vials from the same liraglutide lot have returned figures 0.8 percentage points apart. Janoshik reported 97.8%; Medutest reported a lower number on a shallower gradient with a tighter integration threshold. Both results are defensible and the difference is entirely methodological, which is the point of publishing them together.

The Journal purchased 1200 vials from the same lot and had 12 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.

Reported figures across the set were: highest 97.8%, lowest 94.8%, median close to the midpoint of the two. Every report in the set stated a detection wavelength, and 12 of them stated an integration threshold. The Journal has asked the laboratory to publish thresholds on all reports as a matter of routine.

Four independent services — Janoshik, Medutest, and the two others operating in this market — carry the analytical burden for a trade that has no release-testing obligation of its own. Each is paid by whoever submits the sample, which is usually the vendor. The Journal’s position is that a submitter-pays model is workable but that its incentive structure has to be stated whenever a result is quoted as verification.

"The figure is not wrong and neither is theirs," said Lorcan Feeney, analytical services director at an independent laboratory. "They are answers to two different questions and the trade has decided to pretend there is only one question. Print the gradient and the argument disappears."

TFC’s dossier has been updated with this determination and with the date of the request for comment. The dossier’s "what we could not verify" box records the questions still outstanding.