Two different systems exist in United States law for approving a copy of an existing medicine, and which one applies to a GLP-1 turns on a sentence of federal regulation that counts amino acids. The regulation is blunt about it: a protein is “any alpha amino acid polymer with a specific, defined sequence that is greater than 40 amino acids in size.”[1] Above that line a product is a biological product and a copy of it is a biosimilar, licensed under the Public Health Service Act. Below it, a product is a drug and a copy is a generic, approved under the Food, Drug, and Cosmetic Act. The dates those copies actually arrived are in the generics timeline article.
Nothing in that sentence is about complexity, difficulty or how a molecule is manufactured. It is a length threshold, and the peptides used for weight management sit close underneath it.
Which side of the line each molecule falls on
Two of them state their own size on their own labels. Pramlintide is described as a synthetic 37-amino acid polypeptide.[2] Exenatide is described as a 39-amino acid peptide amide.[2] Those are one and three residues under the threshold. The semaglutide and tirzepatide labels give a molecular formula and a molecular weight instead of a residue count, so no count is asserted for them here — what their regulatory placement is can simply be read off the register: every approved semaglutide, tirzepatide, liraglutide, exenatide and pramlintide product appears in the Orange Book, which lists drugs.[3]
One GLP-1 receptor agonist crossed the line, and it shows what crossing it looks like. Dulaglutide is described on its label as a fusion protein of two identical disulfide-linked chains, each joining a GLP-1 analog sequence to the Fc portion of a modified human IgG4 heavy chain, produced in mammalian cells.[2] It is not in the Orange Book at all. It holds a biologics license, BLA 125469, and a copy of it would have to be a biosimilar.[4] Same receptor, same weekly injection, an entirely different legal machine.
Same molecule, different factory
The awkwardness is not theoretical, and the two approved liraglutide labels put it on the page. Victoza’s says the peptide precursor is produced by a process that includes expression of recombinant DNA in Saccharomyces cerevisiae.[2] Hikma’s generic liraglutide label — approved as an abbreviated application against that same product — says it contains chemically synthesized Liraglutide.[2]
Everything downstream matches. Both labels give the molecular formula as C172H265N43O51 and the molecular weight as 3,751.2 daltons, both list disodium phosphate dihydrate at 1.42 mg, propylene glycol at 14 mg and phenol at 5.5 mg per mL, and both state a pH of approximately 8.15.[2] One is grown in yeast and the other is built residue by residue in a reactor, and the register calls them therapeutically equivalent.
That is precisely the case the generic pathway was not originally built for, and FDA said so in writing. A 2017 Federal Register notice announced draft guidance on when a synthetic peptide referring to a previously approved peptide of recombinant origin should be submitted as an abbreviated application rather than as a new drug application, and it named five substances: glucagon, liraglutide, nesiritide, teriparatide and teduglutide.[5] The guidance was finalized in May 2021.[6]
Nine years on, the census
Three of those five now have approved copies and two do not. The Orange Book product file records eleven abbreviated applications for liraglutide, four for teriparatide and four for glucagon, against zero for nesiritide and zero for teduglutide.[3] A route being open is not the same as a route being used, and naming a molecule in a guidance does not put a copy of it on a shelf.
What the surviving three had to prove is also not what a tablet has to prove. Because the copy is chemically synthesized and the reference was grown, the impurity profile is the battleground rather than the active ingredient. FDA’s approach, as described in a 2025 case study on teriparatide, is that once sameness of the active ingredient is established, residual immunogenicity risk is addressed by characterizing differences in product- and process-related impurities, with limits set on the relative abundance of each impurity and a requirement that any new impurity above a concentration threshold be evaluated for its capacity to drive a T-cell response, using methods that assess both human leukocyte antigen binding and actual T-cell activation.[7]
The underlying worry is specific: a sequence modification introduced by synthesis can create a new T-cell epitope that the reference product does not contain, and of roughly 80 peptide drugs approved for human use, at least five are off-patent and being developed as generic alternatives.[8] Peptide impurities and aggregates arising in manufacture and storage carry consequences for both safety and efficacy that a small-molecule specification does not have to consider.[9] The first generic teriparatide was supported by a side-by-side structural and biological comparison using nuclear magnetic resonance, ion mobility-mass spectrometry, circular dichroism and cell-based bioassays.[10] That is a biosimilar’s evidence package wearing a generic’s label.
What the pathway buys at the pharmacy counter
The practical difference shows up at substitution. A generic carrying an A-rated therapeutic-equivalence code is, by the definition FDA works from, a product that has demonstrated bioequivalence and can be expected to have the same clinical effect and safety profile when administered to patients under the conditions specified in the labeling — which is the basis on which state agencies and pharmacists make a substitution.[11]
A biosimilar does not get that automatically. Under the statute, being highly similar with no clinically meaningful differences makes a product a biosimilar; being substitutable without the prescriber’s intervention requires the additional interchangeability finding, which carries its own switching standard: that the risk of alternating between the two is no greater than the risk of staying on the reference.[12] FDA has since proposed to revise the evidence it expects for that finding, on the reasoning that current analytical technology can characterize a purified therapeutic protein and model its functional effects with high specificity and sensitivity.[12] What a pharmacy is and is not allowed to swap is the subject of the substitution article.
So the generic route is the more useful one to land on, and a peptide copy landing there is genuinely better placed than a biosimilar. That is where the reasoning usually stops, and it should not.
The pathway is not what sets the price
The insulin market ran the experiment. Basaglar, the first long-acting insulin follow-on, took share quickly in a commercially insured population — and an analysis of pharmacy claims from its market entry through 2018 found no evidence that its users had lower out-of-pocket costs than users of the reference product. The authors recommended that cost savings to patients and insurers be evaluated empirically before automatic substitution is promoted.[13] Rapid uptake of a cheaper-in-principle product did not reach the person paying.
The generic side of the GLP-1 class makes the same point from the other direction. Exenatide has exactly one approved abbreviated application, and its national acquisition cost rose 8.2% in the nine months after it first appeared in the federal pricing file, leaving it 11.7% below a brand that is no longer marketed. Liraglutide, with seven labelers listed, fell 63.6% below its brand over two years.[14] Both are generics on the same pathway. The variable that moved was how many companies were selling, which is one reason a single seller’s quote tells a buyer so little — the pattern behind the price variation article.
Europe has taken the second route to the same awkwardness. Synthetic polypeptides referencing biological originator products are not eligible for the biosimilar framework there and must go through the generic or hybrid pathway instead, and the reviewers of that landscape argue for more transparency and global alignment as more complex synthetic products referencing recombinant originators arrive.[15] Two regulators have looked at the same molecules and routed them the same way, for the same reason, and neither is fully comfortable with it.
What none of this makes a compounded vial
A compounded preparation is on neither pathway. It is not a generic, because no abbreviated application was filed, no bioequivalence was demonstrated and no therapeutic-equivalence code was assigned. It is not a biosimilar, because no license was issued. Compounded drugs are not FDA-approved, and the FDA does not review them for safety, efficacy or quality before a pharmacy dispenses them — which means no agency has compared a compounded vial’s impurity profile with anything, by any of the methods described above. What is actually established about the contents is in the vial contents article, and what the phrase not FDA-approved carries in the approval-status article.
The narrow reader-facing conclusion is this. When an approved copy of a GLP-1 peptide does arrive, it will be a generic rather than a biosimilar, it will be substitutable at the counter, and the size of the discount will be decided by how many companies choose to make it rather than by which statute it was approved under. How the figures on this site are established before publication is described in the methodology.