Case Study 19.1 — Compounded Semaglutide, Read as a Quality Story

What this case is for. Chapter 12 owns the access-and-cost dimension of the compounded semaglutide episode: who could get the drug, what it cost, and why a shortage in a medicine that works produces an enormous unmet demand. Chapter 6 owns the hype mechanism: how a drug becomes a cultural event and what that does to claims. This case study owns something narrower and more technical, and it is the part that gets least attention.

Read as a quality story, the episode is the clearest available demonstration of what a legal supply chain does and does not guarantee. Nobody in this story has to be a villain for the lesson to hold. The interesting failures happened without anyone lying.

The situation

Semaglutide is a peptide with an unusually strong evidence base — large randomized trials, hard outcomes, regulatory approvals across multiple indications. On the two-axis picture from §19.10, the molecule sits firmly in the strong-evidence column. That is precisely what makes the episode instructive: the quality problems that follow have nothing to do with whether the drug works.

Demand outran supply. Semaglutide entered official shortage. Under United States law, restrictions that ordinarily prevent compounders from producing what is essentially a copy of a commercially available drug relax when that drug is on the shortage list. A very large compounding market appeared within months.

That market was not homogeneous. At one end were 503B outsourcing facilities operating under current good manufacturing practice, registered with and inspected by the FDA, with documented sourcing for their active ingredient. At the other were arrangements in which the provenance of the active ingredient could not be determined at all. In between sat a very large number of prescribers, telehealth platforms, wellness clinics, and 503A pharmacies, some careful and some not, all operating under the same word: compounded.

Then the shortage resolved, and the permission that had opened the market narrowed again — with a great deal of product in circulation and a great many patients partway through a course of treatment.

Failure one: the salt form

Regulators raised concerns about compounded products containing semaglutide sodium and semaglutide acetate rather than semaglutide base.

This is the kind of distinction that sounds like pedantry to a patient and is not. A salt form is a different chemical entity. It has a different molecular weight, potentially different solubility, different stability behavior, and a different amount of active moiety per unit mass. The clinical trial program that established semaglutide's efficacy and characterized its safety was conducted with the base. Nothing automatically carries that evidence across to a salt. Demonstrating equivalence is a real analytical and regulatory task, and it had not been done.

Notice the structure of this failure. The compounder may have believed in good faith that they had semaglutide. The supplier may have honestly described what they sold. A certificate may have accurately identified the substance as semaglutide sodium — accurately, and in a way that most people in the chain were not equipped to interpret as a difference that mattered.

This is §19.3's identity failure in its most respectable form: not a different peptide, not a counterfeit, but a substance that is nearly the labeled thing and is not covered by the labeled thing's evidence.

Failure two: the delivery device

The approved product is supplied in a pre-filled pen that delivers fixed increments. The compounded version is typically supplied as a vial and a syringe.

That substitution looks administrative. It is not. It silently transfers the measurement task from an engineered device — designed, tested, and manufactured specifically to make a particular error impossible — to a patient at home working from written instructions. Regulators reported receiving adverse event reports involving dosing errors, including errors of large magnitude in which people received substantially more than intended.

Nothing about the molecule changed. Nobody in this failure lied about anything. The failure mode was introduced by the format, and it was introduced at the exact point where the pharmaceutical system had spent years engineering the error out.

This generalizes far beyond semaglutide, and it is the single most transferable lesson in the case: when a product is reformatted, the evidence supporting the original does not automatically reformat with it. Trial data describes a substance delivered in a particular way. Change the way and you have changed part of what was studied.

What the episode does and does not show

It does not show that compounding is illegitimate. Compounding exists for real clinical reasons and serves patients who cannot use a commercial product. Many of the facilities involved were operating lawfully and competently. Some patients who could not otherwise have obtained treatment obtained it.

It does not show that semaglutide is dangerous. The molecule's evidence base is not implicated by any of this, which is the point.

It does show that "compounded" is a description of a legal and manufacturing pathway, not a quality grade and not an efficacy judgment. Neither 503A nor 503B products are FDA-approved. The regulatory attention in compounding runs to how the product is made — facility, process, contamination control — and not to whether it works or is safe for the purpose intended.

And it shows how quality failures arrive in respectable clothing. No wrong molecule. No dishonest seller. A salt instead of a base, and a syringe instead of a pen. Both are the sort of thing that a person evaluating a purchase by its ingredient list would never see.

Discussion questions

1. Semaglutide has excellent trial evidence, and this case describes quality failures in products containing it. Explain, using §19.10's two axes, why those two statements are compatible — and why someone who treats "the evidence is strong" as an answer to "is this vial fine" has made a category error.

2. The salt-form problem could be described as a labeling technicality or as a substantive identity failure. Argue for the second reading. What exactly has to be demonstrated before evidence generated with a base transfers to a salt, and who would have to do it?

3. The delivery-device substitution introduced a failure mode without changing the molecule. Identify two other places in medicine or engineering where changing the format of something known to work introduces risk that the original evidence does not cover. What do your examples have in common with this one?

4. In this episode a great many participants acted lawfully and in good faith, and harms occurred anyway. What does that suggest about strategies for evaluating safety that rely on judging the intentions or professionalism of the parties involved?

5. When the shortage resolved, the permission that had opened the compounding market narrowed again, with patients partway through treatment. Set out the competing considerations a regulator faces at that moment. Which of them are quality considerations and which are access considerations, and what happens when the two point in opposite directions?

6. Suppose a friend tells you they are getting a compounded GLP-1 through a clinic. Using §19.5 and this case, write down the four questions you would want them to be able to answer — framed as things to understand about their own treatment, not as a purchasing checklist. Then say honestly what you would do if they could not answer any of them.