Follistatin 344: Every Registered Trial Tested a Gene, Not the Protein

Follistatin 344 has a human trial record. It's worth being precise about what that record actually tested, because it isn't what the name on a vial suggests.

Five trials, and what each one actually delivers

ClinicalTrials.gov lists five interventional studies that use follistatin itself as the intervention. Each of the five delivers a gene, carried in a viral vector or a plasmid, rather than an injected protein. A viral vector is a disabled virus used only as a delivery vehicle; a plasmid is a small loop of DNA that can carry a gene without one.

Not one study in that census administers follistatin as an injected protein.

Once the gene is delivered, the recipient's own muscle does the rest: it builds the protein, adds its sugar chains and releases it, locally and over months. Injecting the protein produces a different exposure profile entirely. That's why findings from the gene trials can't simply be carried over to an injected protein.

That distinction is the whole point of this piece. A gene therapy product and an injectable protein are different things, delivered differently and behaving differently in the body, even when they share a name.

One name, two stages of the same protein

First, some identity housekeeping, because 344 and 315 aren't two competing products.

Follistatin is encoded by the human FST gene. It is a secreted glycoprotein, a protein carrying sugar chains, rather than a peptide. Because it isn't a small molecule, no CAS number or molecular formula applies, and PubChem's compound database holds no record for it. Its reference identifier is an accession in UniProt, a protein database, instead.

In 1988, Shimasaki and colleagues sequenced follistatin clones from human testis tissue and found two precursor lengths: 344 amino acids and 317. Both include a 29-residue signal peptide, a short leading sequence a cell uses to route a protein and then cuts off. Once it's removed, the 344 precursor yields the mature 315-residue protein, FS-315, and the 317 precursor yields FS-288.

So FS-344 and FS-315 are one gene product at two stages of processing, not two competing versions a researcher chooses between. In human serum, FS-315 was the dominant form measured. In follicular fluid, the fluid surrounding an egg in the ovary, it was undetectable.

The 344 in the product's name and the 344 sequence the trials delivered point to the same gene product. What differs is the form. The trials delivered the instructions for making the protein, not the protein.

The protein, injected directly

The record on injecting follistatin itself is preclinical.

One preclinical study characterized native FS-315 directly as a candidate for injection. Its authors concluded that the protein's own pharmacokinetic and pharmacodynamic properties are poorly suited to injection aimed at broad effects throughout the body. Pharmacokinetics describes how a substance moves through and leaves the body; pharmacodynamics describes what it does there.

The same study built an engineered version. It fused the protein to part of a mouse antibody and removed one of its binding properties. That gave a terminal half-life roughly 100 times longer, meaning the engineered version lingered far longer in the slow, final phase of clearance. Only that engineered version produced an effect in mouse models of muscle wasting; the native protein did not.

A separate study in rats gave follistatin straight into the bloodstream. It left circulation in two distinct phases and accumulated in the liver. The initial half-life was 4.0 minutes, and the terminal half-life was 130.8 minutes.

So the one study that assessed the native protein as an injectable called it poorly suited to the job.

The closest anyone has come, and what it took

If you're looking for the nearest real example of an injected follistatin protein given to a person, there is one, and it still isn't actually follistatin.

ACE-083 is a follistatin-based Fc fusion: follistatin joined to an antibody fragment to change how it behaves in the body. It is not follistatin itself. It was first tested as an injection directly into muscle, in a randomized, placebo-controlled phase 1 trial in 58 healthy postmenopausal women. It later went into two randomized phase 2 trials in neuromuscular conditions. Every one of those trials tested the engineered fusion.

That's worth sitting with. The nearest anyone has gotten to injecting this protein into a person required re-engineering it into something else first.

Why the mismatch matters for reading the evidence

None of this says whether follistatin gene therapy works, or whether an injectable protein version would. This piece doesn't touch that question.

What it does say is narrower, and more useful for reading anything else written about this compound. Every registered human trial under follistatin's own name tested gene delivery, and the nearest protein relative to reach people is an engineered fusion. So if a source describes human results for follistatin without saying which form was used, that's the first thing to check.

One more identity trap is worth knowing. FSTL1 and FSTL3 are separate genes, and much of the recent literature a follistatin search returns concerns those two instead.

The trials and the vial share a name because the gene encodes the protein. They don't share a delivery route, and the registered human trials under that name tested only the gene.

Our full research write-up covers what the registered trials and the preclinical record do and don't establish.