Larazotide: A Peptide That Reached Phase 3 in Coeliac Disease, and the Name It Shares With an Approved Drug

What it is

Larazotide — larazotide acetate, also written AT-1001 or INN-202 — is a synthetic octapeptide: eight amino acids, taken by mouth, designed to block the zonulin receptor and so tighten the junctions between the cells lining the intestine.

Two things make it unusual in this library.

It is not injected and not meant to be absorbed. Almost every compound here is a powder reconstituted for injection whose purpose is to reach the bloodstream. Larazotide's target is the inside surface of the gut, and its intended action is local. That is why trial regimens give it several times a day rather than weekly.

And it went further through clinical development than almost anything else here — a full phase 3 programme — which means its record contains something most entries lack: a real, well-powered test.

The trial record, read from the registry

The phase 3 ran from 29 May 2019 to a primary completion date of 21 July 2022, randomized against a matching placebo, with the primary outcome recorded as the proportion of subjects who are binary responders. The registry gives one reason for its termination — that the sponsor terminated it — and no result. This entry reports that and stops there; a registry that does not explain a decision is not an invitation to infer one.

What the record shows, then, is a compound that was taken seriously enough to fund four phase 2 trials and a 307-patient phase 3, and whose coeliac programme has no published positive endpoint.

The one positive randomized result is in twelve children

In July 2025, Science Translational Medicine published a phase 2a randomized, double-blind, placebo-controlled trial from Massachusetts General Hospital in 12 children with post-COVID multisystem inflammatory syndrome, median age 5.7 years, enrolled during hospitalisation. They received larazotide four times daily for three weeks as an add-on to standard care, with safety follow-up to 24 weeks.

Reported:

• No larazotide-related adverse events.
• Blood spike-protein concentration correlated with interferon-γ (P = 0.004), interleukin-6 (P < 0.0001) and gastrointestinal symptom score (P = 0.003).
• Treated children showed faster resolution of gastrointestinal symptoms, faster clearance of spike antigen, and a faster return to usual activities.

The mechanistic logic is why the trial existed: if viral protein is crossing from a gut reservoir into the blood through loosened tight junctions, a drug that closes those junctions should reduce it. The result is consistent with that, in twelve children, and the authors' own language is conditional — that treatment may be safe and may improve resolution.

It is also worth naming who ran it: the senior author, Alessio Fasano, is the researcher whose group described zonulin in the first place. That is not a criticism — it is the field's founder testing the field's own hypothesis — but a reader is entitled to know that the theory and the trial share an author.

The name collision, which is the practical warning

AT-1001 is two drugs.

It is larazotide's code name. It is also the code Amicus Therapeutics used for migalastat hydrochloride — an oral small molecule for Fabry disease, approved and marketed as Galafold.

Query a trial registry for AT-1001 and both programmes come back interleaved: larazotide's coeliac studies sitting beside migalastat's phase 3 trials, its long-term extensions and its paediatric studies. Counted naively, "AT-1001" looks like a compound with a dozen phase 3 trials and an approved product on the market.

It is not. The approved product belongs to a different molecule, in a different class, for an unrelated genetic disease. The two share nothing but an internal string.

This is the same failure mode as several other entries in this library, where a search matched a word rather than a subject. It is worth generalising: a code name is not an identifier, and any evidence count assembled from one should be read compound by compound.

Where it stands

A mechanistically specific oral peptide, with a mature safety record across roughly a thousand trial participants, no approval anywhere, a terminated phase 3 in the disease it was built for, and one small positive randomized result in a disease nobody anticipated when the programme started. Research continues — 13 of its records are from 2024 onwards, including barrier-protection work in animals and new formulation approaches.

Related: what a peptide is, and — for the opposite case, a compound whose name collision runs in the other direction — the Melanotan I entry, where an approved product exists and the market sells something else under the same name.

Sources

• Yonker LM, Kane AS, Swank Z, et al. Viral spike antigen clearance and augmented recovery in children with post-COVID multisystem inflammatory syndrome treated with larazotide. Sci Transl Med 2025;17(809):eadu4284. PMID 40737433.
• ClinicalTrials.gov, queried 2026-09-15: NCT03569007 (phase 3, 307, terminated), NCT01396213, NCT00492960, NCT00620451, NCT00362856, NCT00386490, NCT00386165, NCT05747534, NCT05022303 — and, under the same AT-1001 code, Amicus Therapeutics' migalastat records including NCT01218659, NCT02194985 and NCT04020055.
• Caliskan AR, Gul M, Yılmaz I, et al. Effects of larazotide acetate, a tight junction regulator, on the liver and intestinal damage in acute liver failure in rats. Hum Exp Toxicol 2021;40(12_suppl):S693–S701. PMID 34791921 — describes larazotide as a synthetic octapeptide reducing tight-junction permeability by blocking zonulin receptors.
• PubMed counts, run 2026-09-15 via the NCBI E-utilities esearch endpoint: larazotide[tiab] — 52; larazotide[tiab] OR "AT-1001"[tiab] OR "INN-202"[tiab] — 98; larazotide[tiab] AND "randomized controlled trial"[pt] — 4; larazotide[tiab] AND (celiac[tiab] OR coeliac[tiab]) — 37; larazotide[tiab] AND 2024:2026[dp] — 13; zonulin[tiab] — 1,133.