What it is
Amlexanox is not a peptide. It is a small heterocyclic molecule, C16H14N2O4, with a molecular weight of 298.29 — a fused three-ring system, with no amino acids in it at all.
It was developed as an anti-allergic and anti-inflammatory compound, and its first US approval had nothing to do with metabolism: APHTHASOL, a 5% paste for mouth ulcers, approved in 1996. A 2 mg mucoadhesive patch followed in 2004. Drugs@FDA lists both as discontinued when checked on 2026-09-21.
It appears in this library because it is now searched, and sold, alongside the metabolic peptides.
Why it moved into diabetes
The reason is a mouse finding. Obesity raises two inflammatory kinases, IKK-epsilon and TBK1, and amlexanox inhibits both. In obese mice, blocking them reduced weight, insulin resistance, fatty liver and inflammation. An old, cheap, already-approved drug with that profile is an obvious candidate for repurposing, and the University of Michigan tested it.
The trial, read from the posted results
NCT01975935 — randomised, double-blind, placebo-controlled, 42 obese adults with type 2 diabetes and fatty liver, 25 mg amlexanox tablets or matching placebo for 12 weeks. The results are posted on ClinicalTrials.gov, and the paper appeared in Cell Metabolism in 2017 (Oral et al.).
| Measure (posted means) | Amlexanox, start → 12 weeks | Placebo, start → 12 weeks | Reported p |
|---|---|---|---|
| HbA1c | 7.68% → 7.44% | 7.77% → 7.76% | 0.05 |
| Body weight | 100.95 → 100.81 kg | 100.24 → 97.94 kg | 0.51 |
| Liver fat by MRI | 18.0% → 16.2% (11 measured) | 12.8% → 12.9% (9 measured) | 0.38 |
Three things in that table matter more than the headline.
- The blood-sugar effect is small and sits on the threshold. A 0.24-point fall against 0.01 on placebo, at p=0.05, in 21 people per arm. The paper also reports a significant fall in fructosamine, a shorter-term glucose marker.
- Weight went the wrong way for the claim. The amlexanox group moved by 0.14 kg; the placebo group lost 2.3 kg. Not significant — but a page that repeats the mouse weight-loss result as if it applied to people is contradicted by the only human trial there is.
- Liver fat is a subgroup story. The groups did not differ significantly. The paper reports that a subset of "responders", identified by an inflammatory gene signature in fat biopsies taken before treatment, improved in insulin sensitivity and liver fat. That subset was defined after the trial ended, so it is a hypothesis for the next study, not a finding of this one.
A smaller open-label pilot came first (NCT01842282, 7 people). It was terminated, and the registry's stated reason is funding: the investigators moved on to the placebo-controlled trial above.
What has happened since
Nothing larger. Of the five ClinicalTrials.gov records under the name, the two Michigan studies are the whole metabolic programme; the others concern oral lichen planus, oral mucositis and a cirrhosis study that lists amlexanox among several interventions. PubMed indexes 256 amlexanox records, 39 of them mentioning diabetes or obesity, and only two tagged as clinical trials — both the Michigan work. Checked 2026-09-21.
So the repurposing idea has one small positive signal on blood sugar, a null result on weight, and no follow-up trial in nearly a decade. That does not make the biology wrong. It does mean the evidence in people is one 12-week study of 42.
How it fits with the rest of this library
Amlexanox acts on inflammation inside fat tissue, a different lever from the incretin drugs. For the approach that has produced large weight-loss results in people, see the GLP-1 analogue semaglutide and the newer oral candidate orforglipron. For other metabolic compounds sold without a human trial behind the claim, the NNMT inhibitor 5-amino-1MQ and the exercise-mimetic SLU-PP-332 follow the same pattern. The broader question of inflammation is set out on the inflammation hub.
Last checked
2026-09-21. Product status from Drugs@FDA through the openFDA API; trial records and posted results from the ClinicalTrials.gov v2 API; the abstract of Oral et al. (PMID 28683283) read on PubMed; chemical identity from PubChem.