Irisin
Irisin (cleaved ectodomain of fibronectin type III domain-containing protein 5, FNDC5)
Written by Aaron CuhaReviewed Sep 2026
Also known as: FNDC5 fragment, Fibronectin type III domain-containing protein 5 (precursor)
A muscle-derived protein fragment that browns white fat in mice. No study has given irisin to a human, and whether commercial blood tests for it measure irisin at all was challenged in 2015 and answered only by mass spectrometry.
Overview
Irisin was described in 2012 as a hormone cut from the muscle membrane protein FNDC5 in response to exercise, which then acts on white fat to make it burn energy like brown fat. That paper was in mice, with some human cell work. The claim that followed, that exercise releases irisin and irisin delivers the metabolic benefits of exercise, is the reason the word appears on supplement labels and in blood test panels.
In people, the record is measurement, not treatment. No published study and no registered trial has administered irisin to a human; a ClinicalTrials.gov search on 11 September 2026 returned dozens of studies that measure serum irisin after exercise, diet or disease and none that gives it. Most of those measurements used commercial ELISA kits. In 2015 Albrecht and colleagues tested four of the polyclonal antibodies those kits rely on and found they bound non-specific serum proteins and produced no band of the expected size against recombinant irisin, concluding that all previous data from those kits was called into question. The same year the laboratory that discovered irisin answered with tandem mass spectrometry using heavy-isotope internal standards and reported that human irisin does circulate, at about 3.6 ng/mL in sedentary people and about 4.3 ng/mL in people doing aerobic interval training. Both papers are described in full below. The mass spectrometry result settles that the peptide exists in human blood; it does not rehabilitate the kit-based literature, and the exercise-induced change it found is roughly 19 percent.
Meta-analyses of exercise training and circulating irisin disagree with each other in direction. A 2015 analysis of 3 randomised trials (173 participants) found training lowered irisin; a 2021 analysis of 7 resistance training trials (282 people) found a non-significant rise; a 2026 analysis of 50 randomised trials (1,780 participants with overweight or obesity) found a moderate rise. All of them pool assays of contested specificity, and none measures any clinical outcome.
Everything else, including the integrin receptor, the bone effects, the sarcopenia rescue and the Alzheimer's model work, is in mice. Irisin is not approved anywhere and has no clinical development programme. Material sold as irisin is unregulated.
Mechanism of action
In mice: exercise raises the transcriptional coactivator PGC-1alpha in muscle, which induces FNDC5; the extracellular part of FNDC5 is cleaved and released as irisin, which acts on white adipocytes to induce UCP1 and a brown-fat-like thermogenic programme, raising energy expenditure and improving glucose handling. A 2018 Cell paper identified alpha-V integrins as irisin receptors in mouse bone and fat. Mouse work also reports effects on osteocytes, on age-associated muscle loss and on synaptic plasticity in Alzheimer's disease models. In humans, the mechanism is contested at the first step. Human FNDC5 carries a mutation of the conserved ATG start codon to ATA, which in transfected cells yields about 1 percent of the full-length protein; FNDC5 mRNA in human muscle biopsies did not change with endurance or strength training in two exercise studies; and recombinant irisin did not brown human preadipocytes in vitro where BMP7 did. The mass spectrometry study reports that human irisin is nonetheless translated from that non-canonical start codon. No receptor has been demonstrated in human tissue, and no downstream effect of irisin has been shown in a person.
Human evidence
No study has administered irisin to a human. Human data consist of measurements of circulating irisin in observational and exercise studies, most of them using commercial ELISA kits whose specificity was challenged in 2015, plus one mass spectrometry study that established the peptide is present in human plasma.
- Mass spectrometry (Cell Metabolism 2015): human irisin detected at about 3.6 ng/mL in sedentary people and about 4.3 ng/mL in people doing aerobic interval training, using heavy-isotope internal standards. This is the only assay in the literature whose specificity is not disputed.
- Antibody validation (Scientific Reports 2015): four commercial ELISA antibodies cross-reacted with non-specific serum proteins and did not detect recombinant irisin at the expected size. Every circulating irisin number obtained with those kits carries this caveat.
- Human muscle biopsies (PLoS One 2013): FNDC5 mRNA unchanged by endurance or strength training in two separate human exercise studies.
- Meta-analysis, 2015: 3 randomised trials, 173 participants, training lowered circulating irisin (d -0.46).
- Meta-analysis, 2021: 7 randomised resistance training trials, 282 people, non-significant rise (d 0.58, p 0.17), ELISA only.
- Meta-analysis, 2026: 50 randomised trials, 1,780 adults with overweight or obesity, training raised circulating irisin (SMD 0.62).
- ClinicalTrials.gov, searched 11 September 2026: dozens of registered studies measure serum irisin as an outcome of exercise, diet or supplementation. None administers irisin.
What this does not tell you: Nothing here shows that irisin does anything in a person. The exercise studies measure a blood analyte, not fat browning, energy expenditure, weight, bone or cognition, and the analyte was measured with kits of contested specificity in almost every case. The three meta-analyses do not agree on whether training raises or lowers it. The mass spectrometry study shows presence and a modest exercise response; it does not show function. There is no dose, no safety data and no pharmacokinetics for administered irisin in humans, because it has never been administered.
Reading the research record
This page exists to report a dispute, and the dispute has two sides with arguments worth stating properly. The objection came in two parts. In 2013 a German group showed that the human FNDC5 gene has lost the conserved ATG start codon, replaced by ATA, and that a human construct using ATA yields about 1 percent of the full-length protein; they also found no change in FNDC5 mRNA in human muscle after training and no browning of human fat cells by recombinant irisin. In 2015 Albrecht, Norheim, Thiede and colleagues went after the measurements: the four commercial polyclonal antibodies underpinning the ELISA kits bound the wrong proteins in human and animal sera and did not detect recombinant irisin at the expected size, so, in their words, all previous data obtained with those kits was called into question. Their title called irisin a myth.
The rebuttal came the same year from the laboratory that discovered irisin, which has an evident stake in the molecule's existence and also the deepest expertise in it. Jedrychowski, Wrann, Paulo and colleagues avoided antibodies entirely and used tandem mass spectrometry with heavy-isotope labelled standard peptides, finding irisin in human plasma at about 3.6 ng/mL in sedentary people and 4.3 ng/mL in people doing interval training, translated from the non-canonical ATA codon. That answers the existence question. It does not answer the function question, it does not validate any kit-based study, and the 0.7 ng/mL exercise response it documents is a long way from the effect sizes implied when irisin is sold. The 2021 Endocrine Reviews survey of the field treats both positions as live. We report both and do not pick a winner.
The economic context is simple. Irisin is an endogenous protein fragment with no composition-of-matter patent available, so no sponsor has a reason to run the trial that would settle whether giving it to a person does anything, and none has. The absence of that trial is not evidence for or against. Meanwhile the biomarker literature keeps growing on the strength of kits whose specificity was never independently re-established, which is why any page or product that quotes a circulating irisin concentration from a commercial ELISA should carry the Albrecht caveat.
The evidence, charted
Fig. 1 · evidence composition
4of 10 citations (40%) are in people
Counted from the citation list on this page. The Human count is the same number shown in the badge at the top. Both understate any literature larger than the sources we cite.
Fig. 2 · evidence over time
Evidence spans 7 distinct years, 2012 to 2026, counted from the citation list on this page.
Fig. 3 · legal status at a glance
US
Not approved
UK
Approved
AU
Not approved
CA
Not approved
Approved in 1 of 4, prescription route in 0, not approved in 3. A jurisdiction's classification is a regulatory fact, not a verdict on the science; see Legal status below for the full text and any notes.
Fig. 4 · dose response
No human dose response curve exists
We draw this figure where the data supports it. For this compound in humans it does not, so the panel stays empty rather than borrowing an animal curve and implying it transfers.
Key studies & citations
- Animal2012
A PGC1-alpha-dependent myokine that drives brown-fat-like development of white fat and thermogenesis
The original report. In mice, exercise-induced PGC-1alpha raised FNDC5 in muscle; its cleaved product, named irisin, induced UCP1 and a brown-fat programme in white fat, increased energy expenditure and improved obesity and glucose homeostasis. NIH and non-US government funded.
Nature - In vitro2013
Evidence against a beneficial effect of irisin in humans
Human FNDC5 has an ATA rather than ATG start codon; a human construct with ATA produced about 1 percent of full-length protein. FNDC5 mRNA in human muscle biopsies from two exercise studies was unchanged by endurance or strength training, and recombinant FNDC5 or irisin did not brown human preadipocytes where BMP7 did. Non-US government funded.
PLoS One - In vitro2015
Irisin, a myth rather than an exercise-inducible myokine
Four commercial polyclonal antibodies used in irisin ELISA kits showed prominent cross-reactivity with non-specific proteins in human and animal sera on Western blot; with recombinant glycosylated and non-glycosylated irisin as positive controls, no band of the expected size appeared in any biological sample. A FNDC5 signature at about 20 kDa was found in human serum by mass spectrometry but was not detected by the antibodies. The authors concluded that all previous data from commercial irisin ELISA kits was called into question. NIH and non-US government funded.
Scientific Reports - Human2015
Detection and Quantitation of Circulating Human Irisin by Tandem Mass Spectrometry
From the laboratory that discovered irisin. Mass spectrometry with heavy stable isotope internal standards found human irisin at about 3.6 ng/mL in sedentary people and about 4.3 ng/mL in people doing aerobic interval training, and reported it is mainly translated from the non-canonical ATA start codon. The authors' conclusion: human irisin exists, circulates and is regulated by exercise. NIH and non-US government funded.
Cell Metabolism - Review2021
Progress and Challenges in the Biology of FNDC5 and Irisin
The field-level review that treats the start codon objection, the antibody objection and the mass spectrometry rebuttal as open questions rather than settled ones.
Endocrine Reviews - Human2015
Chronic Exercise Training and Circulating Irisin in Adults: A Meta-Analysis
3 randomised trials (173 healthy untrained participants): at least 8 weeks of training was associated with lower circulating irisin than control (Cohen's d -0.46, 95 percent CI -0.76 to -0.15). 9 non-randomised studies (113 participants) showed no change (d -0.04). All studies used immunoassays.
Sports Medicine - Human2021
Effect of Chronic Resistance Training on Circulating Irisin: Systematic Review and Meta-Analysis of Randomized Controlled Trials
7 randomised trials, 282 people, all using ELISA kits by inclusion criterion: a non-significant increase after at least 8 weeks of resistance training (d 0.58, 95 percent CI -0.25 to 1.40, p 0.17). The authors call for mass spectrometry in future studies.
International Journal of Environmental Research and Public Health - Human2026
Effects of Different Training Modalities on Circulating Irisin Levels in Overweight and Obesity Adults: A Systematic Review and Meta-analysis of Randomized Controlled Trials
50 randomised trials, 1,780 adults with overweight or obesity: exercise training was associated with higher circulating irisin than passive control (standardised mean difference 0.62, 95 percent CI 0.39 to 0.85). The direction is opposite to the 2015 meta-analysis. Irisin was measured as a biomarker; no clinical outcome was pooled.
Sports Medicine - Animal2018
Irisin Mediates Effects on Bone and Fat via alpha-V Integrin Receptors
Identification of alpha-V integrins as irisin receptors in mouse osteocytes and fat cells. The receptor has not been demonstrated in human tissue.
Cell - Animal2019
Exercise-linked FNDC5/irisin rescues synaptic plasticity and memory defects in Alzheimer's models
In mouse Alzheimer's disease models, FNDC5/irisin restored synaptic plasticity and memory. Mouse only.
Nature Medicine
Frequently asked questions
Does exercise raise irisin in humans?
By mass spectrometry, the one assay whose specificity is not disputed, people doing aerobic interval training had about 4.3 ng/mL against 3.6 ng/mL in sedentary people, a difference of roughly 19 percent. Meta-analyses of exercise trials using commercial kits disagree: a 2015 analysis found training lowered irisin, a 2026 analysis of 50 trials found it raised it. The kits those analyses pool were shown in 2015 to bind non-specific serum proteins.
Has irisin ever been given to a person?
No. No published study and no registered trial has administered irisin to a human. Every human irisin study measures the level in blood after exercise, diet or disease. Effects on fat browning, bone, muscle and memory are all from mice.
Is irisin a myth?
That was the title of a 2015 paper showing that commercial irisin antibodies do not detect irisin specifically. A mass spectrometry study the same year, from the group that discovered irisin, found the peptide in human plasma at a few nanograms per millilitre. So the peptide exists in humans. Whether it does anything in humans has not been tested.
Should I trust an irisin blood test?
If it is a commercial ELISA, the 2015 antibody study found the kits it examined cross-reacted with other serum proteins and did not detect recombinant irisin at the expected size. No independent revalidation of those kits has been published. Mass spectrometry is the only method with established specificity and it is not offered as a routine test.
Why is there no irisin drug or trial?
Irisin is a fragment of a human protein, so nobody can own the molecule, and no sponsor has funded an administration trial. That leaves the question open in both directions: without a trial, neither benefit nor harm in people has been looked for.