FGF21 and Exercise: What the Exerkine Science Really Shows

Exercise does far more than burn calories. Contracting muscle and metabolically active organs release chemical signals that affect glucose handling, inflammation, fat metabolism, blood vessels, and the brain. These signals, called exerkines, help explain why regular movement can improve cardiometabolic health even when the scale barely moves.

Fibroblast growth factor 21, or FGF21, is one of the most interesting—and easiest to misread—exerkines. The liver produces most circulating FGF21. Exercise can alter it, and FGF21 has plausible roles in glucose and lipid regulation. But the useful conclusion is not "make FGF21 go up." In people with metabolic disease, chronically high FGF21 may signal resistance to its effects, much as high insulin can signal insulin resistance.

FGF21 is an exercise signal, not an exercise prescription

FGF21 is a hepatokine: a hormone-like signaling protein released mainly by the liver. Reviews of the field place it alongside muscle-derived signals such as IL-6, irisin, and BDNF in the wider endocrine response to exercise. Türkel et al. (2022) and Jin et al. (2024) describe credible downstream biology involving glucose uptake, fatty-acid oxidation, and insulin sensitivity.

That biology does not yet tell you whether to train at a particular heart rate for a particular number of minutes to "optimize FGF21." Human trials have not established an FGF21 dose-response curve for exercise, a threshold intensity for secretion, or proof that an exercise-driven FGF21 change causes a specific improvement in HbA1c, insulin resistance, triglycerides, or liver histology.

A 2026 human study by Zhang et al. directly examined whether exercise-induced FGF21 contributes to delayed improvement in glucose control after acute exercise in people with type 2 diabetes. It is the most directly relevant study here because it asks a functional question in the population of interest: does the signal relate to post-exercise glycemic recovery? The published abstract available through PubMed does not report the sample size, effect estimate, fold change, or p-values needed to judge the magnitude and precision of that relationship. The study supports FGF21 as a candidate mediator, not as settled causal proof.

The apparent contradiction: training can lower FGF21

The strongest corrective to FGF21 hype comes from Stine et al. (2023). In patients with biopsy-proven nonalcoholic steatohepatitis, supervised exercise training markedly decreased serum FGF21. The study's biopsy confirmation matters: metabolic liver disease is often classified with imaging or blood markers that can misjudge disease severity.

At first glance, a decrease after training seems to clash with the idea that acute exercise may raise FGF21. The two findings can fit the same homeostatic model. Acute exercise may temporarily increase FGF21 as part of a coordinated stress and fuel-management response. Chronic training may lower an already elevated baseline as metabolic dysfunction improves.

This interpretation remains an explanation, not a proven chain of causation. Still, it matches the broader metabolic liver disease literature. Pezzino et al. (2025) describes elevated FGF21 in obesity and insulin resistance as a compensatory response consistent with FGF21 resistance. A high baseline level may therefore reflect a body asking harder for a signal rather than receiving more of its benefit.

That distinction changes the practical reading of blood tests and exercise headlines. A transient rise in FGF21 after one workout and a lower resting FGF21 after months of training are different endpoints, measured at different times, in different physiological states. Treating either number as a universal scorecard would be like judging a fire alarm without asking whether there is a fire.

Intensity research is useful, but it has not solved the question

Researchers are testing whether exercise format changes metabolic responses in high-risk groups. In a study of 25 candidates for sleeve gastrectomy—22 of them women—Enríquez-Schmidt et al. (2024) compared moderate-intensity continuous training with high-intensity interval training. The investigators assessed glycemia, insulin measures, HOMA-IR, lipids, HbA1c, transaminases, and FGF21. The available abstract establishes the clinically relevant comparison but does not provide enough result detail to claim that either program optimized FGF21 or produced a superior metabolic outcome.

Hovsepian et al. (2022) also studied all-extremity HIIT in overweight and obese women, measuring FGF21 alongside athero-protective factors and endothelial function. It is relevant because it pairs an intensity-defined intervention with cardiovascular endpoints, but the accessible summary does not provide the FGF21 values or statistical comparisons. It cannot support a numerical HIIT prescription for FGF21.

Mouse work can explain receptor biology without answering that clinical question. Moharamzadeh et al. (2025) compared exercise intensity and diet in chemically induced type 2 diabetic mice and examined FGF21 and its co-receptor, beta-Klotho. Such studies can generate mechanisms worth testing in people. They cannot establish human intensity thresholds because chemically induced rodent diabetes does not reproduce the full pathophysiology of human type 2 diabetes.

Exerkines extend beyond the liver

The larger exerkine story is more credible than any claim about one miracle molecule. Exercise affects signaling across organs, and some pathways have clearer mechanistic steps than FGF21 does today. Tero-Vescan et al. (2026) reviews evidence that contracting skeletal muscle can increase kynurenine aminotransferases, which convert circulating kynurenine into kynurenic acid. This can reduce kynurenine uptake into the brain, where inflammatory metabolism can otherwise contribute to production of quinolinic acid, a neurotoxic metabolite. The pathway offers a plausible link between exercise, mood, fatigue resistance, and neuroprotection, although mechanistic plausibility still exceeds hard clinical-endpoint evidence.

The muscle-brain axis also matters in type 2 diabetes. Xenos et al. (2025) reviews shared pathways connecting sarcopenia and cognitive decline, including insulin resistance, hyperglycemia, mitochondrial dysfunction, oxidative stress, and inflammation. It identifies several exercise-responsive molecules, including IGF-1, irisin, BDNF, FGF21, and SPARC. These pathways help explain why preserving muscle may matter for brain health; they do not prove that changing any one exerkine prevents Alzheimer's disease.

Human training studies also show that exercise can change inflammatory biology in people with established disease. In a pilot secondary analysis of a registered randomized trial, Garneau et al. (2023) assigned male patients with coronary artery disease to 12 weeks of twice-weekly moderate-to-vigorous continuous training or HIIT. The type 2 diabetes subgroups were small: six in HIIT and five in continuous training. Circulating cytokines fell after training, but those numbers make the study a preliminary signal and a feasibility foundation for a larger trial, not a final verdict about either protocol.

What to do on Monday morning

Do not choose workouts to chase FGF21. Choose a program you can repeat, then use ordinary clinical outcomes to judge whether it helps: fitness, strength, glucose control, blood pressure, lipids, symptoms, and—when relevant—liver care guided by a clinician. Moderate continuous work and intervals both belong in research and practice; the best choice depends on health status, injury risk, preference, and adherence.

FGF21 is worth watching because it gives exercise physiology a sharper molecular explanation. It is not yet a biomarker that can prescribe your interval length or prove why your glucose improved after a workout. The most informative future trials will measure defined exercise programs, FGF21, and hard metabolic outcomes in people with type 2 diabetes or metabolic liver disease. Until then, exercise remains highly evidence-based. FGF21 remains a promising explanation for part of its biology.