ATX-304 for Health & Longevity

Evidence Review created on 08/27/2026 using AI4L / Opus 5

Also known as: O304, O-304, ATX304

Motivation

ATX-304 is an experimental oral medication built to switch on the enzyme that cells use to detect when their fuel is running low. The same switch is thrown by hard exercise and by going without food, so a compound that flips it directly has drawn attention from people who follow the biology of aging and who already train, fast, and track their own metabolic markers.

The molecule was discovered in Sweden and first developed as a treatment for type 2 diabetes. It has since been carried forward for obesity and related disorders of metabolism, and it has been given to people in two short, company-run studies. Only a handful of compounds reach this cellular switch directly rather than by making cells work harder for energy, and ATX-304 is the first of that group to produce measurable metabolic changes in humans.

This review examines what is known about ATX-304: how it is thought to work, what was measured in laboratory animals and in the people who have taken it, what the safety record covers and what it does not, how it has been dosed in trials, and where the evidence base is thin or contested.

Benefits - Risks - Protocol - Conclusion

Recommended Reading

This section collects high-level sources on ATX-304 itself and on pharmacological activation of AMPK (AMP-activated protein kinase, the enzyme cells use to sense low energy), the target the compound was designed around.

• PAN-AMPK activator O304 improves glucose homeostasis and microvascular perfusion in mice and type 2 diabetes patients - Steneberg et al., 2018

  The founding paper: it describes the compound, its proposed action, and the 28-day phase IIa trial. Authored and funded by the developer, Betagenon AB, a direct financial interest.

• New insights into activation and function of the AMPK - Steinberg & Hardie, 2023

  A narrative review of the shared target, AMPK: how activators are classified, why earlier drug candidates stalled, and which tissue effects are desirable versus hazardous. Written by researchers with no stake in ATX-304.

• O304 is a mitochondrial uncoupler which extends C. elegans lifespan and induces vasorelaxation of rat mesenteric arteries - Li et al., 2025

  An independent Chinese group re-characterizes the compound, reporting mitochondrial uncoupling, lifespan extension in the nematode Caenorhabditis elegans, and blood-vessel relaxation — the most directly longevity-framed work available.

• The AMPK activator ATX-304 alters cellular metabolism to protect against cisplatin-induced acute kidney injury - Katerelos et al., 2024

  An Australian laboratory with no commercial stake tests the same compound in kidney injury and maps how it rewires cellular fuel use, giving an outside read on the proposed mechanism.

• Activate AMPK For Optimal Metabolic Health - Robert Friedman

  A plain-language overview of the shared target, AMPK, and of how age lowers its activity. Covers dietary activators rather than ATX-304, and is published by a supplement retailer.

Only one priority platform is represented. No relevant content on ATX-304 or on direct AMPK activators was found on foundmyfitness.com, peterattiamd.com, hubermanlab.com, chriskresser.com or lifespan.io — the compound has not yet reached the independent health-media literature — so peer-reviewed primary papers and a narrative review stand in for expert commentary.

Grokipedia

No Grokipedia article exists for ATX-304.

Examine

No Examine article exists for ATX-304. ATX-304 is an investigational prescription-track drug rather than a dietary supplement, and Examine.com does not typically cover prescription or investigational medications.

ConsumerLab

No ConsumerLab article exists for ATX-304. ConsumerLab tests retail dietary supplements and does not typically cover prescription or investigational medications, which ATX-304 is.

Systematic Reviews

No systematic reviews or meta-analyses for ATX-304 were found on PubMed as of August 27, 2026.

Neither side of the trade-off is represented: there is no evidence synthesis on the claimed metabolic benefits of ATX-304, and none on its principal risk, the consequences of sustained mitochondrial uncoupling.

Mechanism of Action

ATX-304 raises the activity of AMPK, the enzyme that senses when a cell’s fuel is running short. Most drugs that activate AMPK do so indirectly, by making energy scarcer. ATX-304 instead protects AMPK’s activating tag — a phosphate group on threonine 172 — from being stripped off by the enzyme PP2C (protein phosphatase 2C, which switches AMPK back off), so activation persists without ATP (adenosine triphosphate, the cell’s energy currency) being drained. It works across all twelve AMPK subunit combinations, hence “pan-AMPK”, but only in cells that carry the upstream activating enzyme LKB1 (a kinase that places the activating tag). Downstream, activated AMPK moves GLUT4 (the muscle glucose transporter) to the cell surface, blocks fat synthesis through acetyl-CoA carboxylase, and restrains mTOR (a growth-signalling hub), the pattern seen with exercise (Steneberg et al., 2018).

A competing account has emerged. An independent Chinese group — and, earlier, the discovering Umeå laboratory itself — reports that ATX-304 also acts as a mild mitochondrial uncoupler, letting mitochondria burn fuel without capturing it as ATP, and that this manufactured demand, not phosphatase blockade, drives much of the effect (Li et al., 2025; Norlin et al., 2023). The developer’s 2026 conference data counter that AMPK activation occurs with ATP preserved. Both readings remain live.

In rodents ATX-304 is orally available, has a long plasma half-life, and does not cross the blood-brain barrier, so its action is peripheral. Its metabolizing enzymes have not been published.

Historical Context & Evolution

ATX-304 began as O304, discovered at Umeå University in Sweden by Helena Edlund, Thomas Edlund and the late Olof Karlsson, and developed by their company Betagenon AB. The original intent was narrow and therapeutic: a treatment for type 2 diabetes that would raise muscle glucose uptake without forcing the pancreas to secrete more insulin.

The move toward health optimization came from what the compound did beyond glucose. In diet-induced obese mice it raised cardiac stroke volume without enlarging the heart; in dogs single doses lowered blood pressure; and in aged lean mice it improved running endurance while lowering exercise lactate (Steneberg et al., 2018). In aged mice it reversed established insulin resistance and restored exercise capacity (Ericsson et al., 2021). That pattern — the physiological signature of training, produced by a tablet — is what recast it as a candidate longevity agent.

Its predecessors shape how the field reads it. AICAR (an early AMPK-activating compound) was limited by poor oral absorption. Later direct activators reproduced a hazard seen in people carrying PRKAG2 mutations (a gene coding for an AMPK subunit): cardiac glycogen accumulation and heart enlargement. Toxicology and the human trials found no such effect with ATX-304, and no change in left-ventricular mass, though exposure has been brief. Steinberg and Hardie review both the promise and the unresolved hazards of the class (2023).

Betagenon was acquired by Amplifier Therapeutics, a Cambrian Bio pipeline company, which renamed the compound and redirected it toward obesity and diseases of aging.

Expected Benefits

High 🟩 🟩 🟩

No benefit reaches High: no clinical endpoint or validated clinical surrogate has been reproduced in more than one human trial — the two completed studies each measured a different set of surrogates over four to eight weeks.

Medium 🟩 🟩

Lower Fasting Glucose and Improved Insulin Sensitivity

In the 28-day TELLUS trial, 65 people with type 2 diabetes on metformin received 1,000 mg of ATX-304 daily or placebo. Fasting plasma glucose fell in the treated group, and insulin resistance, measured as HOMA-IR (a calculation combining fasting glucose and fasting insulin), fell with it. The proposed route is insulin-independent glucose uptake into muscle. The effect emerged only between day 21 and day 28, so the trial may have been too short to show its full size (Steneberg et al., 2018).

Magnitude: In a post hoc analysis restricted to participants whose day-1 fasting glucose lay between 7 and 13.3 mM, fasting plasma glucose fell by 0.60 mM (millimoles per litre, roughly 11 mg/dL) from day 1 to day 28 versus 0.10 mM on placebo, P = 0.010 between groups (P is the probability that a difference this large would arise by chance alone). HOMA-IR fell within the treated group only, P = 0.0097.

Reduced Blood Pressure

The same 28-day trial recorded a fall in systolic and diastolic blood pressure in the treated group while placebo drifted upward. The proposed mechanism is AMPK-driven relaxation of small blood vessels; single doses lowered blood pressure acutely in dogs, and an independent group later showed the compound relaxes isolated rat arteries. All participants had type 2 diabetes and took metformin, so the change in metabolically healthier individuals is unknown (Steneberg et al., 2018; Li et al., 2025).

Magnitude: Mean absolute reduction of 5.8 mmHg systolic and 3.8 mmHg diastolic at day 28, against increases of 1.2 and 0.9 mmHg on placebo; within-group P = 0.030 and P = 0.009.

Reduced Liver Fat and Visceral Fat

The 8-week phase 1b study in 23 adults with obesity and prediabetes reported significant reductions in liver fat, measured by MRI-PDFF (magnetic resonance imaging that quantifies the liver’s fat fraction), and in visceral fat. This matches mouse work in which the compound cut body fat, liver steatosis (fat build-up in the liver) and fibrosis (scarring) while shifting the liver toward burning fat (Holm et al., 2025). The human data were presented at a conference by the developer and are not peer-reviewed.

Magnitude: Direction is a reduction in both liver fat fraction and visceral fat, statistically significant at P < 0.05 after eight weeks at 400 mg daily; the company’s release reports no outcome figure for either endpoint.

Improved Blood Lipids and Adiponectin

In the same 8-week study, plasma triglycerides fell and adiponectin (a hormone released by fat tissue that tracks with insulin sensitivity) rose. Mouse work found parallel falls in blood cholesterol with remodelling of lipid and cholesterol transport in the liver (Holm et al., 2025). Because both markers moved in a small, developer-run study that has not been peer-reviewed, and because triglycerides respond readily to diet, the finding needs independent replication.

Magnitude: Triglycerides fell and adiponectin rose, both at P < 0.01 over eight weeks at 400 mg daily; the company’s release reports no outcome figure for either marker.

Increased Resting Metabolic Rate

Resting metabolic rate — the energy the body burns at rest — rose in the 8-week phase 1b study. This is the outcome most directly tied to the mitochondrial-uncoupling account of the compound: mitochondria burn more fuel to hold the same output. Weight loss was minimal at this exposure, which the developer attributes to the dose being below the level modelled from animal work (Norlin et al., 2023). Whether the rise persists beyond eight weeks is untested.

Magnitude: Resting metabolic rate rose 8% over eight weeks at 400 mg daily, P < 0.01.

Low 🟩

Improved Microvascular Perfusion

Blood flow through small calf vessels, imaged by scan, improved over 28 days in the treated group (Steneberg et al., 2018). It is graded Low because significance depended on the statistical model used, and splitting the group by starting blood flow put the response in the slow-flow half.

Magnitude: Direction is an increase in the blood-flow surge signal (the rise in flow after a brief squeeze of the calf) versus placebo, P = 0.026 across the treated group as a whole; splitting that group by starting blood flow put the change in those perfusing slowly at baseline (P = 0.034) and showed none in those already perfusing quickly. The trial reports this endpoint as an MRI signal parameter and gives no outcome figure.

Speculative 🟨

Improved Endurance and Cardiac Output

In aged mice, endurance and cardiac output rose; heart weight was 15% higher after six months, free of fibrosis and glycogen — read as training-like adaptation (Ericsson et al., 2021). No human data.

Preserved Beta-Cell Function

In diabetic mice the compound preserved insulin secretion, beta-cell mass and pancreatic insulin content (Norlin et al., 2023), and blocked diet-induced islet stress-gene changes (López-Pérez et al., 2021). Rodent evidence only.

Slowed Kidney Aging and Protection from Kidney Injury

Two independent laboratories report reduced age-related kidney fibrosis and cellular senescence (non-dividing cells that linger) in old mice, and protection from chemotherapy-induced kidney injury (Zhu et al., 2022; Katerelos et al., 2024). No human data.

Extended Lifespan

The compound extended lifespan in the nematode Caenorhabditis elegans through the AMPK–mTOR axis and autophagy (the cell’s recycling of its own worn-out parts) (Li et al., 2025). No mammalian lifespan study exists.

Reduced Pain Sensitivity

Three studies from one independent laboratory report that oral O304 reduces mechanical pain sensitivity in mice after disc puncture and after surgical incision (Das et al., 2022). No human data.

Reduced Formation of Abdominal Aortic Aneurysm (a bulge in the body’s main artery)

In a mouse model, O304 activated AMPK in vascular smooth muscle, preserved the contractile cell state and reduced aneurysm formation and blood pressure (Sun & Du, 2024). Rodent and cell-culture evidence only.

Benefit-Modifying Factors

• LKB1 status: ATX-304 raises AMPK activity only in cells expressing the upstream kinase LKB1. Inactivating variants in STK11 (the gene that encodes LKB1) should therefore blunt or abolish the response in carriers.

• PRKAG2 variants: Variants in this AMPK subunit gene alter how readily the enzyme is activated and how muscle stores glycogen, so they plausibly shift both the size of the metabolic response and the cardiac margin of safety. Untested for this compound.

• Baseline microvascular perfusion: Improvement in calf blood flow appeared only in participants whose perfusion was slower at baseline; those already perfusing well showed no change. Baseline impairment appears to be a prerequisite for this benefit.

• Baseline glucose and insulin resistance: In the phase IIa trial the treated group happened to start with higher fasting glucose but lower insulin resistance, which the authors judged to have limited the observable effect. Benefit should scale with the degree of baseline insulin resistance.

• Sex: Unknown. The published rodent work used male animals almost exclusively, and neither human study has reported outcomes split by sex, so sex-based differences in response cannot be estimated.

• Pre-existing conditions: Both human studies enrolled people with type 2 diabetes or prediabetes and obesity. Metabolically healthy individuals with normal glucose, lipids and blood pressure have less room to improve on every measured endpoint.

• Age: Preclinical benefit was largest in aged mice, where insulin resistance and cardiac decline were already established. In humans, no participant cohort has been reported by age band, so an age gradient is inferred from animals only.

Potential Risks & Side Effects

High 🟥 🟥 🟥

No risk reaches High: no adverse outcome has been documented as a clinical endpoint or validated clinical surrogate in more than one human trial — reported events in both completed studies were mild and occurred at placebo-like frequency.

Medium 🟥 🟥

Blood Pressure Lowering and Hypotension (low blood pressure) in Combination

The blood-pressure fall that counts as a benefit in hypertension is a hazard for anyone already at or below normal pressure, or taking antihypertensive medication. The mechanism is direct relaxation of small blood vessels, so the effect is additive with existing agents rather than competing with them. It appeared within 28 days in people with type 2 diabetes, was not accompanied by any change in heart rate, and reverses on discontinuation (Steneberg et al., 2018).

Magnitude: Mean reduction of 5.8 mmHg systolic and 3.8 mmHg diastolic over 28 days at 1,000 mg daily; larger falls would be expected during concomitant antihypertensive therapy.

Mild Treatment-Emergent Adverse Events

Across both human studies, treatment-emergent adverse events were predominantly mild and occurred at a frequency similar to placebo, with no new signals during the open-label extension. Evidence basis is two small trials of 65 and 23 participants lasting 28 days and 16 weeks. The developer has not published a categorised adverse-event table for either study, so the specific complaints, their frequencies and their severity distribution remain undisclosed (Steneberg et al., 2018).

Magnitude: Direction is a low event burden — mild events at placebo-comparable frequency at 400–1,000 mg daily for up to 16 weeks; the company’s release and the phase IIa report give no event rates or severity breakdown.

Low 🟥

Hypoglycemia (low blood sugar) in Combination with Glucose-Lowering Drugs

ATX-304 drives glucose into muscle without insulin, so it can add to insulin or sulfonylureas (oral drugs that force the pancreas to release insulin). Neither trial reported hypoglycemia, but both enrolled only participants on metformin, which rarely causes low blood sugar (Steneberg et al., 2018).

Magnitude: Not quantified in available studies. No trial has enrolled participants on insulin or sulfonylureas, so no event rate exists for the combination in which the risk would arise.

Thermogenic and Cardiovascular Effects of Uncoupling ⚠️ Conflicted

Uncoupling drugs classically raise body temperature and heart rate; the historical example proved lethal. Independent work calls ATX-304 an uncoupler (Li et al., 2025), yet phase 1b monitoring found no core-temperature or heart-rate rise. Net reading: the effect is real but too small to cause events at tested doses.

Magnitude: Resting energy expenditure rose 8% over eight weeks with no measurable change in core body temperature or 24-hour heart rate; no adverse-event figure exists because no such events occurred.

Speculative 🟨

Cardiac Glycogen Accumulation and Remodelling

PRKAG2 mutations produce glycogen-loaded, enlarged hearts through chronic AMPK activation. Animal toxicology and the 28-day trial found no such change (Steneberg et al., 2018); six-month dosing raised aged-mouse heart weight 15% (Ericsson et al., 2021).

Blunted High-Intensity Performance

The compound shifts muscle toward burning glucose rather than storing it as glycogen. Because glycogen fuels high-intensity effort, chronic use could reduce sprint and heavy-lifting capacity. No study has measured this.

Reproductive and Developmental Toxicity

No reproductive or developmental toxicology has been published, and no trial enrolled pregnant or breastfeeding participants. The basis for concern is absence of data, not any positive finding.

Tumour Growth Support Under Sustained AMPK Activation

Sustained AMPK activation can help tumour cells survive metabolic stress, alongside its opposite, growth-restraining role (Penugurti et al., 2024). No human data exist; the basis is mechanistic.

Adulterated or Misidentified Gray-Market Material

ATX-304 is sold by unregulated “research chemical” vendors, where reported problems include wrong salt forms, unverified identity and undisclosed dose variance. No analytical survey of these products exists, so the failure rate is unmeasured.

Risk-Modifying Factors

• PRKAG2 and STK11/LKB1 variants: Carriers of PRKAG2 variants already predisposed to cardiac glycogen storage sit closest to the theoretical cardiac hazard; loss-of-function STK11/LKB1 variants instead reduce responsiveness, lowering both benefit and risk.

• Baseline blood pressure: Individuals with systolic pressure below about 110 mmHg, or on two or more antihypertensive agents, face the greatest risk of symptomatic low blood pressure from the compound’s vasodilatory effect.

• Sex: Unknown. Rodent studies used male animals almost exclusively, and neither human trial has reported adverse events split by sex, so sex-based differences in the safety profile cannot be assessed.

• Pre-existing conditions: Type 1 diabetes, insulin or sulfonylurea use, heart failure, established cardiomyopathy (disease of the heart muscle), uncontrolled overactive thyroid, and moderate-to-severe liver or kidney impairment all sit outside the enrolled trial populations and outside the safety evidence.

• Age: Everyone studied was an adult under standard trial age caps. Older adults with reduced cardiac reserve, blunted temperature regulation and several concurrent medications carry more exposure to the blood-pressure and thermogenic effects than the trial populations did.

• Baseline biomarkers: Low baseline blood pressure, low fasting glucose, elevated liver enzymes or reduced kidney filtration all shift the risk-benefit balance unfavourably, since none of these states was represented in the studied cohorts.

Key Interactions & Contraindications

• Metformin (prescription): Studied in combination and additive on glucose lowering; both human trials required background metformin. Severity: monitor. Consequence: greater-than-expected fall in fasting glucose. No dose change was needed in trials.

• Insulin and sulfonylureas (prescription, e.g. glipizide, glimepiride, gliclazide): Additive insulin-independent glucose uptake. Severity: caution. Consequence: hypoglycemia. Mitigation: reduce the sulfonylurea or insulin dose and increase glucose self-monitoring before adding.

• Antihypertensives (prescription, e.g. lisinopril, losartan, amlodipine): Additive vasodilation on top of the compound’s own 5.8/3.8 mmHg fall. Severity: caution. Consequence: symptomatic low blood pressure, dizziness on standing. Mitigation: monitor home readings, separate dose timing.

• GLP-1 receptor agonists (prescription, e.g. semaglutide, tirzepatide — drugs mimicking a gut hormone that curbs appetite): Combined in mouse obesity models by the developer. Severity: monitor. Consequence: additive weight loss and glucose lowering. Human combination data do not exist.

• Diuretics (over-the-counter and prescription, e.g. hydrochlorothiazide, caffeine-based products): Severity: caution. Consequence: additive blood-pressure fall and volume depletion. Mitigation: maintain fluid and electrolyte intake, especially with heat exposure or sauna use.

• Non-steroidal anti-inflammatory drugs (over-the-counter, e.g. ibuprofen, naproxen): Severity: monitor. Consequence: these reduce kidney blood flow, opposing the compound’s kidney-protective signal in animals and complicating interpretation of kidney monitoring. Mitigation: avoid routine daily use.

• AMPK-activating supplements (berberine, Gynostemma pentaphyllum, alpha-lipoic acid, metformin-mimetic extracts): Severity: caution. Consequence: additive glucose lowering and, for berberine, additive blood-pressure reduction. Mitigation: introduce one agent at a time.

• Blood-pressure-lowering supplements (beetroot or nitrate concentrates, magnesium at high intake, garlic extract, omega-3 fatty acids): Additive with the compound’s own vasodilation. Severity: caution. Consequence: symptomatic low blood pressure. Mitigation: stagger introduction and track standing readings.

• Other mitochondrial uncouplers and thermogenic agents (2,4-dinitrophenol, high-dose ephedrine or synephrine combinations): Severity: absolute contraindication. Consequence: uncontrolled thermogenesis and hyperthermia. Mitigation: none; do not combine.

• Other interventions — heat exposure and endurance training: Sauna, heat training and prolonged endurance work each add heat load and vasodilation to a compound that modestly raises energy expenditure. Severity: caution. Consequence: heat intolerance, faintness. Mitigation: reduce session intensity during initiation.

Populations who should avoid ATX-304:

• Anyone outside a supervised clinical trial: the compound is investigational and holds no marketing authorisation in any jurisdiction
• Pregnancy and breastfeeding: no reproductive or developmental toxicology has been published
• People under 18 years: no paediatric exposure exists
• Type 1 diabetes, or type 2 diabetes on insulin or sulfonylureas without dose adjustment
• Baseline systolic blood pressure below 100 mmHg, or symptomatic orthostatic hypotension (a fall in blood pressure on standing that causes dizziness or fainting)
• Established cardiomyopathy, heart failure of New York Heart Association Class III–IV, or a known PRKAG2 variant
• Liver impairment of Child-Pugh Class B or C (a standard grading of how far liver function has failed)
• Kidney impairment with eGFR below 45 mL/min/1.73 m² (estimated glomerular filtration rate, a calculated measure of how fast the kidneys filter blood)
• Uncontrolled overactive thyroid, or any condition impairing temperature regulation
• Recent myocardial infarction (heart attack) within 90 days, or unstable angina (worsening chest pain from reduced blood flow to the heart)

Risk Mitigation Strategies

• Trial enrolment over self-administration: The only route with characterised material, dose control and monitoring. Mitigates the gray-market risks of wrong identity, wrong salt form and undisclosed dose variance, none of which has been analytically surveyed.

• Third-party analytical testing of any material obtained outside a trial: Independent identity and purity testing by mass spectrometry, plus a certificate of analysis for the specific lot, mitigates misidentified or adulterated product.

• Stay within studied exposures: The phase 1b study used 400 mg daily for eight weeks; the phase IIa used 1,000 mg for 28 days. Staying in that range mitigates the untested thermogenic margin at higher doses.

• Home blood pressure logging before and during use: Twice-daily seated and standing readings for the first four weeks, against a two-week pre-treatment baseline, mitigates symptomatic low blood pressure from additive vasodilation.

• Antihypertensive and glucose-drug review before starting: Reducing sulfonylurea or insulin doses, and reviewing antihypertensive load, mitigates hypoglycemia and low blood pressure, the two interaction risks with human evidence behind them.

• Allow 14 days before judging response or escalating: Steady-state plasma levels were not reached until day 14, and glucose effects appeared only after day 21. This mitigates dose escalation driven by apparent non-response.

• Track resting heart rate and morning temperature: A wearable or oral thermometer log detects the thermogenic drift that uncoupling would produce, mitigating the theoretical hyperthermia risk before it becomes symptomatic.

• Avoid concomitant thermogenic or uncoupling agents: Eliminating dinitrophenol, high-dose stimulant thermogenics and aggressive heat protocols during use mitigates uncontrolled heat load, the class’s historical cause of death.

• Baseline and 12-week cardiac assessment: An ECG (electrocardiogram, a recording of the heart’s electrical activity) and, where accessible, an echocardiogram mitigate the theoretical cardiac glycogen and remodelling hazard PRKAG2 biology raises.

Therapeutic Protocol

No clinical practice protocol exists: ATX-304 is unapproved, and no physician can prescribe it outside a trial. What follows is what the trials did.

• Studied doses: 1,000 mg once daily as an oral suspension for 28 days in the phase IIa trial; 400 mg once daily as an oral formulation for 8 weeks in the phase 1b, followed by an 8-week open-label extension.

• Time of day: Both trials used once-daily morning dosing. The long half-life makes timing largely irrelevant to exposure, so consistency of timing matters more than the hour chosen.

• Half-life: Long. Steady-state plasma concentration was not reached until day 14 of daily dosing in people, implying an effective half-life measured in days rather than hours.

• Single versus split dosing: Both trials used a single daily dose. Because the compound accumulates to steady state over two weeks, splitting the dose offers no exposure advantage and has never been tested.

• Competing approach — exercise and energy restriction: Vigorous exercise and fasting activate the same enzyme without a drug, with far larger and better-replicated outcome evidence, though they demand time and adherence that a tablet does not.

• Competing approach — indirect activators: Metformin, berberine and Gynostemma pentaphyllum extract raise AMPK activity indirectly. They are cheap and available, with decades of human data for metformin, but they do not reproduce the direct-activation profile.

• Competing approach — appetite-based obesity drugs: GLP-1 receptor agonists produce far larger weight loss but remove lean mass alongside fat. The developer’s stated rationale for ATX-304 is muscle-sparing fat loss, tested so far only in mice.

• Who popularized each approach: Direct pan-AMPK activation was developed by Helena Edlund’s laboratory at Umeå University and Betagenon AB; the longevity framing comes from Cambrian Bio and its founder James Peyer, both of whom hold a financial stake.

• Genetic polymorphisms: No pharmacogenetic data exist. Mechanistically, LKB1/STK11 status governs whether the compound works at all, and PRKAG2 variants would be expected to modify both response and cardiac safety.

• Sex-based differences: Unreported. Rodent studies used male animals almost exclusively, and neither human trial has published results split by sex, so no sex-specific dosing guidance can be derived.

• Age considerations: Preclinical response was largest in aged mice. In humans no age-stratified data exist, and older adults on multiple blood-pressure medications sit closest to the additive hypotension risk.

• Baseline biomarkers: Response tracked with baseline impairment: those with slower baseline microvascular perfusion improved, those without did not. Higher baseline insulin resistance also predicted a larger measured effect.

• Pre-existing conditions: Both trials required either type 2 diabetes on metformin, or obesity with prediabetes. Nothing is known about response in metabolically healthy people, who form much of the interested audience.

Discontinuation & Cycling

• Intended duration: Framed as chronic therapy for cardiometabolic disease, not a course. Longest human exposure to date is 16 weeks, so no evidence supports or refutes indefinite use.

• Withdrawal effects: None reported. Neither trial described rebound in glucose, blood pressure or weight after stopping, though neither followed participants systematically after the last dose.

• Tapering: Not applicable pharmacologically. The long half-life means plasma levels decline gradually over days on their own, producing a self-tapering washout without a stepped reduction schedule.

• Practical washout: Because steady state takes about 14 days to build, full clearance occupies a comparable period. Blood-pressure and glucose effects should be assumed present for at least two weeks after the last dose.

• Cycling: Never studied. No data indicate tolerance develops, so the usual rationale for cycling is absent; conversely, no data show that continuous use preserves the effect either.

• Restarting: Any restart re-enters the same two-week accumulation phase. Judging response or adjusting dose before day 14 of a restart repeats the error the trial data specifically warn against.

Sourcing and Quality

• Regulatory status of supply: ATX-304 has no marketing authorisation anywhere. Legitimate supply exists only through clinical trials; every retail source is a gray-market “research chemical” vendor operating outside pharmaceutical manufacturing regulation.

• Identity and salt form: Vendors list the compound under several codes — O304, O-304, ATX-304 — and salt form is often unstated. A different salt changes the mass of active compound per stated milligram, silently altering dose.

• What to look for: A lot-specific certificate of analysis from an independent laboratory, identity confirmed by mass spectrometry and nuclear magnetic resonance, a stated purity above 98%, and a residual-solvent and heavy-metal screen. Vendor-supplied documents are not independent.

• Third-party testing: No independent analytical survey of retail ATX-304 has been published, and neither ConsumerLab nor Examine covers it. Purchasers who want verification must commission their own testing from an analytical laboratory.

• Formulation: The phase IIa trial used an oral suspension its authors called for replacing (Steneberg et al., 2018), and the developer has since reformulated. Retail powders replicate neither formulation, so absorption from them is unknown.

• Reputable sources: There are none. No compounding pharmacy legitimately compounds an unapproved investigational compound, and no supplement brand carries it. The only defensible source is trial participation.

Practical Considerations

• Time to effect: Slow. Steady-state blood levels took 14 days; glucose fell only between days 21 and 28; fat, lipid and metabolic-rate changes were measured at eight weeks. Nothing meaningful should be expected inside two weeks.

• Common pitfall — dose escalation during the lag: Because nothing happens for two weeks, users escalate. The trial pharmacokinetics show this drives exposure well beyond the studied range before the first dose has taken effect.

• Common pitfall — expecting weight loss: Weight loss was minimal in the human study. The developer attributes this to sub-threshold exposure; either way, weight is not what the human data show changing.

• Common pitfall — unmonitored concomitant use: Adding it to existing antihypertensives, glucose-lowering drugs or thermogenic supplements reproduces exactly the two interaction risks that have human evidence behind them.

• Regulatory status: Investigational everywhere. It is not a dietary supplement, not approved off-label anywhere, and not lawfully sold for human consumption; retail sale relies on “research use only” labelling.

• Cost and accessibility: Trial access is limited to European sites and specific enrolment criteria, so it is effectively inaccessible. Gray-market pricing is unregulated and bears no relation to any assay of content.

• Structural cost bias: Exercise, metformin and generic antihypertensives cost payers almost nothing; a novel branded agent would not. Insurers and national health systems therefore have a systematic incentive to favour the cheap comparators, which shapes trial funding and guideline treatment of this class.

Interaction with Foundational Habits

• Sleep: Direction is most likely none. The compound is peripherally restricted and does not cross the blood-brain barrier in rodents, so no central effect on sleep is expected; neither trial reported sleep disturbance. The one indirect route — raised resting energy expenditure lifting night-time core temperature — has not been measured.

• Nutrition: Direction is potentiating. It pushes glucose into muscle and shifts the liver toward burning fat, compounding the same shift produced by carbohydrate restriction or time-restricted eating — and compounding hypoglycemia risk when fasting is combined with a glucose-lowering drug. Adequate protein matters if muscle-sparing is being relied on; no food-effect study exists.

• Exercise: Direction is uncertain and potentially blunting. The compound reproduces training signals in mice — endurance, stroke volume, lower exercise lactate (Ericsson et al., 2021) — but drugs that pre-activate an exercise pathway can blunt training adaptation, as metformin does. Timing around workouts is untested; the long half-life keeps exposure continuous.

• Stress management: Direction is indirect at most. No cortisol or stress-axis data exist for the compound, and peripheral restriction argues against a direct central effect. The practical consideration runs the other way: heat, dehydration and sleep debt amplify the blood-pressure fall, so periods of high physiological stress are the wrong time to initiate.

Monitoring Protocol & Defining Success

Baseline testing should be completed before the first dose and should establish more than a single snapshot, because the compound’s own effects are slow and modest. Two weeks of home blood-pressure readings, a fasting metabolic panel, a lipid panel, liver and kidney function, and body composition together define the starting point against which any change is judged. An electrocardiogram at baseline addresses the theoretical cardiac hazard raised by PRKAG2 biology. Ongoing monitoring should repeat blood pressure daily for the first four weeks, then weekly; repeat blood panels at 4 weeks, 12 weeks, and then every 3 to 6 months; and repeat imaging-based body composition and liver fat at 6 and 12 months. Nothing should be judged before day 14, since plasma levels have not stabilised until then.

Biomarker	Optimal Functional Range	Why Measure It?	Context/Notes
Fasting plasma glucose	4.4–5.0 mM (79–90 mg/dL)	Primary human-validated effect	Fast 10–12 h; conventional labs call anything under 5.6 mM normal, a wider band than the functional target
Fasting insulin	2–5 mIU/L	Detects insulin resistance before glucose moves	Draw with glucose from the same fasting sample; conventional labs accept anything up to about 25 mIU/L, far above the functional target
HOMA-IR	Below 1.0	The endpoint that improved in the phase IIa trial	Calculated from fasting glucose and insulin, not a separate draw; conventional practice flags insulin resistance only at about 2.5, well above the functional target
HbA1c	4.8–5.2%	Three-month glucose average, filters daily noise	HbA1c is glycated haemoglobin; conventional labs accept up to 5.6%. Non-fasting
Triglycerides	Below 0.9 mM (80 mg/dL)	Fell significantly in the phase 1b study	Fast 12 h and avoid alcohol for 72 h; pairs with the full lipid panel. Conventional labs flag only above 1.7 mM (150 mg/dL)
Adiponectin	Above 10 µg/mL	Rose significantly in the phase 1b study	Rarely ordered; establish an individual baseline, since assay ranges differ between laboratories
ALT	Below 25 U/L (men), below 20 U/L (women)	Liver stress and fat-driven injury	ALT is alanine aminotransferase, a liver enzyme; conventional upper limits near 40 U/L are far looser. Pair with AST (aspartate aminotransferase) and GGT (gamma-glutamyl transferase), two further liver enzymes
Liver fat fraction	Below 3%	The imaging endpoint that changed in humans	Measured by MRI-PDFF; conventional practice calls anything below 5% normal, a looser threshold than the functional target. Ultrasound elastography is a cheaper but less quantitative substitute
Creatinine and eGFR	eGFR above 90 mL/min/1.73 m²	Kidney safety, and the animal-model claim	Avoid heavy protein or creatine loading for 48 h before the draw; conventional labs treat anything above 60 mL/min/1.73 m² as normal
Blood pressure (home, seated and standing)	110–120 / 70–78 mmHg	The clearest human effect and the clearest interaction risk	Morning and evening, same cuff; record standing readings to catch postural drops
Resting heart rate	50–65 bpm	Detects thermogenic drift from uncoupling	bpm is beats per minute; conventional labs accept 60–100 bpm, far wider than the functional target. A wearable overnight average is more reliable than a spot reading
Core body temperature	Within 0.3 °C of personal baseline	The safety signal the drug class is judged on	No fixed target; track deviation from an individual two-week baseline. Same time each morning
Creatine kinase	Below 200 U/L	Muscle integrity, given the muscle-sparing claim	Creatine kinase is a muscle enzyme; avoid measuring within 72 h of hard training
Lactate (fasting)	Below 1.5 mM	Guards against the uncoupling failure mode	No established target on this drug; track change from personal baseline. Rest 15 min before the draw
Body composition	No fat-free mass loss	Distinguishes fat loss from muscle loss	DEXA is dual-energy X-ray absorptiometry; use the same scanner and same hydration state each time

Qualitative markers worth tracking alongside the laboratory work:

• Standing tolerance — lightheadedness on rising is the earliest sign of the additive blood-pressure effect
• Heat tolerance — new discomfort in saunas, hot showers or warm weather
• Endurance session perceived effort at a fixed heart rate
• Recovery from heavy resistance training, and whether session loads hold
• Energy stability across the day, and appetite change
• Sleep onset latency and night-time thermal comfort

Emerging Research

• REWIRE-1: The developer’s planned phase 2 study of higher exposures, aimed at muscle function and lipid metabolism — the exercise-mimetic claim tested directly in people for the first time. No registration number (NCT ID) exists: a ClinicalTrials.gov search on 27 August 2026 returned nothing for ATX-304, O304 or the sponsor.

• REWIRE-2: The planned phase 2 proof-of-concept study in weight loss for people with obesity, testing whether the muscle-sparing fat loss seen in mice reproduces in humans at higher exposure. Also unregistered on ClinicalTrials.gov as of 27 August 2026.

• Phase 1b full publication: The 23-participant results exist only as conference abstracts 1782-P and 1788-P and a company release. Peer review will show whether the reported effect sizes and adverse-event profile survive scrutiny.

• Combination with appetite-suppressing drugs: The developer presented mouse data at ENDO 2025 on ATX-304 alone, with semaglutide, and after semaglutide withdrawal, targeting muscle preservation (Amplifier Therapeutics, 2025). Human combination trials have not been announced.

• Liver disease: Preclinical work in progressive fatty liver disease shows reduced steatosis and fibrosis with a shift toward fat burning (Holm et al., 2025). This is the indication with the strongest animal-to-human bridge so far, given the phase 1b liver fat signal.

• Evidence that could weaken the case — mechanism dispute: If the compound’s effects prove to be driven mainly by mitochondrial uncoupling rather than phosphatase blockade (Li et al., 2025), its safety margin at higher doses narrows and it inherits the hazards of a drug class with a lethal history.

• Evidence that could weaken the case — AMPK and cancer: Sustained AMPK activation helps tumour cells survive metabolic stress in some settings, a dual role reviewed by Penugurti et al., 2024. Long-term activation in people with undiagnosed malignancy has never been studied.

• Evidence that could weaken the case — no mammalian lifespan data: Lifespan extension rests on nematodes (Li et al., 2025). No rodent lifespan study has been registered or reported, and the compound is not listed as an intervention in any published aging-intervention programme.

Conclusion

ATX-304 is an experimental oral medication that switches on the enzyme cells use to sense low fuel — the same switch thrown by hard exercise and by going without food. It is the first compound of its kind to reach people and produce measurable metabolic change. In two short studies it lowered fasting blood sugar and blood pressure, reduced fat in the liver and around the organs, improved blood fats, and raised the energy burned at rest by a small amount. Its safety record so far shows mild complaints at the same rate as placebo, with no rise in body temperature or heart rate.

That record is thin. No result has been repeated in a second study, the longest anyone has taken it is about four months, and almost every animal finding that makes it interesting for long life — better endurance, healthier aging kidneys, longer life in worms — has no human counterpart. Most of the research, including all of the human work, was funded and reported by the company that owns the compound, and the strongest results have been announced at meetings rather than published for review. Independent laboratories dispute how it actually works, and that dispute matters because one of the two explanations carries a hazard the other does not. It cannot be bought in a form anyone can verify.

Top - Benefits - Risks - Protocol