AICAR (Acadesine)

Chemical identity

The name does not resolve to the molecule that gets administered. A PubChem query for AICAR returns CID 65110: 5-aminoimidazole-4-carboxamide ribonucleotide, CAS 3031-94-5, formula C9H15N4O8P, molecular weight 338.21 g/mol, also indexed as ZMP. That is the phosphorylated intracellular species. Every study described on this page administered the dephosphorylated nucleoside, acadesine, which PubChem holds separately at CID 17513: CAS 2627-69-2, formula C9H14N4O5, molecular weight 258.23 g/mol, InChIKey RTRQQBHATOEIAF-UUOKFMHZSA-N. The FDA Global Substance Registration System returns UNII 53IEF47846 for acadesine and carries the same CAS, the same PubChem CID, and ChEMBL identifier CHEMBL1551724. The nucleoside enters cells and is phosphorylated by adenosine kinase to the nucleotide, and it is the nucleotide that resembles AMP closely enough to act at AMP-activated protein kinase.

Catalogue and development codes attached to CID 17513 include AICA riboside, Z-riboside, Arasine, ARA-100, GP-1-110, NSC-105823, SCH-900395 and ATH-001. All denote one nucleoside. It is not a peptide, and product pages that file it under that heading have the chemistry wrong: there is no amide backbone, only an imidazole carboxamide joined to a ribose. The molecule is also endogenous. AICAR the ribonucleotide is the penultimate intermediate of de novo purine biosynthesis and the substrate of the bifunctional enzyme ATIC, which means a measurable concentration circulates in people who have never taken anything.

The published human record is cardiac surgery

Five randomised, placebo-controlled, double-blind trials in coronary artery bypass graft surgery ran through the 1990s. Mangano pooled individual patient data from all five in JAMA in 1997, across 81 medical centres in the United States, Canada and Europe: 4,043 patients evaluable for efficacy, 2,031 on placebo and 2,012 on acadesine at 0.1 mg per kg per minute by intravenous infusion for seven continuous hours and also in the cardioplegia solution. Perioperative myocardial infarction was 27 per cent lower on drug (odds ratio 0.69, 95% CI 0.51 to 0.95, P=.02), cardiac death through postoperative day 4 was 50 per cent lower (OR 0.52, 0.27 to 0.98, P=.04), and the combined outcome 26 per cent lower (OR 0.73, 0.57 to 0.93, P=.01). Cerebrovascular accident was not significantly reduced.

A 2006 report in the Journal of the American College of Cardiology extended the follow-up. At 54 institutions, 2,698 patients were randomised, 1,346 to placebo and 1,352 to acadesine on the same regimen. Perioperative myocardial infarction occurred in 100 patients, and among those, two-year mortality was 27.8 per cent (15 of 54) on placebo against 6.5 per cent (3 of 46) on acadesine, P=0.006. The retrieved abstract does not state how that cohort relates to the five trials pooled in 1997. The first author, Mangano, has a financial position on the record: a 2008 Drugs in R&D profile states that he founded PeriCor Therapeutics, which took a worldwide sublicence to acadesine in January 2005, and that he had acquired that licence from Metabasis in November 2000.

RED-CABG tested the finding prospectively and did not reproduce it. Newman and colleagues randomised intermediate- to high-risk patients 1:1 at 300 sites in seven countries, median age 66, to the same seven-hour infusion or placebo, with a primary composite of all-cause mortality, non-fatal stroke, or mechanical support for severe left ventricular dysfunction through postoperative day 28. A prespecified futility analysis stopped the trial at 3,080 of a projected 7,500. The primary outcome occurred in 75 of 1,493 on placebo (5.0 per cent) and 76 of 1,493 on acadesine (5.1 per cent), odds ratio 1.01, 95% CI 0.73 to 1.41, with no differences in key secondary endpoints. Results posted to the registry record serious adverse events in 327 of 1,442 against 329 of 1,457.

Where the forty-four per cent came from

Narkar and colleagues published the endurance work in Cell in 2008, from the Salk Institute. Male C57Bl/6J mice, eight weeks old, received AICAR at 500 mg per kg per day intraperitoneally for four weeks with no exercise training, and were then put through a treadmill test. The paper reports that treated mice ran roughly 23 per cent longer and roughly 44 per cent further than vehicle-treated animals, with 15 to 20 mice per group in the running experiments. A separate six-day arm at 250 mg per kg per day, also intraperitoneal, supplied the gene-expression result: global analysis of quadriceps found 32 genes linked to oxidative metabolism up-regulated by drug alone. Epididymal fat mass to body weight ratio fell and oxygen consumption rose, without a change in body weight.

One sentence in that paper's abstract has travelled further than its methods. It calls the compound the orally active AMPK agonist. The methods administer AICAR only by intraperitoneal injection. The one oral route in the paper is oral gavage of the comparator GW1516 at 5 mg per kg per day; AICAR is never given by that route. That phrase is the origin of the oral-activity claim that recurs across vendor product pages and aggregator sites, and it describes a route the experiment did not use for the molecule it names.

No comparable measurement exists in a person. A ClinicalTrials.gov search across AICAR and acadesine returns ten registrations. Five involve administration of the compound: cardiac surgery, chronic lymphocytic leukaemia, the myelodysplastic and AML study, Lesch-Nyhan disease and type 2 diabetes. The remaining five, in non-small cell lung cancer, glioblastoma, obesity-associated inflammation, a sunitinib contractility study and the AICA-ribosiduria supplementation trial, return on the keyword rather than as administration studies. None of the ten registers an endurance, exercise-capacity, VO2 or body-composition endpoint. The forty-four per cent figure is real, is properly cited, and belongs to sedentary mice given 500 mg per kg per day for four weeks. Healthy men in the only route-comparison study received single doses up to 100 mg per kg, and leukaemia patients single doses up to 315 mg per kg; the difference is in the repetition and the route, not the per-dose magnitude.

What the pharmacokinetics establish about the oral route

Dixon and colleagues published the first human pharmacokinetics in 1991, from Gensia Pharmaceuticals. The study was placebo-controlled and double-blind in healthy men across 10, 25, 50 and 100 mg per kg given both orally and intravenously, with four subjects on active drug and two on placebo at each dose level and a one-week washout between routes. Post-infusion plasma concentrations declined biphasically with a harmonic mean terminal half-life of 1.4 hours. Total plasma clearance was 2.5 L per hour per kg, mean residence time 0.7 hours, steady-state volume of distribution 1.6 L per kg. The drug was not protein bound. Renal clearance was 0.2 L per hour per kg, with 8 per cent of the intravenous dose excreted unchanged. Oral bioavailability in solution was reported as less than 5 per cent.

A radiolabel study from the same group followed in 1993. Four healthy males received 25 mg per kg of 2-14C-acadesine over fifteen minutes. Total plasma radiocarbon declined multiexponentially with an apparent terminal half-life of about a week, while intact acadesine was measurable for only two hours. Uric acid, the end product of human purine metabolism, was the major plasma metabolite and accounted for all plasma radiocarbon at six hours. Acadesine 5'-monophosphate was confined to red cells and represented 30 per cent of total blood radiocarbon at the end of infusion. Two-week recovery accounted for 48 per cent of dose, 44 per cent in urine and 4 per cent in faeces, with 5 per cent of the dose appearing in urine as intact drug. Drew and Kingwell's 2008 review states that the poor oral bioavailability precludes application in insulin resistance or glucose intolerance.

Measured effects in metabolic and haematological settings

Babraj and colleagues infused the compound into three groups in 2009: six healthy young men aged 23 plus or minus 3, eight older men aged 59 plus or minus 4, and eight men with type 2 diabetes aged 62 plus or minus 4. Each received a six-hour 2-deoxyglucose infusion with AICAR at 10 or 20 mg per kg per hour from hour three to hour six, with quadriceps biopsies at 0, 3 and 6 hours. At 10 mg per kg per hour, 2-deoxyglucose uptake rose 2.9-fold in young men, 1.8-fold in older men and 1.6-fold in men with diabetes; at 20 mg per kg per hour it rose 2.5-fold in older men and 2.2-fold in men with diabetes, and the young group was not tested at that dose. At three hours, AMPK activity and AMPK, acetyl-CoA carboxylase and AS160 phosphorylation were unchanged; ERK1/2 phosphorylation rose at both doses and correlated with uptake.

Boon and colleagues gave a continuous intravenous infusion at 0.75 mg per kg per minute to ten male type 2 diabetic patients, mean age 64, and tracked kinetics with stable isotopes and muscle biopsies. Plasma glucose rate of appearance fell while rate of disappearance was unchanged, plasma glucose concentration declined, and both appearance and disappearance rates for non-esterified fatty acids fell. Muscle AMPK phosphorylation was not increased in that study either, though acetyl-CoA carboxylase phosphorylation rose. Both studies therefore recorded a metabolic effect without detecting increased phosphorylation of AMP-activated protein kinase, the enzyme the compound is widely presented as activating, in skeletal muscle. In Babraj the effect and the null were in the same tissue, which is the sharper result; in Boon the measured effect was hepatic glucose output, so the muscle biopsy speaks to a different compartment.

Oncology supplied the repeated-dose safety data. Van Den Neste and colleagues ran an open-label phase 1/2 study in 24 patients with relapsed or refractory chronic lymphocytic leukaemia, given four-hour intravenous infusions: eighteen in Part I at single doses from 50 to 315 mg per kg, six in Part II. The trial established 210 mg per kg as a single dose to be both the maximum tolerated dose and what the authors designated the optimal biological dose. Grade 2 or higher hyperuricaemia occurred commonly, which the authors classed as not clinically significant and which resolved with prophylactic allopurinol; transient anaemia and thrombocytopenia were likewise so classed. Renal impairment was reported, and the event the authors did class as clinically significant was transient infusion-related hypotension. A French phase 1/2 study in higher-risk myelodysplastic syndrome, acute myeloid leukaemia with 20 to 30 per cent marrow blasts and chronic myelomonocytic leukaemia type 2, NCT01813838, terminated with five participants enrolled; the registry records the reason as renal toxicity.

Endogenous production and the detection problem

Because the molecule circulates naturally, anti-doping laboratories cannot treat its presence as a finding. Thomas and colleagues quantified urinary AICAR in doping-control samples from 499 athletes in 2010 and reported a mean of 2,186 ng/mL with a standard deviation of 1,655 ng/mL, differing by gender, sport type, and whether collection was in or out of competition, at a limit of quantification of 100 ng/mL. Sobolevsky and colleagues took a different approach in 2022, measuring the urinary ratio of AICAr to the related purine metabolite SAICAr across 5,517 athlete samples: median 3.3 with a 99th percentile of 9.3 in men, and 4.2 with a 99th percentile of 14 in women.

Carbon isotope ratio measurement is what separates the two sources. Piper and colleagues validated a gas chromatography/combustion/isotope ratio mass spectrometry method in 2014 against a reference population of 63 males and females and reported that a single oral administration remained detectable for more than 40 hours. Wang and colleagues published a two-dimensional liquid chromatography purification in 2024 and administered 3 grams, which the abstract characterises as a low dose, to two Asian male volunteers; the retrieved record does not state the route. Samples exceeded the published threshold within 16 hours, and the authors read the gap against Piper's 40 hours as dose-dependence in the detection window. Equine work by Wong and colleagues across 1,470 post-race urine samples from Australia, France and Hong Kong proposed a 600 ng/mL screening cut-off, and reported roughly 4.5 hours of detection after 2 grams intravenously in one mare.

The 2026 WADA Prohibited List names the compound at S4.4.1, under metabolic modulators, among activators of AMP-activated protein kinase alongside BAM15 and MOTS-c. Substances in class S4.4 are non-Specified Substances and prohibited at all times, in and out of competition. USADA's athlete guidance states that it is not approved for therapeutic use in humans anywhere in the world, is not available as a medication, and that a therapeutic use exemption is therefore unavailable. That page also warns that excessive or mistargeted AMPK activation can cause neurodegeneration or prevent cells from dividing, and attaches no citation to the statement.

The one setting where sustained elevation has been characterised

ATIC catalyses the final two steps of de novo purine synthesis and uses the ribonucleotide as its substrate. Loss of that enzyme produces AICA-ribosiduria, an autosomal recessive condition defined biochemically by accumulation of AICA-riboside in urine. Ramond and colleagues reported three new cases from two independent families in 2020, with long-term follow-up on the single patient described fifteen years earlier. Across those four patients they describe severe-to-profound global neurodevelopmental impairment, severe visual impairment from chorioretinal atrophy, ante- and postnatal growth impairment, and severe scoliosis, with dysmorphic features in all four and frequent early-onset epilepsy that can be pharmacoresistant. The authors write that data in the literature point toward a cytotoxic mechanism of the accumulated AICA-riboside.

Four patients with a congenital enzyme defect are not a model of intermittent administration to an adult, and that literature quantifies no threshold, no exposure-response relationship and no reversibility. It is nonetheless the only published human circumstance in which chronically elevated concentrations of this molecule have been characterised at all, and the direction of the finding is not benign. An interventional study of purine supplementation in AICA-ribosiduria, NCT06845501, is registered as recruiting ten participants.

What is not known

The published human exposure is essentially one shape: a seven-hour intravenous infusion during cardiac surgery, tested in roughly seven thousand patients across two eras and giving opposite answers. Nothing establishes what repeated administration does over weeks. The longest schedule in any registered study was five doses in the leukaemia trial, and a second registered haematology study was terminated at five participants with renal toxicity recorded as the reason; no study has characterised chronic dosing, reproductive or developmental effects, carcinogenicity, or effects in healthy people beyond single-session pharmacokinetics in small groups of men. Exercise capacity, the property the compound is known for, has never been an endpoint in a registered human trial; a ClinicalTrials.gov search across both names returns ten studies and none measures it. Two human studies that did detect a metabolic effect recorded no increase in AMP-activated protein kinase phosphorylation in skeletal muscle; in Babraj the effect and the null were in the same tissue, while in Boon the measured effect was hepatic, so the mechanism in people is not established. The oral route is not unstudied. It was measured: under 5 per cent bioavailability in the only trial to compare routes, which means nothing about oral administration can be inferred from the intravenous record. Endogenous production sets a floor no one can quantify per individual: urinary concentrations in athletes vary by more than an order of magnitude, and there is no published account of what an added exogenous load does to purine turnover over time in a person with normal ATIC function. The only human setting in which sustained elevation has been characterised is a congenital enzyme defect described in four patients, which reports severe outcomes and no dose-response information at all. No regulator has approved the molecule for any use in any jurisdiction.

Questions