Cardarine (GW-501516)

Chemical identity

Three trade-style names and three catalogue codes all point at the same molecule. Cardarine, Endurobol and GW501516 resolve to a single PubChem record, CID 9803963, which carries CAS 317318-70-0, the formula C21H18F3NO3S2, a molecular weight of 453.5 g/mol, InChIKey YDBLKRPLXZNVNB-UHFFFAOYSA-N, UNII 7I2HA1NU22 and the ChEMBL identifier CHEMBL38943. Under IUPAC rules the compound is 2-[2-methyl-4-[[4-methyl-2-[4-(trifluoromethyl)phenyl]-1,3-thiazol-5-yl]methylsulfanyl]phenoxy]acetic acid. GW-501516, GW1516 and GSK-516 sit in the synonym list attached to that same CID, so the codes and the trade-style names are not distinct materials.

Structurally this is a small phenoxyacetic acid joined through a thioether to a thiazole ring carrying a trifluoromethylphenyl group. It is neither a peptide nor a steroid, and it has no reported androgen-receptor pharmacology. The molecule is nonetheless sold and discussed inside the selective androgen receptor modulator category, and the United States Anti-Doping Agency addresses that misclassification directly in its athlete guidance, describing the compound as a PPAR-delta agonist and not a SARM. Oliver and colleagues published the first description of the molecule in 2001, from GlaxoSmithKline's metabolic disease and nuclear receptor discovery groups.

The published human record is four trials, all measuring lipids

Sprecher and colleagues reported the first administration to people in 2007. Healthy volunteers were allocated placebo (n=6) or GW501516 at 2.5 mg (n=9) or 10 mg (n=9), orally once daily for two weeks, hospitalised and sedentary throughout. HDL cholesterol rose in both treated groups, at P=0.004 for 2.5 mg and P<0.001 for 10 mg, measured against a fall of 11.5 plus or minus 1.6 per cent in the placebo group. Serum triglyceride trended downward at 10 mg without reaching significance (P=0.08), while triglyceride clearance after a fat-feeding challenge improved on drug (P=0.02).

Risérus and colleagues ran a double-blind randomised study in 2008 with three parallel groups of six moderately overweight men, giving 10 mg daily for two weeks against a PPAR-alpha comparator and placebo. Reported reductions were 30 per cent for fasting triglycerides, 26 per cent for apolipoprotein B, 23 per cent for LDL cholesterol and 11 per cent for insulin, with liver fat content down 20 per cent and urinary isoprostanes down 30 per cent. HDL cholesterol was unchanged in that trial, a null result where the other three reported increases. This trial is held separately from the other three; the figures are not averaged. Christodoulides and colleagues later reported plasma FGF21 rose 32 per cent in a six-subject delta-agonist group; matching dose, duration, comparator, group size and authorship indicate the same cohort, though neither report states this.

Ooi and colleagues published a randomised crossover study in 2011 in thirteen dyslipidaemic men with central obesity, six weeks on 2.5 mg daily with a two-week washout, tracking apolipoprotein kinetics with stable isotopes. The largest trial is Olson and colleagues in 2012: 268 patients with HDL cholesterol below 1.16 mmol/L given 2.5, 5.0 or 10.0 mg or placebo for twelve weeks. Reported changes at 10 mg were HDL cholesterol up to 16.9 per cent higher and apoA-I 6.6 per cent higher, with LDL cholesterol down 7.3 per cent, triglycerides down 16.9 per cent, apoB down 14.9 per cent and free fatty acids down 19.4 per cent.

Four studies appear on ClinicalTrials.gov under this intervention and none has results posted. The 2012 paper prints the accession NCT00258899, which the registry returns as not found. The nearest existing registration is NCT00158899, a GlaxoSmithKline phase 2 dose-response trial in the same population, 424 enrolled, completed June 2006. That record carries two different treatment durations in two different fields: the official title says eight weeks, while the primary outcome measure reads change from baseline in fasting plasma HDL cholesterol at the end of twelve weeks of double-blind treatment, with a registered timeframe of twelve weeks. The record contradicts itself, which weakens rather than supports treating it as a study separate from Olson 2012.

A second registration, NCT00388180, covers 71 overweight and obese subjects for twelve weeks, with body fat levels as its registered primary outcome measure, and no publication reporting it was located. A fourth, NCT00318617, is a terminated GlaxoSmithKline study in 30 healthy male subjects over four weeks, measuring energy-related chemicals and heart contraction by MRI. An intervention-name query misses it because the registry spells the drug GW510516X, with two digits transposed. No human study has published an endurance, exercise-performance or body-composition result. Registered body composition and registered cardiac energetics both remain unreported, and twelve weeks is the longest published human exposure.

Development stopped on data never published in full

Mackenzie and Lione's 2013 review in Life Sciences records the sequence plainly: after clinical trials in dyslipidaemia, GlaxoSmithKline announced that further research would be discontinued because preclinical work in rodents linked the compound to widespread tumour development. The primary data behind that decision are two abstracts presented to the Society of Toxicology in 2009, one on rats from Geiger and colleagues and one on mice from Newsholme and colleagues. Both come from the sponsor's safety assessment groups in King of Prussia and Ware, with the studies run at Huntingdon Life Sciences.

Neither abstract has been published as a full peer-reviewed paper and neither is indexed in PubMed. Mitchell and Bishop-Bailey state the non-publication directly in their 2019 letter in Pulmonary Circulation. That same letter prints, in its own reference list, an archive URL for the meeting's abstract volume. The volume downloads intact and both abstracts sit on one page, numbered 895 and 896. Species, strain, route, duration, dose levels and organ-by-organ tumour findings are stated in them. This material was never peer-reviewed, but it is readable, and it is recorded in the ledger below.

The rat study dosed Han Wistar rats by oral gavage for 104 weeks. Males received 0, 5, 15 or 30 mg/kg/day for the first six weeks and 0, 5, 20 or 40 mg/kg/day for the remainder; females received 0, 3, 10 or 20 mg/kg/day throughout. Neoplastic findings occurred in multiple tissues at all doses: liver, urinary bladder, thyroid, tongue, stomach, skin, Harderian gland, testes, ovary and uterus. The heaviest finding is in the females. Mortality was increased at every dose tested, and uterine endometrial adenocarcinoma contributed to death in a high proportion of those animals, with uterine polyp and endometrial adenocarcinoma recorded from 3 mg/kg/day, the lowest dose given to females.

The mouse study dosed CD1 mice by oral gavage for 104 weeks at 0, 10, 30, 60 or 80 mg/kg/day, with survival decreased at 30 mg/kg/day and above. Neoplasms considered test-article related occurred in liver and stomach, with combined squamous cell tumours at all doses. One figure circulates alongside these studies that the abstracts do not contain: an incidence reaching ninety per cent. That number is recorded below as untraced. The dose levels can now be read, so the competing arguments about safety margins can be checked against real numbers instead of asserted against missing ones.

Where the endurance figures come from

Narkar and colleagues published the treadmill work in Cell in 2008. Adult male C57BL/6J mice received GW1516 at 5 mg per kg per day for four weeks, orally, in cohorts of nine, of which six per group were put through the running test at week five. The remaining three per group were held back from the treadmill so that muscle changes could be separated from the acute run. In sedentary animals the result was null: the authors describe running performance as unchanged by GW1516 treatment. That sentence sits in the same paper that produced the figure now attached to the compound everywhere.

Pairing the identical dose and duration with four weeks of treadmill training changed the outcome. The paper reports running time increased by 68 per cent and running distance by 70 per cent relative to trained controls. The circulating claim of a seventy per cent endurance gain traces to that arm and carries the training requirement with it, which the claim as repeated does not. A separate figure of roughly 44 per cent also circulates under this compound's name; in the paper it belongs to AICAR, an AMPK activator given at 500 mg per kg per day intraperitoneally for four weeks and tested in sedentary animals.

Wang and colleagues published the doubling result in PLoS Biology in 2004, and the mouse that ran up to twice the distance of its wild-type littermate was transgenic, expressing an activated form of PPAR-delta in skeletal muscle. Drug treatment of wild-type mice in that paper elicited a similar type I fibre gene expression profile; the running distance was a property of the genotype, not the compound. That paper carries a correction printed in January 2005 which adds a competing-interest disclosure for the senior author, naming a consulting relationship with a company that received milestone payments from GlaxoSmithKline for the development of GW501516.

The preclinical metabolic findings

Oliver and colleagues dosed male obese rhesus monkeys, mean age 12.9 years, orally at 0.1 mg per kg for four weeks and then 0.3, 1.0 and 3.0 mg per kg over sequential four-week periods. At the top dose they reported HDL cholesterol 79 per cent above baseline, triglycerides 56 per cent lower, LDL particle number 36 per cent lower and fasting insulin 48 per cent lower. The design escalates dose within each animal against its own vehicle baseline, with no parallel control group. Six animals entered the study and five reached the 3.0 mg per kg step, one having been withdrawn as a precaution after showing reduced food intake at 1.0 mg per kg.

Tanaka and colleagues reported in 2003 that the compound induced fatty acid beta-oxidation genes in rat L6 myotubes and in mouse skeletal muscle, that high-fat-fed mice showed attenuated diet-induced obesity and insulin resistance, and that plasma glucose and insulin fell in genetically obese ob/ob mice. Krämer and colleagues worked in primary cultured human muscle cells in 2007 and reported AMPK phosphorylation alongside increased fatty acid transport, fatty acid oxidation and glucose uptake; small interfering RNA showed the glucose-uptake effect required AMPK but not PPAR-delta itself. Those are cell and rodent findings and were not tested against a performance endpoint in a person.

Tumour biology is contested in both directions

Gupta and colleagues treated Apc-Min mice, which are predisposed to intestinal polyposis, with GW501516 and reported a significant increase in the number and size of intestinal polyps, the most prominent effect being a fivefold rise in polyps larger than 2 mm. That 2004 report in Nature Medicine is the clearest published tumour-promoting result for this specific ligand. The polyps were counted in a genetically sensitised strain over a short period, which is a promotion model and not a carcinogenicity design.

Evidence pointing the other way exists and uses different tools. Marin and colleagues reported in 2006 that ligand activation of PPAR-beta inhibited colon polyp multiplicity in azoxymethane-treated wild-type mice and produced no such effect in receptor-null animals; the ligand used there was GW0742, a related compound and not this one. Foreman and colleagues reported in 2011 that PPAR-beta/delta mRNA and protein expression was lower, and cyclin D1 protein higher, in human colon adenocarcinomas than in matched non-transformed tissue. The cell-line arms of that study also used GW0742, the same ligand substitution that applies to Marin.

The sponsor's own 104-week mouse study speaks directly to this question. Its abstract states that the results demonstrate an increase in proliferation in certain epithelial cell populations but do not support a role for PPAR-delta in colon carcinogenesis. Both 2009 abstracts add that some of the tumour types observed have not been reported with either PPAR-alpha or PPAR-gamma agonists, and may reflect tumour promotion mediated through PPAR-delta agonism. The 104-week data therefore cut two ways at once: against a colon mechanism, and toward receptor-mediated promotion at squamous, hepatic and endocrine sites.

Regulatory and anti-doping status, and material in circulation

The 2026 WADA Prohibited List names the compound explicitly at S4.4.1, under metabolic modulators, listing peroxisome proliferator-activated receptor delta agonists with the full chemical name followed by GW1516 and GW501516 in parentheses, alongside AMPK activators and Rev-erb-alpha agonists. Substances in class S4.4 are prohibited at all times, in and out of competition, and are non-Specified Substances. WADA issued a public alert in March 2013 warning anti-doping organisations about the compound, stating that development had been terminated on serious preclinical toxicities and that clinical approval had not been and would not be given.

No regulator has approved the molecule as a medicine in any jurisdiction, and the sponsor ended development in 2006 and 2007. USADA's guidance states the compound is not legally permitted in any medication, supplement or food, which makes a therapeutic use exemption unavailable. That same guidance also states the compound has not undergone human studies, which does not match the four published trials described above. The discrepancy stands unresolved here. The trials exist and are indexed, and the anti-doping guidance is the more widely read document.

Material bearing the name reaches consumers through supplement channels. A Korean Ministry of Food and Drug Safety laboratory screened 60 dietary supplements bought online and directly from international vendors in 2020 and reported GW501516 present at 3.49 per cent by mass in one product, within a set of 20 products found to contain SARMs at high concentrations. Breuer and colleagues reported in 2024 on two adverse analytical findings investigated as possible transfer through intimate contact; screening 361 clinical ejaculate specimens returned one containing GW1516 at roughly 48 ng/mL together with its sulfoxide and sulfone metabolites, while the athletes' urine samples carried those metabolites at 4 to 14 pg/mL.

What is not known

The decisive toxicology was never peer-reviewed. Tumours in rats and mice after 104 weeks of dosing are documented only as two 2009 conference abstracts from the sponsor. Those abstracts are readable in the archived meeting volume and state species, strain, route, duration, dose levels and tumour sites, but they report no incidence figures, no group sizes and no no-observed-effect level, and no full study report has ever appeared. On the human side, the published exposure ceiling is twelve weeks in 268 patients, all endpoints lipid or lipoprotein; there is no published human pharmacokinetics, no half-life, no dose-proportionality data, no hepatic or renal safety series, no endocrine measurement, and no results posted to ClinicalTrials.gov for any of the four registrations. No human study has published a result for exercise capacity, endurance, body composition or muscle fibre type, which is the entire basis on which the compound circulates; body composition was registered as the primary outcome of NCT00388180 and cardiac energetics as the primary outcome of the terminated NCT00318617, and neither was ever reported. The receptor's role in tumour biology is unsettled two decades after the question was raised: one ligand accelerated adenoma growth in Apc-Min mice, a related ligand attenuated chemically induced colon carcinogenesis in wild-type mice, and the sponsor's own 104-week mouse study found no support for a colon mechanism while recording squamous and hepatic tumours at every dose. Nothing here establishes carcinogenicity or its absence in a person, and no regulator anywhere has approved the molecule for any use.

Questions