Meldonium (Mildronate)

Identity, and what one substituted atom does

Meldonium has one compound record and one obvious trap, and the trap is hydration. The record is CID 123868: CAS 76144-81-5, molecular formula C6H14N2O2, molecular weight 146.19 g/mol, InChIKey PVBQYTCFVWZSJK-UHFFFAOYSA-N, UNII 73H7UDN6EC, cross-referenced to ChEMBL2104708 and ChEBI 131843, with the IUPAC name given as 3-[(trimethylazaniumyl)amino]propanoate. The World Health Organization assigns the ATC code C01EB22, filed under other cardiac preparations. At 146 daltons this is a zwitterionic small molecule and not a peptide, whatever company it keeps on research-chemical listings. Those fields all describe the anhydrous form. The same synonym record carries dihydrate entries too, including a European Pharmacopoeia reference standard, and PubChem indexes meldonium dihydrate separately as CID 6918082, C6H18N2O4, 182.22 g/mol — a 25 per cent difference in mass on anything weighed.

Little separates meldonium from the molecule it displaces. Gamma-butyrobetaine, the immediate biosynthetic precursor of L-carnitine, is 4-(trimethylammonio)butanoate. Meldonium is the aza analogue: the carbon adjacent to the quaternary nitrogen is replaced by a second nitrogen, producing 3-(2,2,2-trimethylhydrazinium)propionate. One substitution converts a substrate into a competitor. The compound mimics an endogenous metabolite, competing for the same enzyme and the same transporter that handle the real thing.

Meldonium was developed in 1974 by Ivars Kalvins at the Latvian Institute of Organic Synthesis, initially to improve fertility and growth in livestock, and was marketed as Mildronate; that account is Malm and Svensson's 2026 review (PMID 42344756). A query against Drugs@FDA on both names returned no matches, and no centralised European marketing authorisation was located. Published sources describe its European status three different ways. Schobersberger and colleagues' 2017 review states that it is registered in most Baltic countries. A 2024 review in Cureus states that it holds no marketing authorisation within the European Union and is registered only in ex-Soviet republics. Malm and Svensson state that it is authorised through the European decentralised procedure in selected member states including Latvia, Lithuania and Poland. Three Baltic states are European Union members, so those statements cannot all be read literally, and this record does not attempt to reconcile them.

Carnitine subtraction, and an inhibitor its own chemists called poor

Two targets account for the documented pharmacology. Gamma-butyrobetaine hydroxylase catalyses the final step of L-carnitine biosynthesis, and the organic cation transporter OCTN2 carries carnitine into tissue. Dambrova and colleagues' 2016 review in Pharmacological Research describes meldonium as acting on both, with the consequence that tissue carnitine and long-chain acylcarnitine content fall, mitochondrial fatty-acid oxidation declines, and glucose metabolism takes up more of the load. Because oxidising glucose consumes less oxygen per unit of ATP than oxidising fatty acids, the same authors frame the compound as an anti-ischaemic agent rather than a stimulant of any kind.

Measured carnitine depletion in people comes from a small Latvian study. Liepinsh and colleagues gave healthy non-vegetarian volunteers 500 mg orally twice daily and sampled plasma and urine weekly. After four weeks average plasma L-carnitine had fallen by 18 per cent, plasma γ-butyrobetaine had roughly doubled, urinary excretion of both had risen, and plasma meldonium averaged 20 µM. Dietary meat intake partially offset the carnitine reduction, which the authors monitored deliberately. The indexed abstract does not state how many volunteers took part, and this record does not supply a number the source did not report.

The chemistry literature is blunter about potency. Tars and colleagues, working at the Biomedical Research and Study Centre in Riga and publishing in the Journal of Medicinal Chemistry in 2014, describe meldonium as a relatively poor inhibitor of γ-butyrobetaine hydroxylase that requires high daily doses. Their paper synthesised 51 analogues and reported new inhibitors roughly two orders of magnitude more potent, with six co-crystal structures. The group that has characterised the target most closely is therefore also the group reporting the weakest potency for the compound in circulation.

The registered human record is seven studies long

A ClinicalTrials.gov intervention search across meldonium and mildronate returned seven registered studies in total, the same seven under either name. Two are phase 2 acute ischaemic stroke trials from Xijing Hospital in China, one completed with 227 participants and one of unknown status with a planned 240. Two are phase 4 Russian studies of a combination product in post-COVID asthenia, at 30 and 160 participants. One is a completed phase 1/2 study in 60 patients with metastatic renal cell carcinoma and treatment-associated fatigue. One is a phase 1 study of 48 participants at Riga Stradins University concerned with EPA and DHA levels. The last, enrolling by invitation with a 2033 completion date, concerns medically supervised performance-enhancing substances in elite athletes.

The largest efficacy trial in the literature is absent from that list. MILSS I, published in Medicina (Kaunas) in 2011, randomised 512 patients with chronic coronary heart disease across 72 centres in four countries to placebo or one of four meldonium arms, with change in bicycle-ergometry exercise time at 12 weeks as the primary endpoint. Mean change was −2.12±108.45 s at 100 mg daily and 11.48±62.03 s at 300 mg, against −7.10±81.78 s on placebo; neither differed significantly from placebo. The 1000 mg arm changed by 35.18±53.29 s, reported at P=0.002. The 3000 mg arm produced a smaller change than the 1000 mg arm.

The dose-response rises and then falls. A figure measured at one exposure therefore cannot be carried to another in either direction. The standard deviations run wide as well: in the 100 mg arm the spread is fifty times the mean change. Of the four active arms, only one separated from placebo at the trial's 12-week primary endpoint, and the trial is not registered on ClinicalTrials.gov.

Zhu and colleagues ran the registered stroke trial, NCT01831011, and published it in Clinical Drug Investigation in 2013. It was randomised, double-blind and active-controlled rather than placebo-controlled: 113 patients received meldonium injection and 114 received cinepazide injection, both on a background of aspirin. At three months there was no significant difference on the primary endpoint, modified Rankin Scale scores of 0–1 or 0–2, at P=0.52 and P=0.07. Secondary endpoints on the NIHSS and the Barthel Index at 15 days showed no significant differences either. The authors' stated conclusion was equivalence to the active comparator, not superiority to nothing.

The athlete evidence is one non-randomised trial that pointed the wrong way

Malm and Svensson's 2026 narrative review in Frontiers in Sports and Active Living identifies a single placebo-controlled trial of meldonium in healthy athletes: a Russian-language study by Goloborodko and Razinkin, published in 2021, in 20 elite biathletes and cross-country skiers, ten men and ten women, mean age 20.1±0.6 years, over 15 days. Allocation was by exercise-test performance rather than randomisation, leaving the meldonium arm with a baseline VO2max near 66 mL/min/kg against 56 in the placebo arm. The reviewers read that gap as a probable sex imbalance that precludes between-group comparison. Neither the review nor its reference entry for that study states a dose or a route of administration.

Across the 15 days the meldonium group's VO2max, VO2 at anaerobic threshold, oxygen pulse and maximum load declined, while the placebo group's did not; respiratory exchange ratio rose, consistent with a shift towards carbohydrate oxidation but not with improved maximal performance. Blood lactate five minutes after maximal exercise did not differ between groups. Self-rated mood on the SAN scale fell significantly in the meldonium arm and rose in the placebo arm. The reviewers flag physiologically implausible changes in the placebo arm — simultaneous reductions in resting and maximal heart rate — as grounds to distrust the measurements in both arms.

The other human study in this space is uncontrolled. Chainikov and colleagues, publishing in the Ural Medical Journal in 2015, gave 500 mg twice daily for 21 days to 25 male youth ice-hockey players and followed them to day 42 using Russian neuropsychological batteries. Three parameters had higher means at day 42, two were lower and three unchanged. The review reports the comfort outcome two incompatible ways: its summary table states that the number of players rating themselves at high comfort fell from seven to one, while its running text states that the proportion fell from 12 per cent at baseline to 0 per cent at day 42. With 25 players those two cannot both hold, and the Russian primary was not retrieved to adjudicate. There was no placebo arm and no control group. Malm and Svensson's summary of the field is that the reputation rests on anecdote, theoretical mechanism, and extrapolation from ischaemic patients and animal models.

An excretion profile that broke result management

Prevalence was established before the ban, by measurement. Görgens and colleagues at the German Sport University Cologne screened 8,320 routine doping control urine samples over six months and confirmed 182 positives, 2.2 per cent, at concentrations from 0.1 to 1,428 µg/mL. Findings split 135 in-competition to 47 out-of-competition and 97 male to 85 female. The screening was carried out while the substance sat on WADA's monitoring programme and was not yet prohibited.

Meldonium entered the Prohibited List on 1 January 2016. In the 2026 list it sits at S4.4.3, metabolic modulators, prohibited at all times in and out of competition, and that list states that substances in classes S4.3 and S4.4 are non-Specified Substances — the classification carrying the longer default sanction and less latitude at a hearing. WADA's own 2016 Anti-Doping Testing Figures report records 515 meldonium occurrences among adverse analytical findings that year in Table 21, amounting to 71 per cent of everything reported in the S4 class. Press summaries of the same report circulated a figure of 497; the table reads 515.

Then the excretion problem arrived. Görgens and colleagues determined urinary profiles in healthy volunteers and found detection windows of up to 65 days after a single 500 mg oral dose in five volunteers, and up to 117 days after 500 mg twice daily for six days in five more, at a limit of quantification of 10 ng/mL. Rabin and colleagues, writing from WADA and Russia's Federal Medical Biological Agency, gave 32 healthy volunteers 1.0 or 2.0 g orally daily for three weeks and reported that steady state took several days to reach and that elimination ran over several months.

Liepinsh and colleagues offered a mechanism for that in mice. Because meldonium, carnitine and γ-butyrobetaine compete for OCTN2-mediated transport, the drug accumulates in muscle during treatment and then leaves by relatively slow diffusion, which the authors identify as the determinant of the unusually long elimination period. In April 2016 WADA published a statement acknowledging that it could not establish how quickly the substance clears, and result-management guidance built on urinary concentration bands of 1 and 15 µg/mL and a cutoff date of 1 March 2016 followed. Sanctions already imposed were reopened.

A dietary residue route also exists and has been measured. A veterinary preparation licensed for livestock in some jurisdictions dissociates to meldonium, and residues have been detected in milk. Guddat and colleagues gave three volunteers 100 mL of milk spiked at 500 ng/mL and recorded peak urinary concentrations of 7.5 ng/mL after a single dose and 18.6 ng/mL after five consecutive days (PMID 34448364). Those peaks sit roughly two orders of magnitude below the lower of the two result-management bands.

Safety, as far as it has been characterised

Short-term tolerability has been examined once under a modern protocol. Zhao and colleagues ran a randomised, open-label phase 1 study of single escalating and multiple intravenous doses of 250, 500 and 750 mg in healthy Chinese subjects, reporting linear pharmacokinetics across that range, minor accumulation on repeat dosing, no gender difference, primarily renal excretion and no serious adverse events; the indexed abstract does not state how many subjects took part. In the 227-patient stroke trial the incidence of serious adverse events was similar between the meldonium and cinepazide arms. Neither study was designed to detect anything beyond its own duration.

One animal signal points at the liver. Bjørndal and colleagues fed 40 C57BL/6 mice high-carbohydrate diets containing meldonium, tetradecylthioacetic acid, or both for 21 days, confirmed carnitine and acylcarnitine depletion in plasma, and reported meldonium-induced hepatic triacylglycerol accumulation associated with higher estimated delta-6 desaturase activity and reduced elongation of n-6 polyunsaturated fatty acids (PMID 39957749). Co-treatment cut the triacylglycerol rise by 80 per cent. Malm and Svensson separately note that the trimethylhydrazinium moiety belongs to a chemical class associated with hepatotoxicity and possible genotoxicity, and that long-term toxicology to contemporary international standards is scarce.

Human pharmacokinetics were characterised by Peng and colleagues in 2010, by intravenous route. Elimination half-life after single doses of 250, 500 and 1,000 mg was 5.56±1.55, 6.46±1.07 and 6.55±1.17 hours; on 500 mg twice daily it rose to 15.34±3.14 hours with a Cmax of 25.50±3.63 µg/mL, which the authors read as accumulation. Those figures describe plasma after injection. They do not describe the months-long tissue clearance the anti-doping excretion studies measured.

What is not known

Nothing in the published record establishes that meldonium improves physical performance in healthy people. The single placebo-controlled trial in athletes was non-randomised, ran 15 days, allocated by baseline performance, and reported declines in VO2max, oxygen pulse and mood in the treated arm; it is published in Russian in a journal with no PubMed index, and neither its dose nor its route of administration is recorded in the review that reanalysed it. No dose-ranging, no crossover and no adequately powered randomised trial in trained or untrained healthy adults exists in any language that could be retrieved. The mechanism is characterised only as far as two targets and their immediate metabolic consequences: no study has shown that carnitine lowering causes any performance or clinical outcome, and the published work establishes only that the two occur together. Human pharmacokinetics rest on two intravenous studies whose sample sizes are absent from the indexed abstracts, and no absolute oral bioavailability figure could be traced to any primary source. Total body clearance remains imprecisely defined — the WADA-authored excretion study reports elimination over several months without a terminal half-life — which is why result management for the 2016 findings was built on concentration bands and a calendar cutoff. Long-term toxicology to contemporary international standards was not located; the hepatic steatosis signal comes from a 21-day mouse study, and reproductive, developmental, carcinogenicity and chronic-exposure data were not found in any species. Legal status varies by jurisdiction, the compound holds no United States approval, and its European registration position is described three incompatible ways by the reviews that mention it.

Questions