GLP-1 Drugs and L-Carnitine

Why this pairing has no name, and is not a product

Most pages in this section of the library exist because a community coined a name for a mixture. This one does not. There is no name here, no marketed combination, and no assembled thing to describe — two substances simply get mentioned in the same conversation, and this page exists to explain why and to say what the published record does and does not contain about the connection.

Nothing in the published literature proposes the pairing. A PubMed search on 1 September 2026 for carnitine AND semaglutide returned 5 records, for carnitine AND tirzepatide returned 0, and for carnitine AND retatrutide returned 1 — a metabolite-profiling paper in which acylcarnitines are analytes rather than anything anybody administered. A ClinicalTrials.gov API v2 query on the same day for carnitine with semaglutide as interventions returned a total count of 0, and the same query with tirzepatide returned 0.

The conversation that does exist is about a documented problem and an undemonstrated answer to it. Weight loss driven by these drugs takes lean mass along with fat mass,^[2] carnitine is involved in fatty-acid transport into mitochondria, and someone put those two sentences next to each other. That is the whole provenance, and it is why nothing further down this page reads like an instruction: an amount, an order or an interval would have to be invented from nothing.

What "GLP-1" means here, and what L-carnitine is

"GLP-1" is a drug class, not a molecule, and its three members in this library differ in mechanism and in regulatory status. Writing "GLP-1" and then making a claim is the error this section exists to prevent.

• Semaglutide is a glucagon-like peptide-1 receptor agonist and nothing else. It is FDA-approved.
• Tirzepatide is a dual agonist at the glucose-dependent insulinotropic polypeptide and glucagon-like peptide-1 receptors. It is routinely called "a GLP-1" and it is not one. It is FDA-approved.
• Retatrutide is a triple agonist at the glucose-dependent insulinotropic polypeptide, glucagon-like peptide-1 and glucagon receptors. It has no approval anywhere and its published human record is a phase 2 programme.

A finding about one of the three is a finding about that one. The largest GRADE-rated synthesis of obesity drugs makes the point numerically: it separates subcutaneous semaglutide, oral semaglutide and tirzepatide into three different estimates at moderate-to-high certainty, and files retatrutide among "emerging agents" at very low to low certainty.^[2]

L-carnitine is one molecule among several that share the name. L-carnitine, acetyl-L-carnitine and propionyl-L-carnitine are three different compounds with three separate literatures, and L-carnitine-L-tartrate — the salt most exercise research uses — is not an equal mass of L-carnitine. Its own approval is narrow: it is approved for carnitine deficiency, and for nothing else. That approval does not reach weight loss, does not reach body composition, and above all does not reach the drugs named above, whose approvals are separate and whose indications are their own. The same molecule is also sold as an over-the-counter supplement, which is a statutory category rather than a second approval.

What the evidence says about giving them together

No published study has administered L-carnitine together with any drug in this class, and here are the searches. On 1 September 2026 PubMed returned 25 records for L-carnitine, levocarnitine or carnitine combined with "glucagon-like peptide-1 receptor agonist", semaglutide, tirzepatide, liraglutide, exenatide or dulaglutide; 10 when that set was narrowed with combination, combined, co-administration, co-administered, "administered together", adjunct and supplementation; and 1 when restricted with the randomized controlled trial publication-type filter — a single record, PMID 32424935, and the next paragraph explains why it is a false lead. Adding clinical trial[pt] to that filter returns the same one record. Opening the wider set shows three recurring kinds of record, none of which is a co-administration study: metabolomics papers measuring acylcarnitine panels as biomarkers, rodent pharmacology, and reviews.

Two of those records look like hits and are not. The first is a 2018 rat study of moderate brain contusion injury whose PubMed title joins the two names with the word "and" — its own methods state that L-carnitine in drinking water or exendin-4 by the intraperitoneal route were given, in separate arms, and its results are reported arm by arm.^[1] The second is a 2020 randomized trial of liraglutide in type 1 diabetes whose "carnitine" is carnitine palmitoyltransferase-1 gene expression in adipose tissue, not carnitine given to anybody.

One human record does connect the two, and it runs in the opposite direction. A 2025 case report describes a 34-year-old man with multiple acyl-CoA dehydrogenase deficiency, an inherited disorder of fatty-acid oxidation, who had been treated with semaglutide injections for type 2 diabetes. Hypoglycaemic symptoms appeared after he was switched to the oral formulation, his mean blood-free carnitine fell significantly, and liquid chromatography-tandem mass spectrometry found a complex of carnitine with salcaprozic acid sodium at m/z 423.24 in urine exclusively while he was taking the oral form.^[8] Both halves belong here, and neither of them is this page's premise. It is the only published human record in which a product in this class measurably moved carnitine status, and it moved it down — and it is a single patient with a rare inherited disease, in whom the mechanism the authors propose is complexation by salcaprozic acid sodium, the absorption enhancer that makes the oral tablet work. That is an excipient in one formulation, not the GLP-1 mechanism and not a property of the injectable form. It is a reason to know the record exists, not evidence for adding anything.

One study is registered and it is both the honest anchor and the honest caveat. ClinicalTrials.gov record NCT07393360, retrieved directly on 1 September 2026, registers a trial of a food for special medical purposes containing essential amino acids, carnitine, arginine and sucrosomial minerals, for preservation of appendicular skeletal muscle mass during a weight-loss programme with GLP-1 receptor agonists. Its sponsor is the maker of the product, it has been recruiting since 11 December 2025 toward an estimated 144 participants, its estimated completion is December 2026, and its results field reads false. It has reported nothing. And carnitine sits inside a multi-ingredient sachet, so even when it does report, it will not be able to attribute any effect to carnitine specifically.

That is why the first graded claim above sits at the bottom of the scale. Nothing has been administered together, in any species, for any endpoint, so there is no design for a higher grade to describe. A well-evidenced problem on one side and a small, well-evidenced effect on an unrelated outcome on the other do not meet in the middle and become an evidenced answer.

Where the evidence is weak

The problem is documented; carnitine as its solution is not. The lean-mass side is quantified — tirzepatide reduced fat mass the most, by 25.7%, and lean mass the most, by 8.3%, in a network meta-analysis of 262 randomized trials.^[2] The carnitine side does not address it: the largest meta-analysis of L-carnitine supplementation, 43 randomized trials, reports about a kilogram of weight loss and a kilogram of fat mass, with body-fat percentage and waist circumference both unchanged,^[4] and neither it nor the earlier nine-trial meta-analysis reports a lean-mass or fat-free-mass outcome at all.^[5] The outcome the pairing is discussed for has never been measured for the supplement.

Two meta-analyses of the same trials are not two findings. The 43-trial and nine-trial syntheses draw on an overlapping pool of the same published studies, so counting them as independent replication doubles the apparent evidence without adding a participant — and the meta-regression in the smaller of them found the magnitude of weight loss significantly decreasing with duration of consumption (p=0.002), which points the wrong way for a programme that runs a year or more.^[4][5]

The lean-mass concern is contested from inside its own literature. The review most often cited for it reports lean-mass reductions ranging from 40% to 60% of total weight lost in some studies and about 15% or less in others, and concludes on imaging-based evidence that the muscle changes appear adaptive, commensurate with ageing, disease status and the weight achieved.^[3] Both halves belong here: the concern is real enough to have been studied, and the study of it did not conclude that something needs to be added.

L-carnitine is not inert, and what it moves most clearly is not body composition. Over 24 weeks of supplementation in healthy aged women, plasma trimethylamine-N-oxide rose roughly tenfold, while C-reactive protein, interleukin-6, tumour necrosis factor-α, the selectins, both adhesion molecules measured and the lipid profile were all unchanged.^[6] Both halves again: the metabolite moves substantially, and the atherosclerosis markers measured alongside it did not. Whether that metabolite rise causes harm is unsettled, and the compound page for L-carnitine carries that question in full.

Dosing and protocol ranges live on the individual compound pages, not here. That standing rule costs this page nothing, because there is no joint figure anyone could report even if the rule did not exist.

Questions to bring to a provider

The productive version of this conversation is not "should I add carnitine." It is "what specifically am I trying to protect, and what has been shown to protect it."

• Which molecule is actually under discussion — the GLP-1 receptor agonist, the dual agonist, or the unapproved triple agonist? The largest GRADE-rated synthesis rates them separately, giving subcutaneous semaglutide, oral semaglutide and tirzepatide their own estimates at moderate-to-high certainty and filing retatrutide among "emerging agents" at very low to low certainty.^[2] Their regulatory positions differ too: two are approved and the third is approved nowhere.
• If the concern is muscle, what would measure it: a scan, grip strength, a functional test at a fixed interval? The 43-trial meta-analysis of L-carnitine reports weight, body mass index and fat mass, with body-fat percentage and waist circumference unchanged;^[4] neither carnitine meta-analysis reported a lean-mass or fat-free-mass outcome to compare against.
• The review of lean-mass change concludes, on magnetic-resonance-based evidence, that the muscle changes appear adaptive and commensurate with ageing, disease status and the weight lost.^[3] What would make adding anything necessary against that conclusion, and how would the answer be measured?
• Does existing kidney disease, a seizure history, or warfarin change the calculus, given that L-carnitine's approved labelling carries precautions on all three?
• Is the carnitine product under discussion L-carnitine, acetyl-L-carnitine, propionyl-L-carnitine, or the tartrate salt? A result in one is not a result in another, and neither meta-analysis on this page reports its result broken out by carnitine form.

Anti-doping status

None of the four substances named on this page is prohibited by name in the World Anti-Doping Agency Prohibited List in force for 2026. That List, the 2026 Monitoring Program and the 2026 Explanatory Note were retrieved on 1 September 2026 from the World Anti-Doping Agency's own file server, by way of the United States Anti-Doping Agency's prohibited-list page, and searched directly: the string "carnitin" returned zero occurrences in all three documents, and "semaglutide", "tirzepatide" and "retatrutide" each returned zero occurrences in the List itself. This paragraph is uncited because it is a reading of those three documents on that date rather than of any research paper.

Two provisions bear on this page anyway, and both are routinely misread.

Monitoring is not prohibition. Item 6 of the 2026 Monitoring Program reads "Markers of Semaglutide and Tirzepatide: In and Out-of-Competition", and the 2026 Explanatory Note clarifies that the urine monitoring of semaglutide includes also the monitoring of tirzepatide. The Monitoring Program exists so that patterns of use can be detected in substances that are not on the Prohibited List. It is about markers, it names two of the three class members and not the third, and nothing in it makes any of them prohibited.

The provision that reaches carnitine is a rule about a route, not about a molecule. Prohibited method M2.2 prohibits "Intravenous infusions and/or injections of more than a total of 100 mL per 12-hour period except for those legitimately received in the course of hospital treatments, surgical procedures or clinical diagnostic investigations", and the same class states that all its prohibited methods are non-Specified except methods in M2.2, which are Specified Methods. It catches a volume delivered by a route, whatever is in it, says nothing about carnitine as a substance, and does not reach an oral product.

And there is a trap for anyone who searches this themselves. Meldonium is prohibited at section S4.4.3, and meldonium is not L-carnitine: it works by inhibiting carnitine biosynthesis, and a 2026 narrative review of its performance literature finds no high-quality randomized placebo-controlled evidence of consistent ergogenic benefit in healthy athletes, describing the existing human performance studies as at high risk of bias.^[7] A PubMed search on 1 September 2026 for carnitine with "prohibited list" or "World Anti-Doping" returned 8 records, and the ones connecting carnitine to the Prohibited List at all are about meldonium. A reader who searches "carnitine WADA" lands on a different molecule with the opposite mechanism.

Retatrutide has no approval at any regulator, which is the condition section S0 is written around: it reaches any pharmacological substance not addressed by another section of the List and with no current approval by any governmental regulatory health authority for human therapeutic use, expressly including drugs under clinical development. Not being named on the List is not the same as being permitted. The List is reissued annually and its sections renumber between editions, so a competing athlete should confirm the year in force directly rather than relying on this page.