L-Ornithine and Cancer: What the Evidence Shows
Ornithine is sold over the counter for sleep, exercise recovery and liver support. It is also the carrier the urea cycle runs on and the direct precursor of the polyamines that tumours need in order to divide. The research literature points in both directions at once, and no therapeutic trial of ornithine in people with cancer has ever been run. This page sets out what has actually been tested.
Most people taking ornithine are taking it for sleep, for exercise recovery, or because they have read that it supports the liver. It is cheap, it is a normal constituent of the human body, and it is sold without a prescription almost everywhere.
It is also one of the more interesting molecules in cancer metabolism, for a reason that has nothing to do with any of those uses: it sits at a fork, and the two branches lead in opposite directions. One disposes of nitrogen. The other builds the polyamines that rapidly dividing cells depend on. Nothing in a capsule chooses between them.
1 / The Fork
One Molecule, Two Enzymes, Opposite Consequences
Ornithine transcarbamylase condenses ornithine with carbamoyl phosphate to make citrulline, and the nitrogen heads toward urea and out of the body. Ornithine decarboxylase takes the same molecule and strips its carboxyl group, producing putrescine, then spermidine, then spermine — the polyamines, without which no cell divides quickly.
ODC is one of the most studied targets in oncology. Its inhibitor, DFMO or eflornithine, is an approved drug in neuroblastoma. That is the measure of how seriously the polyamine branch is taken as a cancer dependency.
Which branch dominates is not a property of ornithine. It is a property of the cell receiving it, and tumours are not ordinary cells. Lee and colleagues showed in Nature in 2023 that pancreatic ductal adenocarcinoma runs ornithine aminotransferase in its forward direction — an activity normally confined to infancy — to manufacture ornithine from glutamine, because arginine is depleted in the tumour microenvironment and the usual supply route is closed. Silencing or inhibiting OAT collapsed the polyamine pools and suppressed tumour growth.
Not all tumours obtain it the same way. Work on neuroblastoma published in 2025 used isotope tracing and quantitative modelling to identify circulating ornithine as an important tumour precursor, with much of the arginine-to-ornithine conversion happening outside the tumour. Pancreatic cancer emphasises local synthesis; neuroblastoma draws on the circulation. Where a tumour gets its ornithine determines whether what you swallow can reach it.
2 / The Experiments That Reduced Tumour Growth
Two Models, Both Involving Ammonia
These are the results that make ornithine interesting in oncology, and they are real.
Colorectal cancer, 2023
Bell and colleagues found ammonia accumulating in colorectal tumours and driving CD8 T cells into exhaustion. Daily intraperitoneal L-ornithine monohydrochloride reduced MC38 and CT26 syngeneic tumour growth and liver-metastatic burden in mice, with tumour ammonia and proliferation markers declining. The benefit disappeared in nude mice and after CD8 depletion, which is the control that makes the mechanism immune-dependent rather than direct.
Liver cancer, 2024
Elaimy and colleagues, having shown across 48,476 cirrhosis patients that baseline serum ammonia predicts who develops hepatocellular carcinoma, treated Myc/p53-driven HCC-bearing mice with ornithine monohydrochloride. Liver weight and tumour burden fell, along with tumour ammonia, CD44 and SLC4A11 expression, and ALDH activity. A separate graft experiment in immunodeficient mice supported reduced tumour weight under high-ammonia conditions.
Two qualifications belong immediately next to those results rather than at the end of the page.
3 / The Experiments That Went the Other Way
Promotion, Nothing, and a Dose Reversal
This is the part of the literature least represented in popular writing about ornithine, and it is the part that most changes the picture. Each of these is a direct ornithine experiment, not an inference from a related compound.
| Study | Exposure | Result |
|---|---|---|
| Ushmorov, Hack & Dröge, 1999 MCA-105 fibrosarcoma, cancer cachexia model |
0.2 mg/day i.p. for one week, versus 20 mg/day | The low dose improved muscle respiratory-chain activity and redox measures with only marginal tumour effects. The hundred-fold higher dose significantly increased tumour size (P < 0.05). |
| Tharp et al., 2024 Nature Cancer; orthotopic PyMT breast cancer |
2 g/kg daily oral gavage | Did not independently alter tumour growth. In a myeloid-Tgfbr2-knockout experiment, ornithine removed the tumour-mass reduction the knockout produced. In culture, 0.5–4.5 mM impaired CD8 metabolic function and viability and reduced killing of tumour organoids. |
| Yuan et al., 2024 Cell Discovery; gastric cancer xenograft |
Separately administered ornithine arm | Promoted xenograft growth. Dose, route, schedule and sample size for that arm could not be verified from the inaccessible supplement. |
| Pan et al., 2026 Nature Communications; 4T1 bone metastasis |
100 µM in culture; 10 mg/kg i.p. daily for 14 days | Increased PMN-MDSC proportions in mouse bone marrow and human PBMC cultures, and in bone-metastatic tumours in vivo. Ornithine did not directly increase cancer-cell colony formation or migration. |
| Gonzalez & Byus, 1991 Two-stage skin carcinogenesis, 19 weeks |
2% ornithine in drinking water | Dietary arginine restriction reduced papillomas by about 40–50%. Ornithine in the drinking water reversed that reduction. It was not shown to initiate cancer independently. |
Direct ornithine administration experiments with null or adverse outcomes. The Ushmorov result is the only published dose-reversal within a single model.
The Ushmorov finding deserves to be read twice, because it is the single most relevant experiment on this page for anyone taking a supplement. Same molecule, same model, same route, same week. At one dose it improved markers of cachexia. At a hundred times that dose it made tumours bigger. That is not a reason to pick the low dose — a single model does not establish a safe threshold, and the human equivalents of either figure are unknown. It is a reason to stop treating “ornithine” as a thing with a direction.
Parenteral nutrition studies, which are often cited and should not be
A set of studies from the 1990s substituted ornithine for arginine in total parenteral nutrition in tumour-bearing rats, and found lower tumour weights with ornithine than with arginine. In Grossie's 1996 experiment, eight-day tumours weighed 14.1 ± 3.3 g with arginine, 12.7 ± 1.4 g with citrulline, 8.8 ± 2.0 g with ornithine and 8.9 ± 2.1 g with ordinary chow.
4 / Ornithine and T Cells
A Forty-Year-Old Problem That Has Not Been Resolved
If ornithine helps in cancer by relieving an immune brake, as the colorectal work suggests, then it is worth knowing that a separate literature reports ornithine doing the opposite to T cells directly.
| Study | Concentration | Finding |
|---|---|---|
| Dröge et al., 1985 J Immunol |
9 mM | Nearly abolished early cytotoxic T lymphocyte activation in culture; adding it later had less effect. D-ornithine did not reproduce the effect, so it is not a generic osmotic or nitrogen artefact. |
| Jänicke & Dröge, 1985 | 9 mM | Suppressed CTL generation in mixed leukocyte culture. Reversibility and limited general toxicity argued against simple killing of responder cells. |
| Susskind et al., 1986 | 10 mM | Arrested CTL development without killing precursors; suppressor-cell activity persisted. |
| Susskind et al., 1991 | not recovered | CTL differentiation inhibited, but lymphokine-activated killer activity against P815 and YAC-1 targets was not suppressed. Ornithine does not inhibit all effector populations equally. |
| Tharp et al., 2024 Nature Cancer |
0.5, 1.5, 4.5 mM | At increasing ornithine-to-arginine ratios, impaired CD8 metabolic function and viability; higher ornithine reduced killing of tumour organoids. The modern replication of the 1980s finding, in a cancer model. |
| Menjivar et al., 2023 eLife |
1.15 mM, 72 h | In arginine-free medium, partially restored CD8 proliferation but not activation. No enhancement in arginine-replete medium. |
Direct ornithine effects on T cells. Most of the older work used allogeneic stimulation rather than a tumour model.
Two honest caveats. The 9 to 10 mM concentrations in the 1980s work are high — far above anything ordinary oral supplementation would be expected to produce in tissue — and several of those records rest on abstracts rather than full texts. But Tharp's 2024 experiments saw impairment from 0.5 mM upward, in a cancer model, published in a leading journal, and the effect there tracked the ratio of ornithine to arginine rather than ornithine alone. That is a mechanism which supplementation could plausibly move.
And the myeloid compartment
In tumour-associated macrophages, ornithine is converted by ODC into polyamines, and the polyamines are required for M2 polarisation — they drive expression of M2 genes and suppress M1 activation. Pan's 2026 work adds the neutrophilic arm: 10 mg/kg daily expanded PMN-MDSCs, the suppressive myeloid population, in bone marrow and bone-metastatic tumours. The polyamine problem is not only a tumour-cell problem.
5 / The Human Evidence
There Is No Therapeutic Trial. There Is a Tracer Study.
This is the shortest section on the page and the most important one.
Registry checks used indexed ClinicalTrials.gov, UMIN and ISRCTN searches rather than a live registry export, so this is a bounded search finding rather than proof of universal absence. But the absence is large and consistent.
What human ornithine measurements show instead
There is a substantial observational literature measuring ornithine concentrations in people with cancer. It does not converge.
| Cancer | Study and size | Direction |
|---|---|---|
| Breast | Zhang 2020; 735 cases / 735 controls | Higher ornithine, lower odds. OR per SD 0.88 (0.79–0.97). But 95% of controls had benign breast lumps, and samples were taken after disease developed. |
| Breast | Hu 2016; 33 / 32 | Ornithine higher in cancer. Opposite direction to the larger study. |
| Colorectal | Rothwell 2023; EPIC, 654 future cases / 654 controls, prospective | No significant association. OR per SD 0.88 (0.69–1.13); excluding the first ten years, 1.03 (0.78–1.36). The best-designed entry in the table, and it is null. |
| Colorectal | Hu 2021 (63 / 63) versus Coradduzza 2023 (50 / 52) | Lower in cancer, and higher in cancer, respectively. Directly contradictory. |
| Prostate | Ladurner 2025; 1,078 men undergoing biopsy | Higher in clinically significant disease, 63.7 versus 50.3 µM. The reported 0.84 AUC belongs to a multi-metabolite model, not to ornithine. |
| Bladder | Liu 2024 (95 / 96) versus Giorgione 2026 (14 patients) | Higher ornithine with lower odds in one; higher than controls in the other. Contradictory again. |
| AML | 1999 series; 40 untreated / 24 controls | Higher in cancer: 108.7 ± 5.8 versus 78 ± 6 µM, P < 0.001. |
Circulating ornithine in cancer. Several additional entries in the underlying inventory did not survive multiple-testing correction.
6 / What You Would Actually Be Taking
Three Preparations, and They Are Not Interchangeable
L-ornithine-L-aspartate (LOLA)
A salt of two amino acids, marketed for hepatic encephalopathy since the 1990s and available on prescription across much of Europe, Asia and Latin America. It is not FDA-approved in the United States. It works on two fronts: ornithine enters the urea cycle as the OTC substrate, and both amino acids feed glutamate, which skeletal muscle uses to fix ammonia into glutamine via glutamine synthetase. In a cirrhotic patient whose liver cannot make urea, the muscle route matters more than the hepatic one.
Ornithine phenylacetate
A deliberate two-part design. Ornithine drives glutamine synthesis; phenylacetate then conjugates that glutamine into phenylacetylglutamine, which the kidney excretes. The nitrogen leaves the body rather than being parked in a pool that can release it again.
L-ornithine on its own
Sold as a sleep and bodybuilding supplement, typically at 500 mg to 2 g. This is the one people can buy, and it is the one with no clinical trial evidence of any kind in liver disease or cancer.
One practical finding worth having from that trial: 40 patients with overt hepatic encephalopathy randomised to intravenous or oral LOLA alongside standard care showed no difference between routes. Ammonia fell by 55.4 ± 32.58 µM on the drip and 60.75 ± 13.82 µM by mouth (p = 0.511), with both groups improving. For a drug whose oral bioavailability has long been questioned, that is reassuring as far as it goes — though 40 patients is 40 patients.
7 / Where the Reputation Comes From
The Liver Evidence Is Weaker Than It Looks
Ornithine's good name comes almost entirely from hepatology, so it is worth seeing what that evidence actually says.
And the enthusiastic literature has a narrow authorship. The 2019 meta-analysis most often cited for LOLA in prevention of overt encephalopathy — six trials, 384 patients, RR 0.38 (0.23–0.62) — is a single-author paper by Roger Butterworth, who is also co-author of the review making the case for LOLA in fatty liver disease. That is not an accusation of anything. It is a reason to read the primary trials rather than the reviews, and to notice when a field's advocacy rests on a small number of people.
In fatty liver, thinner still
The human data for LOLA in NAFLD, as assembled by its advocates, amounts to: an uncontrolled observational study of 463 patients given oral LOLA for 30 to 90 days, reporting falls in ASAT of 46.6%, ALT of 40.6% and γ-GT of 60.8%; an open-label randomised multicentre study in 72 NASH patients showing a dose-related ALT reduction over 12 weeks; and an uncontrolled study of 78 patients looking at hepatic microcirculation.
Transaminases in fatty liver fall with weight loss, with alcohol cessation, with regression to the mean, and with the simple fact of being enrolled in a study and told to take something. An uncontrolled 40–60% drop is close to uninterpretable. There is no blinded, placebo-controlled trial of LOLA in fatty liver disease with histological endpoints.
The best test anyone has run
STOP-HE was a phase 2b study of intravenous ornithine phenylacetate in 231 hospitalised cirrhotic patients with elevated ammonia and overt hepatic encephalopathy, at three dose levels, against placebo, on top of standard care. Properly powered, properly designed, in exactly the population where ammonia is the agreed cause of the problem.
It missed its primary endpoint. Time to clinical improvement was 47 hours against 64 on placebo, p = 0.129. Ammonia fell significantly — 30.8 versus 11.8 µM, p = 0.017. The drug was safe and well tolerated, with fewer deaths in the treatment arm than on placebo, 11 versus 15, and the higher dose groups did better than the lowest. There is a real signal in there.
8 / If You Take Ornithine
What Follows From All This, and What Does Not
Ornithine is cheap, sold without a prescription, and now has research threads pointing at it in diseases that frighten people. Everything about that combination invites someone reading this to go and buy some, or to stop. So let me be precise about what the evidence does and does not support.
What is established
Ornithine-based drugs lower ammonia in people. That much is solid, by both routes, in the thousands of patients who have taken LOLA and in the 231 of STOP-HE. Ornithine is a normal constituent of the body and of the diet, and healthy-volunteer tolerance studies exist.
What is not
That lowering ammonia improves any clinical outcome — the best-designed trial missed its endpoint. That ornithine has any effect on cancer in humans, in either direction. That the mouse results generalise. That any particular supplement dose is either effective or safe in the presence of a tumour.
What points the wrong way
Direct tumour promotion in a gastric model. Reversal of a protective dietary effect in skin carcinogenesis. Expansion of suppressive myeloid cells. Impaired CD8 function at concentrations from 0.5 mM. And a hundred-fold dose increase turning a cachexia benefit into significantly larger tumours in the same model.
What nobody knows
Which tumour and host features predict benefit versus harm. What oral supplementation does to ornithine concentrations inside a tumour. Whether the ammonia-clearance effect and the polyamine effect can be separated. Whether any of this applies at sleep-supplement doses.
If you are taking ornithine for sleep or training and have no cancer diagnosis, none of the above is an argument that you are harming yourself. It is an argument that the confident claims made for ornithine in cancer — in both directions — are running well ahead of what has been tested.
9 / Open Questions
What Would Actually Settle This
- Reproducible dose–response within a single model. The Ushmorov reversal is one experiment from 1999. Whether a therapeutic window exists, and where, has not been re-tested in the twenty-six years since.
- Both immune endpoints measured together. One experiment measuring ammonia relief of T cell exhaustion and the ornithine-to-arginine ratio effect on CD8 function, in the same animals. The two literatures have never met.
- What oral dosing does to tumour interstitial ornithine. Every mouse experiment used injection. Nobody has measured what swallowing it does to concentrations where a tumour sits.
- Ornithine with ODC blockade. Supply the carrier and block the polyamine branch with DFMO, which is approved in neuroblastoma so the toxicology exists. If the polyamine branch is what limits ornithine's usefulness, blocking it should turn a modest effect into a large one. If it changes nothing, the polyamine worry can be set down. The combination appears never to have been tested.
- Any human trial at all. Standalone oral ornithine, with plasma and where possible tumour-interstitial measurement, in a defined cancer population, with clinically relevant endpoints and investigators independent of the original findings.
Until some of that exists, the accurate description of ornithine in cancer is the one at the top of this page: context-dependent, bidirectional, and untested in people.
Last updated: 7 October 2026. Literature searched through 7 October 2026.
No comments:
Post a Comment