Does creatine boost anti-tumour immunity through dendritic cells?

---

The claim that spread is not in the paper

A widely shared thread in late September 2026 summarised a new UCLA study as showing that creatine supplementation activates dendritic cells, and that depleting those dendritic cells abolished the anti-tumour effect — the step that would make dendritic cells the causal mechanism rather than a correlate.

We read the paper. That depletion experiment does not exist in it. The word "depletion" does not appear. There is no dendritic-cell depletion arm, no anti-CD11c antibody, no diphtheria-toxin receptor model, no genetic ablation of the dendritic-cell compartment.

This matters more than a typical overstatement, because the missing experiment is the load-bearing one. Without it, the study shows that creatine transport supports dendritic-cell activation and that creatine slows tumour growth in mice, but not that the first causes the second. Those are different claims, and only the weaker one is supported.

Evidence tier: 1 — primary paper read directly, full text and figures.

The paper itself is real, competent, and more cautious than the thread. Kang, Elsten-Brown, Wang and colleagues, from Lili Yang's lab at UCLA, published it in iScience in April 2026 (PMID 42006288). What follows is what it actually reports, followed by the published evidence pointing the other way — which is substantial, human, and was absent from the thread entirely.

What the study actually did

Three things, in mice and in human cells in a dish.

One. Dendritic cells take up creatine through the SLC6A8 transporter, and that uptake rises when the cells are activated. Blocking the transporter pharmacologically, or knocking out the gene, reduced markers of dendritic-cell activation and reduced their ability to stimulate T cells in co-culture.

Two. In mouse tumour models, creatine administration slowed tumour growth relative to vehicle.

Three. Human monocyte-derived dendritic cells cultured with creatine showed the same activation pattern as the mouse cells.

Each of those is a reasonable result. The problems are in the details of how they were produced, and every one of those details makes the jump to a human taking creatine powder longer than the thread implied.

The route was injection, not drinking water

Creatine was given by intraperitoneal injection at 10.5 mg per mouse per day. Not in the drinking water, not by oral gavage. No oral dosing arm was tested in vivo anywhere in the paper.

This is not pedantry. Creatine's oral bioavailability, first-pass handling, and the plasma curve it produces are the entire question for anyone considering a supplement. An intraperitoneal bolus bypasses all of it and produces a peak the gut cannot. A study that only injects has said nothing about the supplement.

The knockout was whole-body, not immune-specific

The SLC6A8 knockout used was the whole-body strain (JAX #021072) — not a conditional knockout restricted to dendritic cells or to the immune compartment.

SLC6A8 is expressed in brain, muscle, heart, kidney and gut. A whole-body knockout animal is different from a wild-type animal in many ways that have nothing to do with dendritic cells, including in ways that plausibly affect tumour growth on their own. Attributing the tumour phenotype in that animal to its dendritic cells specifically requires the conditional knockout, and that experiment was not done.

The human experiments used roughly 500× a supplemented plasma level

The human monocyte-derived dendritic-cell work used creatine at 50 mM.

Plasma creatine after a standard supplementation protocol sits in the region of 100 µmol/L — around 0.1 mM. The dish concentration is therefore about 500 times what supplementation achieves in blood. Concentrations that far above physiological range routinely produce effects that do not replicate at achievable exposures, in either direction. The human data in this paper establish that the pathway exists in human cells; they do not establish that it is engaged by a scoop of creatine.

One claim in the thread we could not locate at all

The thread stated that treated mice had more IFN-γ-producing T cells in vivo. We could not find that result in the paper. The T-cell stimulation data are from co-culture assays, not from tumour-infiltrating lymphocytes in the treated animals. If the in vivo measurement exists we could not find it in the figures, supplementary figures, or text.

The dose, scaled honestly

Scaling 10.5 mg per mouse per day to a 70 kg human by body-surface area — the standard method, using the mouse-to-human Km ratio of 12.3 — gives roughly 2 to 3 g per day.

That is worth stating plainly because the thread described the dose as "relatively high." It is not. It is an ordinary creatine maintenance dose, at or below what a large fraction of people already take. The mouse dose is not the reason to be cautious here. The route, the knockout design, the missing depletion experiment, and the literature below are.

The evidence pointing the other way

This is the part the thread omitted, and it is the part that should change what anyone does.

Creatine promoted metastasis in mice, with a human platelet signal

A 2026 paper in Nature Communications reported that exogenous creatine supplementation promoted tumour metastasis across multiple mouse models, through megakaryocyte creatine kinase B and STAT5B signalling (PMID 42449108).

The finding that deserves attention is not the mouse metastasis. It is that the same paper demonstrated hyperactive platelets in healthy human volunteers given creatine. That is a human measurement, in healthy people, at supplemental exposure — a category of evidence the UCLA paper does not contain at all.

And in 2021, independently

A 2021 Cell Metabolism paper reported that creatine promotes cancer metastasis via Smad2/3 activation, and its authors explicitly urged caution about dietary creatine in this context (PMID 33811821).

Two independent groups, five years apart, reporting pro-metastatic effects. That is not a fringe result being weighed against a strong one.

The one randomised controlled trial in cancer patients was null

Creatine has been tested in people with cancer. The N02C4 Alliance trial randomised 263 patients with cancer anorexia/weight-loss syndrome to creatine or placebo, double-blind (PMID 28475678).

It was null on its primary endpoint and null on survival.

One null RCT in a specific population does not refute a mechanism. But it is the highest tier of evidence available on this question, it is in humans, and it is not encouraging. Any account of "creatine and cancer immunity" that cites mouse work and omits the randomised human trial has selected its evidence.

Evidence tier: 1 — randomised, double-blind, placebo-controlled, n=263.

The pharmaceutical industry is betting the other way

This is the detail that should give the strongest pause. *Ompenaclid (RGX-202) is an SLC6A8 inhibitor*** — it blocks the same creatine transporter the UCLA paper is about — and it is in human oncology trials as an anti-cancer agent.

The UCLA paper uses SLC6A8 inhibition as a tool to suppress dendritic-cell activation. A drug developer is giving SLC6A8 inhibition to cancer patients on the thesis that blocking creatine transport helps. Those two positions cannot both be straightforwardly right, and only one of them is being tested in humans with a commercial and regulatory apparatus behind it.

A practical hazard that has nothing to do with immunity

Creatine supplementation raises serum creatinine. This is an analytical and physiological artefact, not kidney injury — creatinine is creatine's breakdown product, so taking more of the parent raises the metabolite.

Published estimates put the shift at roughly +0.14 mg/dL serum creatinine, corresponding to about −10.75 mL/min in estimated GFR.

For most people that is a scare on a lab report and nothing more. In oncology it is a dosing hazard, because several drugs are dosed on renal function: carboplatin (via the Calvert formula, which uses GFR directly), methotrexate, and pemetrexed. A falsely low eGFR can produce a falsely low dose. Anyone taking creatine who is also receiving renally dosed chemotherapy needs their oncology team to know, and that is true regardless of how the immunology turns out.

What the prior literature actually establishes

Creatine and anti-tumour immunity is not a new idea, and the new paper is less of a departure than it was presented as.

• Di Biase et al., 2019, J Exp Med — creatine uptake regulates CD8 T-cell anti-tumour immunity (PMID 31628186). Same laboratory. This is the direct predecessor of the 2026 paper, not independent corroboration of it.
• Peng & Saito, 2023, Front Immunol — creatine supplementation enhances anti-tumour immunity via macrophage ATP production (PMID 37662917).

So: a favourable mouse literature, largely from two groups, one of which has now published the same direction three times; against it, two independent pro-metastatic mouse papers, one human platelet signal in healthy volunteers, one null randomised trial, and a pharmaceutical programme pointed the opposite way.

Declared interest

UCLA has filed a patent application related to this work. This is normal, it is disclosed in the paper, and it is not misconduct. It is also relevant context when a university's own communications describe the finding enthusiastically, and it belongs in any honest summary.

What we think this means

Creatine is one of the best-characterised supplements there is, with a long safety record in healthy people for its established uses. Nothing here changes that.

What the UCLA paper does not support is the specific claim that circulated: that creatine works through dendritic cells to fight tumours, in a way that should influence what a person with cancer does. The experiment that would establish the mechanism was not performed. The route was injection. The human cell work used a concentration supplementation cannot reach. And the surrounding literature, including the only randomised human trial and the only human data at supplemental doses, runs against the enthusiastic reading.

For anyone with active cancer or a cancer history, the pro-metastatic findings and the creatinine/chemotherapy-dosing interaction are the two things worth raising with an oncologist — before, not after, starting or continuing creatine. That conversation is warranted by the published evidence as it stands. Starting creatine for its anti-tumour immunology is not.

This is not medical advice. It is a reading of the primary literature, and the primary literature on this question is mouse-heavy, mixed in direction, and thin in humans.