Does Metformin Treat or Prevent Cancer?

Metformin is the most-prescribed drug for type 2 diabetes, it costs very little, and for about fifteen years it has carried a second reputation: that it also fights cancer. You will see it stated flatly — that metformin is "associated with suppressed tumour growth", or that it works by "suppressing age-related chronic diseases like certain cancers".

The first half of that is true in a laboratory. The second half has been tested in people, and this page is about what happened when it was.

Metformin does inhibit the growth of cancer cells in culture, and it does slow tumour growth in mouse models. This is not fringe science and nobody disputes it. The National Cancer Institute's own drug dictionary describes metformin as having "potential antineoplastic activities" and says it "may exert antineoplastic effects" through effects on a cell-growth pathway called mTOR.

Read those two words again — potential, and may. That is a careful description of a laboratory mechanism. It is not a statement that the drug treats cancer in people, and NCI does not say that it does.

That mechanism was a good enough reason to run trials. So trials were run.

Breast cancer — MA.32

The largest test was MA.32, a phase 3 randomised, placebo-controlled, double-blind trial across Canada, Switzerland, the US and the UK. It enrolled 3,649 patients with high-risk non-metastatic breast cancer, gave them 850 mg of metformin twice a day or a matching placebo for five years, and followed them to October 2020. It was published in JAMA in 2022.

The result: invasive disease-free survival events occurred at 2.78 per 100 patient-years on metformin versus 2.74 on placebo — a hazard ratio of 1.01. Deaths: 1.46 versus 1.32 per 100 patient-years, hazard ratio 1.10. For the hormone-receptor-negative group, the trial was stopped early for futility.

The authors' conclusion, in their own words: adding metformin "did not significantly improve invasive disease-free survival".

Grade 3 non-blood-related side effects were more common on metformin — 21.5% versus 17.5%.

Did it at least prevent new cancers?

A follow-up analysis published in the Journal of Clinical Oncology in 2023 looked at whether metformin stopped these patients developing new primary cancers. 184 did: 102 on metformin, 82 on placebo. The hazard ratio was 1.25 — not statistically significant, but pointing the wrong way. The authors concluded that metformin "did not reduce the risk of new cancer development".

Prostate cancer — twice

The STAMPEDE platform trial tested metformin added to standard care in 1,874 men with metastatic hormone-sensitive prostate cancer, reporting in Lancet Oncology in 2025. Median survival was 61.8 months on standard care and 67.4 months with metformin — a difference that did not reach significance (hazard ratio 0.91, p=0.15). The investigators wrote that they "did not find significant evidence of an overall survival benefit".

There is a genuine finding buried in that trial worth stating fairly: metformin did significantly reduce the metabolic side effects of androgen-deprivation therapy. That is a real benefit. It is a different claim from treating the cancer, and it is worth not letting a press headline blur the two.

The MAST trial tested metformin in 408 men on active surveillance for low-risk prostate cancer, reporting in 2025. No significant difference in progression-free survival. In a pre-specified subgroup, metformin was associated with increased pathological progression among obese participants.

This is the most useful part of the story, because the pattern repeats across medicine.

Before the trials, observational studies reported that people taking metformin got less cancer — with reductions running from 20% to over 90% depending on which study you read. Those numbers were extraordinary. They were also an artefact of how the groups were counted.

Imagine sorting people with diabetes into "metformin users" and "non-users" based on whether they ever filled a prescription. Someone in the metformin group had to survive long enough to collect it. If they had died in the three months between diagnosis and their first prescription, they would have been filed as a non-user instead. So the metformin group is quietly credited with a stretch of time in which, by the very logic of how the group was defined, nobody in it could possibly have died.

Epidemiologists call that "immortal time". Hand a group of people a period of guaranteed survival and they will, of course, appear to live longer. The drug did nothing; the bookkeeping did it.

The studies designed to avoid this trap found no effect. When the randomised trials finally reported, they agreed with the careful studies rather than the flawed ones.

Not exactly, and that is what makes it effective. "Metformin suppresses tumour growth" is true in a dish and in a mouse. What the claim omits is that this was then tested in people, five times, in around 6,250 patients — and it did not work.

Established in the laboratory, tested in patients, and it failed. All three parts are the story. A summary that gives you only the first is not lying to you, but it is leaving out the part that would change your mind.

No major oncology guideline body — NCI, ASCO, ESMO, NICE or NCCN — recommends metformin to prevent or treat cancer.

If you have type 2 diabetes and your doctor has prescribed metformin, keep taking it; controlling diabetes during cancer treatment matters in its own right. If you are considering metformin because of the cancer claim, that is worth a frank conversation with your oncologist first. It is a real drug with real contraindications — reduced kidney function, and situations involving contrast imaging among them — and a boxed warning about lactic acidosis.

The bigger lesson is portable. When you see a headline about a cheap old drug that turns out to fight cancer, the question to ask is not "is there a mechanism?" but "was it tested in people, and what happened?" Our guide to reading cancer statistics and headlines covers the rest of that checklist.

Every figure above is from the trial publication or regulator named beside it, not from a summary of one. Where the honest answer is that nobody knows, this page says so rather than filling the gap.

Educational information, not medical advice. Do not start, stop or change any medication or supplement because of a web page — including this one. If you are on cancer treatment, tell your oncology team about everything you take, including supplements. Spotted an error? Please email [email protected] or use /corrections.