Metformin

Also called: Glucophage.

Chapter 14 of The Cancer Evidence Handbook, Part Three. Last reviewed 2026-09-16. How this works.

Treating breast cancer Grade D

No human evidence of benefit, or none that survived testing.

A very large trial gave this diabetes drug to women with breast cancer and found it made no difference.

A trial randomised 3,649 patients and found nothing.

How this grade was reached

The claim

The claim is that metformin, a cheap and very widely used diabetes drug, treats cancer. It came from observing that people with diabetes taking metformin seemed to get less cancer.

What it does

Several mechanisms were proposed, and the observational signal was large and consistent across a great many studies. This entry is in the book because of what happened when that signal was tested.

The evidence

MA.32 randomly assigned 3,649 people without diabetes, with high-risk non-metastatic breast cancer, to metformin or placebo alongside standard therapy, double-blind, across four countries, enrolled from 2010 and followed to 2020.1

It found no effect on invasive disease-free survival and no effect on overall survival. The event rates in the hormone-receptor-positive group were 2.78 per 100 patient-years on metformin against 2.74 on placebo.1

A later analysis of the same trial records that this negative result is consistent with a large group of smaller randomised trials across other settings. The observational signal did not survive being tested, anywhere.3

The trial's lead investigator said so publicly: that the results tell us metformin is not effective against the most common types of breast cancer, and that off-label use for those types should be stopped. The same statement notes that the negative result underscores the need for randomised trials before observational findings are put into practice.2

One thing survives, narrowly. Exploratory analyses suggested benefit in HER2-positive disease, specifically in carriers of a particular genetic variant, and the investigators state that these require replication. That is a hypothesis, not a finding, and the public statement above explicitly does not cover HER2-positive disease.12

The studies this rests on

  • MA.32, metformin versus placebo in high-risk non-metastatic breast cancer (NCT01101438)1

    A randomised trial, double blind, phase 3, measured survival, 3649 participants.

    no effect on invasive disease-free survival and no effect on overall survival.

What has not been shown

The one surviving signal is an exploratory finding in a genetic subgroup, and the investigators themselves state it requires replication. Nobody has replicated it.

This applies to: the HER2-positive genetic subgroup.

Established by trial scope, checked 2026-09-16. Result: the finding is an exploratory subgroup analysis within a single trial, flagged by its own investigators as requiring replication.1

What to ask

Is there any reason to take metformin for my cancer if I do not have diabetes?

When to call your team, and what to tell them

Whatever you are taking or thinking of taking, the people treating you can only weigh it if they know about it. That is true when nothing has been written down here, and it is most true then.

The register records no warning signs for metformin. That is not a statement that nothing can go wrong, and anyone taking a prescription drug outside its licence should tell their oncology team.

Find a trial instead

The open question after MA.32 is a narrow one about a genetic subgroup, and a trial is where it would be answered.

Search ClinicalTrials.gov for recruiting studies of metformin

A trial is the same decision with monitoring, a known preparation and a doctor who knows what else you are taking.

Also recorded for this entry

Does it reach the body?

The register holds no human pharmacokinetic measurement for metformin against a laboratory concentration. This entry rests on a large randomised trial rather than on a concentration argument. What that question means, for every compound here.

Does it clash with a cancer treatment?

The register holds no measured interaction between metformin and a cancer treatment class. Nobody having looked is not the same as nothing being there.

Every recorded interaction, and how to turn it into a request for your pharmacist.

References

  1. Goodwin PJ, Chen BE, Gelmon KA, et al. Effect of metformin vs placebo on invasive disease-free survival in patients with breast cancer: the MA.32 randomized clinical trial. JAMA 2022;327(20):1963-1973.

    doi:10.1001/jama.2022.6147 · NCT01101438

    How this was reached: taken from the book's notes, chapter 14, note 1. The author names are abbreviated in the source that was read, and have not been filled in here.

  2. Statement by Pamela J. Goodwin issued by the Canadian Cancer Trials Group when the MA.32 results were presented at the 2021 San Antonio Breast Cancer Symposium, reproduced by the Breast International Group and in subsequent reporting.

    No persistent identifier recorded.

    How this was reached: taken from the book's notes, chapter 14, note 2.

  3. Goodwin PJ, Ennis M, Chen BE, et al. Exploration of body mass index and circulating metabolic factors as predictors of metformin benefit in the Canadian Cancer Trials Group MA.32. JNCI Cancer Spectrum 2026;10(4):pkag045.

    doi:10.1093/jncics/pkag045

    How this was reached: taken from the book's notes, chapter 14, note 3. The author names are abbreviated in the source that was read, and have not been filled in here.