In studies of men with diabetes, men taking metformin seemed
to have low incidence of prostate cancer and fewer deaths from prostate cancer. I stress the word "seemed" because all of the evidence for it came from observational studies rather than randomized clinical trials (RCTs). The problem with observational studies is "selection bias." Selection bias means that those taking the drug did so for a reason, and the reason may have affected prostate cancer incidence or progression because of some other reason, Observational studies can only show association
. Only a randomized trial can prove causality
. For example, we can easily observe that people who weigh less are more likely to get breast cancer (an association), but that does not mean that if one loses weight, one will get breast cancer (a causal factor). It is merely because almost all the people who get breast cancer are women and women, on the average, weigh less than men, that there is an association. In spite of this problem of "selection bias," some patients and doctors jumped to the conclusion that metformin can prevent progression. Others (like me) argued that because side effects are mild, there is probably little risk. Let's examine the evidence for causation instead of association.
124 patients were randomly assigned to receive ADT+bicalutamide or metformin+ADT+bicalutamide. Newly diagnosed patients may have been diagnosed with either:
- High-risk localized prostate cancer
- Prostate cancer in distant sites (stage M1)
- Prostate cancer in pelvic lymph nodes only (stage N1)
After short follow-up (18 months, median), 23% of the patients had died. The effects of metformin were:
- Time to castration resistance was delayed in the high-risk group and in those with stage N1
- Time to castration resistance was not slowed significantly in men staged M1, especially no effect in those with a high volume of metastases.
- There was no effect on PSA
- There was no effect on survival
So metformin may slow progression among men who may be cured by radical therapy (removing or irradiating the prostate with or without pelvic lymph nodes) anyway. It is possible that with larger sample size and longer follow-up there may be an effect on survival among metastatic men, but the lack of a PSA response suggests that won't happen.
28 patients with recurrent prostate cancer were given either metformin or observation for 8 weeks. All patients had a short PSA doubling time and a high body mass index. As metformin or placebo continued for 24 more weeks, bicalutamide (50 mg/day) was given to both groups.
- After the initial 8 weeks, PSA dropped in the metformin group
- By 32 weeks, however, there was no difference in PSA
- The trial was ended early for futility
100 patients with metastatic castration-resistant prostate cancer (mCRPC) were randomly assigned to get docetaxel chemotherapy with or without metformin. By the end of chemotherapy (up to 10 infusions, every 3 weeks):
- There was no difference in PSA response between the 2 groups
- There was no difference in objective response
- There was no difference in clinical progression-free response
- There was no difference in overall survival
- There was a higher incidence of diarrhea with metformin
36 patients were randomized to receive metformin+ADT or ADT-alone (+ placebo). Patients were either recurrent (n=15) or newly diagnosed with metastatic prostate cancer (n=21). All were beginning lifelong ADT for the first time. After 28 weeks:
- There was no difference in PSA response between the 2 groups
- There was no difference in metabolic syndrome (see "PRIME RCT" below) between the 2 groups
The researchers conclude that drugs that reduce metabolic syndrome in diabetics do not reduce metabolic syndrome among ADT users.
This RCT was conducted to detect any particular tissue effects metformin might cause. 20 patients were randomized to receive metformin or placebo for 4-12 weeks prior to surgery. Analysis of post-prostatectomy tissue revealed:
- There were no differences in any of the biomarkers of hypothetical biochemical benefit for prostate cancer (see below)
- prospective trial (not randomized)
25 patients with metastatic castration-resistant prostate cancer (mCRPC) who were already taking abiraterone received metformin. After 12 weeks:
- Only 12% were free of progression (35% progression-free survival was the benchmark for an effect)
- Most had PSA progression; almost half had radiographic progression
- There was higher-than-expected gastrointestinal toxicity (nausea, diarrhea, loss of appetite)
(Update 5/24/2022) The MA.32 RCT
among 3,649 women with breast cancer proved that metformin has no effect on breast cancer survival. There were, however, statistically significant increases in serious (Grade 3) toxicities among those taking metformin vs placebo: 21.5% vs 17.5%, respectively. While we cannot extend this result to prostate cancer, it does increase skepticism.
Purported Anti-Carcinogenic Biochemical Mechanisms
Metformin has been used for many years in diabetic people to reduce blood sugar (it blocks glucose production by the liver), counteract the effects of metabolic syndrome, maintain insulin sensitivity, and diminish appetite. Its effects are mediated through its ability to activate an enzyme called AMPK.
There are several hypotheses about how metformin might exert an anti-carcinogenic effect in prostate cancer. All of the hypotheses are based on lab tests rather than clinical data. Some of its purported effects might arise because metformin activates an enzyme called AMPK.
- AMPK in turn activates an enzyme called Acetyl Coenzyme A Carboxylase (ACC) which regulates fatty acid oxidation - the chief source of energy for the prostate cancer cell.
- AMPK also activates a protein (p53) that is an important tumor suppressor.
- AMPK inhibits an enzyme called mTOR. mTOR inhibitors prevent cancer protein synthesis and reduce Cyclin D1 activity. This leads to cell-cycle arrest.
Other proposed anticarcinogenic mechanisms are that metformin:
- increases Cleaved Caspase 3 (CC3), which is needed for apoptosis of mutated cells
- decreases Insulinlike Growth Factor (IGF-1) that allows energy utilization by the cancer. However, because more specific IGF-1 inhibitors have been found to have no effect on prostate cancer: lisitinib, figitumumab, and cixitumumab, this is an unlikely mechanism.
- decreases testosterone formation (it is known to slightly decrease testosterone in women with polycystic ovary syndrome)
- decreases free testosterone formation by increasing sex hormone binding globulin - SHBG.
Nguyen et al.
showed that men taking metformin had no differences in serum or tissue levels of CC3, Cyclin D1, IGF-1, testosterone, SHBG, or mTOR inhibition. This leaves metformin without a plausible mechanism by which it could slow prostate cancer progression.
Lower detection among diabetic men using metformin is a confounder
So why have so many observational studies of metformin in diabetic men found an association with reduced prostate cancer progression? And why has the association failed to be observed in non-diabetic men? A Stockholm study provides a plausible explanation: Beckmann et al.
reported that diabetic men using metformin were less likely to get a biopsy for elevated PSA compared to a matched sample of men who did not use metformin. This suggests that the lower incidence of prostate cancer among men taking metformin is simply that their prostate cancer was less likely to be detected.
Contradictory Evidence from Observational Studies
Because so many patients have relied on observational studies to make a metformin treatment decision, it's worth seeing just what those studies say. All observational studies have been conducted among diabetic men.
is large. It encompasses 30 cohort studies, covering 1.7 million diabetic men. While there was no association with incidence of prostate cancer (no protective effect), there were positive associations with overall survival, prostate cancer-specific survival, and recurrence-free survival. No correction for risk factors or patient matching was possible.
This UK database analysis
of over 55,000 diabetic men, using inverse probability weighting to account for selection bias, found there was no association between metformin use and the detection of any kind of cancer.
This secondary analysis of the REDUCE RCT
compared 194 metformin users to 205 non-users of diabetic medications and 141 who used some other diabetic medication. All were diabetic and had at least one biopsy. After correction for all risk factors, there were no associations for incidence of prostate cancer, neither high grade nor low grade.
Merrick et al.
reported on 65 diabetic men at Wheeling Hospital treated with metformin (median, 6 years), compared to 88 diabetic men treated with another antidiabetic medication, and 881 non-diabetic men who were biopsied. There were no significant differences in prostate cancer diagnosis, Gleason score, number of positive cores, or risk group based on metformin usage. Nor did diabetes make a difference.
A study at Mayo
looked at recurrences among men following prostatectomy. There were 323 diabetic metformin users and 562 diabetic non-users. After 5 years of follow-up, and after correction for known confounders, there were no differences in biochemical recurrence, progression, or all-cause mortality. Neither were there any differences in postprostatectomy pathological findings: Gleason score, stage, positive surgical margin rate, or tumor volume.
Taira et al,
reported on 126 diabetic men at Wheeling Hospital treated with metformin (median, 6 years), compared to 144 diabetic men treated with another antidiabetic medication, and to 2,028 non-diabetic men. All received brachytherapy. There were no 15-year differences in biochemical failures, prostate cancer-specific mortality.
A 1:5 case-control study
of diabetic men diagnosed with prostate cancer was conducted in Ontario. There were 1,104 men who had high-grade and 1,117 men who had low-grade PCa after prostatectomy. In addition, there 3,524 men diagnosed with a biopsy only. Metformin use made no difference in prostate cancer incidence, detection of high-grade PCa, low-grade PCa, or biopsy-detected PCa.
A SEARCH database study
looked at diabetic men who underwent prostatectomies, 156 used metformin, 215 didn't. There was no association found between metformin use, dose or duration of use and time to biochemical recurrence, High metformin dose was associated with earlier castration resistance, metastases, and PC-specific mortality.
A 1:10 case-control UK Database study
looked at 536 diabetic men who had used metformin and 203 diabetic men who hadn't. Metformin use was not associated with prostate cancer incidence. In fact, prostate cancer risk increased in proportion to the number of metformin prescriptions.
Tan et al.
used the SEER database to see if there was an association between metformin, statins, or the combination of the two on prostate cancer mortality in high-risk men. They found that metformin alone had no association. However, statins alone and the combination did have an association. The association disappeared in men with advanced (Stage IV) prostate cancer.
Possible Metformin Danger
In a secondary analysis
of two randomized clinical trial databases, there were 486 patients treated with radiation and ADT. Follow-up was over 10 years. 10-year biochemical recurrence-free survival was:
- 73% if they used metformin
- 85% if they did not use metformin
Metformin was associated with inferior biochemical outcomes.
Should I take metformin?
All of the higher-level evidence so far is consistently showing that there is no benefit in taking metformin for prostate cancer. Also, a plausible mechanism for a beneficial effect is so far lacking. However, all the RCTs so far have been small and short-term, so it is possible that a very large trial with long follow-up, like STAMPEDE, might yet prove there is a small effect, or metformin might prove useful if used early enough, as in men on active surveillance, or in combination with other substances (e.g., statins). There are several ongoing randomized clinical trials (see below).
While metformin does not have serious side effects in most men, it does have gastrointestinal side effects (diarrhea, cramps, nausea, vomiting, and flatulence). It should be avoided in men with known contraindications: lactic acidosis, metabolic acidosis, poor liver or kidney function, and hypoglycemia. There are many drug/supplement interactions that should be carefully checked.
Metformin has been recalled
repeatedly by the FDA because of carcinogenic (NDMA) impurities.
STAMPEDE - ARM K
: Metformin+SOC vs SOC for locally advanced and newly-diagnosed metastatic patients. Results expected: 2024
Enzalutamide vs Metformin+Enzalutamide for mCRPC. Results expected 2022
: Metformin vs Placebo for Active Surveillance. Results expected 2023
: Metformin+atorvastatin vs placebo for recurrent men: Trial terminated for no expected benefit in recurrent men.
: Metformin vs placebo to prevent metabolic syndrome in men starting ADT: Results expected 2022