Our read is that taking metformin is well supported, particularly for individuals with metabolic health concerns, though its broader longevity benefits remain uncertain.
Metformin, a molecule derived from a French lily, is noted for its mechanism of AMPK activation, which can reduce hepatic glucose output and potentially benefit fatty liver.
While it shows consistent benefits in diabetics and can be prescribed for standard indications and other benefits, its efficacy for extending human lifespan in non-diabetics is not yet convincingly established.
Lifestyle, especially diet, is considered a more powerful factor for health optimization than molecules like metformin, and concerns exist regarding its side effects and potential to abrogate exercise benefits.
Andrew Huberman suggests that supplements, metformin, and GLP-1 agonists can be taken concurrently, but GLP-1s should be introduced separately from metformin to manage potential nausea. Peter Attia notes that gastrointestinal side effects are often manageable with a slow ramp-up.
David Sinclair states that Metformin can cause metabolic issues and can prevent muscle gain. Peter Attia, Andrew Huberman, and Paul Saladino collectively claim that evidence for metformin extending human lifespan is not convincing, and research is shifting towards rapamycin. Rhonda Patrick states that Metformin abrogates the beneficial effects of daily exercise and inhibits mitochondrial adaptations and improvements in cardiorespiratory fitness after aerobic exercise training in older adults. Paul Saladino claims that Metformin has significant side effects like lowering B12 and does not treat the root cause of diabetes, and that long-term studies on metformin have not demonstrated expected benefits in reducing risks of cancer, diabetes, or cardiovascular disease. Peter Attia and Andrew Huberman note that Metformin failed in the ITP for geroprotective potential. Rhonda Patrick states that Valter Longo would not take metformin due to concerns about blocking central cellular metabolism pathways. Peter Attia claims that the TAME trial will not assess exercise function or cardiorespiratory fitness and that Metformin should not be administered to patients upon arrival at the hospital for COVID-19 treatment due to risk of lactic acidosis. Paul Saladino and David Sinclair state that Metformin carries risks including fatal lactic acidosis, upset stomach, and lack of hunger. Andrew Huberman notes that observational studies may be limited by a lack of matching for medication use between groups. Paul Saladino claims that a pre-diabetic woman with a high grain intake and frequent oil use may have high fasting insulin and be inappropriately prescribed metformin.
The verdict would change if clear dose equivalence between metformin and rapamycin/rapalogs for mTOR inhibition were established, or if the optimal timing for starting geroprotective interventions like senolytics or metformin became certain. Further, if metformin's precise impact on muscle and mitochondrial function were fully known, or if its safety and efficacy for longevity in non-diabetic individuals were definitively established through studies like TAME, the assessment would shift. Additionally, if it were known whether TRIIM Trial treatments reset the entire body's aging clock, or if metformin's potential benefit in non-diabetics clearly outweighed its downside as a weak mitochondrial toxin, the verdict could be altered. Finally, a better understanding of metformin's impact on the microbiome and its potential endocrine-disrupting effects would also influence the assessment.
The intervention improves the primary outcome at standard doses in healthy adults.
The effect size is large enough to matter clinically, not just statistically.
Benefits hold across the populations where it's been tested.
Mechanistic and trial evidence converge on a real, replicable effect.
The effect size is large enough to matter clinically, not just statistically.
Most of the support comes from short or small studies.
Most of the support comes from short or small studies.
Animal-model results don't translate to the human protocol being recommended.
Animal-model results don't translate to the human protocol being recommended.
Animal-model results don't translate to the human protocol being recommended.