Metformin misses the mark?

Metformin, a medication commonly used for type 2 diabetes and blood glucose control, has also been a staple in conversations surrounding longevity over the last couple of decades. Indeed, in a premium article published last year, we evaluated in detail the evidence for alleged lifespan effects beyond the context of diabetes. To make a long story short, we were unimpressed.

But the current state of evidence hasn’t stopped researchers from conducting clinical trials on potential longevity-related or glycemia-independent benefits of metformin. A recent randomized controlled trial (RCT) known as MET-PREVENT set out to test metformin’s effect on physical performance in older adults diagnosed with probable sarcopenia (i.e., age-related loss of skeletal muscle mass) and physical pre-frailty or frailty.¹ The results were underwhelming, to say the least, but do they really close the door on this possible metformin use-case?

Considering the conclusion from our previous premium article, it would be easy for us to file these results away as another nail in the coffin for metformin as a geroprotective molecule. But I’d argue that this is exactly when it’s worth leaning in, because the real value here isn’t simply in seeing the headline result and nodding in agreement, it’s in understanding why the results landed where they did.

While it’s tempting to dismiss studies that don’t tell us what we want to hear—or blindly accept those that do—there’s a crucial discipline in giving them both the attention required for appropriate critical appraisal.

Why would metformin seem promising in frailty or sarcopenia?

Metformin has been proposed to exert a host of cellular effects that could, in theory, translate to improved muscle health, making it a compelling candidate in discussions about preventing frailty and sarcopenia. These mechanistic impacts range from modulating mTOR activity and regulating mitochondrial complex 1 to exerting senostatic effects (essentially, slowing the accumulation of senescent cells).^2,3 Add to this metformin’s role in dampening the production of proinflammatory cytokines, and you’ve got a pharmacological profile that seems promising for counteracting the multifaceted drivers of muscle decline.⁴

Observational studies have fueled this optimism. In people with type 2 diabetes, metformin use has been associated with a slower loss of muscle mass compared to alternative therapies. Even more intriguing, data suggest that both the prevalence and incidence of frailty, as measured by cumulative deficits, and the rates of frailty-related outcomes, such as falls, are lower among diabetic metformin users than among diabetic non-users.^5,6

But observational data are subject to the influence of confounding variables, especially in the context of diabetes. It’s plausible those on metformin could have their diabetes under better control than those who don’t take metformin, allowing users to engage in more physical activity due to the absence of diabetes-associated joint stiffness and peripheral nerve damage, thereby protecting their muscles from atrophy. 

Thus, we’re left with a straightforward yet critical question: do the promising observational signals hold up when examined more rigorously through an RCT in individuals without type 2 diabetes? 

This is exactly where the MET-PREVENT study, led by Witham and colleagues, stepped in. Could metformin improve physical function in older adults who are frail or pre-frail, even in the absence of overt metabolic disease?

About the study

The MET-PREVENT trial was a four-month, double-blind, placebo-controlled RCT designed to assess whether metformin could improve physical performance in frail, older adults. Participants had a mean age of 80.4 years and met inclusion criteria indicative of probable sarcopenia and physical pre-frailty or frailty (e.g., maximum hand grip strength <16 kg for women and <27 kg for men, 4-meter walk speed of <0.8 m/s), and thus, this group was at high risk for adverse outcomes such as falls and fractures. 

Participants were randomized to receive either immediate-release metformin at a commonly used dose of 500 mg three times per day (n=36) or a matched placebo (n=36). Randomization was stratified by sex and by baseline walk speed (≤0.6 m/s vs. >0.6 m/s), and the primary endpoint was the change in 4-meter walk speed from baseline, adjusted for initial values, as this test is regarded as a metric of frailty according to European Working Group on Sarcopenia in Older People (EWGSOP) guidelines. 

After four months, in the intention-to-treat population, the findings were strikingly clear: metformin showed no statistically significant benefit in improving the primary endpoint. Relative to a baseline mean 4-meter walk speed of just 0.59±0.17 m/s in the metformin group, the post-treatment mean was virtually unchanged (0.57±0.19 m/s), and results in the placebo group were essentially identical to those of the metformin group (from a baseline mean of 0.60±0.26 m/s to a post-treatment mean of 0.58±0.24 m/s). This corresponded to a non-significant adjusted treatment effect of 0.001 m/s (95% CI: -0.06–0.06; P=0.96). 

Even in the per-protocol analysis (limited to participants with at least 80% adherence) results were consistent with the primary analysis. Subgroup analyses by age, sex, baseline walk speed, and insulin resistance also failed to reveal any meaningful differential effects, and secondary outcomes utilizing other sarcopenia metrics painted a similar picture as the 4-meter walk test. In all, the results revealed no significant differences between groups in any measures of strength, mobility, muscle mass, or quality of life.

Where does this leave us?

Metformin didn’t improve physical performance, muscle mass, or quality of life in older adults with frailty and probable sarcopenia. It would be easy to view these findings as yet more proof against the idea of metformin as a longevity intervention or a treatment with benefits beyond glycemic control. After all, that interpretation would certainly align with the opinion that we had expressed about metformin before this study was published, and it’s tempting to accept any result that seems to validate our own biases. 

But as stated at the outset, we must apply a critical eye, especially when study results affirm our beliefs. And when we apply such an eye to the MET-PREVENT trial, we see that this trial simply wasn’t a good test of the question it sought to answer, and the reason the results landed where they did may have had nothing to do with whether or not metformin truly has any potential in reducing frailty. 

Moreover, to play the devil’s advocate, the negative result doesn’t invalidate the underlying biological mechanisms; it simply tells us that, in this population, with this dose and duration, no clinical benefit was observed. The challenge is now to dig into why an intervention failed and what we can learn from the failure.

Too little, too late?

First, let us consider the population selected for this trial. MET-PREVENT focused on older adults (mean age of 80.4) with sarcopenia and physical pre-frailty or frailty. The overall mean gait speed, just 0.59 m/s, was well below the threshold for maintaining robust functional independence. In other words, these individuals were already deep into the physiological decline that characterizes advanced aging.

By this stage, mitochondrial dysfunction, a high burden of senescent cells, neuromuscular disintegration, and chronic systemic inflammation are often entrenched and, in many cases, irreversible by metabolic modulation alone. Metformin’s strongest evidence for benefit comes from earlier in the metabolic deterioration continuum, in settings like insulin resistance, prediabetes, or pre-frailty, where there’s still a degree of plasticity in metabolic and cellular pathways. In such populations, interventions that target metabolic dysfunction might still be able to meaningfully shift the trajectory of decline, but in the severely frail participants in MET-PREVENT, metformin was likely deployed too late in the game to make a significant difference.

Moreover, even for the population selected, the trial was underpowered to answer deeper mechanistic questions. While the study conducted exploratory subgroup analyses, it wasn’t powered to detect meaningful differences in these subgroups. When a trial is designed to detect a moderate treatment effect across an entire heterogeneous population, it often lacks the granularity to uncover effects within specific subgroups, such as by sex, metabolic status, inflammatory markers, or other relevant factors.

Finally, a central challenge highlighted by the MET-PREVENT trial is a misalignment between the biology an intervention is intended to affect and the clinical outcomes used to test it. The primary endpoint—change in 4-meter walk speed over four months—is a validated marker of frailty and predicts morbidity and mortality, particularly for detecting changes due to interventions that directly target systems important for walking performance. For example, one study on supplementation with leucine-enriched protein and vitamin D in older adults with sarcopenia was associated with improvements in gait speed in 4–8 weeks, consistent with effects on muscle protein synthesis and neuromuscular function.⁷ Metformin, however, is hypothesized to act further upstream, through cellular energy sensing and metabolic signaling, and could be expected to exert effects more gradually and indirectly. Evaluated over just four months using an acute functional endpoint, null results may therefore reflect a mismatch between biology and study design rather than an absence of biological effect.

Bottom line

The MET-PREVENT trial reminds us of an uncomfortable but essential truth in longevity science: promising mechanisms don’t always translate to clinical outcomes, especially when assessed in the wrong population, over too short a timeframe, using crude endpoints. Metformin’s null effect in this trial doesn’t mean the drug is useless for aging, but it does tell us that deploying it in severely frail, sarcopenic 80-year-olds for four months isn’t the answer. As such, this trial was not a good test of metformin’s potential. So, the real insight here isn’t in writing off metformin, it’s in refining the questions we ask and the tools we use to answer them. Longevity interventions require thoughtful targeting, mechanistic validation, and long-term thinking. Anything less risks missing the point entirely.

For a list of all previous weekly emails, click here. 

podcast | website | ama

References

1. Witham MD, McDonald C, Wilson N, et al. Metformin and physical performance in older people with probable sarcopenia and physical prefrailty or frailty in England (MET-PREVENT): a double-blind, randomised, placebo-controlled trial. Lancet Healthy Longev. 2025;6(3):100695. doi:10.1016/j.lanhl.2025.100695
2. Kulkarni AS, Aleksic S, Berger DM, Sierra F, Kuchel GA, Barzilai N. Geroscience-guided repurposing of FDA-approved drugs to target aging: A proposed process and prioritization. Aging Cell. 2022;21(4):e13596. doi:10.1111/acel.13596
3. Moiseeva O, Deschênes-Simard X, St-Germain E, et al. Metformin inhibits the senescence-associated secretory phenotype by interfering with IKK/NF-κB activation. Aging Cell. 2013;12(3):489-498. doi:10.1111/acel.12075
4. Cameron AR, Morrison VL, Levin D, et al. Anti-inflammatory effects of metformin irrespective of diabetes status. Circ Res. 2016;119(5):652-665. doi:10.1161/CIRCRESAHA.116.308445
5. Sumantri S, Setiati S, Purnamasari D, Dewiasty E. Relationship between metformin and frailty syndrome in elderly people with type 2 diabetes. Acta Med Indones. 2014;46(3):183-188. https://www.ncbi.nlm.nih.gov/pubmed/25348180
6. Wang CP, Lorenzo C, Habib SL, Jo B, Espinoza SE. Differential effects of metformin on age related comorbidities in older men with type 2 diabetes. J Diabetes Complications. 2017;31(4):679-686. doi:10.1016/j.jdiacomp.2017.01.013
7. Rondanelli M, Cereda E, Klersy C, et al. Improving rehabilitation in sarcopenia: a randomized-controlled trial utilizing a muscle-targeted food for special medical purposes. J Cachexia Sarcopenia Muscle. 2020;11(6):1535-1547. doi:10.1002/jcsm.12532