Clinical perspectives and concerns of metformin as an anti‐aging drug

Abstract As percentages of elderly people rise in many societies, age‐related diseases have become more prevalent than ever. Research interests have been shifting to delaying age‐related diseases by delaying or reversing aging itself. We use metformin as an entry point to talk about the important molecular and genetic longevity‐regulating mechanisms that have been extensively studied with it. Then we review a number of observational studies, animal studies, and clinical trials to reflect the clinical potentials of the mechanisms in lifespan extension, cardiovascular diseases, tumors, and neurodegeneration. Finally, we highlight remaining concerns that are related to metformin and future anti‐aging research.


| INTRODUC TI ON
The life expectancy in many countries is projected to exceed 85 years by 2030. 1 Globally, one quarter of the population is expected to be in their 60s or older in 2050. 2 Nonetheless, the unprecedented longer life expectancy heralds a staggering number of people living with age-related diseases and considerable burdens on the social, economic, and health care systems worldwide. There is a pressing need to combat the challenges posed by age-related diseases and increase people's health span, the amount of time that people are alive and healthy. However, while age-related diseases can often coexist, current delivery of health services and research efforts have continued to deal with the diseases ineffectively in an insular fashion. 3 In contrast, mechanisms that account for the phenotypes of old age, such as impaired metabolism, dysregulated immune profile, and abnormal DNA methylome, underlie many chronic diseases. 4 Therefore, modifying the mechanisms of aging directly seems to be a more productive approach to fundamentally curb the growth of chronic diseases.
A growing number of studies have indicated that metformin, a well-known drug against type 2 diabetes (T2DM), can potentially stall aging and delay the onset of age-related diseases in humans.
In this review, we survey a selection of articles about metformin to give an overview of the anti-aging mechanisms that have been studied with it and the extent of their effectiveness. Then we highlight remaining questions and concerns that are preventing approved use of metformin as an anti-aging drug and future research directions.

| E X TEND ING HE ALTH S PAN
Almost all life forms constantly sit on a balance between production and maintenance, and under low nutrient conditions when reproduction is more challenging; in order to ensure reproductive success, increasing somatic maintenance is necessary to prolong the reproductively competent period and, consequently, lifespan. 5 Hence, calorie restriction (CR) without malnutrition is one of the most reliable approaches to extend both lifespan and health span in various vertebrate and non-vertebrate species. However, CR is difficult to sustain and implement, since individuals must remain in a state of hunger and endure feelings of starvation, fatigue, and irritations.
Besides, individuals who practice CR are more susceptible to viral infections 6 and resistant to wound-healing, 7 both of which impede its widespread use. Alternatively, metformin can act as a CR mimetic by triggering the nutrient sensing pathways that sense and respond to the changing intracellular and extracellular energy and nutrient levels without actually restricting calorie intake. 8 Metformin acts on the 5′-AMP-activated protein kinase (AMPK) ( Figure 1). The activating capacity of the AMPK signaling pathway declines with aging, and its decline disturbs autophagy, increases cellular stress, and promotes inflammation, which further provokes many age-associated diseases, such as cardiovascular disease, diabetes, and cancer. 9,10 Correspondingly, increased activation of the AMPK pathway has been shown to extend lifespan in animal models in response to CR and pharmaceutical agents, such as metformin. 11 AMPK phosphorylates and activates Unc-51 like autophagy activating kinase 1 (ULK1) of the ULK complex that promotes autophagy as well as activates the forkhead box O-class (FOXO) transcription factors that transactivate the genes involved in detoxification, autophagy, tumorigenesis suppression, and energy homeostasis. 5 Furthermore, metformin attenuates the endoplasmic reticulum (ER) stress and oxidative stress often caused by nutrition overload, aging, and reactive oxygen species (ROS) production. ER and oxidative stress contribute to chronic inflammation, and AMPK inhibits nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), the major regulator of innate and adaptive immunity, and relieves ER and oxidative stress by promoting the expression of mitochondrial uncoupling protein (UCP-2). UCP-2 has an active role in inhibiting the production of reactive oxygen species (ROS) in mitochondria, suppressing the ROS produced by the reduced nicotinamide adenine dinucleotide phosphate (NAD(P)H) oxidase, and inducing the expression of thioredoxin (Trx) by activating FOXO3. 12 Another ground for metformin and nutritional signals to interact is the gut microbiota. Microbial nutrition sensors such as the phosphotransferase system (PTS) controls the downstream transcription factor Crp in response to the nutritional landscape in the gut.
Agmatine, whose level is regulated by Crp, is important for the effects of metformin by interacting with lipid and peroxisome metabolism. In Caenorhabditis elegans, metformin treatment decreased lipid droplet size and increased peroxisomal abundance when the Crp gene is not deleted. The group also used computer modeling to predict whether the PTS-Crp axis is present in T2DM humans treated with metformin, and the predicted agmatine production capacity was significantly higher. 13 Metformin is also implicated in DNA methylation, which is the process of adding a methyl group to the fifth carbon on cytosine facilitated by DNA methyltransferases (DNMT). DNA methylation usually leads to gene silencing by interacting with transcription mechanisms, and global hypomethylation and promoter hypermethylation are often observed in aged people. 14 Metformin activates AMPK and upregulates let-7, a family of microRNAs that bind to and degrade H19, a long noncoding RNA that is normally downregulated in adults. H19 causes aberrant methylation by binding to and inhibiting s-adenosylhomocysteine hydrolase (SAHH), which hydrolyzes s-adenosylhomocysteine (SAH), an inhibitor of DNMT3B. 15 Quite a few studies have corroborated the proposed life-extending effects of metformin. Metformin extended health span and lifespan F I G U R E 1 Metformin could decrease the incidence of age-related diseases. Calorie restriction and activation of IGF-1 receptor (IGF-1R) and insulin receptor (IR) lead to downstream effects such as decrease of pro-apoptotic effect, ROS (reactive oxygen species) detoxification, and inhibition of cell cycle arrest, which help protect against age-related diseases. Metformin elicits the same effects by activating AMPK and inhibiting mTORC1

| C ARD I OVA SCUL AR D IS E A S E S
Cardiovascular diseases (CVDs) are the leading cause of death globally, claiming around 17.9 million lives per year. 19 Aging is the dominant CVD risk factor. 20 In the United States, roughly 70%-75% of people who are 60-79 years old are afflicted with CVDs. 21 Dyslipidemia, insulin resistance, and chronic inflammation that commonly occur in older people make them more susceptible to CVDs. AMPK activation by metformin can suppress fatty-acid desaturase (FADS) genes, reducing the circulating levels of lipid metabolites and low-density-lipoprotein (LDL) cholesterol. 22 Metformin also improves insulin sensitivity, helps weight loss, and reduces perceived hunger and food intake. 23 This can be attributed to the interaction between the GDNF family receptor α-like (GFRAL) receptor in the central nervous system and growth differentiating factor 15 (GDF-15), whose expression is increased most prominently in the liver and gastrointestinal system by metformin. 24 Metformin's protective effects have been confirmed in both animal and human studies. Chronic low doses of metformin given to mice that had poor lipid-clearing capabilities and age-related atherosclerosis led to reduced subendothelial inflammation and pro-inflammatory cytokines levels. 28 Bovine aortic endothelial cells exposed to clinically relevant amounts of metformin had increased activities of nitric oxide synthase (eNOS), endothelium-derived nitric oxide (NO), and AMPK, while no such effects were observed in AMPK knockout mice. NO and eNOS maintain vascular homeostasis and its integrity, supporting the vascular-protective effects of metformin. 29  gene was also inactivated, suggesting that a feed-forward response to continuously suppress H19 can be established by metformin. 15 In addition, the 11 differentially methylated CpG sites mentioned earlier were related to multiple tumor-related genes: SIX3 downregulation in lung cancer due to promoter methylation was rescued by metformin; POFUT2 is linked to glioblastoma and adenocarcinoma; MUC4 is implicated in pancreatic cancer; KIAA1614 is related to colon cancer; lastly, UPF1 is associated with genome stability. The differentially methylated regions revealed the gene EPHB1, whose under-expression leads to gastric carcinoma and invasion of colorectal cancer cells, and SERP2, which is positively correlated with body mass index (BMI) and abnormal glucose tolerance as well as colorectal cancer. Pathway enrichment analysis found the unfolded protein response, which is involved in metformin-induced apoptosis in acute lymphoblastic leukemia. 18 In 2005, a case control study first discovered metformin reduced risk of cancer in diabetic patients. 42 43 Diabetic patients who took metformin also had 7% less chance of getting hepatocellular cancer for each incremental year they took metformin, and it was attributed to inhibited proliferation and cell cycle arrest induced by metformin. 44 Nevertheless, evidence from randomized control trials has been largely inconclusive. 45,46 Additionally, a study compared metformin with rosiglitazone and sulfonylureas, and metformin did not reduce malignancy rates in people. 47 Multiple meta-analyses also did not support metformin reducing cancer incidence. 48,49 Work is still needed to resolve these inconsistencies. had a similar risk of developing AD to individuals without diabetes (AOR 1). Furthermore, patients who had been exposed to various anti-diabetic drugs and had received more than 60 prescriptions of metformin had higher risk of developing AD (AOR 1.71, 95% CI 1.12-2.60), which was attributed to the production of A-β peptides, a hallmark for AD. However, the risk did not increase with the number of metformin prescriptions, and patients who had taken metformin exclusively did not have increased risk (AOR 1.00, 95% CI 0.55-1.81, Parkinson's disease (PD) affects around 1% of people aged >60 years. 60 Loss of dopaminergic neurons is characteristic of PD and leads to imbalance between dopamine and acetylcholine.

| NEURODEG ENER ATION
Over-activation of cholinergic system causes motor and cognitive disturbances. Hence, current PD drugs either provide more dopamine or reduce the amount of acetylcholine to restore the balance, working as a remedy instead of neuroprotective agents. [61][62][63] Parkinson's disease, diabetes, and dementia share mitochondrial bioenergetics disorder and abnormal protein folding in their pathogenesis, and several studies have found metformin to alleviate PD.
An analysis of a cohort of 800,000 people from the Taiwan National Health Insurance database showed that T2DM patients had 2.2-fold risk of PD, and metformin-inclusive sulfonylurea therapy reduced the risk (HR 0.78 relative to diabetes-free, 95% CI 0.61-1.01). 64 The reason has to do with metformin's ability to reduce α-synuclein re-

| CON CLUD ING REMARK S AND FUTURE PER S PEC TIVE S
Although current research gives promise to metformin as an antiaging drug, there are still concerns that need to be highlighted, and they apply not only to research into metformin but to other antiaging mechanisms and drug research as well.
First, despite the positive outcomes from many studies, it is not uncommon to find a change in dosage turning the result from life-extending to life-ending. When a low dose of metformin (0.1%) was given to middle-aged male mice with their diet, their lifespans were extended by 5.83% on average, but a higher concentration (1%) became toxic. 17 In another study, although metformin activated AMPK and suppressed lipid storage in fruit flies, their lifespan did not increase. At higher doses (25 and 50 mM) metformin reduced the survival rates. The authors reasoned the causes to be excessive starvation, disrupted intestinal fluid homeostasis, or metformin toxicity. 68 In the PD study with mice models, although metformin increased BDNF and GDNF levels, the high dosage (400 mg/kg) killed all the mice. 66 The issues with dosage along with physical and genetic differences between humans and animals make scaling the positive lab results for human use a tricky matter. In the study that showed metformin's beneficial effects for treating CVDs in mice, the dosage was 200 mg/kg, which is definitely not applicable to humans. 31 Hence, conducting human clinical trials may be a more straightforward approach to finding a safe and effective dosage for human use.
Encouragingly, when anti-diabetic doses of metformin were given to 12 pre-operative endothelial cancer patients and comparison of their tissue samples before and after the operation was made, the same effects observed in vitro were found-increased AMPK phosphorylation, decreased tumor cell proliferation, and decreased H19 levels. 15 Another issue standing in the way relates to the side effects associated with chronic use of drugs. About 25% of patients treated with metformin have gastrointestinal side effects associated with the phenotype of organic cation transporter 1 (OCT1). 69 Besides, chronic use of metformin can cause dose-dependent vitamin B12 deficiency, increasing the risk for anemia and neuropathy. 70,71 Lactic acidosis has been reported as a side effect of metformin, but there have been controversies. In the study using diabetes model mice to study AD-like brain changes, metformin did not further increase the serum lactate concentrations. 59 Whether this holds true in non-diabetic mice or humans is yet to be seen. The issues with side effects can be addressed in four ways. The first is to selectively take supplements, such as vitamin B12, to make up for the loss. The second is to reduce the dose and increase the interval between every dose, but more studies are needed to calibrate the balance between safety and anti-aging potency. The third is taking a variety of anti-aging drugs (also known as drug cocktail therapy), each with a very low dose, instead of taking only metformin, since the side effects are dose-dependent. Although cultured cells and a mice study both showed metformin combined with insulin reduced A-β peptide levels, 58,59 the GPRD study showed long-term combined exposure to metformin and other anti-diabetic drugs increased the risk of AD, while using only metformin did not show any difference. 72 Therefore, this method requires further validation as well.
The fourth way is to find analogs with fewer side effects. Currently, metformin has few available analogs with well-studied and tolerated side effects. Phenformin and buformin, two biguanide drugs like metformin, were withdrawn from the market due to fatal lactic acidosis. 73 Mito-metformin, synthesized by adding a positively charged triphenylphosphonium group to metformin, showed 100-fold to 1000-fold anti-proliferative effects on cancer cells depending on alkyl chain length, but how the dramatically improved potency will impact healthy cells is poorly understood at the moment. 74 Future research should also work to elucidate how gender in- Understanding this complex network of interactions will not only promote further understanding of metformin but also help develop more anti-aging measures targeting epigenetic modifications. In addition, SNPs in the human genome also affect the efficacy of drugs ( year-old people who were randomized to take metformin and placebo in either order each for six weeks with a two-week washout period in between. As the number of subjects was small and the duration short, MILES effectively revealed many transcriptomic and metabolomic changes in the muscle and adipose tissue. 82 The Glucose Lowering In Non-diabetic hyperglycaemia Trial (GLINT) is intended to evaluate the performance of metformin in reducing CVD risks by following 20,000 hyperglycemic but non-diabetic patients for five years. A one-year feasibility RCT enrolling 249 elderly, obese, and with high CVD risk (mean 28.8%, SD 8.5%) participants was concluded in 2018, and metformin improved several CVD risk indicators and decreased vitamin B12 levels. However, it also revealed problems such as a high rate of trial discontinuation (30% by six months). 83 The Targeting Aging with Metformin (TAME) trial is a large placebo-controlled trial that is designed to enroll 3000 subjects to test whether metformin delays age-related diseases. 84 The TAME trial received FDA approval in 2015, and after receiving all the required budget in 2019, it was set to start at the end of the same year. The TAME trial may make metformin the first approved drug for anti-aging, but, more importantly, since it is not testing metformin against a single disease but a collection of age-related ones, it establishes aging as a medical condition that can be intervened or treated instead of an irreversible process outside human control.
The shift in the notion of aging will enable future anti-aging clinical to trials proceed with much more ease. 85

ACK N OWLED G M ENTS
We sincerely appreciate the support from the Lars Blound institute of Regenerative Medicine in Qingdao. We thank Jessica Mar for her insightful comments in language revising and proofreading during the composition of the manuscript.

CO N FLI C T S O F I NTE R E S T
Nothing to disclose.