Metformin is a biguanide molecule, in use for the treatment of T2DM since the 1950s [
79]. Despite initial concerns over the ability of biguanides to induce lactic acidosis [
80], metformin has proven to be a safe and effective member of the therapeutic armamentarium in the management of T2DM. It effectively lowers plasma glucose concentrations when concentrations are high, but it is rare for metformin to be associated with hypoglycaemia when used as a single agent [
81]. Unlike insulin and some other medications used for treating diabetes, it does not cause weight gain. Lactic acidosis is very rare side effect (and usually precipitated by intercurrent illness); gastrointestinal disturbance limits tolerability for 5–10% of users [
82].
Despite nearly 70 years of use, the mechanisms of action of metformin continue to be debated. Metformin inhibits gluconeogenesis in the liver [
83] in patients with T2DM and elevated blood glucose concentrations, but in patients without T2DM, new evidence suggests that metformin may increase endogenous glucose production [
84]. Multiple other pathways have been identified by which metformin might exert beneficial effects on age-related pathophysiology – particularly on mechanisms relevant to energy utilisation. Many of these mechanisms may be relevant to skeletal muscle function. Firstly, metformin inhibits mitochondrial complex 1, and this in turn may reduce the production of reactive oxygen species [
85,
86], thought to be involved in cellular damage and mitochondrial dysfunction. Secondly, metformin activates adenosine monophosphate (AMP)-activated kinase (AMPK), the ‘master regulator’ of energy utilisation by cells. AMPK has effects on autophagy and stress resistance, and in stimulating mitochondrial biogenesis [
87]. A further AMPK-driven action of metformin is to inhibit the mammalian target of rapamycin complex (mTORC). The mTORC pathway is critical to multiple ageing processes, a key example of which is regulation of autophagy. Constitutive mTORC upregulation inhibits autophagy, including in skeletal muscle cells; in animal models the loss of autophagy leads to a skeletal myopathy that is reversible with the mTORC inhibitor rapamycin [
88,
89]. Although high mTORC activity is associated with shorter lifespan in multiple species, mTORC also plays important roles in the maintenance of skeletal muscle mass and prevention of atrophy [
90] suggesting that mTORC inhibition could have deleterious effects on skeletal muscle. The net effect of mTORC inhibition may depend on both the starting activity and the degree of inhibition – skeletal muscle from sarcopenic rats was found to have highly active mTORC, and partial inhibition of mTORC was able to increase muscle mass in this animal model [
91].
Metformin may not act directly on myocytes in a clinically important way however. Human skeletal muscle uptake of metformin appears low in comparison to hepatic, renal and intestinal uptake in radioactive tracer studies using
11C-labelled metformin [
92]. It is therefore not certain that sufficient metformin reaches skeletal muscle to mediate the aforementioned effects directly within myocytes.
Thirdly, metformin inhibits pro-inflammatory cytokine production [including interleukin (IL)-1 and tumour necrosis factor (TNF)-alpha], and intracellular pathways activated by inflammation [via inhibition of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB)] [
93‐
95]; it has also been shown to increase circulating levels of irisin, a key myokine [
96]. Fourthly, metformin suppresses cellular senescence in multiple tissues, including skeletal muscle, which is a major driver of ageing in mammals via its impact on mitochondrial complex I, AMPK, mTOR and NF-κB. Metformin’s senostatic action has been shown to improve frailty (including improvements in skeletal muscle mass and function in a mouse model of premature ageing [
97]) and led to improvements in muscle strength and endurance in a sarcopenic mouse model, along with reduction in a suite of pro-inflammatory cytokines [
98].
Finally, metformin also precipitates changes to the gut microbiome [
99]. This has been proposed as an indirect mechanism of action for some of the effects of metformin, and although the precise biological pathways involved remain unclear, a range of actions have been postulated, including changes to bacterial-derived pro-inflammatory compounds, changes to production of circulating bacterial metabolites with actions on AMPK and indirect effects to stimulate release of GLP-1 by the gut [
100]. Not all effects of metformin on skeletal muscle may be beneficial however; metformin has been shown to reduce gastrocnemius mass in mice in one study, an effect likely driven by induction of myostatin transcription [
101].
3.2 Evidence from Observational and Intervention Studies
Few studies have examined the relationship between metformin use and skeletal muscle health – such studies are challenging because metformin is currently indicated for use in T2DM – a condition which is itself implicated in accelerated ageing and muscle dysfunction. Observational data from the Osteoporotic Fracture in Men (MrOS) study suggested that men with T2DM treated with metformin showed slower loss of muscle mass measured by DXA over time than those not treated with insulin-sensitising agents [
102]; similar results were found among women with T2DM in the Study of Osteoporotic Fractures (SOF) [
103].
One randomised controlled trial suggests promising effects of metformin on physical function [
104]. A total of 120 participants aged 60 years and over with prefrailty (measured and without diabetes were randomised to receive 16 weeks of metformin 500 mg three times a day or matching placebo. Walk speed increased significantly in the treatment group compared with placebo (by 0.13 m/s, which exceeds the minimum clinically important difference for this measure [
105]). No differences were seen for quality of life as measured by the EQ5D tool, handgrip strength or myostatin level; weight change was not reported, and dropout rates were high (≥ 20%) in both arms. The MASTERS trial [
106] also found no augmentation of muscle strength when metformin 850 mg twice daily was added to a 14-week resistance training programme for older men and women, and indeed the muscle hypertrophic response measured by both computed tomography and by DXA to exercise was blunted by metformin when compared with placebo. Similar findings were seen in a small but detailed trial of metformin (up to 1 g twice daily versus placebo) given to healthy older people undertaking 12 weeks of aerobic exercise training. In this trial, metformin prevented the expected improvement in whole-body insulin sensitivity measured by oral glucose tolerance test, and attenuated the expected improvement in maximal oxygen uptake and skeletal muscle mitochondrial respiration from aerobic training [
107]. The focus of these trials was on augmenting the effect of training rather than standalone effects of metformin (the effect of metformin alone was not studied), and although the trial included people with a range of physical capabilities, it did not specifically target those with sarcopenia. These results suggest however that careful targeting may be needed to avoid deleterious effects of metformin by targeting only those populations most likely be benefit (perhaps those with over-active mTOR as discussed above, or those not willing or able to undertake exercise training).
A number of other trials are under way that seek to test the effects of metformin on skeletal muscle function. Of particular note are a randomised, placebo-controlled trial enrolling participants with impaired glucose tolerance [
108], testing metformin up to 1 g twice daily versus placebo for 2 years, and a trial (MET-PREVENT) [
109] enrolling older people with probable sarcopenia, testing metformin 500 mg three times a day versus placebo for 4 months. The results of these trials are necessary to test whether the potential advantages of metformin outweigh any adverse effects of metformin on appetite, food intake, weight and exercise benefits.
3.3 Sulphonylureas and Insulin Secretagogues
Sulphonylureas are another class of drug that have been used for decades to treat DM [
110]. Glinides are a more-recently introduced, molecularly distinct class of compounds with faster onset and offset of action [
111]. Both work by closing adenosine triphosphate (ATP)-sensitive potassium channels (K
ATP channels), a mechanism that parallels the effect of ATP on channel closure. The closure of the K
ATP channel triggers membrane depolarization, calcium influx and insulin release from pancreatic beta cells [
112]. A number of different subtypes of the K
ATP channel exist, with differential expression across a wide range of tissues, including skeletal muscle [
113]. Individual members of the sulphonylurea and glinide drug classes bind with differential potency to these subtypes of K
ATP channel [
114], which may lead to effects outwith the primary mechanism of action based on insulin release.
Although many of the actions of sulphonylureas and glinides are therefore attributable to the actions of endogenous insulin, additional actions may also be mediated by direct effects on skeletal muscle K
ATP channels. Not all sulphonylureas bind to skeletal muscle K
ATP channels, but for those that do (e.g. glibenclamide), this may represent a further mechanism of action. Open K
ATP channels are postulated to protect skeletal muscle against energy depletion in fatigue [
115]; closure of these channels has also been shown to precipitate muscle atrophy in animal models [
116]. Conversely, sulphonylureas may also enhance glucose uptake by enhancing glucose transporter type 4 (GLUT-4) expression at the cell membrane [
117].
Because sulphonylureas and glinide medications lower blood glucose and precipitate hypoglycaemia, studies examining the effect of these medications on muscle function in patients without T2DM have not been performed. Observational studies in patients with T2DM are difficult to interpret; in many studies, sulphonylureas are the comparator class of medication, and it is difficult to dissect out whether sulphonylureas have deleterious effects on muscle function or whether comparator agents have beneficial effects [
101‐
103]. For example, falls were more frequent in patients with T2DM taking sulphonylureas than other drugs for T2DM in a recent observational study [
118]. Skeletal muscle weakness is known to be an important risk factor for falls [
119], but this study was not able to ascertain how much of the effect was attributable to skeletal muscle dysfunction and how much to hypoglycaemic episodes. The current evidence does not support a beneficial effect of these drug classes on muscle function, and their hypoglycaemic effects in people without diabetes mean that they are unlikely to be practical to test further.
3.4 Thiazolidinediones (Glitazones)
Thiazolidinediones (glitazones) activate the peroxisome proliferator-activated receptor (PPAR) gamma, triggering a complex set of metabolic changes centred around lipid metabolism, primarily in adipocytes but also in a wide range of other tissues, including skeletal muscle [
120]. Theoretically, these changes in lipid metabolism, including reductions in free fatty acid concentrations, could reduce intramyocellular lipid deposition, leading to improvements in mitochondrial function and insulin sensitivity [
53,
54] that could be beneficial to skeletal muscle function. Human studies show a reduction in intramyocellular lipid deposition in patients with T2DM treated with glitazones [
121]. However, data from laboratory studies on glitazones is less encouraging; in a young non-diabetic mouse model, pioglitazone did not improve mitochondrial function or grip strength alone or as an adjunct to exercise training [
122].
Observational data from the MrOS study suggested that men with T2DM treated with glitazones showed slower loss of total and appendicular lean mass over time as measured by DXA than those not treated with insulin-sensitising agents [
102]. Uncontrolled muscle biopsy studies enrolling patients with T2DM show that 12 weeks of pioglitazone 30 mg once daily improved skeletal muscle mitochondrial respiration measured by high-resolution respirometry of muscle biopsies, whereas 12 weeks of rosiglitazone 4 mg once daily worsened mitochondrial respiration [
123]; this was despite both agents improving insulin resistance. Beneficial effects were also seen in a randomised trial of pioglitazone 45 mg once daily versus diet for 6 months in patients with T2DM. Both groups showed similar improvements in HbA1c and insulin sensitivity, but the pioglitazone group additionally showed increased levels of mitochondrial gene expression and AMPK phosphorylation [
124]. Similar increases in mitochondrial proteins involved in oxidative phosphorylation were observed in a placebo-controlled trial of pioglitazone 15 mg once daily given for 6 months [
125]. However, another placebo-controlled randomised controlled trial (RCT) using a higher dose of pioglitazone (30–45 mg once a day for 12 weeks) in T2DM failed to find any improvement in maximal ATP synthesis measured using
31P magnetic resonance spectroscopy [
126] despite a decrease in intramyocellular lipids in the pioglitazone arm relative to placebo.
Although no clinical trial has yet evaluated the effect of glitazone therapy on skeletal muscle mass or strength in humans as a standalone therapy, pioglitazone has been tested as an adjunctive therapy in older people with impaired physical function undergoing a weight loss programme. In this trial, 30 mg per day of pioglitazone was associated with less loss of muscle mass measured by computed tomography (CT) and DXA over the 16 week follow-up [
127], although the difference compared with the control group did not reach statistical significance. In women, the pioglitazone group showed a greater increase in leg press power with resistance training than did the placebo group, but this effect was not seen in men [
128]. Glitazones may have additional beneficial effects on skeletal muscle beyond effects on lipid metabolism; for example, troglitazone up to 600 mg per day improved skeletal muscle capillarisation and glucose disposal to a greater extent than metformin up to 2550 mg per day in patients with T2DM [
57]. The broader balance of benefits and risks from glitazone therapy may not support use of these agents to prevent or treat sarcopenia however. Although glitazone therapy can reduce cardiovascular events in selected populations [
129], these agents also increase the risk of clinically apparent heart failure and of fractures [
129,
130], which in turn may reduce physical function and worsen sarcopenia.
3.5 Glucagon-Like Peptide 1 (GLP-1) Receptor Agonists
GLP-1 is released by cells in the gut in response to a meal and enhances the release of insulin from pancreatic beta cells and inhibits release of glucagon by pancreatic alpha cells in response to circulating glucose [
131]. GLP-1 receptors are also present on a range of other cell types, including adipocytes and endothelial cells [
132,
133]. As well as effects mediated by insulin release, GLP-1 receptor agonists improve endothelial function in healthy individuals and stimulate angiogenesis and endothelial cell proliferation [
134]. These effects are also seen in patients with T2DM and in other insulin-resistant states such as obesity [
135,
136] and may be mediated in part by restoration of vascular insulin sensitivity [
137]. In addition, GLP-1 receptor agonists reduce expression of myostatin and pro-inflammatory cytokines in animal models [
138].
In patients with T2DM, GLP-1 receptor agonists cause significant weight loss [
139] – an action often viewed as beneficial, as many patients with T2DM have obesity. However, a recent review of studies of body composition suggests that up to 50% of the weight loss seen from GLP-1 receptor agonist use in patients with T2DM is attributable to a loss of lean body mass [
140]. It is not known at present whether GLP-1 receptor agonist use in patients with or without T2DM causes a clinically significant change in skeletal muscle strength; the loss in total body weight might be expected to facilitate improved mobility, but any loss of lean body mass might attenuate or even negate this benefit. The most relevant evidence comes from three recent randomised controlled trials of liraglutide. The FIGHT trial enrolled 300 patients with chronic heart failure; T2DM was not an inclusion criterion in this trial. Participants had been recently hospitalised and had poor endurance (the mean 6-min walk distance at baseline was only 220 m) [
141]. No difference was seen in 6-min walk distance between the group receiving liraglutide 1.8 mg once a day and the placebo group after 6 months of treatment (between-group difference 5 m;
p = 0.79) but skeletal muscle strength was not tested in this trial. A small trial comparing liraglutide 1.8 mg once a day with placebo in 24 participants with T2DM measured maximal and submaximal measures of endurance but not measures of skeletal muscle strength [
142]. Six months of treatment did not lead to a clinically or statistically significant improvement in maximal oxygen uptake (VO
2 max) in the liraglutide group (liraglutide versus placebo: 18.0 versus 17.0 ml/min/kg;
p = 0.31). Additionally, no significant between-group differences were seen in the improvement in cycle ergometry time or 6-min walk distance during the study. Finally, the large (
n = 3731) SCALE randomised trial of liraglutide 3.0 mg once a day versus placebo to facilitate weight loss in patients with obesity showed a significant improvement in self-reported physical function measured using the SF-36 health status questionnaire over the 1-year follow-up, but objective measures of physical performance were not obtained in this trial [
143]. It is possible that adjuvant interventions to preserve muscle mass and strength (e.g. resistance exercise and provision of adequate protein intake [
144]) may need to be given alongside GLP-1 receptor agonists if they are to be used effectively for sarcopenia, and research on this class of treatments may be best targeted to patients with sarcopenic obesity rather than those of normal or low body mass index. To date, there have been no trials reporting the effect of GLP-1 receptor agonists on skeletal muscle mass or strength in older people with sarcopenia; no such trials are currently registered at either ISRCTN.com or Clinicaltrials.gov
3.6 Dipeptidyl Peptidase 4 (DPP4) Inhibitors
DPP4 inhibitors (‘gliptins’) exert their effects by preventing the inactivation of GLP-1 by the Dipeptidyl peptidase 4 (DPP4) enzyme [
145]. There is therefore considerable overlap between the potential actions of DPP4 on skeletal muscle and the actions of GLP-1 receptor agonists, and the data presented above in relation to GLP-1 receptor agonists are likely to be relevant. Animal models suggest that DPP4 inhibitors increase circulating concentrations of the myokine irisin as well as peroxisome proliferator-activated receptor γ coactivator-1-alpha [
146], a key component of the muscle response to exercise.
Two small observational studies suggest an association between DPP4 inhibitor use and better indices of muscle mass and strength in older people with T2DM. One study showed faster walk speed, greater muscle strength measured by handgrip dynamometry and higher fat free mass measured by bioimpedance (adjusted for height squared) in DPP4 users than those on sulphonylureas [
147], although it is unclear whether this was due to beneficial effects of DPP4 inhibitors or deleterious effects of sulphonylureas. Another study found that patients with T2DM taking DPP4 inhibitors had a slower decline in skeletal muscle index measured by DXA than those not taking DPP4 inhibitors [
148]. In contrast, pharmacovigilance studies have found a higher rate of arthralgia and myalgia (including rare reports of rhabdomyolysis in conjunction with statin medications) in users of DPP4 inhibitors [
149]; the pathophysiological mechanisms underpinning this observation are unclear.
3.7 Sodium–Glucose Co-transporter 2 (SGLT2) Inhibitors
Sodium–glucose co-transporter 2 (SGLT2) inhibitors work by partly blocking reabsorption of filtered glucose in the proximal convoluted tubule of the kidney [
150]. Although the SGLT2 transporter is localised to the kidney, a series of knock-on metabolic effects in patients with T2DM have been described. These include a degree of weight loss (thought to be due to glucose excretion and fluid depletion), but also a reduction in blood pressure as a result of volume depletion, sodium excretion, weight loss and reduction in renin–angiotensin–aldosterone system (RAAS) activity [
151,
152]. Intriguingly, in other organ systems, SLGT2 inhibitors also appear to drive changes that are potentially relevant to skeletal muscle dysfunction, for example, increasing lipolysis and ketogenesis, reducing inflammatory cytokines in liver [
153], the aforementioned reduction in RAAS activity [
154] and upregulation of AMPK via the sestrin pathway with consequent inhibition of mTOR in cardiac myocytes [
153]. Off-target effects via inhibition of the related SGLT1 transport protein may also mediate reductions in oxidative stress seen in cardiac tissue [
155]. Although SGLT2 does not appear to be expressed in skeletal muscle, the related transporter SGLT3 is expressed near the neuromuscular junction [
156]. Its physiological function is unclear, but it may play the role of a glucose sensor [
157]; furthermore, phlorizin (the parent compound from which currently marketed SGLT2 inhibitors are derived) is able to inhibit SGLT3 [
158]
Although few studies have assessed the effects of SGLT2 inhibitors in preclinical models, one recent study has shown that canagliflozin reduces inflammatory cytokine concentrations in skeletal muscle and improves skeletal muscle contractile force in an obese mouse model [
159]. Similar improvements in skeletal muscle size and strength were seen with administration of luseoglifozin in another mouse model [
160]. However, there is currently a lack of evidence from studies in humans. SGLT2 inhibitor use is associated with a reduction in both fat mass and lean body mass (measured using a range of techniques) similar to that caused by GLP-1 receptor agonists; between 20% and 50% of the weight loss seen with these agents may be due to loss of lean mass [
140]. Preferential loss of fat mass may however improve the ratio of lean to fat mass, which would be expected to improve strength:weight ratio. Data on the effect of SGLT2 inhibitors on muscle strength in humans are lacking, but there is at least one trial underway testing the impact of SGLT2 inhibitors (empagliflozin 10 mg once daily versus placebo for 52 weeks) on both fat free mass (measured by bioimpedance) and muscle strength (handgrip strength and five times sit to stand test) in older patients with T2DM [
161]. In patients with heart failure, it is also possible that SGLT2-inhibitor-mediated improvements in cardiac function could translate into improved exercise capacity (a recent meta-analysis suggests a modest 20 m improvement in 6-min walk distance with SGLT2 inhibitors in patients with heart failure [
162], and hence into increased physical activity with consequent benefit on deconditiong. This hypothesis requires testing however. To date, no trials have been conducted using SGLT2 inhibitors as a treatment for sarcopenia in patients without T2DM.
3.8 Insulin
Insulin is a key anabolic hormone, and acts both by the classical insulin receptor and indirectly by stimulating release and availability of insulin-like growth factor 1 (IGF-1) [
163‐
165]. Insulin stimulates glucose uptake into skeletal muscle [
166], thus lowering circulating glucose concentrations. In addition, it inhibits glucose production via effects on glycogenolysis and gluconeogenesis, but also inhibits lipolysis and promotes fat synthesis [
167,
168]. Perhaps most importantly for skeletal muscle health, insulin stimulates muscle protein synthesis and reduces protein degradation and autophagy [
169]. Insulin also exerts important direct effects on the microvasculature, causing vasodilatation and reversing the endothelial dysfunction commonly seen in T2DM [
45].
In patients with T2DM, insulin concentrations are often already supranormal, and treatment with exogenous insulin does not appear to increase protein synthesis, reduce muscle protein breakdown or improve mitochondrial function [
170‐
172]. Nevertheless, in a cohort of Japanese patients with T2DM, decreases in HbA1c resulting from insulin treatment were associated with increases in skeletal muscle mass (measured by bioimpedance) and gait speed and, after adjusting for confounders, were associated with a 1.4% increase in skeletal muscle index over the 1-year follow-up period [
77]. These findings suggest that, in patients with type 2 diabetes mellitus, it may be the reduction in hyperglycaemia, rather than direct anabolic signalling of exogenous insulin, that is of benefit. A similar association between insulin use and a slower decline in skeletal muscle mass measured by bioimpedance over a 3-year follow-up period was found in a cohort of older people in Augsburg, Germany [
173], although measures of muscle function (grip strength and timed up and go) did not show comparable benefit. One small study of 40 patients found no change in handgrip over the first 6 months after starting insulin therapy [
174].
The benefits of insulin treatment on skeletal muscle may be more apparent in patients with type 1 diabetes mellitus (T1DM) [
16]; in patients with T1DM, muscle protein breakdown is inhibited by insulin, and several studies suggest an increase in muscle mass over the first few months after starting insulin therapy [
175,
176], although it is unclear whether this is due to direct anabolic effects of insulin or to the reduction in hyperglycaemia. The effect on muscle strength of starting insulin therapy has not been studied in humans. The profound hypoglycaemia caused by administration of insulin to patients without DM means that the effects of insulin on muscle strength or muscle mass in patients without DM have not been able to be studied to date in humans.