Diabetic peripheral neuropathy (DPN) affects ∼50% of the 500 million people with type 2 diabetes worldwide and is considered disabling and irreversible. The current study was undertaken to assess the effect of metformin on peripheral neuropathy outcomes in type 2 diabetes. Participants with type 2 diabetes (n = 69) receiving metformin were recruited and underwent clinical assessment, peripheral nerve ultrasonography, nerve conduction studies, and axonal excitability studies. Also concurrently screened were 318 participants who were not on metformin, and 69 were selected as disease control subjects and matched to the metformin participants for age, sex, diabetes duration, BMI, HbA1c, and use of other diabetes therapies. Medical record data over the previous 20 years were analyzed for previous metformin use. Mean tibial nerve cross-sectional area was lower in the metformin group (metformin 14.1 ± 0.7 mm2, nonmetformin 16.2 ± 0.9 mm2, P = 0.038), accompanied by reduction in neuropathy symptom severity (P = 0.021). Axonal excitability studies demonstrated superior axonal function in the metformin group, and mathematical modeling demonstrated that these improvements were mediated by changes in nodal Na+and K+conductances. Metformin treatment is associated with superior nerve structure and clinical and neurophysiological measures. Treatment with metformin may be neuroprotective in DPN.

Article Highlights
  • We aimed to assess whether peripheral neuropathy outcomes for patients with type 2 diabetes were better on metformin treatment.

  • Metformin and nonmetformin groups with type 2 diabetes were matched for demographic and metabolic factors.

  • Clinical neuropathy scores, peripheral nerve ultrasonography, nerve conduction studies, axonal excitability, and mathematical modeling findings were all superior in the metformin group. Mathematical modeling suggested this was due to superior nodal Na+and K+conductances.

  • Treatment with metformin may be neuroprotective in diabetic peripheral neuropathy.

Type 2 diabetes affects >500 million people worldwide (1). Diabetic peripheral neuropathy (DPN) occurs in ∼50% of individuals with type 2 diabetes and causes muscle weakness, neuropathic pain, and sensory loss (2). There is no definitive treatment of DPN, and previous studies have failed to demonstrate any significant influence of strict glycemic control on peripheral neuropathy outcomes (3–6).

Metformin is an oral antihyperglycemic agent with a range of metabolic and cellular actions. It has been shown to lower fasting and postprandial glucose levels, increase insulin sensitivity, inhibit hepatic gluconeogenesis (7,8), enhance the activity of cellular insulin receptor substrate 2, and increase glucose cellular uptake via GLUT-1 (9). Neuroscience studies have suggested that metformin may slow aging and protect against neurodegenerative and cerebrovascular disease by modulating mitochondrial metabolism, insulin sensitivity, and activation of AMPK (10–14).

The current study was undertaken to assess the effect of metformin on peripheral neuropathy outcomes in type 2 diabetes. The primary end point was assessment of nerve morphology, which was undertaken using peripheral nerve ultrasonography, a clinically validated technique (15) that has reliably revealed prominent changes in DPN (16,17) and in other forms of peripheral neuropathy (18,19), with an increase in nerve cross-sectional area (CSA) the most characteristic pathological finding. In addition to standard clinical assessment and nerve conduction studies, mechanistic investigations were also undertaken using specialized axonal excitability techniques that provide information on the activity of voltage-gated ion channels, energy-dependent pumps, and exchangers that play a role in impulse conduction in peripheral nerves (20). These studies were undertaken to provide insight into the potential basis for metformin-induced effects on peripheral nerve function.

The current study recruited 69 participants with type 2 diabetes receiving metformin therapy as part of standard clinical care from the Diabetes Centre, Prince of Wales Hospital, Sydney. Studies were approved by the Human Research Ethics Committee of the South-Eastern Sydney Local Health District. All assessments were undertaken with informed consent. Exclusion criteria included age <18 years, use of glucagon-like peptide-1 receptor agonists (21), peripheral neuropathy due to other causes, including peripheral nerve inflammatory disease, vitamin B12 deficiency, and prior exposure to neurotoxic medication, including immunotherapy and chemotherapy. In addition, 318 individuals with type 2 diabetes who were not receiving metformin at the time of obtaining informed consent or in the month before obtaining consent were concurrently screened. From this group, 69 participants were selected, matched to the metformin group for age, sex, diabetes duration, BMI, HbA1c, serum triglyceride and cholesterol levels, and use of other oral antihyperglycemic agents, namely sodium–glucose 2 cotransporter inhibitors, dipeptidyl peptidase-4 inhibitors, and sulfonylurea treatments. All participants underwent peripheral nerve ultrasonography of the median and tibial nerves in conjunction with clinical assessment of peripheral neuropathy and lower-limb nerve conduction studies. Axonal excitability studies were also performed on all participants, with analysis of excitability data undertaken in comparison with data from 39 age-matched control subjects.

Data from electronic and hard-copy medical records over the previous 20 years were also collected to assess the duration of the metformin use in the metformin group and to assess whether there had been prior prescription of metformin in participants who were assigned to the nonmetformin group. Initiation date and total daily dose of metformin was recorded in the metformin group. For the nonmetformin group, duration of any prior metformin therapy, total daily dose, and time since metformin cessation were recorded. Systolic blood pressure at the time of neuropathy assessment was recorded for all participants.

Median and tibial nerve ultrasonography was performed by a single ultrasonographer. Images were obtained with a 10–18 MHz linear array transducer (MyLabOne, Esaote, Genoa, Italy) using consistent gain and focus settings and the musculoskeletal factory preset (acoustic power 100%, line density set at medium, dynamic range set at 14, and persistence set at 1). Depth settings were altered as required, and the probe was maintained at a 90° angle. Peripheral nerve CSA was calculated by three freehand traces of the inner margin of the endoneurium, with the mean value used (22). Median nerve CSA was measured at a location one-third of the length of the forearm proximal to the wrist in the participant’s dominant hand, after tracking the nerve proximally from the wrist, and captured again away from a potential entrapment site (22). Tibial nerve CSA was measured at 5 cm proximal to the medial malleolus, away from any entrapment site (23). The nerve was tracked proximally along with tibial vessels from the ankle.

Clinical severity of peripheral neuropathy was calculated using the modified Toronto Clinical Neuropathy Scale (mTCNS) (24) and the Total Neuropathy Score (TNS) (25). The mTNCS consists of 6 symptom scores and 5 sensory examination scores, each valued from 0 to 3, with a maximum possible score of 33. The TNS comprises eight sections: sensory and motor symptoms, sensory and motor examination, deep tendon reflexes, and sural tibial nerve amplitudes on nerve conduction studies (Natus, Middleton, WI). The TNS can yield a maximum score of 4 for each section and 32 in total. A higher score for both the TNS and mTCNS represents more severe peripheral neuropathy. The TNS scores further subdivide into four grades of severity: TNS 0–1 = grade 0, 2–8 = grade 1, 9–16 = grade 2, 17–24 = grade 3, and 25–32 = grade 4. Nerve conduction studies were undertaken on the tibial motor and sural sensory nerves using a Medelec Synergy system (Oxford Instruments, Surrey, U.K.) with conventional electrode placement as outlined in previous studies (26).

Median motor axonal excitability assessment was performed on each participant using the TROND protocol and Qtrac software (Digitimer, London, U.K.), with nonpolarizing surface electrodes (Ambu, Sydney, Australia) and a DS5 Isolated Bipolar Current Stimulator (Digitimer). TROND is the consensus protocol for the automated threshold tracking technique developed to assess peripheral nerve excitability measures and first displayed at a neurophysiology conference in Trondheim, Norway. According to this standard protocol, skin temperature was maintained >32°C, and compound muscle action potentials (CMAPs) were recorded from the abductor pollicis brevis muscle. Excitability values obtained in the disease cohorts were compared with healthy control participants (n = 39; 22 men, 16 women), with a mean age of 64.6 ± 1.2 years, and a mean BMI of 23.9 ± 0.5 kg/m2.

Stimulus-response curves were produced using 1-ms test pulses to obtain the highest CMAP amplitude, and 40% of maximal target response was calculated. Threshold current to achieve this target response was tracked in strength-duration behavior, threshold electrotonus, and recovery cycle. Strength-duration relationship was assessed by tracking responses over 0.2-, 0.4-, 0.8-, and 1-ms stimulus durations. Strength-duration time constant was calculated using the Weiss law as an indirect measure of persistent Na+ conductances at the node of Ranvier (27). Rheobase is defined as the minimum current (of infinite duration) required to elicit a response from the nerve. Threshold electrotonus was assessed using subthreshold depolarizing and hyperpolarizing conditioning currents, providing information on voltage-gated nodal and internodal conductances. Threshold changes were assessed after 1-ms test pulses during or after subthreshold conditioning currents of 100 ms at +40% (depolarizing) or −40% (hyperpolarizing) of the control threshold. Threshold percentage change was tracked at 10-ms intervals and depolarizing threshold change at 10–20 ms. S2 accommodation is the segment of depolarizing threshold electrotonus where threshold reduction is limited and approaches control levels. Hyperpolarizing threshold change was tracked at 10–20 ms, 20–40 ms, and 90–100 ms. The recovery cycle plotted threshold changes through 2- to 200-ms conditioning intervals after the supramaximal stimulation of 1-ms duration. During the early part of the recovery cycle, known as the relative refractory period, reduced excitability occurs due to Na+ channel inactivation. Superexcitability then occurs associated with a decrease in the threshold, reflecting the activity of fast K+ channels. Subexcitability is the final period of the recovery cycle where activation of slow K+ channels typically results in an increase in the threshold (28).

Statistical Analysis

A single trained investigator undertook analysis of ultrasound images, nerve conduction waveforms, and axonal excitability recordings, blinded to all participant data, including clinical and treatment histories. Data were analyzed for all measured variables using SPSS Statistics 26.0 for Windows, with statistical significance defined as P < 0.05. Participants were deidentified, and analysis of relevant variables was performed using a coded system. Normality of data was tested with the Shapiro-Wilk test. Where relevant, results are reported as mean ± SE. Multiple regression analysis was also undertaken to assess the influence of confounders including age, sex, duration of disease, HbA1c, estimated glomerular filtration rate (eGFR), and BMI.

Mathematical Modeling

To investigate the effect of metformin on peripheral nerve function, axonal excitability recordings from each group were analyzed using the Bostock model of axonal excitability (20). This is a validated model of the human axon and has been used to assist in the interpretation of axonal excitability studies of metabolic, toxic, inflammatory, and genetic neuropathies (20,29). The model assists in the interpretation of axonal excitability recordings by providing information on potential changes in and around the axonal membrane. This includes parameters such as resting membrane potential, nodal and internodal Na+and K+channel conductances and permeabilities, Na+/K+pump function, and Na+and K+concentrations, as well as other biophysical properties. The model was first adjusted to fit the mean axonal excitability obtained from the normal control group before fitting the mean data from metformin and nonmetformin groups. Using the control group as a baseline, modeling analyses involved changes in a single or combination of parameters in an iterative fashion to objectively fit simulated excitability data to the mean recorded data of the disease groups as closely as possible using a least squares approach.

Data and Resource Availability

The data and resources that support the findings of this study are not publicly available due to institutional ethics review board restrictions.

Participant demographic data are summarized in Table 1. Participants were divided into metformin and nonmetformin groups and were matched for age, sex, diabetes duration, BMI, eGFR, HbA1c, serum triglyceride and total cholesterol concentration, and use of sodium–glucose 2 cotransporter inhibitors, dipeptidyl peptidase-4 inhibitors, and sulfonylurea treatments. In the metformin group, the mean duration of metformin therapy was 162 ± 14 months, and the mean daily dose was 1,523 ± 69 mg. In the nonmetformin group, 33 participants had never received metformin according to clinical history and medical records. For the remaining 36 participants, the mean time since cessation of metformin was 33 ± 6 months.

Table 1

Participant demographics

P value
On metformin (n = 69)Not on metformin (n = 69)Metformin and nonmetforminNonmetformin and control subjects
Age (years) 66.9 ± 1.5 65.3 ± 1.6 0.605 0.217 
Sex (male:female) 42:27 41:28 0.862  
Diabetes duration (months) 183 ± 14 180 ± 15 0.904  
BMI (kg/m230.7 ± 0.7 31.4 ± 0.7 0.458 <0.001 
Mean SBP (mmHg) 132.3 ± 1.7 134.6 ± 2.3 0.419  
HbA1c (mmol/mol) 65 ± 2 70 ± 2 0.112  
HbA1c (%) 8.1 ± 0.2 8.6 ± 0.2 0.114  
eGFR (mL/min/1.73m²) 65.4 ± 2.6 59.7 ± 3.1 0.153  
Chronic kidney disease (Y:N) 29:40 34:35 0.393  
Triglycerides (mmol/L) 2.2 ± 0.3 2.1 ± 0.2 0.662  
Total cholesterol (mmol/L) 3.8 ± 0.2 4.0 ± 0.2 0.268  
On SGLT2 (Y:N) 14:55 9:60 0.253  
On DPP4I (Y:N) 18:51 13:56 0.308  
On sulfonylurea (Y:N) 12:57 9:60 0.477  
Insulin use (Y:N) 37:32 57:12 <0.001  
mTCNS 5.2 ± 0.7 7.2 ± 0.8 0.048 <0.001 
TNS 5.4 ± 0.7 7.8 ± 0.8 0.021 <0.001 
Sural amplitude (µV) 8.4 ± 0.9 6.3 ± 0.8 0.098 <0.001 
Tibial amplitude (mV) 6.5 ± 0.5 5.7 ± 0.6 0.212 <0.001 
Median CSA (mm27.5 ± 0.2 8.1 ± 0.2 0.023 <0.001 
Tibial CSA (mm214.1 ± 0.7 16.2 ± 0.9 0.038 <0.001 
P value
On metformin (n = 69)Not on metformin (n = 69)Metformin and nonmetforminNonmetformin and control subjects
Age (years) 66.9 ± 1.5 65.3 ± 1.6 0.605 0.217 
Sex (male:female) 42:27 41:28 0.862  
Diabetes duration (months) 183 ± 14 180 ± 15 0.904  
BMI (kg/m230.7 ± 0.7 31.4 ± 0.7 0.458 <0.001 
Mean SBP (mmHg) 132.3 ± 1.7 134.6 ± 2.3 0.419  
HbA1c (mmol/mol) 65 ± 2 70 ± 2 0.112  
HbA1c (%) 8.1 ± 0.2 8.6 ± 0.2 0.114  
eGFR (mL/min/1.73m²) 65.4 ± 2.6 59.7 ± 3.1 0.153  
Chronic kidney disease (Y:N) 29:40 34:35 0.393  
Triglycerides (mmol/L) 2.2 ± 0.3 2.1 ± 0.2 0.662  
Total cholesterol (mmol/L) 3.8 ± 0.2 4.0 ± 0.2 0.268  
On SGLT2 (Y:N) 14:55 9:60 0.253  
On DPP4I (Y:N) 18:51 13:56 0.308  
On sulfonylurea (Y:N) 12:57 9:60 0.477  
Insulin use (Y:N) 37:32 57:12 <0.001  
mTCNS 5.2 ± 0.7 7.2 ± 0.8 0.048 <0.001 
TNS 5.4 ± 0.7 7.8 ± 0.8 0.021 <0.001 
Sural amplitude (µV) 8.4 ± 0.9 6.3 ± 0.8 0.098 <0.001 
Tibial amplitude (mV) 6.5 ± 0.5 5.7 ± 0.6 0.212 <0.001 
Median CSA (mm27.5 ± 0.2 8.1 ± 0.2 0.023 <0.001 
Tibial CSA (mm214.1 ± 0.7 16.2 ± 0.9 0.038 <0.001 

Data are presented as n or mean ± SE. DPP4i, dipeptidyl peptidase-4 inhibitor; N, No; SBP, systolic blood pressure; SGLT2, sodium–glucose cotransporter-2 inhibitor; Y, yes.

Peripheral nerve ultrasonography demonstrated significant differences in mean median nerve CSA between the metformin group (7.5 ± 0.2 mm2) and nonmetformin group (8.1 ± 0.2mm2, P = 0.023) (Fig. 1). Similarly, mean tibial nerve CSA was also significantly lower in the metformin group (14.1 ± 0.7 mm2; nonmetformin group, 16.2 ± 0.9 mm2; P = 0.038). Assessment of the clinical severity of peripheral neuropathy revealed that the metformin group had less severe neuropathic symptoms. Mean mTCNS was 5.2 ± 0.8 in the metformin group and 7.2 ± 0.8 in the nonmetformin group was (P = 0.048). Mean TNS 5.4 ± 0.7 was in the metformin group and was 7.8 ± 0.8 in the nonmetformin group (P = 0.021) (Fig. 2). Nerve conduction studies showed there were no significant differences in mean tibial CMAP (metformin group, 6.5 ± 0.5 mV; nonmetformin group, 5.7 ± 0.6 mV; P = 0.212). There was a nonsignificant higher mean sural sensory nerve action potential amplitude in the metformin group (8.4 ± 0.9 mV; nonmetformin group, 6.3 ± 0.8 mV; P = 0.098).

Figure 1

Peripheral nerve ultrasound images of individual participants from the nonmetformin group (left) and metformin group (right). Cross-sectional images were obtained of median (A and B) and tibial nerves (C and D). Left-sided panels (A and C) were taken from a participant receiving glicazide as monotherapy (TNS, 13; mTCNS, 8), whereas the participant in right-sided panels (B and D) was receiving metformin, again as monotherapy (TNS, 6; mTCNS, 6).

Figure 1

Peripheral nerve ultrasound images of individual participants from the nonmetformin group (left) and metformin group (right). Cross-sectional images were obtained of median (A and B) and tibial nerves (C and D). Left-sided panels (A and C) were taken from a participant receiving glicazide as monotherapy (TNS, 13; mTCNS, 8), whereas the participant in right-sided panels (B and D) was receiving metformin, again as monotherapy (TNS, 6; mTCNS, 6).

Close modal
Figure 2

Group comparison of peripheral nerve ultrasonography and clinical neuropathy scores for metformin and nonmetformin groups. Values are expressed as mean ± SE for median nerve CSA (A), mean tibial nerve CSA (B), mean mTCNS (C), and mean TNS (D). B: Upper limit of normal tibial nerve CSA is shown as 12.4 mm2 (23).

Figure 2

Group comparison of peripheral nerve ultrasonography and clinical neuropathy scores for metformin and nonmetformin groups. Values are expressed as mean ± SE for median nerve CSA (A), mean tibial nerve CSA (B), mean mTCNS (C), and mean TNS (D). B: Upper limit of normal tibial nerve CSA is shown as 12.4 mm2 (23).

Close modal

To provide insights into the mechanisms underlying the morphological and clinical benefits of metformin, axonal excitability recordings were undertaken in all participants (Fig. 3). These demonstrated significant differences in parameters reflecting the behavior of voltage-gated ion channels, particularly nodal Na+ and K+channels. There was lower mean stimulation for 50% of the maximal response in the metformin group (4.8 ± 0.3; nonmetformin group, 5.8 ± 0.3; P = 0.023) and lower mean rheobase (metformin, 3.2 ± 0.2; nonmetformin, 3.8 ± 0.2; P = 0.027), which is typically increased in severe neuropathy. Data obtained in the metformin group demonstrated a higher mean magnitude of subexcitability (metformin, 11.4% ± 0.5%; nonmetformin, 10.2% ± 0.5%; P = 0.020) and S2 accommodation (metformin, 21.5 ± 0.5; nonmetformin, 20.1 ± 0.5; P = 0.037), both of which reflect the behavior of nodal slow K+channels. No significant changes were noted in any other excitability parameters, including refractoriness, superexcitability, depolarizing, or hyperpolarizing threshold electrotonus.

Figure 3

Group comparison of axonal excitability recordings, demonstrating mean axonal excitability data for metformin and nonmetformin groups. Values are expressed as mean ± SE for stimulation for 50% of maximum response (A), rheobase (B), subexcitability (C), and S2 accommodation (D).

Figure 3

Group comparison of axonal excitability recordings, demonstrating mean axonal excitability data for metformin and nonmetformin groups. Values are expressed as mean ± SE for stimulation for 50% of maximum response (A), rheobase (B), subexcitability (C), and S2 accommodation (D).

Close modal

Mathematical modeling of axonal excitability data in the nonmetformin group demonstrated that the changes in excitability values were best explained by a reduction in nodal Na+permeability (nonmetformin: 3.4; control: 4.2, cm3s−1 × 10−9). This single change reduced the discrepancy between the nonmetformin and control group by 79%. Furthermore, combining this reduction in nodal Na+permeability with a decrease in nodal slow K+conductances reduced the discrepancy by 83% (nonmetformin: 45.9; control: 47.0 nanosiemens). The alterations in the nonmetformin group reduced the resting membrane potential to −82.9 mV from the control value of −82.7 mV. For the metformin group, modeling demonstrated that there was less severe reduction in nodal Na+ (metformin: 3.6; control: 4.2, cm3s−1 × 10−9) and slow K+(metformin: 47.4; control: 47.0 nanosiemens) conductances compared with the nonmetformin group. The resting membrane potential with these two changes was reduced to −83.0 mV. This particular modeling combination reduced the discrepancy between the metformin group and the control group by 71% and suggests that metformin’s mechanism of action is to mitigate the reduction in axonal nodal Na+permeability that occurs in DPN and to facilitate an increase in nodal slow K+conductances, resulting in a mild relative hyperpolarization of the nerve membrane potential compared with the nonmetformin group.

Multiple regression analysis was performed to assess whether metformin total daily dose or duration of therapy influenced clinical, morphological, or electrophysiological outcomes in both the metformin and nonmetformin groups. In both groups, total daily dose or duration of current or prior metformin therapy did not significantly influence peripheral neuropathy outcomes. In the nonmetformin group, time since cessation of metformin therapy also did not influence clinical, morphological, or electrophysiological outcomes. It was noted that insulin use was different between both groups, possibly due to inadequate glycemic control. Therefore, a multiple regression analysis was also performed to assess whether insulin use or HbA1c influenced peripheral neuropathy outcomes. The regression analysis demonstrated no significant correlation of measures of clinical neuropathy, peripheral nerve morphology, or axonal excitability with either insulin use or HbA1c.

The current study has demonstrated superior neuropathy outcomes in individuals receiving metformin therapy compared with those who were not prescribed metformin, with groups closely matched for a wide range of demographic, metabolic, and treatment variables. The better outcomes were noted in nerve morphology, assessed using nerve ultrasonography of both upper- and lower-limb nerves, and supported by less severe neuropathy symptoms. Axonal excitability recordings also demonstrated a potential mechanism for the benefits of metformin, with superior nerve Na+and K+channel function in the metformin group. Although nerve conduction studies are a surrogate marker of structural changes typically relating to loss of axons or injury to myelin, axonal excitability recordings provide information on the physiological integrity of surviving nerve fibers. The superior excitability findings noted in the metformin group suggest that metformin therapy is associated with ongoing physiological benefits in surviving fibers, irrespective of prior structural injury.

However, a limitation of the study was the cross-sectional assessment of participants, and therefore, long-term randomized studies are needed to assess whether there is a sustained benefit of metformin on DPN outcomes. Another limitation is that insulin use was not matched between groups. However multiple regression analysis did not demonstrate an effect of insulin use on peripheral neuropathy outcomes in either group. Despite having assessed prior treatment records over a 20-year period, it is possible that there were other metabolic, lifestyle, or treatment factors that influenced the outcomes, including a prior history of smoking, which was not recorded. We also did not specifically record the reasons for lack of metformin use in the nonmetformin cohort. It is possible that metformin was not prescribed due to participant intolerance, renal impairment, or lactic acidosis. Use of medications to treat other elements of the metabolic syndrome, including antihypertensive medication, statins, and fenofibrates were also not matched between groups, and it remains possible that differences between the groups in prescription of these medications may have impacted neuropathy outcomes. Future prospective studies should include matching for these variables and documentation of reason for lack of metformin use on participant recruitment as these factors may influence peripheral neuropathy and study outcomes. Although DPN most commonly presents with a distal symmetrical polyneuropathy, in some cases proximal nerve dysfunction can occur first. A further limitation of the current study is that changes in minimum latencies and persistence of F wave studies were not recorded. If these had been performed, they may have provided further insights into the potential effects of metformin on conduction in proximal nerve segments (30).

Our findings differ from previous human studies of metformin use in DPN, with some studies suggesting greater duration and dose of metformin use are risk factors for more severe DPN, especially in younger patients (31,32). Metformin’s bioavailability is 50–60%, and its plasma half-life is ∼2.5 h. Although metformin does not produce toxic metabolites, it is nevertheless renally cleared, and its use is contraindicated in patients with creatinine clearance <15 mL/min due to the risk of lactic acidosis (33).

The current study did not demonstrate a dose-dependent relationship with neuropathy outcomes, but further studies are required to investigate whether higher doses and longer duration of use confer greater neuroprotection. Large studies, such as the UK Prospective Diabetes Study trial, support metformin use in ensuring long-term glycemic control and in reducing macrovascular and microvascular complications. These studies do, however, note the association with vitamin B12 deficiency as a potential risk for the development of peripheral neuropathy and suggest vitamin B12 supplementation for those patients on metformin (32). Of note, vitamin B12 deficiency may lead to peripheral neuropathy on its own, which would suggest that metformin-treated patients would have inferior outcomes on neuropathy measures, which is the converse of the findings in the current study. The current study did not collect data on vitamin B12 stores, active vitamin B12 levels, nitrous oxide use, or vitamin B supplementation. These data would be important for future prospective studies on metformin use in DPN.

Data from animal studies provide support for our own findings and suggest that metformin may be neuroprotective in DPN, with an increase in myelin-based protein, neural and vascular endothelial growth factors, and other anti-inflammatory factors, such as interleukin 10, and that higher doses are associated with a reduction in peripheral neuropathy (34). Evidence from animal studies also suggests that metformin may alleviate hyperalgesia and allodynia in DPN via activation of AMPK and reduction of oxidative stress (35).

Multiple studies have demonstrated benefits of metformin in streptozotocin-treated rats in reversing markers of DPN, including neural sensory and motor abnormalities, nerve conduction studies, and neuropathic pain (35–38). Metformin is known to act through numerous pathways, including increasing insulin sensitivity, inhibition of hepatic gluconeogenesis, antagonism of glucagon effects, and inhibition of mitochondrial respiration (39). Systemically, metformin has been shown to increase AMPK activation with a subsequent reduction in peripheral nerve inflammation and oxidate stress in animal studies (36,40), which may explain the improvements in nodal Na+and K+conductances noted in the current study. AMPK impairment has been implicated in obesity, metabolic syndrome, and inflammation (41). Another mechanism for the findings in the current study may be the reversal of Schwann cell dysfunction, which is a well-established component of diabetic neuropathy pathophysiology (42–44) and which may also contribute to altered ion channel localization (45). Recent studies using models of nerve injury and other neurological disorders have shown that metformin administration improves remyelination capacity of Schwann cells and the subsequent restoration of axonal function (46,47).

Nerve morphology changes are well-established pathological features in DPN, particularly in those individuals with other metabolic risk factors (48). The current study suggests that metformin is associated with superior nerve morphology. The mechanism by which metformin may achieve this could be via reduction in the intracellular osmotic load at the node of Ranvier, secondary to a reduction in depolarization due to metformin’s beneficial effects on nodal Na+and K+conductances. Previous studies have demonstrated a reduced Na+gradient in animal models of DPN, which is associated with nerve fiber swelling (49). The current study provides human evidence of reduced Na+conductances in the nonmetformin group, which was associated with an increase in nerve size. In contrast, there was improvement in Na+conductances in the metformin group that was associated with superior nerve morphology, indicated by a reduced nerve size. Previous studies have demonstrated that findings of increased peripheral nerve CSA are also observed in combination with increased intraneural blood flow (50). Because metformin is able to cross the blood-brain and blood-nerve barrier, future studies are needed to demonstrate whether metformin through its various mechanisms of action may have direct or indirect effects in reduction of intraneural blood flow in DPN (11–14).

In conclusion, the current study therefore provides evidence that metformin treatment in DPN is associated with superior morphological, electrophysiological, and clinical outcomes, which may be mediated by improvements in voltage-gated nerve ion channel function.

Acknowledgments. The Total Neuropathy Score was provided to A.V.K. by Professor David Cornblath and John Hopkins University. The authors are grateful to the staff and patients of the Diabetes Centre at Prince of Wales Hospital, Sydney.

Funding. This research was supported by an Australian Government Research Training Program Scholarship.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.D. was involved in study design, recruitment, data collection, data interpretation, and manuscript composition. T.I. was involved in study design, data interpretation, discussion, and manuscript composition. L.L.W. was involved in recruitment, data collection, and discussion. D.T. was involved in data collection and discussion. A.M.P. was involved in recruitment, data interpretation, and discussion. K.-L.M. was involved in recruitment, data interpretation, and discussion. N.C.G.K. was involved in manuscript composition, data interpretation, and discussion. A.V.K. was involved in study design, data interpretation, and manuscript composition. All authors approved the final version of the manuscript. A.V.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Magliano
DJ
,
Boyko
EJ
,
IDF Diabetes Atlas 10th Edition Scientific Committee
.
IDF Diabetes Atlas.
10th ed. Brussels,
International Diabetes Federation
,
2021
2.
Pop-Busui
R
,
Boulton
AJM
,
Feldman
EL
, et al
.
Diabetic neuropathy: a position statement by the American Diabetes Association
.
Diabetes Care
2017
;
40
:
136
154
3.
Feldman
EL
,
Nave
K-A
,
Jensen
TS
,
Bennett
DLH
.
New horizons in diabetic neuropathy: mechanisms, bioenergetics, and pain
.
Neuron
2017
;
93
:
1296
1313
4.
Gaede
P
,
Vedel
P
,
Larsen
N
,
Jensen
GVH
,
Parving
H-H
,
Pedersen
O
.
Multifactorial intervention and cardiovascular disease in patients with type 2 diabetes
.
N Engl J Med
2003
;
348
:
383
393
5.
Duckworth
W
,
Abraira
C
,
Moritz
T
, et al.;
VADT Investigators
.
Glucose control and vascular complications in veterans with type 2 diabetes
.
N Engl J Med
2009
;
360
:
129
139
6.
Ismail-Beigi
F
,
Craven
T
,
Banerji
MA
, et al.;
ACCORD trial group
.
Effect of intensive treatment of hyperglycaemia on microvascular outcomes in type 2 diabetes: an analysis of the ACCORD randomised trial
.
Lancet
2010
;
376
:
419
430
7.
Johnson
AB
,
Webster
JM
,
Sum
CF
, et al
.
The impact of metformin therapy on hepatic glucose production and skeletal muscle glycogen synthase activity in overweight type II diabetic patients
.
Metabolism
1993
;
42
:
1217
1222
8.
Consoli
A
,
Nurjhan
N
.
Contribution of gluconeogenesis to overall glucose output in diabetic and nondiabetic men
.
Ann Med
1990
;
22
:
191
195
9.
Gunton
JE
,
Delhanty
PJD
,
Takahashi
S-I
,
Baxter
RC
.
Metformin rapidly increases insulin receptor activation in human liver and signals preferentially through insulin-receptor substrate-2
.
J Clin Endocrinol Metab
2003
;
88
:
1323
1332
10.
Valencia
WM
,
Palacio
A
,
Tamariz
L
,
Florez
H
.
Metformin and ageing: improving ageing outcomes beyond glycaemic control
.
Diabetologia
2017
;
60
:
1630
1638
11.
Demaré
S
,
Kothari
A
,
Calcutt
NA
,
Fernyhough
P
.
Metformin as a potential therapeutic for neurological disease: mobilizing AMPK to repair the nervous system
.
Expert Rev Neurother
2021
;
21
:
45
63
12.
Campbell
JM
,
Stephenson
MD
,
de Courten
B
,
Chapman
I
,
Bellman
SM
,
Aromataris
E
.
Metformin use associated with reduced risk of dementia in patients with diabetes: a systematic review and meta-analysis
.
J Alzheimers Dis
2018
;
65
:
1225
1236
13.
Jiang
T
,
Yu
J-T
,
Zhu
X-C
, et al
.
Acute metformin preconditioning confers neuroprotection against focal cerebral ischaemia by pre-activation of AMPK-dependent autophagy
.
Br J Pharmacol
2014
;
171
:
3146
3157
14.
Lin
Y
,
Wang
K
,
Ma
C
, et al
.
Evaluation of metformin on cognitive improvement in patients with non-dementia vascular cognitive impairment and abnormal glucose metabolism
.
Front Aging Neurosci
2018
;
10
:
227
15.
Walker
FO
,
Cartwright
MS
,
Alter
KE
, et al
.
Indications for neuromuscular ultrasound: expert opinion and review of the literature
.
Clin Neurophysiol
2018
;
129
:
2658
2679
16.
Narayan
S
,
Goel
A
,
Singh
AK
,
Thacker
AK
,
Singh
N
,
Gutch
M
.
High resolution ultrasonography of peripheral nerves in diabetic patients to evaluate nerve cross sectional area with clinical profile
.
Br J Radiol
2021
;
94
:
20200173
17.
Goyal
K
,
Aggarwal
P
,
Gupta
M
.
Ultrasound evaluation of peripheral nerves of the lower limb in diabetic peripheral neuropathy
.
Eur J Radiol
2021
;
145
:
110058
18.
Fionda
L
,
Di Pasquale
A
,
Morino
S
, et al
.
Changes of clinical, neurophysiological and nerve ultrasound characteristics in CIDP over time: a 3-year follow-up
.
J Neurol
2021
;
268
:
3011
3019
19.
Liu
L
,
Ye
Y
,
Wang
L
, et al
.
Nerve ultrasound evaluation of Guillain-Barré syndrome subtypes in northern China
.
Muscle Nerve
2021
;
64
:
560
566
20.
Kiernan
MC
,
Bostock
H
,
Park
SB
, et al
.
Measurement of axonal excitability: consensus guidelines
.
Clin Neurophysiol
2020
;
131
:
308
323
21.
Issar
T
,
Tummanapalli
SS
,
Borire
AA
, et al
.
Impact of the metabolic syndrome on peripheral nerve structure and function in type 2 diabetes
.
Eur J Neurol
2021
;
28
:
2074
2082
22.
Dhanapalaratnam
R
,
Issar
T
,
Poynten
AM
,
Milner
K-L
,
Kwai
NCG
,
Krishnan
AV
.
Diagnostic accuracy of nerve ultrasonography for the detection of peripheral neuropathy in type 2 diabetes
.
Eur J Neurol
2022
;
29
:
3571
3579
23.
Fisse
AL
,
Katsanos
AH
,
Gold
R
,
Krogias
C
,
Pitarokoili
K
.
Cross-sectional area reference values for peripheral nerve ultrasound in adults: a systematic review and meta-analysis–Part II: lower extremity nerves
.
Eur J Neurol
2021
;
28
:
2313
2318
24.
Bril
V
,
Tomioka
S
,
Buchanan
RA
,
Perkins
BA
,
mTCNS Study Group
.
Reliability and validity of the modified Toronto Clinical Neuropathy Score in diabetic sensorimotor polyneuropathy
.
Diabet Med
2009
;
26
:
240
246
25.
Cornblath
DR
,
Chaudhry
V
,
Carter
K
, et al
.
Total neuropathy score: validation and reliability study
.
Neurology
1999
;
53
:
1660
1664
26.
Kimura
J
.
Electrodiagnosis in Diseases of Nerve and Muscle: Principles and Practice.
Philadelphia
,
FA Davis
,
1983
27.
Burke
D
,
Kiernan
MC
,
Bostock
H
.
Excitability of human axons
.
Clin Neurophysiol
2001
;
112
:
1575
1585
28.
Bostock
H
,
Cikurel
K
,
Burke
D
.
Threshold tracking techniques in the study of human peripheral nerve
.
Muscle Nerve
1998
;
21
:
137
158
29.
Liang
C
,
Howells
J
,
Kennerson
M
,
Nicholson
GA
,
Burke
D
,
Ng
K
.
Axonal excitability in X-linked dominant Charcot Marie Tooth disease
.
Clin Neurophysiol
2014
;
125
:
1261
1269
30.
Andersen
H
,
Stålberg
E
,
Falck
B
.
F-wave latency, the most sensitive nerve conduction parameter in patients with diabetes mellitus
.
Muscle Nerve
1997
;
20
:
1296
1302
31.
Yang
R
,
Yu
H
,
Wu
J
, et al
.
Metformin treatment and risk of diabetic peripheral neuropathy in patients with type 2 diabetes mellitus in Beijing, China
.
Front Endocrinol (Lausanne)
2023
;
14
:
1082720
32.
Hashem
MM
,
Esmael
A
,
Nassar
AK
,
El-Sherif
M
.
The relationship between exacerbated diabetic peripheral neuropathy and metformin treatment in type 2 diabetes mellitus
.
Sci Rep
2021
;
11
:
1940
33.
Graham
GG
,
Punt
J
,
Arora
M
, et al
.
Clinical pharmacokinetics of metformin
.
Clin Pharmacokinet
2011
;
50
:
81
98
34.
Lós
DB
,
Oliveira
WHd
,
Duarte-Silva
E
, et al
.
Preventive role of metformin on peripheral neuropathy induced by diabetes
.
Int Immunopharmacol
2019
;
74
:
105672
35.
Ma
J
,
Yu
H
,
Liu
J
,
Chen
Y
,
Wang
Q
,
Xiang
L
.
Metformin attenuates hyperalgesia and allodynia in rats with painful diabetic neuropathy induced by streptozotocin
.
Eur J Pharmacol
2015
;
764
:
599
606
36.
Hasanvand
A
,
Amini-Khoei
H
,
Hadian
M-R
, et al
.
Anti-inflammatory effect of AMPK signaling pathway in rat model of diabetic neuropathy
.
Inflammopharmacology
2016
;
24
:
207
219
37.
Wang
S
,
Kobayashi
K
,
Kogure
Y
, et al
.
Negative regulation of TRPA1 by AMPK in primary sensory neurons as a potential mechanism of painful diabetic neuropathy
.
Diabetes
2018
;
67
:
98
109
38.
Haddad
M
,
Eid
S
,
Harb
F
, et al
.
Activation of 20-HETE synthase triggers oxidative injury and peripheral nerve damage in type 2 diabetic mice
.
J Pain
2022
;
23
:
1371
1388
39.
Rena
G
,
Pearson
ER
,
Sakamoto
K
.
Molecular mechanism of action of metformin: old or new insights?
Diabetologia
2013
;
56
:
1898
1906
40.
Bierhaus
A
,
Fleming
T
,
Stoyanov
S
, et al
.
Methylglyoxal modification of Nav1.8 facilitates nociceptive neuron firing and causes hyperalgesia in diabetic neuropathy
.
Nat Med
2012
;
18
:
926
933
41.
Alcocer-Gómez
E
,
Garrido-Maraver
J
,
Bullón
P
, et al
.
Metformin and caloric restriction induce an AMPK-dependent restoration of mitochondrial dysfunction in fibroblasts from fibromyalgia patients
.
Biochim Biophys Acta
2015
;
1852
:
1257
1267
42.
Kalichman
MW
,
Powell
HC
,
Mizisin
AP
.
Reactive, degenerative, and proliferative Schwann cell responses in experimental galactose and human diabetic neuropathy
.
Acta Neuropathol
1998
;
95
:
47
56
43.
Kim
ES
,
Isoda
F
,
Kurland
I
,
Mobbs
CV
.
Glucose-induced metabolic memory in Schwann cells: prevention by PPAR agonists
.
Endocrinology
2013
;
154
:
3054
3066
44.
Eid
AA
,
Ford
BM
,
Block
K
, et al
.
AMP-activated protein kinase (AMPK) negatively regulates Nox4-dependent activation of p53 and epithelial cell apoptosis in diabetes
.
J Biol Chem
2010
;
285
:
37503
37512
45.
Rosenbluth
J
.
Intramembranous particle distribution at the node of Ranvier and adjacent axolemma in myelinated axons of the frog brain
.
J Neurocytol
1976
;
5
:
731
745
46.
Huang
Z
,
Lin
J
,
Jiang
H
, et al
.
Metformin promotes Schwann cell remyelination, preserves neural tissue and improves functional recovery after spinal cord injury
.
Neuropeptides
2023
;
100
:
102348
47.
Houshmand
F
,
Barati
M
,
Golab
F
, et al
.
Metformin-induced AMPK activation stimulates remyelination through induction of neurotrophic factors, downregulation of NogoA and recruitment of Olig2+ precursor cells in the cuprizone murine model of multiple sclerosis
.
Daru
2019
;
27
:
583
592
48.
Issar
T
,
Tummanapalli
SS
,
Borire
AA
, et al
.
Impact of the metabolic syndrome on peripheral nerve structure and function in type 2 diabetes
.
Eur J Neurol
2021
;
28
:
2074
2082
49.
Brismar
T
,
Sima
AA
,
Greene
DA
.
Reversible and irreversible nodal dysfunction in diabetic neuropathy
.
Ann Neurol
1987
;
21
:
504
507
50.
Borire
AA
,
Issar
T
,
Kwai
NC
, et al
.
Sonographic assessment of nerve blood flow in diabetic neuropathy
.
Diabet Med
2020
;
37
:
343
349
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.