Emerging Nonpharmacologic Interventions to Treat Diabetic Peripheral Neuropathy

Abstract

Significance:

Diabetic peripheral neuropathy (DPN), a complication of metabolic syndrome, type I and type II diabetes, leads to sensory changes that include slow nerve conduction, nerve degeneration, loss of sensation, pain, and gate disturbances. These complications remain largely untreatable, although tight glycemic control can prevent neuropathy progression. Nonpharmacologic approaches remain the most impactful to date, but additional advances in treatment approaches are needed.

Recent Advances:

This review highlights several emerging interventions, including a focus on dietary interventions and physical activity, that continue to show promise for treating DPN. We provide an overview of our current understanding of how exercise can improve aspects of DPN. We also highlight new studies in which a ketogenic diet has been used as an intervention to prevent and reverse DPN.

Critical Issues:

Both exercise and consuming a ketogenic diet induce systemic and cellular changes that collectively improve complications associated with DPN. Both interventions may involve similar signaling pathways and benefits but also impact DPN through unique mechanisms.

Future Directions:

These lifestyle interventions are critically important as personalized medicine approaches will likely be needed to identify specific subsets of neuropathy symptoms and deficits in patients, and determine the most impactful treatment. Overall, these two interventions have the potential to provide meaningful relief for patients with DPN and provide new avenues to identify new therapeutic targets.

Introduction

Clinically symptomatic diabetic peripheral neuropathy (DPN) is insidious and largely irreversible, with no known curative treatments available for patients (Coppini et al, 2006; Zochodne, 2019). There is much to be learned about the underlying pathogenesis, and effective therapeutics discovery remains tenuous. Tight glycemic control can prevent or slow neuropathy progression (Ang et al, 2014; Skyler, 1996). Alarmingly, diabetes and obesity are rapidly increasing, leading to an increased prevalence of DPN.

Common consequences of DPN include axon degeneration, loss of sensation, pain, autonomic dysfunction, and gait disturbances (Feldman et al, 2017). Overall, these significantly reduce patients' quality of life with DPN. Below, we outline and discuss the limitations of current approaches for DPN but also identify emerging ideas that may lead to new lifestyle changes or identify new drug targets. The review also highlights the overlap of these new approaches that help refine putative mechanisms to target DPN.

Glycemic control in type 1 diabetics is managed via daily injections of prandial and basal insulin (Mathieu et al, 2017). For most type 2 diabetic patients, Metformin is commonly used with other approaches, including insulin (American Diabetes Association, 2020). Achieving tight glycemic control is difficult, as reducing blood glucose levels while avoiding hypoglycemic episodes is challenging for patients.

Current pharmacologic therapies only treat DPN-associated pain symptoms. Approximately 20% of patients with DPN develop pain symptoms described as burning, stabbing, or electrical pain (Tesfaye et al, 2013). First-line therapies for pain associated with diabetes include anticonvulsants (pregabalin or gabapentin), norepinephrine and serotonin reuptake inhibitors, or tricyclic antidepressants (Feldman et al, 2019; Issar et al, 2021). Alpha lipoic acid is an over-the-counter supplement that reduces neuropathic pain in clinical studies (Agathos et al, 2018; Papanas and Ziegler, 2014). Topical treatments include 0.075% capsaicin cream and a 5% lidocaine patch (Yang et al, 2019). This is not an exhaustive list, as additional medications or herbal supplements used off-label can be used to treat painful DPN (Heydari et al, 2016; Ostovar et al, 2020).

Overall, these medications are limited in number, have less than ideal efficacy, and can have psychological and physiological side effects that limit their usefulness. Recently approved approaches to improve glycemic control, including incretin agonists and glucose transporter inhibitors, may eventually help slow DPN through improved glycemic control (Nowak et al, 2022; Sivakumar et al, 2021). These approaches related to DPN have yet to be intensively explored (Issar et al, 2021; Jolivalt et al, 2011; Zhang et al, 2022).

Nonpharmacologic Treatments of Diabetic Neuropathy

Neuromodulation

One emerging approach for the treatment of DPN is the use of neuromodulation techniques. Neuromodulation includes a variety of electromagnetic devices designed to manipulate neuronal activity, including transcutaneous electric nerve stimulation (TENS). TENS has been reported to improve pain and cold sensation in patients with DPN (Bosi et al, 2013).

Importantly, pain reductions were transient, and the trial failed to reach its primary endpoint of improved nerve conduction velocity (NCV). Reichstein et al reported a higher proportion of responders in both painful and nonpainful DPN to high-frequency muscle stimulation than TENS, as defined by improvement in one or more sensory symptoms. However, the degree of improvement did not differ between responders (Reichstein et al, 2005). Of note, the increased efficacy of high-frequency muscle stimulation may be explained, in part, by the modulation of neurotrophins, cytokines, and other non-neuronal actions elicited by muscle activity that will be discussed below.

High-frequency spinal cord stimulation (HF-SCS) is a new neuromodulatory approach to treating DPN and other chronic pain conditions. HF-SCS requires surgical implantation of an electromagnetic stimulating device providing either tonic or bursting input to the spinal cord. This approach includes surgical risks, and there is no clarity on the efficacy or beneficial mechanisms of HF-SCS, with some patients reporting significant improvement in pain and patient quality of life (Daousi et al, 2005; de Vos et al, 2014; Petersen et al, 2021), while others demonstrate more limited (Duarte et al, 2020) efficacy. Most patients with DPN receiving HF-SCS reported improvements in their pain (Daousi et al, 2005; de Vos et al, 2014; Petersen et al, 2021; van Beek et al, 2018). However, it is essential to note that people with diabetes are nearly twice as likely to develop surgery-induced infections after implantation of HF-SCS devices compared with nondiabetic patients (Mekhail et al, 2011).

Improved rigor and experimental designs in clinical trials should help identify putative mechanisms by which HF-SCS mediates positive effects in DPN. In the most extensive randomized control study of HF-SCS in DPN to date, 216 patients with DPN were randomized to receive either the best standard medical care or 10 kHz HF-SCS for 6 months, with 50% pain relief on a visual analog scale without worsening neurological symptoms as a primary outcome, which was met by patients receiving HF-SCS (Petersen et al, 2021).

These and future studies could benefit from a pseudo-sham in which a third patient group receives HF-SCS device implantation without stimulation for 6 months before an initial assessment of primary outcome measures, followed by another 6 months of HF-SCS device activation. These experimental approaches would provide insight into potential placebo from SCS implantation. Still, they would be limited by difficulties with year-long patient follow-up and the increased risk of infection without therapeutic benefit in people with diabetes receiving SCS device implantation.

Despite added risks, these new neuromodulatory approaches may hold promise for an alternative treatment for DPN, or identify new mechanisms or signaling pathways that can be exploited (Gupta et al, 2021). It is plausible that HF-SCS leads to signaling and cellular changes that alter neuronal, glial, and vascular function in the setting of DPN. This area also needs additional reverse translational approaches in rodent models to help identify underlying mechanisms. A recent study in streptozotocin-injected rats suggests that SCS-stimulated rats with diabetes should make improvements in several behavioral assessments relevant to painful DPN (Wang et al, 2022).

Exercise and diabetic neuropathy

Physical activity interventions are feasible nonpharmacologic interventions for DPN and other forms of peripheral neuropathy (Streckmann et al, 2022). A recent meta-analysis concluded that evidence-based recommendations to incorporate physical activity interventions can now be made, and that a combination of endurance and sensorimotor training provides the most benefit (Streckmann et al, 2022). These conclusions are significant as the promotion of physical activity and exercise for diabetes patients has suffered from fear of foot injury. Hopefully, these studies will increase physical activity implementation in clinics treating diabetes patients.

Exercise interventions for DPN can be modeled preclinically to some degree. Exercise has been shown to modify features that contribute to the onset and progression of DPN and neuropathy associated with metabolic syndrome—such as hyperglycemia, accumulation of advanced glycation end-products (AGEs) (Dall'aglio et al, 1983; Kluding et al, 2012), dyslipidemia (Goldhaber-Fiebert et al, 2003), and insulin resistance (Cuff et al, 2003; Dall'aglio et al, 1983).

In addition, other biomarkers of DPN, such as intraepidermal nerve fiber (IENF) density (Groover et al, 2013; Kluding et al, 2012) and NCV (Balducci et al, 2006), can be moderately improved by exercise intervention. In this section, we will discuss the effects of exercise on non-neuronal and neuronal tissues as they relate to the course of DPN.

Beneficial effects of exercise on DPN

As highlighted in Figure 1, exercise is essential in opposing obesity and metabolic syndrome, two significant contributors to peripheral neuropathy associated with metabolic syndrome and type II diabetes. Exercise can improve body composition, which correlates with the incidence of DPN in type II diabetes patients (Oh et al, 2019). In addition, there is an association between NCV sensory thresholds, nerve conduction amplitude, and bone mineral density in patients with DPN (Strotmeyer et al, 2006). Exercise can improve bone mineral density in both males and premenopausal women (Bassey et al, 1998; Kelley et al, 2000; Marques et al, 2012); however, it is unclear whether reduced bone mineral density has a causative or associative relationship with DPN.

FIG. 1.

Benefits of exercise in DPN. Exercise leads to a wide range of disease-modifying effects in DPN. Exercise reduces fasting blood glucose, insulin resistance, and accumulation of AGEs. Exercise also modifies and reduces inflammation, improves mitochondrial function, and increases neurotrophin signaling, thereby contributing to a recovery-permissive environment. AGEs, advanced glycation end-products; DPN, diabetic peripheral neuropathy.

Critical features of exercise on non-neuronal tissues in DPN include its effects on dyslipidemia, insulin resistance, hyperglycemia, AGE accumulation, inflammation, and neurotrophin release from target tissues (Cooper et al, 2016). Dyslipidemia causes mitochondrial damage (Rumora et al, 2018, 2019a), increased reactive oxygen species (ROS) generation, and inflammation in target tissues. These consequences ultimately contribute to neuronal harm associated with DPN.

Moreover, dyslipidemia and insulin are strongly linked and mutually regulate each other (Lombardo et al, 2007; Semple et al, 2009), adding to the consequences of hyperglycemia. Exercise reduces triglycerides in diabetes and metabolic syndrome (Goldhaber-Fiebert et al, 2003), and new evidence is emerging related to the critical roles of saturated and unsaturated fatty acids. Molecular studies have now shown that these fatty acids differentially impact nerve lipid profiles and have different effects on nerve function.

Recent studies in mice suggest that high-fat diet interventions enriched in monounsaturated fats can lessen the negative consequences of the high-fat diet on several DPN endpoints (Rumora et al, 2019b). These studies highlight the need for additional research on fatty acid biochemistry and nerve function. They could lead to the identification of new therapies for DPN, and the ability of exercise to reduce DPN within this area holds promise for using exercise as an intervention (Rumora et al, 2022).

Aerobic exercise can improve insulin resistance in preclinical rodent models of diabetes (Dall'aglio et al, 1983) and clinical studies (Cuff et al, 2003). Exercise reduces insulin resistance and hyperglycemia in type I and type II diabetes (Dall'aglio et al, 1983). This, in turn, results in reduced AGE accumulation in individuals who exercise (Kluding et al, 2012). Regular exercise can decrease inflammatory markers across the life span (Cooper et al, 2016; Petersen and Pedersen, 2005). It is known that during and after exercise, skeletal muscle interleukin (IL)-6 levels increase and appear responsible for rises in other anti-inflammatory cytokines such as IL-10 and IL-1 receptor agonist (IL-1RA) (Pedersen, 2009).

During muscle contraction, IL-6 is released from muscles where it can affect nonmuscle tissues and organs (Petersen and Pedersen, 2005). As an infusion, IL-6 leads to anti-inflammatory effects similar to exercise and reduces pro-inflammatory cytokines, including TNF-α, placing IL-6 as a critical mediator of the long-term anti-inflammatory benefits of exercise (Starkie et al, 2003). These actions of exercise can collectively lead to important secondary actions on peripheral nerve function and improve DPN.

Exercise has direct effects that benefit sensory neurons affected by DPN. For instance, neurotrophin expression in relevant tissues is affected by exercise. The expression of neurotrophin-3 (NT-3) (Gómez-Pinilla et al, 2001) and brain-derived neurotrophic factor (BDNF) (Sleiman et al, 2016) in muscle is increased after exercise. NT-3 and BDNF are both critical in proprioceptive axon growth and maintenance. These fibers are damaged in DPN (Feczko and Klueber, 1988; Muller et al, 2008; van Deursen et al, 1998), resulting in poor proprioception, altered gait, and difficulty with balance (de Oliveira Lima et al, 2021; Khan and Andersen, 2022; Reeves et al, 2021).

Exercise shows improvement in these sensory deficits (Allet et al, 2010), possibly through improved intrafusal spindle innervation due to skeletal muscle neurotrophin release. This remains an understudied area, and more information is needed to elucidate the connection between exercise, neurotrophins, muscle spindle innervation, and gait/balance in DPN (Muller et al, 2008).

Exercise also influences IENF (Tariot et al, 2000) density, which is reduced in DPN and commonly used as a measure of disease severity. Studies of patients with DPN have shown that after aerobic and resistance training, patients with DPN have moderately increased IENF density and sprouting in leg skin biopsies (Kluding et al, 2012) (Singleton et al, 2014).

Several benefits of exercise on peripheral nerves are associated with improvements in sensory neuron mitochondrial health. The unique morphology and length of sensory axons originating from the dorsal root ganglion (DRG) place a massive energetic demand on these neurons. Damage to mitochondria in sensory axons alters energy production, leading to axon degeneration in DPN (Roy Chowdhury et al, 2012; Rumora et al, 2019a; Rumora et al, 2018). Heat shock proteins likely play an essential role, as exercise increases heat shock protein 70 (HSP70) expression in the DRG of diabetic rats (Yoon et al, 2015). HSP70 is important in the process of mitophagy to remove damaged mitochondria (Zheng et al, 2018).

Moreover, exercise can increase important molecules in mitochondrial biogenesis, including peroxisome proliferator-activated receptor gamma-associated coactivator 1 alpha (PGC1α) (Schnyder et al, 2017; Wrann et al, 2013). In addition, overexpression of downstream targets of PGC1α in sensory neurons can reverse the painful behaviors and epidermal axon loss in mice with DPN (Chandrasekaran et al, 2019b; Chandrasekaran et al, 2015). Combined with these favorable findings, there is a need for improved behavioral implementation and education for patients to use exercise to successfully lessen the consequences of DPN.

Dietary impact on DPN

The redox state of sensory neurons is sensitive to diet, as consumption of a diet high in fat and sucrose (e.g., a western diet) predisposes the individual to dyslipidemia and hyperglycemia. Hyperglycemia results in increased glucose utilization and accumulation of toxic metabolites, including sorbitol (Hubinont et al, 1981; Stavniichuk et al, 2012) and the reactive dicarbonyls glyoxal and methylglyoxal (Bierhaus et al, 2012). These toxic metabolites impair ROS scavenging and generation (Amano et al, 2002; Kikuchi et al, 1999; Kuntz et al, 2010; Polykretis et al, 2020). As an important risk factor for DPN, dyslipidemia alters ROS generation, mitochondrial trafficking, and mitochondrial health (Fig. 2) (Lupachyk et al, 2012; Rumora et al, 2019a; Rumora et al, 2018; Viader et al, 2013).

FIG. 2.

Effects of palmitate on DPN. Palmitate is a saturated fatty acid that affects DPN in diverse ways. Palmitate inhibits the trafficking of K_ATP channels, which contribute to insulin release in the pancreas and nociceptive modulation in the peripheral and central nervous systems. Palmitate also inhibits mitochondrial trafficking in the DRG, decreases mitochondrial membrane polarization (Δψ), and affects mitochondrial copy number and bioenergetics. ATP, adenosine triphosphate; DRG, dorsal root ganglion; K_ATP, ATP-gated potassium.

Consequently, ROS signals through several transient receptor potential (TRP) channels and can modify voltage-gated channels, leading to sensory dysfunction (Miyake et al, 2016; Nazıroğlu, 2017; Nishio et al, 2013; Schink et al, 2016; Schlüter and Leffler, 2016; Sözbir and Nazıroğlu, 2016).

Symptoms of DPN may also be improved using dietary interventions. Lipotoxicity in the DRG caused by elevated long-chain saturated fatty acids such as palmitate is not evident in mice supplemented with short-chain fatty acids (Rumora et al, 2019a), and supplementation with unsaturated long-chain fatty acids reverses biomarkers of DPN and the mitotoxic effects of saturated long-chain fatty acids in these rodent preclinical models (Rumora et al, 2019b).

New evidence suggests that very-low-carbohydrate diets can reverse mechanical allodynia in rodent models of metabolic syndrome (Cooper et al, 2018b), reduce peripheral oxidative stress (Cooper et al, 2018a), reduce inflammation (Dupuis et al, 2015; Ruskin et al, 2021), and detoxify pain-contributing metabolites produced as a consequence of hyperglycemia (Enders et al, 2022). Here, we highlight evidence using a ketogenic diet as an intervention for DPN, and identify how ketogenic diets impact the sensory nervous system's health, function, and redox state (Fig. 3).

FIG. 3.

Dietary makeup of a ketogenic diet and standard dietary recommendations. The standard dietary recommendations for macronutrient composition are ∼40% carbohydrate, 30% protein, and 30% fat by calorie. The recommendations for a ketogenic diet are ∼10% carbohydrate (∼50 g), 30% protein, and 60% fat.

Ketogenic diets

Very-low-carbohydrate, ketogenic diets have been used to treat a variety of ailments, ranging from epilepsy and seizures (Bough et al, 2003; Dressler et al, 2010) to neurodegenerative disorders (Brownlow et al, 2013; Taylor et al, 2018) and chronic pain conditions (Cooper et al, 2018b; Di Lorenzo et al, 2019; Ruskin et al, 2021). The underlying principle of these diets is to replace most carbohydrate intake with an increase in fats and oils. This shift to excess fat consumption leads to increased fatty acid oxidation in the liver, producing the ketone bodies acetoacetate and β-hydroxybutyrate (β-HB). These ketone bodies are released from the liver into circulation and serve as a replacement fuel source in extrahepatic tissues.

Since ketogenic diets significantly reduce circulating glucose as a fuel source, it is unsurprising that these diets have previously been used to treat diabetic patients. Before the discovery of insulin in 1921, it was common practice in mild cases of diabetes to reduce hyperglycemia by dietary elimination of carbohydrate intake. More severe cases of diabetes led to the elimination of both carbohydrate and protein intake (Holmden, 1913). Starvation was sufficient to mitigate hyperglycemia using this approach; however, this approach could result in patient weakness and death.

Mosenthal et al (1918) reported that postprandial hyperglycemia was not evident in diabetes patients fed very low-carbohydrate meals similar to patients consuming a standard diet. In 1920, Newburgh and Marsh “ignored the belief concerning the danger of fat in the diet of diabetics,” introducing high-fat, low-carbohydrate diets in weakened diabetes patients resistant to starvation (Newburgh and MARSH, 1920). These diets successfully improved patient outcomes, but the use of high-fat, low-carbohydrate diets as interventions for diabetes has only recently re-emerged.

Safety and tolerability of ketogenic diets

Potential health concerns surrounding the use of a high-fat diet in patients with diabetes are reasonable, as diabetic ketoacidosis is a serious diabetes complication caused by an accumulation of lactate and ketone bodies in the blood. Lactate, acetoacetate, and β-HB deprotonate at physiologic pH and decrease blood pH. Conventional wisdom would suggest that consuming a ketogenic diet would increase circulating ketones, lower blood pH, and possibly put a patient at risk of acidosis. With decreased carbohydrate intake, however, lactate production is reduced (Luukkonen et al, 2020), which likely limits the risk of acidosis.

One study reported a reversal of acidosis in several diabetes patients fed a ketogenic diet (Newburgh and MARSH, 1920). In recent years, there have been several case reports (Chandrasekaran and Rani, 2020; Dressler et al, 2010; Tóth, 2014) and larger studies (Choi et al, 2020; Goday et al, 2016; Westman et al, 2008) in humans highlighting the efficacy and safety of ketogenic diets in diabetic patients. These results are replicated in preclinical models of type I and type II diabetes (Enders et al, 2021; Hussain et al, 2012; Poplawski et al, 2011) and metabolic syndrome (Cooper et al, 2018b). These studies are briefly outlined in Table 1.

Table 1.

Clinical Studies on the Safety and Efficacy of Implementing a Ketogenic Diet to Treat Type I and II Diabetes and Complications Associated with Diabetes

Study design	Population	Outcome measures	Reference
Case study	Type I	Hb-A1C, fasting blood glucose	Tóth (2014)
Intervention	Type II	Hb-A1C, bodyweight	Yancy et al (2005)
Intervention	Type II	HB-A1C, bodyweight, HDL-cholesterol	Westman et al (2008)
Intervention	Type I	Hb-A1C, fasting blood glucose	Leow et al (2018)
Meta-analysis	Type II	Hb-A1C, bodyweight, HDL-cholesterol	Choi et al (2020)

Hb-A1C, hemoglobin A1C; HDL, high-density lipoprotein.

No clinical study to date has examined whether a ketogenic diet can reduce pain or improve sensation in patients with DPN. Two recent clinical studies report that a ketogenic diet benefits migraine sufferers (Bongiovanni et al, 2021) and musculoskeletal pain (Gylfadottir et al, 2022). In a single-arm study, patients with refractory migraines consumed a ketogenic diet for 3 months (Bongiovanni et al, 2021). These patients reported decreased frequency and duration of migraines, and reduced intensity of migraine pain.

This study did exhibit a high patient dropout, which the authors attributed to the strict nature of the diet and difficulty maintaining ketosis. Another pilot randomized control trial removed processed foods from the diets of patients suffering from musculoskeletal pain for 3 weeks, and then randomized patients to a whole-food diet or a low-carbohydrate ketogenic diet for 3 months (Gylfadottir et al, 2022). This study suggested that patients in both arms improved visual analog pain scores and quality of life measures after 3 months. While there were no significant differences between groups in pain reported by the visual analog scale, only patients consuming a ketogenic diet reported reductions in pain interference.

Group differences in pain ratings at baseline were confounding factors in this study. However, these studies suggest that using a ketogenic diet could positively impact chronic pain in human populations. More studies are needed to address using a ketogenic diet as a therapeutic intervention for DPN.

Actions of a ketogenic diet on neuronal activity

More robust preclinical evidence exists for the benefits of using a ketogenic diet for DPN in rodent neuropathy models (Cooper et al, 2018b; Di Lorenzo et al, 2019; Enders et al, 2021; Ruskin et al, 2021). We previously reported that using a high fat-fed mouse model, switching mice to a ketogenic diet, prevents and reverses mechanical allodynia in the hind paw (Cooper et al, 2018b).

In a mouse model of type 1 diabetes, consuming a ketogenic diet could prevent and reverse mechanical allodynia and abnormal thermal sensation (Enders et al, 2021). These studies also revealed that ketone bodies enhance neurite outgrowth from sensory neurons in vitro (Cooper et al, 2018b), and that a ketogenic diet prevents IENF loss and stimulates IENF regeneration in vivo (Enders et al, 2021). Below, we highlight the effects of a ketogenic diet on neuronal tissue (neurons, glia, etc.), and discuss how these mechanisms might work for DPN and chronic pain.

In models of seizure and epilepsy, ketogenic diets reduce neuronal firing rate due in part to the regulation of adenosine triphosphate (ATP)-gated potassium (K_ATP) channels (Fig. 4). When binding ATP, these K_ATP channels close. Through these channels, both acetoacetate and β-HB reduce gamma-amino butyric acid (GABA)ergic neuron firing rates in the substantia nigra (Ma et al, 2007). This effect occurs in the presence of glucose, and ketone bodies exert this inhibitor effect in the presence of glucose, and it could be reversed using the K_ATP antagonist tolbutamide. Masino and colleagues noted that glucose could hyperpolarize and decrease postsynaptic currents in hippocampal neurons (Kawamura et al, 2010).

FIG. 4.

Metabolic contributions of ketone signaling. β-HB is transported into the cell through MCT1, where it inhibits glycolysis and contributes to oxidative phosphorylation. This shifts ATP accumulation away from the plasma membrane and deeper into the cell, releasing the inhibition of K_ATP channels and hyperpolarizing the axon. Ketolysis further contributes to mitochondrial health by improving ATP production and mitochondrial biogenesis while reducing ROS accumulation. Ketone bodies further contribute to direct and indirect scavenging of the toxic glycolytic by-product methylglyoxal. β-HB, β-hydroxybutyrate; MCT1, monocarboxylase transporter 1; ROS, reactive oxygen species.

Tolbutamide reversed these effects, implicating K_ATP channels in these ketone-mediated effects on neuronal excitability. A ketogenic diet also reduces glycolysis, reducing plasma membrane-proximal ATP and opening K_ATP channels. Glycolysis can occur adjacent to the plasma membrane to quickly provide ATP to Na⁺/K⁺-ATPase cellular pumps (Campanella et al, 2005; Epstein et al, 2014). Thus, ketone bodies can lower glucose utilization and glycolytic flux (Puchalska et al, 2019; Valdebenito et al, 2015), increasing the activity of K_ATP channels via reductions in ATP gradients.

K_ATP channel opening reduces current amplitude induced by electrical stimuli, and inhibition of K_ATP channels with glibenclamide leads to the opposite effect (Sarantopoulos et al, 2003). In sensory neurons, sciatic nerve ligation closes K_ATP channels and decreases K_ATP channel expression in the spinal cord (Kawano et al, 2009; Wu et al, 2011). After nerve injury, intrathecal delivery of K_ATP channel activators improves sensory deficits. In rats with bone cancer-associated pain, glibenclamide increases the mechanical sensitivity of the hind paw (Wu et al, 2011; Xia et al, 2014).

These K_ATP channels could serve as a new pharmacologic target, as their channel is sensitive to the concentration of glucose (Henquin, 1990; Henquin, 1988), and hyperglycemia-evoked neuronal depolarization can be reversed by the K_ATP activator diazoxide (de Campos Lima et al, 2019). Palmitate supplementation can alter the membrane localization of the K_ATP channel subunit sulfonylurea receptor 1 (SUR1) (Fig. 2) (Ruan et al, 2018). It is becoming clear that ketone bodies can significantly impact cellular glucose utilization and glycolytic flux (Poplawski et al, 2011; Puchalska et al, 2019; Valdebenito et al, 2015), and help mitigate dyslipidemia (Feinman and Volek, 2006; Volek et al, 2008).

TRP ankyrin 1 (TRPA1) and TRP vanilloid 1 (TRPV1) are ion channels that are responsive to methylglyoxal and toxic metabolites upregulated in diabetes (Andersson et al, 2013), as well as elevations in ROS (Miyake et al, 2016; Nazıroğlu, 2017; Nishio et al, 2013). It is known that methylglyoxal increases neuronal excitability and nociception that involves TRPA1 (Griggs et al, 2017). Increased ROS in the spinal dorsal horn and DRG (Nazıroğlu, 2017; Nishio et al, 2013) can lead to spontaneous activity, and TRPA1 and TRPV1 antagonists can reverse this hyperactivity (Nazıroğlu, 2017; Nishio et al, 2013).

Nociceptive signaling due to toxic metabolites and ROS can be dampened by consuming a ketogenic diet (Fig. 4). Methylglyoxal synthesis occurs at the triose-phosphate step of glycolysis; thus, its synthesis is likely abrogated by decreased glycolytic flux during ketosis. A ketogenic diet and ketone bodies, directly and indirectly, detoxify methylglyoxal in mice (Enders et al, 2022), and can prevent or reverse methylglyoxal-evoked pain behaviors. In addition, the antioxidant benefits of a ketogenic diet are well established (Greco et al, 2016). In support, we previously demonstrated that consumption of a ketogenic diet in mice lowers mitochondrial H₂O₂ in the distal sciatic nerve (Cooper et al, 2018a).

Axonal regeneration and a ketogenic diet

A ketogenic diet may also improve peripheral nerve loss in DPN by stimulating peripheral axon growth. DPN generally results in demyelination of large-diameter axons and/or a loss of small IENFs (Cheng et al, 2013; Christianson et al, 2007; Hsieh et al, 2012; Johnson et al, 1986, 2008; Malik et al, 2005; Tariot et al, 2000). In vitro, Schwann cells cultured in high glucose can dedifferentiate (Hao et al, 2015) and undergo apoptosis (Zhu et al, 2018). In addition, reductions in neurite outgrowth from cultured DRG neurons have been attributed to decreased secretion of nerve growth factor by Schwann cells (Tosaki et al, 2008).

We recently reported that the consumption of a ketogenic diet by mice with DPN leads to increased epidermal nerve fibers in the hind paw (Enders et al, 2021). These regenerative effects likely involve both ketogenic diet-induced reductions in hyperglycemia. Ketone bodies likely have direct actions on sensory nerves since the addition of β-HB in cultured DRG neurons increases neurite outgrowth independent of glucose concentrations in the media (Cooper et al, 2018b). There appears to be a synergistic effect between β-HB and glucose on neurite outgrowth, suggesting that the impact of β-HB on neurite outgrowth may be independent of reduced glycolysis (Cooper et al, 2018b).

Wallerian degeneration provides another mechanism that could play a role in sparing epidermal loss. In this pathway, axonal injury or impaired axonal transport results in reduced synthesis of nicotinamide adenine dinucleotide (NAD⁺) (Chandrasekaran et al 2019a) (Sasaki et al, 2016), an essential cofactor for bioenergetics and the electron transport chain. This pathway has recently been implicated in mouse models of DPN, and its inhibition rescues IENF fiber loss in these models (Cheng et al, 2019). Ketosis may compensate for decreasing axonal NAD⁺ concentrations in DPN and other neuropathies due to its reduced NAD⁺ consumption for the generation of ATP (Fig. 4).

Indeed, Masino and colleagues report elevated NAD⁺ concentrations in the hippocampus of mice fed a ketogenic diet (Elamin et al, 2018). Thus, consuming a ketogenic diet likely increases NAD⁺ concentration in the peripheral nerve, potentially sparing axonal damage in DPN and other nerve injury paradigms. These data from preclinical models provide strong evidence that ketogenic diets can slow or even reverse sensory fiber loss in DPN.

An alternative approach to improve diet adherence may include ingestion of exogenous ketone bodies that are available commercially as ketone esters, ketone salts, or ketone salts supplemented with medium-chain triglyceride oil. Exogenous ketone supplementation recapitulates some of the beneficial effects of a ketogenic diet, including reductions in blood glucose (Kesl et al, 2016) and oxidative stress (Kephart et al, 2017).

Ketone bodies and noncanonical mechanisms

In their canonical role, ketone bodies primarily serve as a bioenergetic fuel source; however, they also activate G-protein–coupled receptor cascades, modulation of inflammation and neurotrophin signaling, and alteration of epigenetic landscapes (Fig. 5) (Puchalska and Crawford, 2017). For instance, β-HB binds to the G-protein–coupled receptor hydroxycarboxylic acid receptor 2 (HCAR2) (Boccella et al, 2019), which is present in nociceptors and immune cells. Administration of dimethyl fumarate, another HCAR2 ligand, can reverse mechanical allodynia (Boccella et al, 2019) and axonal degeneration in preclinical nerve injury models (Bombeiro et al, 2020). These benefits are consistent with the view that ketosis can lead to analgesia and promotes sensory axon growth via a β-HB–HCAR2 signaling pathway.

FIG. 5.

Nonmetabolic contributions of ketone signaling. The primary extracellular signaling pathway for β-HB includes activation of the G-protein–coupled receptor HCAR2, which then activates antioxidant and anti-inflammatory pathways. Alternatively, intracellular β-HB can inhibit the NLRP3 inflammasome. β-HB inhibition of HDACs contributes to increased neurotrophin promoter activity, especially for BDNF. BDNF transcription, translation, and release then contribute to improved neuronal outcomes, especially in proprioceptive sensory neurons. BDNF, brain-derived neurotrophic factor; HCAR2, hydroxycarboxylic acid receptor 2; HDACs, histone deacetylases; LRR, leucine-rich repeat; NLRP3, NOD-, LRR-, and pyrin domain-containing protein 3.

In addition, β-HB has known anti-inflammatory actions via its ability to inhibit the NOD- leucine-rich repeat (LRR)- and pyrin domain-containing protein 3 (NLRP3) inflammasome, presumably through the HCAR2 signaling (Shang et al, 2018; Youm et al, 2015). NLRP3 activation results in the cleavage of pro-IL-1 to IL-1β, which induces painful behaviors (Jia et al, 2017; Sweitzer et al, 1999). These studies suggest that ketone bodies could partly improve sensory neuron function by nonmetabolic regulation of inflammation.

Conclusions

Used alone, current pharmacologic approaches to treat DPN are not enough. As our understanding of nonpharmacologic interventions increases, combinations of drug treatment, lifestyle and dietary changes, or interventional approaches are likely to be required to have sustainable improvements in DPN. It is doubtful that corrections in diabetes alone will be enough to recover from longstanding nerve dysfunction and loss in DPN.

The emerging evidence of exercise and dietary changes to reduce inflammation, stimulate neurotrophin expression, and provide efficient energy sources to fuel nerve growth appears promising and likely convergent. These approaches continue identifying new convergent relationships with dyslipidemia, mitochondrial function, nerve growth, normalized sensory function, and pain reduction. As we learn more about these nonpharmacologic challenges, developing sustainable behavioral changes will be mandatory to ensure that patients can integrate these lifestyle changes into their daily habits.

Abbreviations Used

AGE

advanced glycation end-products

ATP

adenosine triphosphate

β-HB

β-hydroxybutyrate

BDNF

brain-derived neurotrophic factor

DPN

diabetic peripheral neuropathy

DRG

dorsal root ganglia

GABA

gamma-amino butyric acid

Hb-A1C

hemoglobin A1C

HCAR2

hydroxycarboxylic acid receptor 2

HDAC

histone deacetylase

HDL

high-density lipoprotein

HF-SCS

high-frequency spinal cord stimulation

HSP70

heat shock protein 70

IENF

intraepidermal nerve fiber

IL

interleukin

IL-1RA

interleukin-1 receptor agonist

K_ATP

ATP-gated potassium channel

LRR

leucine-rich repeat

MCT1

monocarboxylase transporter 1

Na⁺/K⁺-ATPase

ATP-dependent sodium–potassium pump

NAD⁺

nicotinamide adenine dinucleotide

NCV

nerve conduction velocity

NLRP3

NOD-, LRR-, and pyrin domain-containing protein 3

NOD

nucleotide-binding oligomerization domain

NT-3

neurotrophin 3

PGC1α

peroxisome proliferator-activated receptor gamma-associated coactivator 1 alpha

ROS

reactive oxygen species

SUR1

sulfonylurea receptor 1

TENS

transcutaneous electric nerve stimulation

TNFα

tumor necrosis factor alpha

TRP

transient receptor potential

TRPA1

TRP ankyrin-containing 1

TRPV1

TRP vanilloid-receptor 1

Authors' Contributions

J.E., D.E., and D.E.W. cowrote the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This study was supported by NIH grants RO1 NS043314 (D.E.W.), T32 HD057850 (J.E. and D.E.), and the Kansas Institutional Development Award (IDeA) P20 GM103418 (D.E.W.). Figures are created using BioRender (biorender.com).