Perspective Chapter: Exploring the Potential of Vagus Nerve Neuromodulation as a Personalized and Enhanced Therapeutic Experience for Type 2 Diabetics

Abstract

Glucagon-like peptide-1 (GLP-1) receptor agonists have become the gold standard for the treatment of type 2 diabetes mellitus (T2D). The therapeutic benefits of GLP-1 are marred by compliance, cost, and side effects. Vagus nerve modulation (VNM) holds great potential for current and future neuromodulation therapies in a range of medical conditions. Treatment for T2D using VNM is a potential new area of research. Most VNM studies investigating glycemic control utilize glucose tolerance tests. There are few long-term chronic studies examining both fasting plasma glucose and plasma glucose in the absorptive state. This chapter explores literature involving the use of VNM to enhance glycemic control. Also, results of a novel method of multi-site, multi-frequency sub-diaphragmatic vagal neuromodulation will be reviewed as well as a discussion of mechanisms associated with this VNM technique. This modality holds promise in enhancing glycemic control in the fasting and postprandial states. Multi-site multi-frequency neuromodulation offers a multitude of different therapy parameters for personalized medicine.

1. Introduction

Despite advances in diabetic medications, the use of bariatric surgery and lifestyle modification, there is still an unmet need for novel treatments for type 2 diabetes mellitus (T2D). For example, outside of clinical trials, discontinuation rates of glucagon like peptide 1 (GLP)-1 Receptor Agonist (GLP-1 RA) therapy has been reported to be approximately 45% at 1 year [1, 2] due to unwanted side effects [3], high cost [4], and unwillingness to follow physician’s treatment advice [5]. There are questions regarding the long-term durability of bariatric surgery to treat T2D [6] and as many as 75% of diabetics fail at lifestyle modification [7]. This chapter will explore current treatment options and the advances, and potential advantages, of new non-pharmacological methods in development as treatment options for T2D. These include insulin pump technology, duodenal mucosal resurfacing, gene therapy, external electrical stimulation, peripheral focused ultrasound, direct stimulation to visceral organs, and vagus nerve modulation (VNM) technologies. Benefits of many of these therapies include a decreased need for compliance, decreased long term costs, and decreased unwanted side effects. Finally, this chapter will highlight a novel technique comprising of multi-site, multi-frequency electrical modulation of the sub-diaphragmatic vagus nerve for the treatment of T2D.

1.1 Limitations of current T2D treatment options

According to the American Academy of Clinical Endocrinologist (AACE) Treatment Guidelines, T2D therapeutic options follow an algorithm of three escalating scales of intervention starting with lifestyle modification (LSM), and escalating to monotherapy with pharmacological agents, concurrent dual therapy, triple therapy, and eventually medical and surgical weight loss [8]. These pharmacologic treatments begin with Metformin, and then are combined with other drugs such as sulfonylureas, thiazolidinedione, DPP-4 & SGLT2 inhibitors and GLP-1 RAs. The medications are not without side effects, potentially inducing hypoglycemia, heart failure, ketoacidosis, diarrhea, nausea, abdominal pain, orthopedic fracture and weight gain [9].

As an adjunct to pharmacotherapy, insulin therapy is an effective tool to increase glycemic control, however, there is a risk of hypoglycemia and a large portion of diabetics fears insulin injections [10, 11, 12, 13, 14, 15]. A study by Stotland et al. [16] found that 45% of diabetic subjects avoided injections because of anxiety, phobia and fear. Injection anxiety has been related to poorer treatment adherence, greater psychological distress, a greater incidence of diabetes-related hospitalization and a higher risk of retinopathy and neuropathy [10].

Weight-loss surgery for T2D has demonstrated efficacy as a non-pharmacological treatment for T2D. However, it is not without challenges and potential complications. T2D has shown to be recurrent in 58% of bariatric surgical patients, albeit with sustained reductions in medications and insulin [17]. While a safe option, with mortality rates below 1%, the side effects of some bariatric surgeries still carry the risks of bowel obstructions, staple line dehiscence, sepsis, gallstones, GERD, Barret’s Esophagus, hernia, and dumping syndrome [18]. Non-anatomical sparing bariatric surgeries such as the Roux-Y-Gastric Bypass (RYGB) and its variants, Vertical Sleeve Gastrectomy (VSG) and the Biliopancreatic Diversion with Duodenal Switch (BPD-DS) can cause a number of nutritional complications secondary to these malabsorptive procedures, such as anemias, metabolic bone disease, calcium deficiencies, sarcopenic obesity, and neurologic disorders if a patient is non-compliant with their post-operative nutritional protocols [19].

GLP-1 receptor agonists have gained much recent popularity and have shown promise for the treatment of T2D. In clinical trials HbA1c reduction ranges from −14.8 to −2.7 mmol/mol [20] and weight loss up to 12% from baseline [21]. However, real-world studies have demonstrated lower positive outcomes with this class of drugs. In a retrospective cohort study by Weiss et al. [2], weight and adherence were analyzed at 12 and 24 months in 589 subjects that were prescribed GLP-1 RAs. At 12 months only 33% of patients achieved weight loss of ≥5%, adherence was 65% and the discontinuation rate was 45%. At 24 months 44% of type 2 diabetics achieved weight loss of ≥5%, adherence was 59% and 65% of patients discontinued GLP-1 RAs. Non-adherence to GLP-1 RAs has been observed in multiple other studies (Table 1). Low adherence to new oral forms of GLP-1 RAs is predicted, with only 50% of type 2 diabetics adhering to oral medications as prescribed [25] mainly due to forgetfulness [26]. Approximately 5–10% of type 2 diabetics cannot tolerate GLP-1 RAs [20] due to nausea, vomiting and diarrhea [3].

Study	Study duration	GLP-1 agonist	Adherence (%)
Divino et al. (2019) [22]	1 year	Dulaglutide	36.8–67.2
		Exenatide twice daily	5.9–44.4
		Exenatide once weekly	24.7–44.2
		Liraglutide	22.2–57.5
		Lixisenatide	15.5–40.0
Johnston et al. (2014) [23]	1 year	Exenatide once weekly	78.3
		Exenatide twice daily	50
		Liraglutide once daily	72.2
Uzoigwe et al. (2021) [24]	1 year	Semaglutide once weekly	67.0
		Dulaglutide	56.0
		Liraglutide	40.4
		Exenatide one weekly	35.5
Weiss et al. (2020) [1]	1 year	Once daily GLP-1RA	43.8
		Once Weekly GLP-1 RA	64.2
Weiss et al. (2022) [2]	1 year	Once daily GLP-1RA	59.8
		Once Weekly GLP-1 RA	82.1
	2 years	Once daily GLP-1RA	55.3
		Once Weekly GLP-1 RA	74.1

Table 1.

Multiple studies demonstrate issues with adherence to taking GLP-1 RAs. This is observed in once daily and once week injections. A major problem for lack of adherence is forgetfulness to take medications.

2. Non-pharmacological technologies under development

There are multiple non-pharmacological T2D treatment methodologies in various stages of development. Clinical trials have included next generation closed loop insulin pumps, duodenal mucosal resurfacing, external electrical stimulation, direct stimulation of visceral organs and transcutaneous auricular vagus nerve stimulation. Preclinical tests have involved gene therapies, peripheral focused ultrasound, direct visceral organ stimulation and various approaches to vagus nerve stimulation. The following is a summary of these emerging technologies.

2.1 Next generation closed-loop insulin pumps

For many type-1 diabetics insulin pumps have demonstrated efficacy in increasing targeted glycemic outcomes, decreasing glucose variability and increasing quality of life. Studies have indicated that this technology may be beneficial for insulin delivery for therapy for type 2 diabetics on insulin therapy [27]. However, there are many drawbacks of using pump technology. Hypoglycemic events still remain an issue for both diabetic populations using this technology [27, 28]. Continuous glucose monitoring (CGM) sensor-augmented insulin pump therapy reduces the incidence of hypoglycemia [29],though these systems cannot actively raise plasma glucose (PG). To compensate for unidirectional glycemic control, CGM sensor-augmented pump algorithms incorporate a significant safety floor margin when defining target blood glucose range. This results in average HbA1c levels greater than the recommended <53 mmol/mol [30], at which point diabetes-associated complications develop [31, 32]. Alternatively, bidirectional glycemic control systems have been explored to increase time-in-range. Dual-hormone pumps using both insulin and glucagon enable tighter glycemic control, including reducing the number of hypoglycemia episodes. Challenges associated with the stability of glucagon at room-temperature have prevented widespread acceptance of dual-hormone therapies [33]. A significant drawback for the use of insulin pump technology for type 2 diabetics is the considerably greater amount of daily insulin required for T2D insulin therapy versus that required to treat T1DM (approximately 100 Units for T2D vs. 10 units for T1DM [27]). This would require frequent daily refills of the pump revisor. For these reasons, insulin pump therapy has not been well adopted by type 2 diabetics.

2.2 Duodenal mucosal resurfacing

Duodenal mucosal resurfacing (DMR) is a potential new method of increasing glycemic control for T2D. It has demonstrated positive results in early single-arm, open-label clinical trials [34]. The procedure involves thermal energy ablation of duodenal mucosa using an endoscopic procedure [35]. Two-year results demonstrated a HbA1c level reduction from 69.4 ± −15.8 to 58.5 ± −11.5 mmol/mol [34]. Despite promising early results, the endoscopic DMR procedure carries potential risks of perforation of the intestine and intestinal leaks [36]. Further research is required to establish the long-term safety and effectiveness of DMR.

2.3 Gene therapies

Gene therapy for type 2 diabetes is an emerging field of research and there are several promising approaches that are being investigated. One approach is to use gene therapy to increase the production of insulin in the pancreas [37]. Another approach is to modify genes that regulate blood glucose levels outside of pancreatic function. This involves using gene editing tools such as CRISPR-Cas9 to modify genes associated with insulin resistance or glucose metabolism [38]. These approaches are still in the early stages of research and many further studies are needed to ensure safety and efficacy in type 2 diabetics.

Evidence is accumulating that obesity-associated and/or adipose tissue-derived circulating miRNAs are promising new therapeutic targets for the treatment of obesity and T2D [39]. The development of novel therapeutics, such as miRNA mimetics, anti-miRNA oligonucleotides and exosomes loaded with miRNAs, might allow the clinical application of miRNA-based therapeutics for T2D.

Another exciting branch of gene therapy for the treatment of T2D utilizes opto- and chemogenetic neuromodulation. Optogenetics involves two components, genetic modulation, and optics. The genetic component comprises cellular expression of light sensitive ion channels, or opsins, that are typically isolated from sensory photoreceptors from green algae. There are two major classes of opsins: channelrhodopsins which conduct cations that induce neuronal excitation and anion-conducting channelrhodopsins which induce neuronal inhibition. A cell specific promoter actuator gene is fused to the opsin for expression to specific neuronal populations, allowing for cell type specific excitation or inhibition. Genes coding for opsins are delivered to the cell by generating lines of transgenic animals, transfection, or viral transduction.

A study by Fontaine et al. [40] utilized a non-diabetic transgenic mouse line that employed an acetyltransferase promoter to express Channelrhodopsin-2 (ChR2) in vagal pre-ganglionic afferent nerves innervating the pancreas. When optically stimulating the pancreas, they observed a decrease in blood glucose by 2.8 mmol/l which recovered to baseline following cessation of stimulation. Insulin was increased by 84% but did not reach significance and there was no change in glucagon. Optogenetic stimulation enhanced performance on glucose tolerance tests and there was an observed increase in insulin in these experiments. One drawback was that isoflurane was used in these acute experiments which may have confounded the results.

Studies employing chemogenetic stimulation have exhibited effects on glycemic control. Chemogenetic stimulation has similarities to optogenetic stimulation which requires genetic modification and cell specific stimulation, or inhibition, by utilizing cell specific promoters to express desired genes. However, in contrast to optogenetics, pharmacologically inert compounds are used to active transfected designed mutated receptors. The most common receptors are genetically altered G protein-coupled receptors [41] with GqD and GiD used to activate or silence neurons, respectively. Chemogenetic stimulation has advantages over optogenetic modulation in that it does not require the complexity of introducing a light source inside the animal.

Many chemogenetic-targeted-tissues include extra-neuronal cells such as alpha and beta-cells, L-cells, adipocytes, hepatocytes and skeletal muscle [42, 43, 44, 45, 46, 47]. Effects of chemogenetic stimulation on these tissues involve an enhanced glucose uptake in skeletal muscle, reduced glucagon secretion, increased insulin secretion and beta cell proliferation [42, 43, 44, 45, 46, 47]. Chemogenetic neuronal targets have focused on AgRP and POMC neurons [47, 48]. Modulation of activity in these neuronal populations may increase glycose homeostasis primarily through modulating food intake [47, 48].

Despite promising experimental results using gene therapy, there are many obstacles to overcome for this methodology to be utilized for a treatment of T2D. The use of adeno-associated viruses (AAVs) is a common method to deliver genes in vivo and the most reasonable method to be used clinically for delivery of genes for gene therapy. However, there are inherent risks involved with AAV-mediated gene delivery and many barriers for acceptance for a treatment of T2D. The type of diseases that have been accepted for gene therapy are sever, rare and debilitating [49]; and not metabolic disorders.

Serious adverse events are associated with the delivery of genes using AAVs which are associated with complement immune system activation [50, 51]. Due to this immunosuppressant therapy is required which is a clinical risk. Hepatotoxicity, kidney damage and dorsal root ganglia toxicity have been observed with use of AAVs [52].

2.4 External electrical stimulation

Studies have tested the effects of electrical stimulation delivered to the surface of the skin on increasing glycemic control. A study by Lee et al. [53] examined stimulation in 12 subjects using applications microcurrents at an ultra-low frequency of 0.0007 Hz delivered to the surface of the skin for 3.5 hours per day, 5 days per week for 2–4 months. Results were reported for 6 subjects at the conclusion of the trial. HbA1c decreased in these subjects from 55.2 ± −13.7 to 44.3 ± −15.8 mmol/mol (average ± stdev). Limitations of this study include a lack of a control group, a small sample size, and no statistical analysis.

Catalogna et al. [54] studied peripheral electrical stimulation (PES) in 12 T2D subjects in an interventional, open label, randomized, crossover trial. Two adhesive electrodes were placed bilaterally, one on the anterior aspect of the legs, below the kneecap, and lateral to the anterior crest of the tibia in proximity to the common peroneal and tibial branches of the sciatic nerve. Electrical current was applied for 5 minutes every morning before breakfast for 14 days. They found a significant decrease in average nocturnal PG (from 7.4 ± 0.6 mmol/l at baseline to 6.5 ± 0.5 mmol/l) and FPG decreased from 7.6 ± 0.4 to 6.7 ± 0.4 mmol/l following treatment. There was no change in insulin. A controlled experiment with a larger sample is needed to draw conclusions regarding the use of this technique as a treatment for T2D.

2.5 Peripheral focused ultrasound

Peripheral focused ultrasound (pFUS) has been tested as a novel approach to externally modulate neuronal activity [55]. Cotero et al. [56] evaluated the effects of daily pFUS on restoring glycemic control in rat and mouse models of T2D. Ultrasound was focused on the hepatoportal nerve plexus, known to contain glucose metabolism sensory neurons. The study demonstrated that pFUS increased glycemic control by reversing the onset of LPS-induced hyperglycemia and improved performance on glucose tolerance tests. Interestingly, they found that pFUS modulated firing rates in the hypothalamus at areas that are innervated by vagal projects and altered levels of neurotransmitters regulated to metabolism. In non-diabetic swine they demonstrated increased insulin sensitivity by using hyperinsulinemic-euglycmic clamps. The authors note that there are barriers to the practicality of this system. At home use of this methodology requires large technological advancements. Currently pFUS systems require physicians and technicians to operate the system. Hand-held ultrasound devices with flexible body-conformal ultrasound patches are under development [57], which may help bridge this technical gab. However, the focal depth of these systems is 10–20 mm and would not reach the hepatoportal nerve plexus in humans. Also, daily use of an external device may have problems with adoption in the T2D population.

2.6 Direct stimulation to visceral organs

Stimulation of the duodenum has demonstrated increased glycemic control in rodents [58] as well as in T2D subjects. In an open-labeled, prospective, single-arm study by Aberle et al. [59] electrodes were implanted into the subserosal layer of the anterior duodenal wall in 12 obese type 2 diabetic subjects with a mean HbA1c of 63.9 mmol/mol at baseline. After 12 months of duodenal stimulation there was a − 15.8 mmol/mol decrease in HbA1c and a drop in FPG of 1.8 mmol/l from 9.6 mmol/l at baseline. There was no change in c-peptide at 12 months. Further investigation with adequately powered blinded clinical trials is required to determine the effects of this approach.

Direct hepatic electrical stimulation has demonstrated increased glycemic control in rodents. Chen et al. [60] placed cardiac pacing electrodes on two lopes of the livers in the Zucker obese (fatty) rat (ZDF fa/fa) T2D animal model. The locations of the electrodes on the liver were described as at the “the top and middle lobe”. The rat’s FPG decreased from 17.1 ± 2.2 to 15.9 ± 2.1 mmol/l with 3 hours of stimulation. There was also a slight increase in performance during an oral glucose tolerance test (OGTT) with hepatic stimulation. Limitations to this technique are the possibility of perforation to the liver at the site of the implanted electrodes as well as stimulation induced hypoglycemia.

Carotid sinus nerve (CSN) stimulation has been assessed as an approach of increasing glycemic control. In a study by Sacramento et al. [61] a 50 kHz signal was applied to the CSN in a diet induced model of type 2 diabetes in rat. Stimulation increased performance on OGTTs as well as potentiating insulin sensitivity. However, there was no change in FPG. A limitation to this procedure is possible adverse effects on ventilatory control and chemoreflexes during stimulation [62].

2.7 Vagus nerve modulation

Studies investigating the therapeutic potential of using vagus nerve modulation (VNM) have involved both external and implanted electrodes. Locations of internal stimulation have included cervical and abdominal vagus nerve segments. The auricular segment of the vagus nerve has been widely evaluated as a site for external vagus stimulation.

Transcutaneous auricular vagus nerve stimulation (taVNS) has shown mixed results to increase glycemic control. The physiological concept behind taVNS is excitation of the auricular branch of the vagus nerve (ABVN) which runs close to the external ear [63, 64]. Supportive evidence that transcutaneous auricular stimulation effects vagal function is demonstrated through changes in heart rate variability [65, 66], known to be affected by many vagus nerve stimulation approaches, and fMRI evidence showing increased central nervous system (CNS) activity in central vagal projections areas [67, 68] during taVNS.

Preclinical and clinical studies have evaluated the effects of taVNS on glycemic control. A study by Li et al. [69] examined if taVNS prevents the development of hyperglycemia in the Zucker rats. taVNS was delivered for 30 minutes a day for 34 days. Zucker rats that received taVNS did not develop hyperglycemia and had a lower HbA1c compared to control after 34 days. Insulin receptors were upregulated in hypothalamus, liver, and skeletal muscle.

Clinical studies evaluating taVNS have had mixed results. A study by Kozorosky et al. [70] measured blood glucose, C-peptide, insulin, leptin, glucagon, ghrelin, baroreceptor-heart rate reflex sensitivity and heart rate variability in non-diabetic subjects following a glucose tolerance test involving ingestion of a beverage containing: 9 g protein, 33 g total carbohydrates (10 g glucose), and 9 g fat. The results were confounding. The sham group demonstrated a decrease in blood glucose similar to the taVNS group. In the glucose tolerance experiments taVNS did not affect blood glucose, glucagon, insulin and C-peptide. There was no change in heart rate variability or baroreceptor-heart rate reflex sensitivity (as would be expected with vagus nerve excitation). However, taVNS decreased postprandial plasma ghrelin levels.

In a single blind randomized controlled trial [71] involving 72 subjects with impaired glucose tolerance (baseline HbA1c = 45.4 mmol/mol) taVNS was applied for 12 weeks twice per day. Fasting plasma glucose (FPG), 2-hour plasma glucose (2hPG, during OGTTs) and HbA1c were evaluated. There was a modest decrease in FPG from 6.2 to 5.7 mmol/l after 12 months with taVNS and a decrease in 2hPG from 9.7 to 7.5 mmol/l. However, the sham groups also saw a decrease in FPG from 9.1 to 8.0 mmol/l. There was a very modest, clinically insignificant, decrease in HbA1c from 45.4 to 43.2 mmol/l with taVNS, however, the sham group experienced a drop from 44.3 to 42.1 mmol/mol. Taken together this study demonstrated no (or at least very modest) improvement in glycemic control and the subjects were not diabetic.

Another clinical study [72] demonstrated no effect of taVNS on glycemic control. In this study OGTTs were carried out on fifteen non-diabetic male subjects and demonstrated no increase in glycemic control. Application of taVNS did not affect serum catecholamine levels or a change in heart rate as would be expected during vagus nerve stimulation.

Overall, taVNS studies have either shown a very modest increase in glycemic control or none. Internal stimulation may be required to increase the effectiveness of vagus nerve stimulation for modulation of PG. Indeed, cervical, and sub-diaphragmatic vagus nerve stimulation approaches have repeatedly demonstrated efficacy to modify glycemic control.

The effects of efferent and/or afferent cervical vagus nerve stimulation on glycemia has been evaluated in preclinical studies. In a study by Meyers et al. [73] stimulation was applied to the cervical vagus nerve non-diabetic rats. Experimental conditions involved ligation of the vagus nerve and stimulation at either the cranial or distal segments, as well as stimulation to an intact nerve. When the uncut nerve was stimulated there was an increase in PG 30 minutes following the initiation of stimulation as well as an increase in plasma glucagon but not insulin. When the cranial side was stimulated there was also an increase in PG but no change in insulin or glucagon. Interestingly, when the distal side of the cut vagus nerve was stimulated there was an increase in PG at 30 minutes as well as an increase in glucagon, however, following 120 minutes of stimulation PG decreased with a corresponding increase in insulin and a decrease in glucagon. Despite these intriguing findings, cervical vagus nerve ligation was required, making it unrealistic for human use.

The results of the Mayer et al. study were replicated by stimulation of intact cervical vagus nerve [74]. In addition to experiments in the Mayer et al. study, researchers also accessed glycemic control with an intraperitoneal glucose tolerance test (IPGTT). Results from glucose tolerance test demonstrated decreased glycemic control with a corresponding decrease in insulin. Interestingly, both right and left cervical vagus stimulation caused a decrease in blood insulin. This suggests that stimulation of either celiac branch or accessory celiac branch vagal pathways may affect pancreatic function (anatomical vagal branch depiction can be found on Figure 1).

Figure 1.

Various sub-diaphragmatic vagus nerve branches innervate the liver and the pancreas. The pancreas is innervated by the celiac and accessory celiac vagal branches. The celiac branch stems from the posterior sub-diaphragmatic vagal trunk and the accessory celiac branch bifurcates from the anterior sub-diaphragmatic vagal trunk. The hepatic branch, innervating the liver, stems from the anterior sub-diaphragmatic vagal trunk. Note: Crossing of axon projects between right/posterior and left/anterior vagal trunks also occurs. Image created by biorender.com.

Payne et al. [75] also found similar results as Meyers et al. [73] which was demonstrated by increased glycemic control in a rat model of T2D by selective stimulation of efferent vagal fibers of the anterior sub-diaphragmatic vagal trunk. The Payne et al. methodology is clinically practical compared to the one tested by Meyers et al. in that selective efferent stimulation was achieved by the combination of an electrical nerve blockade signal (stimulation at 26 kHz), in lieu of vagotomy, and stimulation distal to the site of nerve blockade. Also, the modulation was below the level of the heart, decreasing possible cardiac effects. The procedure induced an increase in glycemic control as assessed by OGTTs. Pancreatic function was possibly affected with a trend (p = 0.057) in decreased glucagon during neuromodulation. This may have occurred through modulating accessory celiac branch activity. The group did not report changes in FPG.

In a study by Yin et al. [76] the dorsal sub-diaphragmatic vagus was stimulated in a rat model of T2D. The group applied various frequencies (5, 14, 40, 5000 Hz as well as intermittent stimulation with 10 seconds on and 90 seconds off at 5 Hz) and determined that 14 Hz with a 0.3 ms pulse width, 5000 Hz and intermittent stimulation significantly increase glycemic control. They also found an increase in insulin sensitivity with stimulation at 5 Hz. They did not report changes in FPG.

Radio frequency applied to the walls of hepatic vessels has been studied to induce a sub-diaphragmatic hepatic vagotomy with the goal to decrease sympathetic nervous system tone to increase glycemic control [77]. Experiments using this technique have demonstrated increased glycemic control by enhanced performance on OGTTs. Also, plasma insulin and c-peptide were shown to increase during the OGTT experiments. Despite the positive findings hepatic vagotomy may be problematic. Hepatic vagotomy has been shown to cause negative changes in feeding behavior, increased hypoglycemic episodes, negative effects on liver regeneration, and cause increased metastasis during liver cancer [78, 79, 80, 81].

Methods to increase glycemic control through VNM may not be entirely dependent on modulation of pancreatic or hepatic function. Vagal neuromodulation-induced decreases in systemic inflammation and weight may facilitate controlling glycemic dysfunction. Many studies have shown that inflammation dysregulation is prevalent in T2D [82, 83, 84, 85]. VNS decreases pro-inflammatory cytokines [86, 87, 88] and reduces effects of inflammatory diseases such as rheumatoid arthritis [89]. Utilizing VNS strategies that decrease inflammation may be a target for T2D treatments [90], but further study is required.

Neuromodulation involving electrical-induced conduction block of the anterior and posterior vagal trunks has demonstrated weight loss and increased glycemic control in human subjects. In a study by Herrera et al. [91] HbA1c decreased by −16.9 mmol/mol, from a baseline of 61.7 mmol/mol, and FPG decline by 1.0 mmol/l, from a baseline of 8.4 mmol/l, following 3 years of vagal blockade. This coincided with a 21% loss of excess weight from baseline.

In summary, multiple studies have demonstrated the ability to increase glycemic control by utilization of VNM. The translation of many of the above preclinical studies to human application requires ongoing research, including studies by our research group. Below highlights recent results from our work utilizing a new approach to VNM. This involves a novel method of electrically modulating multiple sites along the vagus nerve to treat glucose dysregulation.

3. Exploring a targeted multisite/multifrequency vagal neuromodulation approach as a treatment for type 2 diabetes mellitus

Since multiple organ systems involved in blood glucose regulation are influenced by different vagus nerve branches, there may be necessity for electrical modulation of multiple vagus nerve branches to achieve optimal effects on plasma glucose (PG) regulation. This idea has led our team to investigate a new dual neuromodulation concept involving combined stimulation of celiac fibers of the vagus nerve innervating the pancreas along with reversible electrical blockade of neuronal hepatic fibers of the vagus nerve innervating the liver (sites of neuromodulation are depicted in Figure 2). This technique is termed Diabetic Block Stimulation Neuromodulation (DBSN) and has shown efficacy of increasing glycemic control in animal models of T2D. The following highlights key findings of DBSN experimental results in rodent and swine models of T2D [92].

Figure 2.

The sites of dual neuromodulation involve the hepatic can celiac sub-diaphragmatic vagus nerve branches. The red arrow indicates the site of electrical neuronal blockade, and the green arrow indicates the cite of stimulation. Image adapted from Waataja (2022) [92]. Image originally created by biorender.com.

The Zucker obese (fatty) male rats (ZDF fa/fa) rat is a well-established rodent model of T2D [93, 94, 95, 96, 97]. Zucker rats are homozygous for a non-functional leptin receptor which causes obesity and insulin resistance [98]. Pancreatic ß-cells have also been shown to fail to respond to glucose in these rats [98]. This model was used to test our hypothesis that DBSN will reversibly increase glycemic control in an animal model of T2D.

Various VNM techniques were tested in Zucker rats. Since hepatic vagotomy has demonstrated a decrease in insulin resistance and celiac stimulation has demonstrated increased insulin release, both VNM techniques were tested during intravenous glucose tolerance tests (IVGTTs). These methods failed to increase glycemic control (Figure 3a). However, their coadministration (vagotomy+stimulation) demonstrated improved performance on the IVGTT compared to sham (Figure 3a). Despite the increased glycemic control observed with vagotomy+stimulation, the procedure would be undesirable clinically in that hepatic vagotomy is non-reversible, the body may adapt to hepatic vagotomy over time and significant unwanted side effects may be associated with hepatic vagotomy [78, 79, 80, 81]. Therefore, in lieu of a hepatic vagotomy, a high frequency alternating current (HFAC at 5 kHz) blocking signal was applied to the hepatic branch which has shown to reversibly block conduction through the sub-diaphragmatic vagus nerve [99]. Application of HFAC to the hepatic branch with concurrent celiac branch stimulation (the DBSN procedure) reversibly increased glycemic control during IVGTTs (Figure 3b).

Figure 3.

As assessed by IVGTT experiments, HFAC+stimulation and DBSN increased glycemic control in the Zucker rat model of T2D compared to controls (a) Application of HFAC+stimulation increased performance on IVGTTs compared to sham, celiac branch stimulation alone or hepatic vagotomy alone. The various neuromodulation techniques were applied 1 hour before the IVGTT and during the 30 minutes course of blood sampling during the IVGTT, (b) DBSN demonstrated improved performance on IVGTTs compared to sham. DBSN was applied 15 minutes prior to the IVGTT and during 30 minutes of blood sapling after the glucose injection. As part of the DBSN experiments, a subsequent IVGTT was performed with the devices turning off to demonstrate reversibility of glucose tolerance following termination of the electrical signals. Also, to note, the IVGTT results from the HFAC+stimulation (DBSN) experiments mimicked the results from the vagotomy+stimulation suggesting that HFAC mimicked a vagotomy.

We next tested DBSN in Alloxan treated swine, a common model used in the study of diabetes [100, 101, 102, 103, 104]. In our preparation a titrated amount of Alloxan was delivered to the swine that partially ablated β-cells. This procedure decreased glycemic control but did not induce an insulin dependent state [92]; indicative of a T2D animal model.

DBSN was delivered to swine by an implantable ReShape Lifesciences Viking DBSN™ system (Figure 4) which increased glycemic control compared to sham by assessment of OGTTs (Figure 5). Interestingly, following the cessation of DBSN signals, a sustained decrease in FPG was observed. Over 3 OGTTs, with 2 days rest between experiments, sham conditions demonstrated a consistent FPG of 6.7 ± 0.8 mmol/l. When DBSN was first applied during an OGTT experiment blood glucose declined to 3.7 ± 0.3 mmol/l. Surprisingly, at the initiation of the subsequent DBSN experiment, FPG remained decreased at 3.8 ± 0.2 mmol/l. Following 16 days of DBSN experiments FPG remained depressed at 3.9 ± 0.2 mmol/l and there were no signs of hypoglycemia.

Figure 4.

Alloxan treated swine were implanted with two Viking DBSN™ systems that consisted of two implanted pulse generators, one to deliver 5000 Hz HFAC to the hepatic branch via 2 leads/cuff electrodes and the other to deliver 1 Hz stimulation to the celiac branch via a second pair of leads/cuff electrodes. Above is an implanted swine during a recharging session between glucose tolerance experiments. Two external mobile chargers were connected to two transmitter coils placed superficial to the skin and above the implanted pulse generators. The mobile chargers transmitted energy through the transmitter coils as well as stimulation parameter to be used in subsequent experiments.

Figure 5.

Alloxan treated swine demonstrated increased glycemic control with DBSN as assessed by OGTTs. DBSN signals were applied 5 minutes after the swine consumed glucose and continued for the course of the OGTT experiment.

We also tested if HFAC applied to the hepatic branch alone would decrease glucose following an OGTT and found that there was no effect of the single modality modulation on glycemic control.

4. Discussion

These experiments demonstrated that DBSN is an effective method of increasing glycemic control in animal models of T2D. The necessity of the addition of concurrent celiac stimulation to hepatic block was demonstrated. Possible DBSN mechanisms are discussed below.

The rational to induce a hepatic branch conduction block to increase glycemic control is fourfold: decreasing the liver’s sensitivity to glucagon with increased hepatic sensitivity to insulin, increasing serum levels of GLP-1, increasing tonic release of pancreatic insulin, and decreasing insulin resistance.

It has been demonstrated that hepatic cholinergic denervation decreases the liver’s sensitivity to glucagon. This was shown though a decrease in glucagon-induced hepatic glucose production following hepatic vagotomy [105]. Sub-diaphragmatic vagotomy has also demonstrated an increase in hepatic insulin sensitivity through an upregulation of hepatic insulin receptors and a stronger insulin-induced activation of AKT following vagotomy [106]. Vagotomized mice in this study also demonstrated elevated levels of GLP-1.

Hepatic vagotomy may enhance tonic pancreatic insulin release. In a study by Lee and Miller [107], ligation of the vagus hepatic branch induced increases in both arterial and portal plasma insulin concentrations. Stimulation of the cranial end of the sectioned hepatic branch suppressed this effect. Interestingly, ligation of the celiac branch abolished the hormonal changes due to the vagotomy and the stimulation. These physiological results support the hypothesis that hepatic afferents emit tonic inhibition of celiac preganglionic vagal efferents. Literature suggests a likely mechanism of this flip in activity which involves a glycemic reflexive arc with central nervous system (CNS) processes mainly located in the brain stem [108]. DBSN’s effects on this reflex arc are as follows.

Many studies indicate involvement of hepatic afferent and celiac efferent activity [108] in a glycemic reflex arc with much CNS activity localized to dorsal medullary neural circuits [109]. Similar reflex arcs have been proposed in various vago-vagal loops such as those involved in cardiovascular/respiratory processes [110, 111, 112, 113, 114, 115, 116], digestive processes [117, 118, 119, 120] and the regulation of inflammation [86]. The glycemic reflex arc is initiated with activation of glucose receptors in the portal vein leading to decreased hepatic branch axonal discharge rates [121]. Hepatic branch fibers project to the nucleus tractus solitarius (NST) in the dorsal medulla and form a glutamatergic synapse with their first order neuronal targets [116, 122]. Many of these target neurons involve the multitude of inhibitory GABAergic cells found in metabolic NTS pathways [108, 123]. These neurons project to sites of preganglionic celiac vagal efferents [109] and exert tonic inhibition [108, 124]. Decreased NTS GABAergic activity leads to a release of tonic inhibition of DMV neurons, leading to increased excitatory vagal efferent celiac tone [108]. This neural circuitry can be appreciated though increased discharge rates of celiac efferent fibers resulting from decreased hepatic afferent discharge rates during delivery of glucose to the hepatic portal vein [121, 125]. Blockade of hepatic afferent activity during delivery of DBSN would lead to high celiac discharge rates, increasing tonic pancreatic insulin release. A graphical depiction of this metabolic reflex arc is demonstrated in Figure 6.

Figure 6.

Diabetes block stimulation Neuromodulation affects multiple sub-diaphragmatic vagus nerve physiological processes involved in plasma glucose regulation (a) Tonic activity of vagus nerve’s hepatic branch decreases pancreatic insulin release. Research has demonstrated that increased hepatic branch activity excites inhibitory neurons in the NTS. These neurons then project to the DMV resulting in decreased celiac branch efferent activity. Decreased celiac branch activity leads to decreased tonic insulin release, (b) Increased portal vein glucose concentration induces a decrease in afferent hepatic axon activity leads to celiac axon activity upregulation. This celiac branch efferent activity has been shown to induce pancreatic insulin release, (c) Blocking conduction through the hepatic branch inhibits hepatic neuron tone and, in turn, increases celiac efferent activity, (d) the combination of celiac branch stimulation and hepatic branch blockade may influence various physiological processes leading to increased glycemic control through the following proposed mechanisms: (1) hepatic branch block decreasing hepatic tone leading to increased tonic insulin release, (2) hepatic branch block decreasing the liver’s sensitivity to glucagon, (3) hepatic branch block increasing expression of hepatic insulin receptors, and (4) celiac stimulation potentiating pancreatic insulin release. Despite the greater number of proposed benefits involving block, stimulation was necessary in both our Zucker rat and alloxan treated swine experiments to achieve significantly increased glycemic control. Images created by biorender.com.

It should be noted that communication between the NTS and DMV is not limited to GABAergic projections. Glutamatergic projections from the NTS to the DMV have also been described [109] and there are hypothalamic projections to the DMV [126] that may modify the reflex (Figure 6). However, the vast number of projects from NTS to DMV are GABAergic [108, 123, 127, 128] and localization of the majority of reflex neuronal circuitry located in the brainstem is supported by electrophysiology experiments [108, 109]. This indicates a simple reflexive GABAergic-induced activity flip between the NTS and DMV. Concurrent electrical block, or ligation, with celiac nerve stimulation was required to improve glycemic control in the Zucker rat experiments and demonstrated efficacy in Alloxan treated swine studies.

The final rational for including hepatic block in the DBSN signal is evidence that it may improve insulin resistance. Hepatic nerve ligation has been shown to decrease insulin resistance through attenuation of hepatic PPARα [129] expression as well as to the aforementioned increase expression of hepatic insulin receptors and activation of hepatic AKT [106], a principle molecule in intracellular insulin signaling pathways.

Stimulation of the celiac branch during hepatic block has importance for induction of pancreatic insulin release in the postprandial and fasting states. One limitation of blocking conduction through the hepatic branch is the disruption of the glycemic reflex arcs in the postprandial period. Nagase et al. [125] demonstrated decreased celiac efferent activity and lower insulin release during IPGTTs following hepatic branch ligation. The addition of celiac stimulation may augment these undesired decreases by increasing celiac efferent tone and inducing insulin release (many studies have demonstrated insulin release as a result of vagus nerve stimulation [130, 131, 132, 133, 134, 135] with the requirement that stimulation occurs cranial to either the celiac, or accessory celiac, branching points). In the case of DBSN, the stimulation site was chosen to occur at the celiac branching point to limit possible off target stimulation effects. Experimental results demonstrate the necessity for adding stimulation to blocking to achieve glycemic in the Zucker rat IVGTT experiments.

In the fasting state, vagus stimulation has demonstrated increased insulin concentration in pancreatic venous and in systemic blood circulation [134]. However, a caveat to vagal stimulation-induced insulin release is the co-release of glucagon [133, 134] which can lead to increased blood glucose levels [73, 133, 134]. Decreased hepatic sensitivity to glucagon following vagus nerve hepatic branch ligation [105] suggests hepatic nerve blockade may nullify the effects of glucagon co-release. There was a decrease in FPG during our DBSN experiments.

It should be highlighted that DBSN demonstrated efficacy with signal initiation 5 minutes following the start of the OGTT. This inspires utilizing DBSN with continuous glucose monitoring (CGM) technology to deliver signals on demand, during glucose spikes. Blunting glucose spikes and decreasing blood glucose variability (GV) has positive ramifications for decreasing diabetic co-morbidities [136, 137, 138, 139, 140, 141].

AI and machine learning may be utilized for DBSN parametric set optimization similar to what has been proposed in other systems such as closed-loop insulin pump technology [142]. DBSN has advantages by offering a fully implanted device and multi-site/multi-frequency modulation gives a multitude of parametric combinations to test to maximize glycemic control with AI and machine learning. CGM technology would also enable AI the to refine the timing of DBSN delivery in relation to glucose spikes by using PG prediction algorithms that are in development [143, 144].

One note of particular interest is the observation of lasting changes in FPG following cessation of DBSN signals. One possible mechanism may be due to neuronal plasticity in the glycemic reflex arc induced by applications of DBSN (Figure 7). Experiments demonstrate long term depression of synaptic strength between vagal afferents and their first order glutamatergic synapse in the NST [145] and DBSN signals impact activity at this synapse. Higher order plasticity is also likely to occur in hypothalamic circuits during different glycemic states [126]. Neuronal plasticity has been demonstrated in similar respiration [110, 111, 112, 113, 114, 115, 116] and digestion control neuronal circuitry [117, 118, 119, 120].

Figure 7.

Multiple plastic glutamatergic synapses in the glycemic reflex arc offers many opportunities for activity induced long lasing changes in to increase glycemic control. The above figure labels 5 of such synapses: (1) between hepatic afferents of the tractus solitarius and the NTS, (2) between projection neurons from the NTS to the hypothalamus, (3) between intra-hypothalamic nuclei, (4) between excitatory synaptic projections between the NTS and DMV, and (5) between projections from the hypothalamus to the NTS. Images created by biorender.com.

Exploiting neuronal plasticity has importance for the design of bio-electronic VNM devices to treat T2D. As observed in synaptic long-term potentiation (LTP) and long-term depression (LTD) simulation protocols, only transient stimulation is required for long lasting changes in synaptic strength which can last for hours, days, months or greater. Utilizing neuronal plasticity may hold great importance in minimizing bio-electronic device energy expenditure aiding in the miniaturization of future devices.

5. Conclusion

The use of vagus nerve neuromodulation to treat T2D has great potential. The localized targeting of specific organs responsible for glycemic control may decrease unwanted side effects, a one-time procedure may decrease long term costs of T2D treatments, and an implantable system decreases the need for patient compliance. A key metric to determine the effectiveness and durability of a therapy to treat T2D is the progression of diabetic medication use. First in human clinical trials involving electrical sub-diaphragmatic VNM to treat type 2 diabetics have demonstrated substantially resolved medication use [91] (results in Table 2). These positive results highlight the importance of continued research into VNM which holds the potential of an enhanced therapeutic experience to treat T2D.

Medication change	Percent of subjects (%)
Able to stop taking diabetes medications	22
Decrease in either the number or dosage of medications	16
No change in either the number or dosage of medications	44
Increase in either the number or dosage of medications	17
Advanced to insulin therapy	0

Table 2.

A three-year clinical trial using sub-diaphragmatic vagus nerve neuromodulation demonstrated delayed progression, and often resolution, of T2D medication requirements [91].