EFFECTS OF VAGAL NEUROMODULATION ON FEEDING BEHAVIOR

Abstract

Implanted vagus nerve stimulation (VNS) for obesity was recently approved by the FDA. However, its efficacy and mechanisms of action remain unclear. Herein, we synthesize clinical and preclinical effects of VNS on feeding behavior and energy balance and discuss engineering considerations for understanding and improving the therapy. Clinical cervical VNS (≤30 Hz) to treat epilepsy or depression has produced mixed effects on weight loss as a side effect, albeit in uncontrolled, retrospective studies. Conversely, preclinical studies (cervical and subdiaphragmatic VNS) mostly report decreased food intake and either decreased weight gain or weight loss. More recent clinical studies report weight loss in response to kilohertz frequency VNS applied to the subdiaphragmatic vagi, albeit with a large placebo effect. Rather than eliciting neural activity, this therapy putatively blocks conduction in the vagus nerves. Overall, stimulation parameters lack systematic exploration, optimization, and justification based on target nerve fibers and therapeutic outcomes. The vagus nerve transduces, transmits, and integrates important neural (efferent and afferent), humoral, energetic, and inflammatory information between the gut and brain. Thus, improved understanding of the biophysics, electrophysiology, and (patho)physiology has the potential to advance VNS as an effective therapy for a wide range of diseases.

Graphical abstract

Introduction

Implanted vagus nerve stimulation (VNS) involves delivering electrical signals from an implanted pulse generator to an electrode placed around the vagus nerve. Given the VN’s widespread anatomical targets, VNS has been used or investigated for a wide range of applications, including cervical VNS (cVNS) for epilepsy and depression, as well as subdiaphragmatic VNS (sVNS) for obesity, FDA-approved in 1997, 2005, and 2015, respectively [1,2]. Obesity affects over one third of US adults [3] and 13% of adults worldwide [4]. Herein, we synthesize clinical and preclinical effects of VNS on feeding behavior and energy balance, and discuss engineering considerations for understanding and improving this therapy.

Vagal Anatomy, Morphology, and Function in Energy Balance

The right and left vagi course through the neck, thorax, abdomen, and may extend to the pelvis [5,6]. In the thorax, they innervate the heart, lungs, and esophagus; in the abdomen, they innervate the stomach, intestines, liver, portal vein, and pancreas, among other organs [7,8]. Visceral efferents originate in the dorsal motor nucleus of the vagus in the brainstem; afferent fibers have cell bodies in the nodose ganglion in the neck and project to the nucleus of the solitary tract. Distributions of vagal fibers are shown in Figure 1 for cats [9], and similar distributions are found in other mammalian species (Table 1) [10–13]. These fiber distributions have important physiological (e.g., afferent vs. efferent signaling) and neural engineering implications (e.g., substantially higher activation and block thresholds for unmyelinated vs. myelinated fibers [14]).

Figure 1. Distributions of fibers in the cervical (A) and subdiaphragmatic (B) vagus nerve (VN) in cat [9].

These data, as well as fiber distributions in other species, are shown in Table 1. All data were obtained from electron micrographs.

Table 1. Distributions of vagal fibers at the cervical and subdiaphragmatic levels in cats (Figure 1 [9]), ferrets [10], rats [11], and humans [12].

The human data combined light microscopy and electron microscopy; the other datasets used electron microscopy. Myel: myelinated. Unmyel: unmyelinated. Aff: afferent. Eff: efferent.

Cervical	Myel aff	Myel eff	Unmyel aff	Unmyel eff	Myel	Unmyel	Aff	Eff
Cat [9]	8.3%	12.2%	53.3%	26.2%	20.5%	79.5%	61.6%	38.4%
Ferret [10]	9.3%	4.1%	78.7%	7.9%	13.4%	86.6%	88.0%	12.0%
Rat [13]	--	--	--	--	16.3%	83.7%	--	--
Human [12]	--	--	--	--	19.0%	81.0%	--	--
Subdiaphragmatic
Cat [9]	0.4%	1.5%	96.3%	1.9%	1.9%	98.1%	96.7%	3.3%
Ferret [10]	0.6%	0.0%	93.7%	5.7%	0.6%	99.4%	94.3%	5.7%
Rat [11]	--	--	--	--	0.5%	99.5%	73.0%	27.0%
Human [12]	--	--	--	--	3.0%	97.0%	--	--

Control of food intake and energy homeostasis involves complex humoral and neural pathways. While it is understood that the VN is important for short-term control of food intake, a role in long-term energy balance is also emerging [8,15]. Subdiaphragmatic vagal afferents signal satiety through increased firing rate – often dose-dependent – in response to presence of food in the stomach, gastric distension, nutrient detection in the portal vein in transit to the liver, and hormones (e.g., leptin, CCK) from nutrient-sensing enteroendocrine cells in the gastric and intestinal walls [7,8,15]. Conversely, among the >30 identified gastrointestinal (GI) neurohormones, only ghrelin is known to stimulate feeding, achieved through inhibition of vagal firing [7,15]. Thus, vagal afferent activity provides negative feedback control of ingestion, resulting in vagal efferent activity to control digestion and nutrient absorption. Vagal efferents can synapse onto post-ganglionic cells that either stimulate or inhibit digestion by controlling release of fluids (pancreatic juices; gastric and bile acids) and motility (gastric emptying, mechanical breakdown, propulsion through the intestines) [7]. Regarding long-term energy balance, the VNs have reduced sensitivity to peripheral inputs in obesity through downregulation of satiety-signaling neuropeptides and corresponding receptors on vagal terminals [8]. Furthermore, vagal fibers may change their phenotype in obesity, shifting from anorexigenic signaling (signaling satiety) to orexigenic signaling (increasing food intake) [8,16].

Low Frequency Vagus Nerve Stimulation and Weight Loss

Clinical reports of weight loss in response to left cVNS are inconsistent and largely anecdotal. A few retrospective, uncontrolled studies reported decreased weight in select patients receiving cVNS to treat epilepsy or depression [17–19], whereas others did not [20,21]. These reports suggest that any weight loss with cVNS is not affected by VNS efficacy in treating the target disease [18,19,21] or by medications, many of which are associated with weight gain or loss [19–21], although baseline body mass index (BMI)¹ may be a predictor for weight loss [19,20]. It remains unclear whether stimulation settings (Table 2) affect weight loss [18,19,21].

Table 2. Parameters for low frequency VNS.

Typical stimulation parameters used in clinical left cVNS for epilepsy and depression [22] and in preclinical VNS studies reporting weight loss (Table 3 and [23]). Clinical amplitudes are subject to patient tolerance and typically decrease with increased pulse width.

	Typical clinical values			Ranges of preclinical values
Frequency	20 or 30 Hz			0.05 to 34 Hz
Pulse width	130 μs	250 μs	500 μs	0.1 to 500 ms
Amplitude	2.25 mA	1.75 mA	1.5 mA	0.17 to 4 V or 0.25 to 5 mA
Duty cycle	10% (30 s ON + 5 min OFF)			Typically always on or 12 hrs/day or 30 s ON + 5 min OFF
Waveform	Asymmetric biphasic pulse			Typically monophasic rectangular pulse

A few case studies in other disease states reported effects of VNS on feeding behavior. A case study of an epileptic patient reported 12% weight loss after 15 months of cVNS, but when cVNS was resumed following a 6 month hiatus, weight remained stable, despite higher current amplitude [24]. One retrospective study reported “freedom from food focus” in two patients receiving VNS for Prader-Willi syndrome [25]. Conversely, a prospective study of VNS to help developmentally disabled individuals with weight and nutrition found increased appetites and body weights in individuals who were below their ideal weight, as well as improved motor control allowing them to feed themselves [26]. Both metabolic and behavioral effects of VNS may contribute to the observed body weight increase; for instance, VNS-induced neural signaling and possible metabolic changes may result in increased appetite, while the weight gain may be caused by the increased appetite, improved motor control, and/or metabolic changes.

While clinical results regarding VNS and weight loss are mixed, most preclinical studies report decreased food intake and decreased weight gain, despite a wide range of study designs:

• Species (rat, pig, dog, rabbit) and strains, which may entail different nerve diameters, neural anatomy, and physiological responses;

• Animal models (healthy [growing or stable weight], epileptic, diet-induced obesity);

• Feeding schedules and diets;

• Control groups (sham-operated, naïve, none);

• Stimulation parameters (Table 2);

• Stimulation locations (cervical, thoracic, subdiaphragmatic caudal/rostral to the hepatic branch; right/posterior, left/anterior, bilateral);

• Durations of VNS (2 to 12 weeks);

• Electrode designs (e.g., monopolar/bipolar).

Preclinical studies of VNS for weight loss are clearly summarized in Table 1 of Val-Laillet 2010 [23], and additional studies are summarized in Table 3 using the same layout. Most of these studies applied VNS to the subdiaphragmatic vagi, although they used parameters comparable to clinical cVNS (Table 2). Table 3 also lists key output measures other than weight and food intake. Studies of changes in hormone secretions, blood glucose levels, gastric motility, neural activity, and other metrics in response to VNS are key to uncovering the mechanisms of VNS effects on energy balance. Such investigations are particularly complex given that the direct or short-term effects of altered vagal activity may result in compensatory responses in other neural pathways and physiological systems. In addition to the studies and citations in the publications listed in Table 3, recent review papers discussed the role of the vagus nerve in gut-brain communications and energy balance (e.g., [7,8,27–31]). Note that half of the publications in Table 1 of Val-Laillet 2010 [23] and Table 3 are from a consistent set of investigators; many lack suitable experimental designs (e.g., insufficient explanations of methods; inadequate design of treatment/control groups; lack of justification for parameter settings), do not include robust data analyses (e.g., lacking or incorrect statistical tests; incorrect, misleading, or unclear conclusions), and lack clear data presentation and prose.

Table 3. Preclinical studies of VNS and weight loss.

Summary of preclinical studies on VNS and weight loss, complementing Table 1 in Val-Laillet et al. 2010 [23]. Note that monopolar/bipolar refers to the number of contacts of the electrode around a given nerve trunk, whereas monophasic/biphasic refers to the polarities of the applied electrical signal. There are many other preclinical papers investigating the effects of VNS on energy balance and gut-brain communication that used acute preparations or did not report changes in weight or food intake. Other output measures for which there were no significant changes are not included herein. cVN=cervical vagus nerve; PW=pulse width; sVN=subdiaphragmatic vagus nerve; VN=vagus nerve; VNS=vagus nerve stimulation.

Animal study	Model	VNS location	VNS parameters	VNS dur’n	Weight	Food intake	Other results
Sobocki et al. 2002 [32]	Rabbit (N=16)	Laparotomy Location? Laterality?	Not reported	3 weeks	↓ weight	↓	Rat groups: ↑ amplitude (not frequency) of gastric contractions. ↑ basal acid output.
Dedeurwaerdere et al. 2003 (abstract) [33]	Rat, genetic absence epilepsy (N=10)	Left cVN	30 Hz, 0.5 ms PW, 60 s ON + 12 s OFF, monophasic	2–3 weeks	↓ weight	Not reported	Regained weight after 1 week without VNS.
Biraben et al. 2005 (abstract) [34]	Pig (N=8)	Anterior & posterior thoracic VN	10 to 30 Hz, 0.5 to 5 mA, 20 to 100 ms PW, 30 s ON + 5 min OFF	5 weeks	↓ weight gain	↓	Brain activity (positron emission tomography).
Dedeurwaerdere et al. 2006 [35]	Rat, Fast-kindling (N=17)	Left cVN	30 Hz, 0.25 to 0.5 mA, 0.5 ms PW, 30 s ON + 1.1 min OFF, 2 hrs/day (?), bipolar	10 days (?)	Weight stabilization	↓	Anti-convulsive effects.
Dìaz-Güemes et al. 2007 [36]	Pig, Large White (N=15)	Laparoscopy Anterior sVN	0.5 Hz, 0.5 V, 10 ms PW, monophasic, bipolar	4 weeks	Same initial weight and weight gain as sham	No effect	Acute group (30 min): No change in heart rate or bispectral index; ↓ arterial pressure.
Gil et al. 2009 [37]	Rat, Wistar, diet-induced obesity (N=18)	Laparotomy Anterior sVN	0.05 Hz, 0.2 V, 10 ms PW, monophasic, bipolar	100 days	Same initial weight and weight gain as sham and naïve control	No effect	↓ epididymal fat pad weight in VNS vs. sham, but not VNS vs. control. ↑ mast cell count in gastric wall.
Gil et al. 2011 [38]	Rat, Wistar, diet-induced obesity (N=24)	Laparotomy Anterior sVN	10 Hz, 0.2 V, 10 ms PW, 12 hrs/day, monophasic, bipolar	6 weeks	↓ weight gain	↓	↓ epididymal fat pad weight and leptin. ↑ ghrelin.
Gil et al. 2012 [39]	Rat, Wistar, diet-induced obesity (N=32)	Laparotomy Anterior sVN	1 Hz, 0.2 V, 10 ms PW, 12 hrs/day, monophasic, bipolar	6 weeks	↓ weight gain	↓	↓ epididymal fat pad weight, blood triglycerides, cholesterol, and leptin. ↑ nesfatin-1.
Banni et al. 2012 [40]	Rat, Sprague-Dawley (N=18)	Left cVN	30 Hz, 1.5 mA, 500 ms PW, 30 s ON + 5 min OFF, bipolar	4 weeks	↓ weight gain	↓	No change in total lipid or fatty acid content in feces. See paper for other outcome measures.
Samniang et al. 2016 [41]	Rat, Wistar, diet-induced obesity (N=16)	Left cVN	20 Hz, 0.5 to 0.75 mA, 0.5 ms PW, 14 s ON + 48 s OFF, bipolar	12 weeks	No effect	No effect	↓ visceral fat, blood pressure, plasma insulin, cholesterol, and triglycerides; other improved metabolic and hemodynamic output measures.
Johannessen et al. 2017 [42]	Rat, Sprague-Dawley (N=21)	Laparotomy Anterior & posterior sVN	30 Hz, 0.5 to 2 mA (increased weekly), 0.5 ms PW, 30 s ON + 5 min OFF, bipolar	6–8 weeks	↓ weight or weight gain (unclear from results)	↓	48 hrs of VNS: hormones in brain, brainstem, plasma; feeding behavior.
Malbert et al. 2017a [43]	Pig, Yucatan mini-pig, diet-induced obesity (N=15)	Laparoscopy Anterior & posterior sVN	30 Hz, 5 mA, 500 ms PW, 30 s ON + 5 min OFF, biphasic, bipolar	12 weeks	↓ weight gain	Not reported	↑ insulin sensitivity; ↑ hepatic, skeletal, and brain glucose uptake (restored fasting glucose metabolism). ↓ SQ fat, but not visceral fat.
Malbert et al. 2017b [44]	Pig, Large White (N=24)	Laparoscopy Anterior & posterior thoracic VN (at diaphragm)	30 Hz, 30 s ON + 5 min OFF, bipolar S1) 1 ms PW: 5 mA S2) 1 ms burst with 14 pulses (25 μs on, 50 μs off): 15 mA S3) S2 with rising amplitude Exponential recharge phase?	2 weeks	Not reported	↓ (S2 and S3)	With S2 and S3: shorter eating bouts; ↓ high fat and high sugar food; ↑ activation of dorsal motor nucleus of the vagus and metabolism of downstream cortical structures.

Effects of VNS on weight could reflect changes in behavior (food intake) and/or metabolism. Half of patients receiving cVNS for depression had a significant reduction in craving in response to images of sweet food between cVNS on and off conditions, while the other half of the patients had an increase in craving. Increased craving for sweets was correlated with lower cVNS ON time, lower stimulation amplitude, and lower BMI [45]. Interestingly, in two porcine studies (6 and 14 weeks of VNS), the treatment group consumed significantly less high-carb food compared to sham-operated animals [23,46]. In a pair-fed study, rats receiving 4 weeks of VNS still had significantly lower weight than sham-operated, pair-fed rats (7.5%) albeit with smaller differences than with free feeding (25%) [40], suggesting important changes in both food intake and metabolism to achieve weight loss with VNS. On the metabolic side, 12 weeks of VNS in obese pigs restored glucose metabolism [43], 4 weeks of VNS in pigs did not change the O₂ consumption or CO₂ production [47], and 15 patients receiving VNS for epilepsy exhibited increased energy expenditure related to changes in the activity of brown adipose tissue [48], although these latter findings may be confounded by a lack of validated methods [43]. Overall, these data suggest that both change in food intake (quantity and content), as well as changes in metabolism, may contribute to weight loss with VNS. Future studies may investigate why preclinical results show decreased food intake and reduced weight gain with VNS, while analogous clinical studies are ambiguous regarding the effects of VNS on energy balance.

Kilohertz Frequency Vagus Nerve Stimulation and Weight Loss

Whereas the aforementioned clinical observations of weight loss in response to cVNS were secondary to a target disease, EnteroMedics Inc. (now ReShape Lifesciences Inc.; St. Paul, MN) developed vBloc® therapy – “Vagal BLocking for Obesity Control” – to promote weight loss by putatively blocking conduction in the subdiaphragmatic VNs using 5 kHz signals. The clinical and preclinical studies discussed above used low frequency signals (≤30 Hz) to elicit action potentials in vagal fibers. However, kilohertz frequency (KHF) signals can instead reversibly block action potential conduction in axons [49]. EnteroMedics conducted two randomized, double-blinded, prospective, controlled, multi-center clinical trials (Figure 2). Although vBloc® failed to meet primary efficacy endpoints in either trial, the therapy received FDA approval in 2015; the FDA Advisory Committee concluded that the 18-month ReCharge results indicated sustained weight loss and that the potential benefits outweighed the risks [50]. Further, an FDA-sponsored patient survey revealed a group of patients accepting of the device’s risks in light of the potential weight loss [50].

Figure 2. Percent excess weight loss (EWL) in EnteroMedics’ clinical trials (mean ± SEM).

Twelve-month data are shown for the first clinical trial, EMPOWER, for all patients (n_ctl=88, n_treat=165) and for patients who used vBloc® at least 12 hours per day (n_ctl=14, n_treat=16) [51]. Data are also shown for the second clinical trial, ReCharge, at 12 months (primary efficacy end point) (n_ctl=77, n_treat=162) [52,53], at 18 months (n_ctl=77, n_treat=162) [53], and at 24 months (n_treat=103) [54]. For both trials, the primary efficacy end point was defined as significantly greater percent EWL in the treatment group by at least 10 percentage points at 12 months. Missing data are assumed to be missing at random, except the 24-month data which exclude patients with missing data.

In the EMPOWER study (2007–2008, n=253), both groups received the same implant, but the treatment group received 5 kHz stimulation at 3 to 8 mA for 5 min ON and 5 min OFF, while the control group received 40 Hz stimulation up to 1 mA during the 5 min ON period, as well as 13 impulses of 1 kHz at 3 mA at t = 0 min and t = 3 min of each ON period. Both groups achieved 16 to 17% excess weight loss (EWL)², and the study failed to meet its primary efficacy endpoint. The stimulation parameters in the control group may have had therapeutic effects, and the observation that both groups had greater weight loss for more hours of device use suggests both potential therapeutic efficacy and placebo effect [51].

In the ReCharge study (2011–2012, n=239), both groups received an implanted pulse generator that required recharging, but the control group did not receive electrode leads to the VNs. In the treatment group, therapy was delivered for at least 12 hours per day, eliminating patient compliance as a confounding factor. However, at 12 months, the study failed to meet the primary efficacy end point of significantly different EWL by at least 10 percentile points (24% vs. 16% EWL), although it was attained at 18 months (23% vs. 10% EWL) (Figure 2) [52,53]. At 24 months, the treatment group had 21% EWL [54]. For comparison, Roux-en-Y gastric bypass results in 83% EWL at 1 year [55].

While vBloc® is approved for sale in the United States, advancement of this therapy requires carefully designed and controlled post-market studies, including stimulation parameter optimization (e.g., effects of duty cycle (Figure 2); increased stimulation amplitudes given the high thresholds of unmyelinated fibers (Table 1 and Figure 1)) and consideration of electrode placement given varied subdiaphragmatic vagal branching patterns [56,57].

EnteroMedics claims to achieve weight loss by blocking conduction in the subdiaphragmatic VNs, thereby causing earlier satiation at meals, reduced food intake, and reduced hunger between meals [51,54,58]. The vBloc® therapy also reduced blood pressure, cholesterol levels, and triglycerides [54]. However, the efficacy of vBloc® and its target neural and hormonal mechanisms of action are unclear. The substantial weight loss in the control groups (Figure 2) revealed important placebo effects from undergoing surgery. Further, the prevalence of gastrointestinal discomfort reported as adverse events in the ReCharge trial at 12 months [52] could also lead to reduced food intake, although they were mostly resolved by 24 months [54]. Computational modeling suggests that the 5 kHz signals may elicit action potentials, rather than blocking conduction [59], which could indicate a mechanism where vBloc® enhances the satiety signals of vagal afferents. Indeed, EnteroMedics’ rat study produced partial block in small myelinated fibers at >1 mA and in unmyelinated fibers at >5 mA, using the same range of amplitudes as clinical vBloc® [60]. However, block thresholds in the preclinical study were likely lower than in humans due to nerve desheathing, smaller nerve diameter in rats than humans, lack of scar tissue encapsulating the electrode, and hook wire electrodes (smaller surface area, and thus higher charge density than the cuff electrodes used clinically). Further, the room temperature thresholds would likely differ from body temperature measurements (e.g., [61–63]). In addition, unlike established literature showing immediate reversal of conduction block [49], the preclinical study found block carryover for minutes after stimulation. On the other hand, certain physiological results are consistent with block of vagal efferents: suppressed plasma pancreatic polypeptide (PP) responses to sham feeding (chew, then spit) in vBloc® patients (Camilleri et al., 2008; Taché and Wingate, 1991), as well as decreased pancreatic exocrine secretions and decreased frequency of gastric contractions in a porcine model [64,65]. However, PP signals satiety, encouraging decreased food intake and increased energy expenditure, so its suppression would be expected to yield weight gain [66–68].

Vagotomy and Weight Loss

It is interesting to consider the effects vagal activation (0.5 to 30 Hz) and putative block (KHF) in the context of vagotomy (transection of the subdiaphragmatic VNs), which inspired vBloc® therapy. Vagotomy was originally used to treat peptic ulcer disease and has since been used to treat obesity, albeit with unclear and uncontrolled results. Weight loss of ~15% over 1 to 4 years is reported [69–72], although half of the 21 patients in one study underwent re-operation using other bariatric surgical techniques due to insufficient or transient weight loss [69]. Vagotomy in conjunction with other bariatric surgery has not increased weight loss [72–74]. Vagotomy was used extensively in preclinical models to study brain-gut communications [7,31]; some studies found reduced body weight with vagotomy [75,76] while others did not [77] or only observed decreased fat stores [78–80]. However, reduced body weight with vagotomy may be due to lack of gastric emptying [8,81].

Parameters and Specificity of Vagus Nerve Stimulation

Studies of VNS for weight loss are paradoxical from two perspectives: vagal fibers can signal both hunger and satiety, and experiments investigated both putative vagal activation and block with the same objective of weight loss. Low frequency VNS (≤30 Hz) elicits action potentials in some nerve fibers within the cuff electrode. Preclinical data show weight loss and decreased food intake with low frequency VNS in multiple animal models with different stimulation locations and parameters. Clinical low frequency cVNS has inconsistent results with respect to weight loss, albeit through uncontrolled studies in heterogeneous populations receiving cVNS for diseases other than obesity. KHF VNS is intended to block conduction in some or all of the nerve fibers within the cuff electrode, although depending on the fiber diameter and electrode-fiber distance, KHF VNS can also elicit action potentials [59]. Clinical KHF VNS resulted in significant weight loss, although the mechanisms of action are unclear, and may include a combination of placebo, GI discomfort causing reduced food intake, vagal block in certain fibers (analogous to vagotomy), and/or vagal activation (as in low frequency VNS).

To determine mechanisms of action and achieve effective and selective VNS for obesity or other applications, many parameters must be considered, as illustrated in the Graphical Abstract, including electrode design (geometry, materials, impedance), and stimulation parameters (frequency, pulse width, waveform, and amplitude). For example, fMRI revealed different brain activation between VNS patients using 250 or 500 μs pulse width vs. 130 μs [82]. The duration, duty cycle, and daily timing of VNS must also be considered in the context of vagal and downstream cortical plasticity and possible desensitization. The anatomical level of stimulation (cervical, thoracic, subdiaphragmatic; primary vagal trunks or smaller downstream branches; unilateral or bilateral) and characteristics of the target fibers (fiber diameter, fiber myelination, and electrode-fiber distance) affect thresholds and thereby the most effective and selective VNS parameters. Finally, differences between preclinical animal models and humans, including nerve diameter, branching, fascicular morphology, fiber distributions, physiology, growing versus stable weight, hedonic desire for food, and duration of stimulation must be considered when translating findings and selecting parameters. Thus, to further develop VNS for obesity, we need better understanding of the target fibers in animal models and in humans, as well as improved methods to selectively stimulate and block specific fiber types. In future studies, these experimental design parameters should be reported in detail, considerations of the translation of the findings should be discussed, and systematic variation of stimulation parameters should be examined, rather than, for example, being constrained to those used for VNS for epilepsy.

New Applications of Vagus Nerve Stimulation and New Treatments for Obesity

The VN is a target for neuromodulatory therapies to treat a wide range of diseases (e.g., inflammatory bowel disease, PTSD, tinnitus, and stroke [83]) given its innervation of multiple organs and its neuro-immuno-humoral roles. In VNS for obesity, insights are emerging into broader underlying physiological phenomena, including inflammation and the gut microbiota [15]. Further, the potential exists for improved understanding of eating disorders and effective treatments by synthesizing findings across different treatment options, including deep brain stimulation [84], bariatric surgery [85], gastric pacing [86], and pharmaceuticals [87]. Finally, as in all neural stimulation therapies, improved understanding of the anatomy, biophysics, neural signaling, and (patho)physiology may translate to other applications of VNS and allow development of optimized electrode designs, stimulation parameters, and patient-specific programming to permit effective and selective modulation of target nerve fibers.

HIGHLIGHTS.

• Vagus nerve stimulation (VNS) for obesity was recently approved by the FDA

• The efficacy and mechanisms of VNS remain unclear

• Preclinical low frequency VNS studies report weight loss; clinical findings vary

• Clinical kilohertz frequency VNS shows weight loss, but unclear effects versus sham

• Need engineering design of vagal stimulation and block for efficacy and selectivity

Abbreviations

BMI

Body mass index

CCK

Cholecystokinin

cVNS

Cervical vagus nerve stimulation

EWL

Excess weight loss

FDA

Food and drug administration

GI

Gastrointestinal

KHF

Kilohertz frequency

PP

Pancreatic polypeptide

PW

Pulse width

sVNS

Subdiaphragmatic vagus nerve stimulation

vBloc®

Vagal blocking for obesity control

VN

Vagus nerve

VNS

Vagus nerve stimulation