Peripheral nerve-mediated glucose lowering: Mechanisms, translational strategies, and future perspectives

Abstract

Glucose homeostasis is crucial for metabolic health, with disruptions leading to hyperglycemia (e.g., diabetes affecting over 589 million adults worldwide in 2025) or hypoglycemia, both associated with severe complications like nephropathy, neuropathy, and cardiovascular disease. The peripheral nervous system (PNS) is a less studied regulator of glucose balance through neural circuits involving the sympathetic, parasympathetic, sensory, and somatic nerves, which interact with organs like the liver, pancreas, and gut. This review aims to synthesize recent advances in PNS-mediated glucose lowering, covering mechanistic foundations, organ-specific pathways, translational interventions (e.g., device-based neuromodulation, pharmacological approaches, closed-loop systems, and regenerative strategies), preclinical/clinical evidence, translational challenges, and future perspectives to provide a roadmap for developing novel, drug-free therapies for dysglycemia.

Key Words: Peripheral nervous system; Glucose homeostasis; Diabetes; Neuromodulation; Vagus nerve stimulation; Focused ultrasound; Closed-loop system; Non-pharmacological intervention; Metabolic flexibility

Core Tip: The peripheral nervous system is an underappreciated regulator of glucose balance. Targeted neuromodulation, like vagus nerve stimulation or focused ultrasound, modulates neural circuits to control liver glucose output, pancreatic hormones, and tissue uptake. Preclinical data show strong metabolic gains; early human trials prove safety. Tackling issues in precision, patient differences, and nerve damage could enable scalable, drug-free treatments for diabetes.

INTRODUCTION

Glucose homeostasis is a tightly regulated process essential for maintaining metabolic health. Disruption of this balance, manifesting as hyperglycemia or hypoglycemia, is closely associated with significant morbidity and mortality. According to the latest estimates by the International Diabetes Federation (2025), over 589 million adults worldwide are living with diabetes, and an additional 1.1 billion of the global population present with prediabetic states such as impaired fasting glucose or impaired glucose tolerance[1]. Persistent hyperglycemia imposes a substantial burden through microvascular and macrovascular complications, including nephropathy, neuropathy, retinopathy, and cardiovascular disease. Conversely, hypoglycemia—although less common—remains a serious, potentially lifethreatening complication, especially in insulintreated individuals, elderly patients, and those in critical care. Severe episodes may result in seizures, cognitive dysfunction, cardiac arrhythmias, and, in extreme cases, death. Importantly, recurrent hypoglycemia can impair counterregulatory hormone responses and diminish autonomic warning symptoms, leading to hypoglycemia unawareness. This condition markedly increases the risk of delayed recognition and treatment of low glucose levels, thereby heightening the likelihood of severe outcomes. These dual threats highlight that effective glucose regulation strategies must not only address hyperglycemia but also prioritize prevention of both symptomatic and unrecognized hypoglycemia[2].

Significant strides have been made in recent years in elucidating the mechanisms by which the peripheral nervous system (PNS) modulates glucose homeostasis, highlighting its multifaceted role in maintaining glycemic stability. Vagus nerve stimulation (VNS) has emerged as a promising approach to enhance insulin sensitivity. VNS exerts its effects through a combination of peripheral and central mechanisms: Efferent cholinergic signaling directly targets pancreatic β-cells and α-cells, promoting insulin secretion while suppressing glucagon release, thereby facilitating glucose uptake and utilization in peripheral tissues; afferent signals from the vagus nerve converge in brainstem nuclei, including the nucleus tractus solitarius (NTS) and dorsal motor nucleus of the vagus (DMV), where central integration modulates autonomic outflow to metabolic organs, coordinating systemic glucose regulation[3]. The sympathetic nervous system (SNS) also exerts a profound influence on glucose regulation, particularly in the context of hepatic gluconeogenesis. Selective inhibition of sympathetic nerve activity has been shown to significantly reduce hepatic gluconeogenesis, leading to a decrease in blood glucose levels. This mechanism primarily involves the attenuation of norepinephrine (NE) release, a neurotransmitter that typically stimulates glycogenolysis and gluconeogenesis in the liver[4]. Peripheral sensory nerves, such as TRPV1 and Piezo2, have garnered attention for their role in detecting nutrients in the gut. These sensory nerves transmit signals to the central nervous system (CNS) via the solitary nucleus in the brainstem, thereby modulating insulin and glucagon secretion to maintain postprandial glucose homeostasis. This sensory feedback loop is crucial for the fine-tuning of glucose levels following nutrient intake. Moreover, the vagus nerve is intricately involved in complex signaling pathways with multiple organs, including the liver, pancreas, and gut, to achieve synchronized metabolic regulation. For instance, the vagus nerve regulates glycogen synthesis in the liver and insulin secretion in the pancreas, facilitating coordinated metabolic responses across organs. Additionally, the role of the SNS in brown adipose tissue (BAT) has been explored, with its activation promoting glucose oxidation to maintain energy balance[5]. Recent advances in PNS-targeted neuromodulation offer novel avenues for glucose regulation. Transcutaneous auricular VNS (taVNS) modulates autonomic tone and peripheral organ function, potentially supporting glycemic control[6]. High-frequency spinal cord stimulation (SCS), though developed for neuropathic pain, can alter sympathetic output and thereby influence hepatic glucose production and peripheral uptake[7]. Optogenetic stimulation of pancreatic β-cells or GLP-1-producing enteroendocrine cells (EECs) has improved insulin secretion and glucose tolerance in preclinical models. Non-invasive focused ultrasound can target vagal or sympathetic pathways without implants[8], and miniaturized wireless microstimulators offer long-term, closed-loop autonomic modulation[9]. These approaches enable precise manipulation of metabolic neural circuits and pave the way for integrated neuro-metabolic control systems.

This review synthesizes recent progress in peripheral nerve-mediated glucose lowering, covering mechanistic foundations, organ-specific neural pathways, and translational interventions (Figure 1). We first outline the physiological and anatomical basis of PNS-mediated glucose regulation, followed by an analysis of well-characterized neural control mechanisms. We then examine current and emerging translational strategies, encompassing device-based neuromodulation, pharmacological interventions, integrated closed-loop systems, and regenerative approaches, and conclude with translational challenges and future perspectives. Our aim is to provide a roadmap for advancing PNS-targeted modulation as a novel therapeutic approach for dysglycemia.

Figure 1 Peripheral nerve-targeted neuromodulation for glucose control. The peripheral nervous system regulates multiple metabolic organs, including the liver, adipose tissue, pancreas, skeletal muscle, and gut, to maintain glucose homeostasis. Key actions include inhibiting hepatic gluconeogenesis and enhancing glycogen synthesis; modulating lipolysis and brown adipose tissue thermogenesis; regulating insulin secretion from β-cells and glucagon release from α-cells; promoting glucose uptake and utilization in skeletal muscle; and enhancing gut hormone (e.g., GLP-1) secretion to improve glucose metabolism. This mechanism provides potential therapeutic targets for neuromodulation in metabolic disorders such as diabetes. BAT: Brown adipose tissue. Created in BioRender (Supplementary material).

PHYSIOLOGICAL AND ANATOMICAL BASIS OF PNS IN GLUCOSE REGULATION

Overview of PNS in metabolic homeostasis

Traditionally, the role of the autonomic nervous system (ANS) in metabolism has often been oversimplified as a dichotomy: The sympathetic nervous system (SNS) mediates fight or flight responses that promote energy mobilization, while the parasympathetic nervous system (PSNS; specifically the vagus nerve) supports rest and digest functions that facilitate energy storage[10]. However, cutting-edge research has fundamentally transformed this perspective, revealing that the PNS functions as a sophisticated network capable of refined, dynamic, and organ-specific metabolic regulation. Far from a simple on/off switch, it constitutes a central command system that continuously senses, integrates, and responds to changes in the internal and external environment to maintain whole-body energy homeostasis[11].

This regulatory network comprises four major components, each playing a distinct yet highly coordinated role in blood glucose regulation. SNS generally exerts antagonistic effects by elevating blood glucose through multiple mechanisms: Stimulating hepatic glycogenolysis and gluconeogenesis (thus increasing hepatic glucose output), suppressing insulin secretion (pancreas), promoting glucagon release, enhancing lipolysis (adipose tissue), and restricting glucose utilization in skeletal muscle[12]. PSNS primarily mediated by the vagus nerve, and it acts to lower blood glucose in the postprandial state through multiple integrated mechanisms: Stimulating the secretion of hormones such as insulin and GLP-1 from the pancreas, enhancing hepatic glycogen synthesis, and modulating gastrointestinal motility and nutrient absorption.

Acting as metabolic sensors, sensory nerves are distributed throughout organs such as the liver, pancreas, gastrointestinal tract, and adipose tissue. They express various receptors (e.g., TRPV1, Piezo2) that detect signals including glucose, free fatty acids, hormones (e.g., insulin, leptin), and mechanical stress. This information is relayed to the CNS via spinal or vagal afferent pathways. Dysfunction in this sensory signaling is a key element in complications such as diabetic peripheral neuropathy and represents an important target for neuromodulation therapies[13].

Although somatic nerves do not directly innervate internal organs, somatic nerves function as the controllers of skeletal muscle contraction—effectively regulating the body’s largest glucose-consuming tissue and thereby exerting substantial influence over glucose disposal.

The contemporary view characterizes peripheral neural regulation of metabolism as a multimodal, bidirectional communication system. This system not only executes central commands but also forms closed-loop feedback circuits via sensory afferents. Furthermore, it engages in complex interactions such as immune-neural crosstalk (e.g., neuromodulation of macrophage function) and gut microbiota-brain axes, enabling precise coordination of glucose homeostasis, energy allocation, nutrient digestion and absorption, and immunometabolic coupling[14].

Organ-specific neural pathways

Regulation of blood glucose by the PNS is achieved through direct innervation of key metabolic organs. Table 1 systematically summarizes the neural innervation, physiological functions, and underlying molecular mechanisms in major organs.

Table 1 Innervation of key metabolic organs and their roles in blood glucose regulation.

Organ	Nerve type	Innervation description	Role in glucose regulation	Key neurotransmitters/pathways
Liver	SNS	Predominantly innervated by the greater splanchnic nerve	Activation stimulates hepatic glycogenolysis and gluconeogenesis, thereby increasing hepatic glucose output	Primarily releases norepinephrine, which acts on α- and β-adrenergic receptors
	PSNS	Innervated by the hepatic branch of the vagus nerve	Activation promotes glycogen synthesis and suppresses gluconeogenesis, leading to reduced hepatic glucose output	Releases acetylcholine, acting primarily on M3 muscarinic receptors
	Sensory nerves	Express various metabosensors	Detect intrahepatic signals such as glucose levels, ATP/AMP ratio, and inflammatory cytokines, and relay this information to the brainstem and hypothalamus	/
Pancreas	SNS	Originates from the celiac ganglion	Activation inhibits insulin secretion from β-cells while stimulating glucagon release from α-cells	Releases norepinephrine, which acts on α2-adrenergic receptors to suppress insulin secretion
	PSNS	Derived from the pancreatic branch of the vagus nerve	Activation (particularly postprandially) strongly stimulates the secretion of both insulin and glucagon, exhibiting a biphasic effect	Primarily releases acetylcholine acting on M3 receptors, promoting insulin secretion via the IP3/PKC signaling pathway. Additionally, neuropeptides such as VIP and PACAP are involved in enhancing secretory responses
	Sensory nerves	Densely distributed throughout the islets of Langerhans and surrounding pancreatic tissue	Detect local insulin and glucose levels, and participate in the feedback regulation of pancreatic islet function	/
Adipose tissue	SNS	Heavily innervates both white and brown adipose tissue	Activation stimulates lipolysis, increasing the release of FFAs, which may indirectly affect hepatic glucose output and muscle glucose utilization via lipotoxicity	Releases norepinephrine, which primarily acts on β3-adrenergic receptors to promote lipolysis
	Sensory nerves	Provide feedback on adipose tissue metabolic status	Detect levels of adipokines such as leptin and adiponectin, and relay energy storage signals to the central nervous system	Express receptors such as LepR and TrkB, the latter being a high-affinity receptor for BDNF
Skeletal muscle	SNS	Innervates blood vessels and muscle fibers	Moderate activation induces vasoconstriction, thereby limiting glucose delivery; excessive activation indirectly suppresses glucose uptake via β-AR-mediated mechanisms	Releases norepinephrine, which acts on α1-adrenergic receptors (causing vasoconstriction) and β2-adrenergic receptors (promoting vasodilation and enhancing glucose uptake). Key mechanisms in contraction-induced
	Somatic motor nerves	Regulate voluntary muscle contraction	Muscle contraction per se serves as the most potent stimulus for glucose uptake and utilization, primarily through AMPK activation and enhanced GLUT4 translocation	/
Gastrointestinal tract	PSNS	Vagus nerve (afferent/efferent) SNS	Enhances intestinal motility, stimulates secretion, and increases nutrient absorption surface area, thereby indirectly modulating the rate of blood glucose elevation	/
	Sensory nerves	Extremely abundant	Play an essential role. They detect nutrients such as glucose, fatty acids, and amino acids, as well as hormones (e.g., GLP-1, PYY, CCK), and transmit these signals via vagal afferents to the NTS. This triggers gut-brain axis reflexes that preemptively regulate insulin secretion (cephalic phase insulin release) and promote satiety	Express a wide range of nutrient-sensing receptors, including but not limited to GLP-1R, CCKAR, SGLT1, and GPR40

SNS: Sympathetic nervous system; PSNS: Parasympathetic nervous system; VIP: Vasoactive intestinal peptide; PACAP: Pituitary adenylate cyclase-activating polypeptide; FFA: Free fatty acid; BDNF: Brain-derived neurotrophic factor; NTS: Nucleus tractus solitarius.

Closed-loop reflex architecture

The organ-specific neural innervation described above does not operate in isolation but is integrated within a sophisticated closed-loop reflex arc, which is central to understanding peripheral nervous glycemic control. This classic reflex arc consists of four essential components, which in the context of metabolic regulation are manifested as follows: Peripheral nerves act as vigilant detectors scattered across key organs like the portal vein, liver, gut, and pancreas. These endings serve as chemical and mechanical sensors, constantly tracking elements such as blood sugar, available nutrients, hormone levels, and various metabolites[15]. In the incoming signal route, these sensory inputs are converted by neurons into electrical impulses, transmitting signals through two principal pathways: Vagal afferents, which convey visceral information from thoracoabdominal organs to the brainstem, and spinal afferents, which relay somatic and visceral input via dorsal root ganglia. Vagal afferents constitute the parasympathetic sensory component, whereas spinal afferents are anatomically distinct and independent of parasympathetic circuits. Collectively, they channel metabolic information to the brain. For vagal paths, the neuron bodies are located in the nodose ganglion, with their extensions connecting to the NTS in the brainstem[16]. Acting as the main hub for processing, the NTS gathers and begins to merge these internal signals. From there, data flows to advanced brain areas like the hypothalamus, including spots such as the arcuate, paraventricular, and ventromedial nuclei that oversee energy and hormone systems, the parabrachial nucleus, and even the cerebral cortex, which handles feelings like appetite. Collectively, these form a core command center that analyzes the inputs thoroughly and crafts suitable outgoing instructions[17].

These coordinated directives are sent out through sympathetic channels starting in areas like the hypothalamus, moving to the spinal cord's intermediolateral column, then to follow-up neurons in sympathetic clusters, and finally to the intended organs. Similarly, parasympathetic routes begin at the dorsal vagal complex, extend through preganglionic vagal fibers to local postganglionic neurons, and reach their destinations. Together, they fine-tune processes like insulin release, liver sugar production, muscle sugar absorption, and digestive operations, bringing blood glucose back to its ideal level[18] (Table 2).

Table 2 Overview of the neural regulation of glucose metabolism.

Component	Description	Key structures/examples	Primary function	Ref.
Sensors	Sensory nerve endings acting as chemoreceptors and mechanoreceptors	Portal vein, liver, intestine, pancreas	Monitor blood glucose, nutrients, hormones, and metabolites	Joly-Amado et al[15], 2022
Afferent pathway	Nerves transmitting sensory signals to the central nervous system	Vagal afferent nerves (via nodose ganglion), spinal afferent nerves	Encode and relay peripheral signals to the brainstem (NTS)	Li et al[16], 2023
Central integration	Brain regions that process sensory information and generate commands	NTS, hypothalamus (ARC, PVN, VMH), parabrachial nucleus, cortex	Integrate signals to regulate energy balance and generate sensations like hunger	Vohra et al[17], 2022
Efferent pathway	Nerves conveying commands from the brain to peripheral organs	Sympathetic pathways, Parasympathetic pathways (dorsal vagal complex)	Regulate insulin secretion, glucose production/uptake, and gastrointestinal function to restore homeostasis	Papazoglou et al[18], 2022

NTS: Nucleus tractus solitarius; ARC: Arcuate nucleus; PVN: Paraventricular nucleus; VMH: Ventromedial hypothalamus.

This closed-loop system explains why blood glucose regulation is both rapid and precise. For example, postprandial nutrient sensing in the gut is transmitted via vagal afferent nerves to the CNS, which can then—even before blood glucose rises significantly—trigger cephalic phase insulin release through vagal efferent pathways, thereby preparing the body for the impending glucose load. Similarly, at the onset of exercise, sensory signals from muscles are relayed to the CNS, which modulates sympathetic output to the liver and pancreas to ensure adequate glucose supply to active muscles.

The PNS is far more than a passive set of wires; it constitutes an active and intelligent regulatory network. This integrated physiological perspective highlights the foundational role of sensory information and central integration in maintaining glucose homeostasis. Furthermore, it provides a physiological basis for integrating such mechanisms with artificial intelligence (AI) and continuous glucose monitoring (CGM) technologies to develop intelligent, real-time feedback control systems[19].

WELL-CHARACTERIZED MECHANISMS OF PNS-ENDOCRINE-IMMUNE REGULATION

Energy metabolic homeostasis relies on the intricate interplay among the nervous, endocrine, and immune systems. Traditional perspectives have predominantly focused on isolated autonomic neural control or endocrine regulation. However, contemporary studies have demonstrated that these mechanisms are interdependent—central neural circuits integrate neural and endocrine signals to coordinate immune responses, thereby fine-tuning blood glucose levels within a dynamic equilibrium[20]. Within the PNS, both the PSNS and SNSs play pivotal roles in glucose homeostasis. These divisions establish tight functional coupling with the CNS through positive and negative feedback loops, collectively contributing significantly to the reduction of blood glucose levels.

Vagal parasympathetic pathways

The PSNS, particularly the vagus nerve, is critically involved in anabolic processes and plays an essential role in energy storage. Efferent fibers of the vagus nerve originating from the DMV project to pancreatic β-cells, where the released acetylcholine (ACh) activates M3 muscarinic receptors, triggers calcium influx, and enhances glucose-stimulated insulin secretion (GSIS)[21]. In addition to its insulinotropic effect, vagal signaling also suppresses glucagon secretion from pancreatic α-cells, thereby exerting a dual glucose-lowering effect. Furthermore, subsequent studies have demonstrated that vagal activity promotes hepatic glycogen synthesis and suppresses gluconeogenesis via both neural and humoral pathways: Insulin acts on hypothalamic AgRP neurons to activate vagal efferents, which project to the liver and stimulate resident macrophages, such as Kupffer cells, to locally release interleukin-6. This, in turn, downregulates the expression of gluconeogenic enzymes such as PEPCK, thereby reducing hepatic glucose output. Notably, the regulation of glycemia by the vagus nerve is not solely dependent on ACh. Evidence suggests that neuropeptides, such as vasoactive intestinal peptide, act as co-transmitters and modulate glucose metabolism through endothelial insulin signaling, highlighting the mechanistic diversity of vagal-mediated metabolic control[22]. In summary, the vagal parasympathetic pathway regulates systemic metabolism by enhancing insulin secretion, suppressing hepatic glucose production, and promoting glycogen storage.

Sympathetic pathways

In contrast to the PSNS, the SNS primarily exerts catabolic actions, lowering blood glucose by increasing energy expenditure. In the liver, sympathetic innervation stimulates glycogenolysis and gluconeogenesis via α1- and β2-adrenergic receptors, which elevate intracellular cAMP levels, activate protein kinase A, and subsequently phosphorylate glycogen phosphorylase. This mechanism is typically engaged during fasting or stress to increase circulating glucose levels in order to meet heightened energy demands. Conversely, suppression of hepatic sympathetic tone—achieved through pharmacological or genetic interventions—can inhibit gluconeogenic pathways and improve insulin sensitivity.

Notably, SNS activation also promotes glucose oxidation through the activation of BAT. Sympathetic fibers innervating BAT release NE, which binds to β3-adrenergic receptors and triggers thermogenesis mediated by UCP1, thereby consuming substantial amounts of glucose and fatty acids[23]. This process is independent of insulin and represents a unique physiological function of the SNS, particularly prominent during cold exposure or administration of adrenergic agonists.

Recent studies have further revealed that the sympathetic axis can directly modulate immune cell function. NE, acting via β2-adrenergic receptors, promotes the polarization of anti-inflammatory M2 macrophages in adipose tissue, thereby suppressing pro-inflammatory cytokine expression and improving insulin sensitivity. For example, Lu et al[24] demonstrated that electroacupuncture suppressed sympathetic activity in epididymal white adipose tissue of mice, markedly reducing neuropeptide Y (NPY) release and alleviating adipose tissue inflammation. However, chronic SNS activation may lead to catecholamine resistance and sympathetic neuropathy, thereby exacerbating metabolic dysfunction. Together, these findings underscore the importance of SNS-immune cell interactions in metabolic regulation. Therefore, the SNS modulates glucose metabolism through two interconnected mechanisms: Directly regulating hepatic glucose production and BAT-mediated glucose disposal, as well as integrating immune-metabolic pathways to influence tissue insulin sensitivity.

Sensory afferent pathways

Afferent nerves convey peripheral metabolic information to the CNS, enabling coordinated regulation of glucose homeostasis through multi-level, bidirectional communication between the PNS and CNS. Within the CNS, the hypothalamic paraventricular nucleus and arcuate nucleus are considered key therapeutic targets for metabolic disorders. Due to their unique structural features, these nuclei can sense and receive nutrient and hormonal signals from peripheral organs such as the liver, adipose tissue, and skeletal muscle, and relay metabolic signals via the ANS.

First, sensory afferent nerves can significantly influence glucose and insulin delivery and exchange efficiency by modulating local tissue blood flow. Studies have shown that gastrointestinal peptidergic sensory neurons release calcitonin gene-related peptide and substance P, inducing vasodilation and thereby enhancing nutrient and hormone exchange. Concurrently, the hypothalamic median eminence dynamically regulates vascular permeability, enabling real-time sensing of circulating metabolites or inflammatory signals, and mediates neuroendocrine output to precisely control systemic glucose homeostasis[25].

Furthermore, recent findings have identified the TRPV1—a molecular sensor expressed on sensory neurons activated by capsaicin—as a potential therapeutic target for improving insulin resistance. TRPV1 activation enhances glucose oxidation and elevates intracellular ATP levels via CAMKK2 and AMPK activation. However, experimental data regarding the effect of capsaicin on intracellular signaling molecules involved in glucose uptake remain limited, and the underlying mechanisms are yet to be fully elucidated[26].

In addition, in the gastrointestinal tract, EECs function as chemoreceptors, releasing peptides such as CCK and GLP-1. These peptides act on vagal afferent fibers, transmitting signals to the NTS in the brainstem, where visceral feedback is integrated with hypothalamic and limbic system inputs to enhance glucose tolerance[27]. Notably, the gut-brain-liver axis operates as a closed-loop feedback system that embodies multi-organ reflex integration: Following nutrient absorption (e.g., glucose and lipids), EECs secrete GLP-1 and CCK, which not only act via endocrine pathways but also activate vagal afferent terminals innervating the gut. Signals are relayed to the NTS and subsequently transmitted via vagal efferents to the liver, where they suppress the expression of key gluconeogenic genes. The resulting reduction in hepatic glucose output lowers blood glucose levels, which in turn attenuates stimulation of intestinal EECs, thereby completing a physiological feedback loop[28].

In summary, sensory afferent pathways represent critical channels of metabolic communication, linking multi-organ function with central regulatory hubs through TRPV1-dependent signaling, chemosensory mechanisms, and vascular control, ultimately ensuring effective glycemic regulation.

TRANSLATIONAL APPROACHES FOR PNS-MEDIATED GLUCOSE MODULATION

Building upon the established roles of peripheral nerves in glucose homeostasis, significant efforts are underway to translate these mechanistic insights into clinically viable therapeutic strategies. This section delineates the major translational approaches, categorized by their underlying technological or biological principles, with a specific focus on their distinct characteristics in modulating glucose metabolism.

Device-based neuromodulation: Instant reversibility

Unlike simple hypoglycemic drugs, device-based neuromodulation technology enables accurate and reversible interventions for glucose regulation. For example, VNS uses a subcutaneously implanted pulse generator and two leads (placed around the anterior vagal trunk near the gastroesophageal junction) to deliver signals inducing intermittent vagal blockage. Follow-up studies have shown that this therapy is effective in improving patients' lipid profiles, glycated hemoglobin (HbA1c) levels, systolic and diastolic blood pressure, as well as their quality of life[29]. Its key advantages include immediate reversibility and adjustable parameters, enabling real-time optimization. Preclinical studies demonstrate VNS reduces hyperglycemia in diabetic rodent models by modulating autonomic outflow[30].

As a non-invasive alternative, taVNS stimulates auricular vagus nerve branches. Systematic reviews confirm its safety profile, with minimal adverse events across clinical applications[31]. SCS targets dorsal spinal structures. Rodent studies demonstrate kilohertz-frequency SCS differentially modulates excitatory and inhibitory neuronal activity in the dorsal horn, suggesting potential for fine-tuning autonomic outflow[32].

Pharmacological neuromodulation: Systemic action

Pharmacological neuromodulation employs chemical agents to modulate peripheral neural activity for glucose regulation, characterized by well-defined stimulatory effects yet limited specificity due to systemic actions. Cholinergic agonists, such as carbachol, enhance GSIS by activating pancreatic M3 receptors, amplifying the first-phase insulin response. In high-fat diet-induced insulin-resistant mice, carbachol (15-50 nmol/kg) normalized glucose tolerance by increasing insulin secretion efficiency (392%-479% potentiation vs controls), though its effects were not islet-specific, as ileal contractility remained unchanged[33]. β-adrenergic blockers like metoprolol reduce hepatic glucose production via β2-receptor antagonism. At high doses (≥ 100 mg/day), metoprolol significantly suppressed terbutaline-induced glucose output [area under the curve (AUC) reduction: 2424-2437 mg/dL·minute] in heart failure patients, whereas carvedilol showed no effect, highlighting receptor-subtype-dependent specificity limitations[34]. TRPV1 agonists (e.g., capsaicin) mitigate hyperglycemia-induced liver injury by upregulating mitochondrial fusion protein OPA1, reducing reactive oxygen species and apoptosis in diabetic mice. However, TRPV1’s broad expression in sensory neurons and hepatocytes results in off-target effects, restricting tissue-specific modulation[35]. These agents demonstrate clear dose-responsive efficacy but face challenges in target selectivity.

Integrated closed-loop systems: Dynamic control

Integrated closed-loop systems enable real-time glucose monitoring and adaptive neuromodulation by combining continuous glucose sensors, control algorithms, and implantable stimulators. These systems dynamically adjust neural stimulation parameters in response to glycemic fluctuations, exemplifying the core feature of real-time glucose sensing and immediate stimulation adjustment[36]. For instance, a rat study demonstrated a self-powered VNS device that modulated gastric peristalsis based on real-time motility measurements, achieving a 38% reduction in body weight through adaptive stimulation[37]. Similarly, in a murine model of autoimmune diabetes, closed-loop pancreatic nerve stimulation (10 Hz, 450 μA) suppressed pro-inflammatory cytokines and autoreactive T-cell proliferation, effectively halting disease progression without adverse effects[38]. Commercial hybrid closed-loop systems, though primarily insulin-delivery focused, validate the translational potential of this architecture: They integrate CGM data with algorithms to autonomously titrate therapy, improving time-in-range (TIR) by 10%-15% in clinical trials[39]. Such systems underscore the feasibility of extending closed-loop principles to neuromodulation, where neural signals could similarly drive precise, glucose-responsive stimulation.

Regenerative and genetic approaches: Long-term effects

Regenerative and genetic strategies offer potential long-term glucose modulation through sustained neurotrophic factor enhancement or precise gene editing, yet face significant risks and ethical challenges. Neurotrophic factors like GDNF demonstrate durable protective effects in diabetic enteric neuropathy. In streptozotocin-induced diabetic mice, GDNF overexpression prevented hyperglycemia-induced neuronal apoptosis and loss of nitrergic inhibitory neurons via PI3K/Akt pathway activation, with benefits persisting for weeks[40]. Similarly, BDNF gene therapy in obese mice promoted thermogenesis and improved insulin sensitivity long-term by upregulating UCP1 in adipose tissue. CRISPR-Cas9 editing further enables permanent genomic alterations. Silencing Fabp4 via CRISPR interference in adipocytes reduced body weight by 20% in high-fat-diet mice, with effects maintained over 6 weeks through suppressed inflammation and improved hepatic metabolism[41]. However, CRISPR systems pose substantial risks, including off-target effects (e.g., unintended DNA cleavage) and delivery challenges due to Cas9’s large size. Ethical concerns surrounding germline editing and long-term safety in humans remain unresolved, limiting clinical translation despite promising preclinical efficacy (Table 3).

Table 3 Typical patents for precise regulation of blood glucose based on the peripheral nervous system.

Number	Name	First inventor	Core mechanism	Function	Ref.
US10722714B2	Methods and systems for glucose regulation	Arnold W Thornton	Modulating peripheral nerves affects insulin/GLP-1 secretion to regulate glycemia	Improve blood sugar control in diabetic patients	Yu et al[42], 2022
US8579891B2	Devices for thermally-induced hepatic neuromodulation	Jonathan Allen Coe	Destroying hepatic sympathetic nerves reduces hepatic glucose production via thermal ablation	Thermal ablation lowers blood sugar levels	Zhang et al[43], 2025
US20200046968A1	Transdermal optogenetic peripheral nerve stimulation	Hugh M Herr	Optogenetic activation of peripheral nerves via percutaneous light delivery from wearables regulates physiological functions	Optogenetic stimulation assists in blood glucose regulation	Kaushik et al[44], 2025
US8538542B2	Nerve stimulation and blocking for treatment of gastrointestinal disorders	Mark B Knudson	Vagus nerve electrical stimulation improves pancreatic exocrine, thereby modulating glucose metabolism	Improves digestion and assists in blood sugar regulation	Zhang et al[45], 2023
US9872985B2	Glucose regulation via electrical stimulation of nerves innervating the liver or pancreas	Robert Butera	Hepatic vagus/visceral nerve stimulation directly regulates hepatic glucose flux	No medication for blood sugar control	Dirr et al[46], 2023
US6885888B2	Electrical stimulation of the sympathetic nerve chain	Ali R Rezai	Sympathetic chain stimulation modulates ganglia, altering metabolic responses	It is widely used for treatment and does not directly target blood sugar	Payne et al[47], 2022

EVIDENCE FROM PRECLINICAL AND CLINICAL STUDIES

Preclinical studies

A range of preclinical studies has shown that peripheral neuromodulation can beneficially modulate glucose metabolism in both small and large animals. VNS approaches have demonstrated positive effects[42-49]. TaVNS in rodent models lowered fasting glucose, improved glucose tolerance, and favorably modulated hormones such as GLP-1 and ghrelin. Other kinds of VNS, including invasive stimulation of the cervical vagus or direct pancreatic branches, have increased insulin secretion, enhanced pancreatic blood flow, and rapidly reduced blood glucose. Ultrasound stimulation has also demonstrated great potential[50-53]. Hepatic peripheral focused ultrasound stimulation (pFUS) has produced consistent metabolic improvements in rodent and swine models of insulin resistance, including reductions in fasting glucose and insulin, improved Homeostasis Model Assessment of Insulin Resistance (HOMA-IR), enhanced glucose tolerance, and increased glucose infusion rates (GIRs) during hyperinsulinemic-euglycemic clamps (HECs)[50]. Optogenetic activation of pancreatic cholinergic vagal efferents has further confirmed the potential of neuromodulation to acutely elevate insulin levels and improve glycemic control[54-56]. Overall, these findings consistently demonstrate that peripheral neuromodulation can improve glycemic control in animal models, providing a strong rationale for clinical translation.

Clinical studies

Clinical evaluation of peripheral neuromodulation for glycemic control remains limited, with only a few trials exploring its feasibility and metabolic effects in humans. Ashe et al[53] reported in a first-in-human feasibility study, hepatic pFUS was applied to patients with type-2 diabetes for 15 minutes daily over three consecutive days. The intervention was well tolerated and resulted in a statistically significant reduction in fasting insulin and HOMA-IR (P = 0.01). Although fasting glucose remained unchanged, it suggests a modest improvement in insulin sensitivity over a short treatment window. In a larger randomized, sham-controlled study[49], Kufaishi et al[49] evaluated cervical transcutaneous VNS (tVNS) in people with diabetes and autonomic neuropathy and found that, while group-level CGM outcomes were neutral, a predefined type-1 diabetes subgroup experienced a clinically meaningful 11.6% reduction in glucose coefficient of variation (CV) after 8 weeks of active stimulation, demonstrating that neuromodulation can reduce glycemic variability in susceptible patients. SCS, primarily developed for painful diabetic neuropathy, has also been associated with secondary metabolic benefits: In a post-study investigation of a randomized trial of 10 kHz SCS[57], significant and sustained pain relief was accompanied by reductions in HbA1c and body weight in patients with type-2 diabetes and elevated baseline values, although these were secondary endpoints. Collectively, these studies indicate that peripheral neuromodulation is technically feasible and generally safe in human populations.

Bench-to-bedside translation issues

Preclinical models have consistently demonstrated substantial metabolic benefits following peripheral neuromodulation interventions. For instance, in rodents subjected to hepatic pFUS, fasting insulin levels significantly decreased and HOMA-IR improved, while in swine, GIR—a HEC measure of insulin sensitivity—was markedly increased (pFUS group showed higher steady-state GIR and elevated GIR AUC vs controls)[50]. In contrast, the first-in-human pFUS feasibility trial in type 2 diabetes patients, applying 15 minutes of hepatic ultrasound for 3 consecutive days—produced only a modest statistically significant reduction in fasting insulin and HOMA-IR (P = 0.01), with no significant change in fasting glucose[53]. A key contributor to this translation gap lies in the choice of outcome metrics. Animal studies utilize high-sensitivity metabolic assessments, such as oral glucose tolerance test, AUC, HOMA-IR, and GIR—that can detect subtle shifts in glucose handling[51]. Clinical studies, however, often depend on composite or long-term outcomes like CGM metrics (CV, TIR) and HbA1c, which may require longer duration to reflect meaningful changes. For example, taVNS in humans did not improve CGM metrics over 8 weeks overall; only in a type 1 diabetes subgroup did CV decline by 11.6%, P = 0.009: A much smaller effect size relative to preclinical outcomes[49,50]. Stimulation parameters and anatomical targeting also differ dramatically. Preclinical protocols typically use precisely targeted, higher-intensity stimuli, activating specific plexuses such as the hepatic portal system for longer duration[51-53]. By contrast, clinical pFUS is limited to short daily sessions and conservative parameters for safety, likely reducing efficacy. Similarly, taVNS in animal models may use extended stimulation protocols at higher intensity[42-45], whereas human subjects underwent limited daily stimulation (4 times/day for 1 week, then 2 times/day for 8 weeks) with self-administered devices[50]. Moreover, model characteristics differ: Animal models often involve young, otherwise healthy subjects with induced or genetic diabetes of short duration, lacking comorbidities. Human participants, conversely, often have established, long-duration type-2 diabetes complicated by autonomic neuropathy, obesity, and cardiovascular disease—factors likely dampening neuromodulatory efficacy. Finally, differences in neuroanatomy, innervation density, and nerve responsiveness between species may hinder translating rodent-targeted interventions to human contexts. Activation thresholds and distribution of sensory afferents in humans may not mirror those in animal models, making direct parameter translation unreliable. In conclusion, although preclinical studies have fully demonstrated that metabolic regulation can be achieved through peripheral neural control, the clinical benefits in human studies remain limited. This disparity mainly stems from differences in outcome indicators, stimulation parameters, subject characteristics, and neurophysiological mechanisms, etc. Therefore, it is of vital importance to conduct an in-depth analysis of the mechanisms and practical obstacles that cause the gap.

Comparative analysis of neuromodulation modalities

Given the variety of peripheral neuromodulation methods explored for metabolic control, a structured comparative framework was crucial to understand their different mechanisms and potential for translation. Each modality engages different neural targets, leading to heterogeneous biological effects and patient suitability. These approaches can be generally divided based on the type of energy delivered and the level of invasiveness into non-invasive electrical stimulation, invasive electrical stimulation, acoustic/mechanical neuromodulation and optogenetic modulation. Such a systematic taxonomy not only facilitates a mechanistic comparison of each technique’s neurophysiological basis, advantages and limitations, but also offered clear concepts for recognizing proper clinical applications in glucose regulation (Tables 4 and 5).

Table 4 Summary of representative neuromodulation and biophysical stimulation interventions for glycemic regulation in experimental and clinical models.

Intervention methods	Experimental subject	Intervention duration	Stimulus parameter	Main results	Ref.
taVNS	ZDF rats with T2D	4 weeks, 30 minutes daily	2 mA, 15 Hz	The blood glucose level in the taVNS group decreased from the second week and remained below the baseline level throughout the observation period	Payne et al[48], 2020
taVNS	Male Balb/C mice with HFD	1 week	1.0 mA, 10 Hz	The elevated fasting blood glucose levels in the HFD mice were ameliorated by taVNS	Kufaishi et al[49], 2025
taVNS	Male albino Wistar rats	2 weeks	6 Hz, 1 ms, 6 V	The fasting glucose level was significantly supressed by VNS therapy	Cotero et al[50], 2022
taVNS	Male ZDF rats with T2D	6 weeks	2 mA, 2 Hz/15 Hz	The fasting blood glucose was decreased (P < 0.05, P < 0.01) and the insulin receptor expression level in the liver, the skeletal muscle and the pancreas was increased (P < 0.05, P < 0.01, P < 0.001)	Wang et al[51], 2025
pVNS	Male Lewis rats with T1D	30 minutes	0-3 mA, 10 Hz	After 30 minutes of stimulation, insulin levels rose significantly (+279 pg/mL)	Chang et al[52], 2023
eVNS	Male Sprague-Dawley rats with T2D	1 hour	15 Hz distal pulses with 26 kHz/4 mA proximal HF block	During 1 hour of eVNS, glycemia decreased in 90% of subjects (-1.25 ± 1.25 mM/hour, P = 0.017)	Ashe et al[53], 2023
aVNS	Male Sprague-Dawley rats	60 minutes	4 mA, 40 kHz	High-frequency stimulation at 40 kHz did not affect blood glucose levels but led to a significant reduction in glucagon levels	Li et al[54], 2021
taVNS	Individuals with T1D or T2D and DAN	8 weeks	60 mA, 25 Hz	No significant changes were observed in CGM metrics between treatment arms, while individuals with T1D and DAN decreased their CV after 8 weeks of tVNS treatment	Yu et al[55], 2021
pFUS	ZDF rats with T2D, HFD or db/db mice, swine	40 days, 3 minutes daily	1.1 MHz, 5 Hz, 125.7 W/cm2	Daily application of pFUS for 3 minutes can maintain the circulating blood glucose of T2D animals to normal levels, increase glucose uptake and glycogen accumulation in peripheral tissues, especially skeletal muscles, and comprehensively improve glucose tolerance and insulin sensitivity	Mansouri et al[56], 2021
LIPUS	C57BL/6J mice with T2D	4 weeks	1 MHz, 1.0 kHz	The mice in the T2D-LIPUS group displayed significantly lower area under the curve of glucose tolerance tests and insulin tolerance tests and fasting serum insulin levels compared to the T2D-sham group	Petersen et al[57], 2025
US	Sprague-Dawley rats with T2D	30 minutes	1 MHz, 15 Hz, 7.21 W/cm2	The blood glucose level decreased within 5 minutes of US and the insulin concentration showed an upward trend	Huang et al[72], 2014
pFUS	Individuals with T2D	3 days	2.28 MHz, 3.7 Hz, 200 µs, 489 mW/cm2	Fasting insulin was lowered, resulting in a reduction of HOMA-IR scores	Nicolai et al[73], 2023
OS	Mice with T1D	Not mentioned	460 nm, 1.4 W/m2	Blue light regulates insulin expression and reduces blood glucose levels in T1D mice	Browning et al[58], 2014
OS	Mice with T1D	40 days, 2 hours daily	730 nm, 10 mW/cm2	Far-red light exposure raised insulin levels, reduced blood glucose levels and glycated hemoglobin concentrations	EAS FHSC[59], 2024
OS	Mice with T1D	15 minutes	2-5 mW/cm2	After 15 minutes of white light irradiation, insulin levels increased and glucose concentration decreased	Lincoff et al[60], 2023

taVNS: Transcutaneous auricular vagus nerve stimulation; ZDF: Zucker diabetic fatty rat; T2D: Type 2 diabetes; HFD: High fat diet; VNS: Vagus nerve stimulation; eVNS: Electrical vagus nerve stimulation; pVNS: Peripheral vagus nerve stimulation; aVNS: Auricular vagus nerve stimulation; T1D: Type 1 diabetes; DAN: Diabetic autonomic neuropathy; CGM: Continuous glucose monitoring; CV: Coefficient of variation; tVNS: Transcutaneous vagus nerve stimulation; pFUS: Peripheral focused ultrasound stimulation; LIPUS: Low-intensity pulsed ultrasound stimulation; US: Ultrasound stimulation; HOMA-IR: Homeostasis Model Assessment of Insulin Resistance; OS: Optical stimulation; EAS FHSC: European Atherosclerosis Society Familial Hypercholesterolaemia Studies Collaboration.

Table 5 Comparison of peripheral neuromodulation techniques for glucose regulation.

Category	Techniques	Primary targets/ depth	Mechanistic modality	Advantages& limitations	Potential population/stage
Non-invasive electrical stimulation[48-52,54,55,62]	taVNS; pVNS; aVNS	Auricular or cervical vagal branches (superficial/subcutaneous)	Transcutaneous or percutaneous electrical activation of vagal afferent fibers	Safe, inexpensive, easily repeatable; limited depth and targeting precision	Mild hyperglycemia, impaired glucose tolerance, metabolic syndrome
Invasive electrical stimulation[53,61,63]	eVNS; SCS	Cervical vagus trunk or thoracic dorsal column (deep)	Implantable electrodes deliver direct neural modulation via electrical pulses	High precision and sustained effects; invasive surgery, infection risk	Advanced or refractory diabetes with neuropathic complications
Acoustic/mechanical neuromodulation[56-59]	pFUS; LIPUS; US	Hepatic hilum, portal vein region, or peripheral nerve plexus (deep tissue)	Mechanical stress on glucose-sensing afferents by focused ultrasound	Non-invasive, spatially selective, real-time image guidance; limited mechanistic resolution	Obese or insulin-resistant individuals, early T2DM
Optogenetic neuromodulation[58-60]	OS	Pancreatic parasympathetic fibers, hepatic vagal branches (experimental)	Optical activation of genetically targeted neurons controlling endocrine output	High cellular specificity; requires genetic modification, currently preclinical	Conceptual or experimental models of neural and metabolic coupling

taVNS: Transcutaneous auricular vagus nerve stimulation; pVNS: Peripheral vagus nerve stimulation; aVNS: Auricular vagus nerve stimulation; eVNS: Electrical vagus nerve stimulation; SCS: Spinal Cord Stimulation; pFUS: Peripheral focused ultrasound stimulation; LIPUS: Low-intensity pulsed ultrasound stimulation; US: Ultrasound stimulation; OS: Optical stimulation; T2DM: Type 2 diabetes mellitus.

TRANSLATIONAL CHALLENGES

The translation of PNS based interventions for glucose regulation from preclinical research to clinical practice faces several substantial challenges. These barriers span scientific, technological, clinical, and ethical domains, each requiring careful consideration to ensure safe and effective application.

Neuroanatomical and functional variability

Significant interspecies differences in neuroanatomy and neural function complicate the extrapolation of animal findings to humans. For instance, vagal innervation patterns to metabolic organs such as the liver and pancreas differ notably between rodents and humans[58]. Rodent models often exhibit more diffuse and accessible neural pathways, whereas human anatomy involves deeper and more variable nerve distributions, which can affect the accuracy and efficacy of neuromodulatory interventions.

Furthermore, considerable interindividual variability in autonomic nerve architecture exists, which is often exacerbated in diabetic patients with underlying neuropathic changes[59]. These discrepancies contribute to inconsistent responses to neuromodulation modalities such as taVNS when moving from standardized animal models to heterogeneous human populations. Personalized mapping of nerve anatomy and function may be necessary to optimize stimulation sites and parameters.

Impact of diabetic neuropathy

The presence of diabetic peripheral and autonomic neuropathy may severely compromise the efficacy of PNS-targeted therapies. Structural and functional impairments, including axonal degeneration and demyelination, can diminish electrical conduction efficiency and alter neurotransmitter release[60]. This neurodegeneration not only reduces the responsiveness to external stimulation but also disrupts endogenous regulatory mechanisms.

Additionally, neuropathy-induced changes in receptor expression and signaling pathways may further distort metabolic responses to neural stimulation[61]. For example, altered adrenergic receptor sensitivity in diabetic patients can lead to aberrant sympathetic outflow, undermining the glucose-lowering effects of sympathetic modulation. Therefore, patient stratification based on detailed neurological assessments is essential to identify suitable candidates for neuromodulatory treatments.

Device-related complications and long-term adaptations

Although implantable neuromodulation devices (e.g., VNS and SCS) are established in neurology and pain management, their long-term safety and efficacy in metabolic disorders remain uncertain. Risks include surgical complications, infection, fibrotic encapsulation, electrode migration, and hardware malfunction. These issues are particularly relevant in diabetic patients, who often exhibit impaired wound healing and increased infection risk.

Moreover, chronic electrical stimulation may induce neural adaptation or tolerance, leading to attenuated therapeutic effects over time[62]. This phenomenon, often referred to as “stimulation fatigue”, necessitates the development of dynamic stimulation paradigms that can adapt to changing neural plasticity. Improving device biocompatibility, refining electrode design, and optimizing stimulation paradigms are active areas of investigation to enhance long-term efficacy.

Lack of stimulation specificity and off-target effects

Achieving targeted neuromodulation without activating unintended pathways remains a significant hurdle. Non-invasive approaches such as taVNS often lack spatial precision, potentially stimulating cardiac or respiratory branches of the vagus nerve and causing adverse effects[63]. These off-target effects not only limit therapeutic windows but also raise safety concerns in vulnerable populations.

Similarly, pharmacological agents like beta-blockers may modulate sympathetic activity but carry cardiovascular risks that limit their use in diabetic populations. Enhancing the specificity of both device-based and pharmacological interventions through advanced targeting technologies or combination therapies is critical for future success.

Regulatory and ethical hurdles

The regulatory pathway for neuromodulation therapies is still evolving. Agencies such as the Food and Drug Administration require robust long-term efficacy and safety data, particularly for implantable systems. The lack of standardized endpoints for metabolic neuromodulation further complicates trial design and regulatory approval. Demonstrating not only glycemic improvement but also reduced morbidity and mortality will be essential for widespread adoption.

Ethical considerations including informed consent in vulnerable populations (e.g., elderly patients with cognitive impairment), data privacy in closed-loop systems, and equitable access to high-cost therapies also warrant serious attention. The potential for neuromodulation to exacerbate health disparities must be addressed through inclusive trial designs and accessible technology platforms.

The translation of PNS-mediated glucose modulation strategies is fraught with multifaceted challenges, ranging from biological variability and neuropathic complications to technical and ethical limitations. Addressing these issues will require interdisciplinary collaboration to develop personalized stimulation protocols, improve device technology, and establish clear regulatory frameworks. Overcoming these barriers is essential to realize the full clinical potential of neuromodulation in diabetes care.

FUTURE PERSPECTIVES

As reviewed in previous sections, PNS is emerging as a promising model for treating diseases such as obesity, diabetes, and other diseases caused by glucose homeostasis disorders, yet its clinical translation requires overcoming current anatomical, technical, and accessibility barriers. Future development must shift towards multi-target, adaptive and personalized solutions. Here we outline the following key directions.

Multitarget synergistic stimulation

Current PNS approaches predominantly target single pathways (e.g., vagal hepatic branches or sciatic nerves), failing to harness the full complexity of neural-glucose crosstalk. Future advances should enable synergistic stimulation for multiple targets. For instance, joint regulation of vagal pancreatic branches (enhancing insulin secretion via M3 receptors) and sympathetic BAT-activating fibers (promoting glucose uptake via UCP1-mediated thermogenesis) could amplify glycemic control beyond monotherapies[64]. Developing multi-electrode arrays capable of selectively activating efferent/afferent fibers across vagal, sympathetic, and somatic nerves will be critical to replicate these physiological networks.

Personalized neuromodulation based on neural phenotypes

Anatomical variability (e.g., 45% of auricular regions innervated by non-vagal nerves) and physiological heterogeneity (e.g., autonomic neuropathy in long-term diabetes) necessitate patient-specific neuromodulation[65]. Future systems must integrate neural phenotyping to tailor stimulation parameters. For example, tVNS efficacy could be optimized by identifying patient-specific targets in the auricle using functional magnetic resonance imaging (MRI)-guided targeting[66], while avoiding regions prone to off-target effects (e.g., laryngeal activation). Genetic profiling may further stratify patients: Those with NPY-receptor polymorphisms might benefit from adjusted sympathetic stimulation protocols to counteract NPY-mediated hepatic gluconeogenesis. This precision approach could resolve the U-shaped efficacy curve dilemma, where standardized parameters yield suboptimal outcomes[67].

For the clinical translation to be possible, it was essential to create a reproducible format for stimulation standardization. High resolution ultrasound along with functional MRI imaging and electrophysiological profiling could potentially define anatomical landmarks and functional thresholds of neural activation which would enable a reference format to guide parameter selection among patients; meanwhile, physiological biomarkers like heart rate variability, circulating GLP-1 and skin conductance were able to offer continuous and objective measures of autonomic and metabolic states which allowed for real time adjustment of stimulation intensity and frequency. Multimodal datasets, when combined with algorithms powered by AI, had the potential to dynamically adjust stimulation parameters according to an individual's physiological state while keeping within the limits for safety. This combined method of standardized personalization did not just account for differences between individuals but also created a reproducible path towards scalable neuromodulation protocols.

AI-driven closed-loop optimization

Real-time adaptation to dynamic metabolic demands is unattainable with open-loop devices. Closed-loop systems integrating CGM, neural activity sensors (e.g., vagal afferent traffic), and AI algorithms will enable autonomous measurement and control. Machine learning models trained on multimodal data (glucose trends, heart rate variability, inflammatory markers) could predict optimal stimulation parameters. For instance, reinforcement learning algorithms might adjust VNS frequency (5-30 Hz) based on Metabolic biomarkers and neurophysiological recordings (like real-time GLP-1 secretion and hepatic glucose output), mimicking physiological feedback. Crucially, such systems must incorporate safety constraints (e.g., capping stimulation intensity to prevent laryngeal spasm) and address latency challenges via edge computing. Early prototypes combining CGM with tVNS reduced HbA1c variability by 60% in pilot trials, underscoring the feasibility of adaptive control[68].

Global accessibility design

To ensure equitable adoption, PNS technologies must transcend cost and infrastructure barriers. Modular, low-power devices using smartphone-based controllers could democratize access. For example, simplified tVNS units with reusable electrodes and Bluetooth connectivity could be deployed in resource-limited settings, leveraging telemedicine for parameter calibration[69]. Additionally, non-invasive alternatives like focused ultrasound (pFUS) or transdermal optogenetics—requiring minimal training—could bypass stimulation of implantable devices and reducing surgical expertise gaps[70].

Regenerative and remodeling neuroengineering

Long-term device efficacy is compromised by fibrotic encapsulation, electrode corrosion, and nerve damage. Biohybrid interfaces merging conductive biomaterials with regenerative therapies offer solutions. Electrodes coated with anti-inflammatory hydrogels (e.g., dexamethasone-eluting polymers) could mitigate fibrosis, while optogenetic tools enable cell-type-specific stimulation without metallic implants. Concurrently, neurotrophic support—via localized delivery of GDNF or NGF—may promote nerve regeneration around implants, as demonstrated in diabetic neuropathy models where GLP-1 agonists enhanced axonal integrity[71]. Future devices should leverage biodegradable electronics that dissolve after restoring endogenous glucose regulation, eliminating surgical extraction risks.

Design principles for future clinical trials

Earlier attempts at using peripheral neuromodulation for regulating glucose were restricted due to tiny sample quantities, brief periods, and diverse participant traits which held back the dependability and applicability of the results. To get past these issues, future randomized controlled experiments ought to utilize multicenter, double-blind, sham controlled setups with lengthened follow-up duration (at least 6 months) and sufficiently large sample amounts to spot both immediate and long-lasting metabolic impacts. These studies should include thorough main endpoints like decreasing HbA1c, TIR and CV together with secondary consequences such as HOMA-IR, GIR, inflammatory signs, autonomic function measures and outcomes reported by patients. Stratification in accordance with the type of diabetes, the length of the disease, the severity of neuropathy and the baseline body mass index would guarantee the identification of subgroups that respond, making it possible for precision medicine methods to be employed and through incorporating strict methodological controls, long-term monitoring and evaluations of biomarkers related to mechanisms. These randomized controlled trials would offer strong evidence regarding the effectiveness and safety of neuromodulation founded on the PNS, which in the end would promote its application in clinical practice.

CONCLUSION

PNS-mediated regulation of glucose homeostasis is an emerging and promising therapeutic paradigm. Evidence from preclinical and early clinical studies shows that autonomic and sensory pathways modulate hepatic glucose output, islet hormone secretion, and peripheral glucose uptake via organ-specific neuroendocrine circuits. Neuromodulation technologies such as VNS and focused ultrasound provide feasible, non-pharmacological intervention routes, yet translation to robust and sustained clinical benefit remains limited by stimulation specificity, neuropathy, and individual variability. Future strategies integrating mechanistic targeting, personalized protocols, and closed-loop control may overcome current barriers, enabling PNS-based interventions to become safe, effective, and scalable tools for managing diabetes and related metabolic disorders.

ACKNOWLEDGEMENTS

We thank the reviewers and editorial team for their insightful comments and the staff of Nanjing University of Chinese Medicine for technical support.