The main focus of this review is to summarize the peripheral nerves involved in acupoint stimulation based on studies related to acupuncture interventions for VCI. The aim is to explore the potential peripheral nerve pathways that may be involved in acupuncture treatment for VCI.

We searched English-language databases, including PubMed, Web of Science, MEDLINE, and Embase. The search terms included subject headings and related terms such as “acupuncture,” “electroacupuncture,” and keywords like “vascular cognitive impairment.” The two sets of keywords were connected using the logical operator “AND.” The search covered the period from the inception of each database up to March 27, 2024, and included all publicly available journal publications. Supplementary Table 1 provides detailed information on the PubMed search strategy.

Articles involving acupuncture and related therapies were eligible, provided that the intervention subjects met the established criteria for VCI. Eligible study types included clinical randomized controlled trials, animal studies, systematic reviews, meta-analyses, and case reports. Studies that did not pertain to acupuncture interventions specifically aimed at VCI were systematically excluded. The criteria ensured that only those studies with a clear focus on the intervention and target population were retained.

The screening and selection process commenced with the use of EndNote X9 to manage the retrieved literature. Initially, the built-in duplication tool was utilized to eliminate duplicate records, followed by manual checks to remove any remaining duplicates not detected by the software. Subsequently, a thorough review of titles and abstracts was conducted to exclude studies unrelated to acupuncture interventions for VCI. The remaining studies were then subjected to a blinded evaluation by two independent reviewers. Any discrepancies or disagreements encountered during the review process were resolved through consultation with a senior researcher, whose expertise ensured that the final selection was both accurate and consistent.

We primarily extracted acupoint information from these studies and compiled it into a table. Additionally, we reclassified the acupoints mentioned in these articles based on the nerves within the anatomical structures underlying each acupoint. The neuroanatomical correlations of the acupuncture points will be referenced from the Chinese higher education “14th Five-Year Plan” textbook Acupoint Anatomy (Shao, 2023) and the Interactive Medical Acupuncture Anatomy (Teton NewMedia, USA, 2016) (Robinson, 2016). Finally, we categorized and summarized the research progress on acupuncture treatment for VCI based on classifying peripheral nerves and explored the underlying mechanisms involved in these pathways.

Through a search of English-language databases, including PubMed, Embase, MEDLINE, and Web of Science, a total of 233 articles were identified (PubMed = 31, Embase = 36, MEDLINE = 58, Web of Science = 108). After removing duplicates, 124 articles remained for the initial screening phase. Of these, 40 articles were excluded based on screening titles and abstracts, leaving 79 articles for full-text evaluation. During the full-text review, two conference abstracts were excluded, resulting in 79 articles that were ultimately included in this review (Figure 1).

The reviewed articles cover a timeline from 2005 to 2024, providing a comprehensive overview of research on acupuncture and its effects on VCI and related conditions. The body of research was composed of 32 experimental animal studies, 14 clinical studies (including 12 randomized controlled trials and 2 non-randomized trials), and 33 literature-based investigations. The animal studies were conducted to explore the mechanisms underlying acupuncture’s effects on cognitive function, often employing rat models of multi-infarct dementia or VD to mimic clinical conditions.

On the clinical side, the randomized controlled trials focused on evaluating the efficacy of acupuncture in improving cognitive function in patients with VCI or other forms of cognitive decline. Furthermore, a substantial portion of the literature consists of comprehensive reviews and syntheses of existing data. Among the 33 literature investigations, 31 are systematic reviews, with 13 of these incorporating meta-analyses to quantitatively assess the efficacy of acupuncture. In addition to these rigorous analyses, there is one narrative review that offers a succinct overview of the field, as well as one Delphi consensus study that gathers expert opinions to guide future research directions. Specific information of all literatures can be found in Supplementary Table 2.

Based on the statistical analysis of acupoints that appeared across all the included studies, we have summarized the acupoints incorporated across all studies under consideration. All fourteen meridians are represented, including extra nerve acupoints and scalp acupoints. As illustrated in Figure 2, The most commonly involved acupoints are often located in the Taiyang Bladder Meridian, followed by the Foot Shaoyang Gallbladder Meridian and the Governor Vessel. We have categorized the acupoints according to the involved nerves, dividing them into groups that encompass acupoints related to upper limb neural networks and their branches, as well as those pertaining to the lower limbs and sacral region innervated by the lumbosacral plexus and its branches. Other categories include acupoints related to the lower limbs and lumbar-dorsal areas innervated by the lumbar plexus and its branches, as well as back-shu points associated with the thoracic nerve plexus and its branches, neck and occipital acupoints connected to the cervical plexus and its branches, facial and cranial acupoints linked to cranial nerves and their branches, and acupoints distributed along the autonomic nervous system. Notably, cranial nerves, lumbosacral plexus, and cervical plexus acupoints are the most prevalent (Figure 3), with specific acupoints and their corresponding neural affiliations detailed in the Table 1.

Our examination of acupoints, categorized by meridians and associated neural networks, reveals their predominant distribution around the peripheral nerves of the limbs, cranial nerves, spinal nerves, and the autonomic nervous system. In 1971, Jiao creatively established sixteen stimulation areas on the scalp. Some studies suggest that scalp acupuncture in specific zones can treat injuries to the corresponding brain regions and even induce molecular changes (Sun et al., 2020). Scalp acupoints predominantly project onto cranial and cervical nerves (Dorsher and Chiang, 2018). Additionally, a subset of the summarized acupoints involves the autonomic nervous system. It is established that acupuncture significantly influences the central autonomic regulatory center, a controller of the autonomic nervous system, as demonstrated by a wealth of animal studies (Li et al., 2022; Ohsawa et al., 1995; Uchida et al., 2010; Sato et al., 1993). Acupuncture’s impact on the autonomic nervous system follows a segmental distribution pattern, with the ability to modulate sympathetic/parasympathetic output. Acupuncture engages in diverse neural regulations, and the following sections will categorize them by different neural systems, exploring the pathways and mechanisms through which acupuncture exerts its effects on VCI (Figure 4).

Peripheral nerves are richly distributed in the body’s meridians and acupoints, and the involvement of the peripheral nervous system is essential for the efficacy of acupuncture. Acupuncture can improve cognitive function by stimulating the brachial plexus nerve, typically through the synergistic action of multiple nerve branches. Acupoints commonly used in VCI acupuncture interventions include LI4 (Hegu), PC5 (Jianshi), PC6 (Neiguan), and Waiguan, traversed by the radial nerve, ulnar nerve, median nerve, and musculocutaneous nerve, respectively. The brachial plexus nerve constitutes a complex anatomical nerve network, including branches such as the musculocutaneous nerve, radial nerve, ulnar nerve, median nerve, axillary nerve, thoracodorsal nerve, and long thoracic nerve, primarily supplying the upper limbs (Kattan and Borschel, 2011; Orebaugh and Williams, 2009). An interaction exists between the brachial plexus nerve and cognitive function. The hippocampus, an essential limbic system component, is believed to play a significant role in brain memory and emotional formation processes. Preclinical evidence suggests that chronic peripheral nerve injuries reduce neurogenesis in the adult hippocampus’s granule cell layer of the dentate gyrus (DG) (Vania et al., 2016; Xia et al., 2020). Clinical studies have found activation of the medial-hypothalamic pathway following brachial plexus nerve injury (BPI), a critical structure for pain processing integration (Wang et al., 2019). Changes occur in other brain regions related to cognitive functions post-brachial plexus nerve injury (p-BPI), with decreased functional connectivity within the primary motor cortex (M1) (Fraiman et al., 2016). Cognitive functions participate in the sensory (Mahmoudaliloo et al., 2011) and functional (De Azevedo et al., 2018) recovery processes of the hand after upper limb peripheral nerve injury, where patients with BPI exhibit dynamic changes in the resting-state brain networks (RSNs) during treatment. Patients with improved muscle strength show significantly increased connectivity in the sensorimotor network (SMN) and salience network (SN) post-treatment. In contrast, all patients exhibit a trend of decreased connectivity in the default mode network (DMN) postoperatively, indicating the role of brain plasticity and compensatory mechanisms in the recovery process of peripheral nerve injury (Bhat et al., 2017). Research analyzing the conduction function of peripheral nerves in elderly individuals with different cognitive levels has found that the conduction velocity of the median motor nerve and the peroneal motor nerve differs significantly (Qian et al., 2022), suggesting damage to fast-conducting fibers related to movement in cognitive impairment patients, with a decline in motor conduction ability. Additionally, the motor conduction velocity of the median and peroneal nerve is positively correlated with the ideographic part of the Montreal Cognitive Assessment (MoCA) and verbal fluency task (VFT) (Amira et al., 2024). Magnetic stimulation of PC6 (Neiguan) and PC (Daling) in healthy subjects shows increased functional connectivity of brain areas associated with higher cognitive functions such as emotion, memory, and language (Dai et al., 2019), with activation of brain regions closely related to cognitive and emotional processes, such as the frontal lobe and temporal lobe (Haili et al., 2022), and significant improvement in information transmission efficiency in the brain cortex during stimulation. Acupuncture at LI4 (Hegu) has been shown to modulate frontal and temporal lobe activity in patients with AD and MCI, while enhancing the connectivity between the hippocampus and the motor cortex (Zheng et al., 2018; Wang et al., 2012). Functional magnetic resonance imaging (fMRI) scans of 24 healthy individuals show that cerebellar activation is typically only observed during acupuncture at the Waiguan acupoint (Wik et al., 2016). Traditionally associated with motor coordination and balance, the cerebellum plays a crucial role in various aspects of higher functions, with emerging research revealing its broader contributions to cognition, emotion, and reward processes (Buckner, 2013; Schmahmann, 2019; Wagner and Luo, 2020). Electrical stimulation of the ulnar nerve at an α frequency (10 Hz) in Brain-Computer Interface (BCI) subjects increases the classification accuracy of left and right-hand motor imagery from 66.41 to 81.57%, with effects lasting for at least two days (Zhang et al., 2020). Stimulation of the ulnar nerve was considered that can improve motor imagery classification accuracy, enhancing the neural function of cognitive impairment patients.

The lumbosacral plexus, originating from L1-S3 spinal nerve roots, controls lower limb sensation and motor function (Di Benedetto et al., 2005). Anatomically, it consists of the “upper” lumbar plexus and the “lower” sacral plexus (Laughlin and Dyck, 2013). The lumbar plexus is located within the psoas major muscle and comprises the anterior branches of the T12 to L4 nerve roots. The nerve branches of the lumbar plexus mainly include the iliohypogastric nerve, ilioinguinal nerve, genitofemoral nerve, lateral femoral cutaneous nerve, obturator nerve, and femoral nerve. The sacral plexus comprises the lumbosacral trunk (L4, L5), all sacral nerves, and the anterior branches of the coccygeal nerve, which originate from the ventral edges of the L4 to S4 nerve roots. The nerve branches of the sacral plexus mainly include the lumbosacral trunk, superior gluteal nerve, inferior gluteal nerve, sciatic trunk/nerve, and pudendal nerve. More research is needed on the relationship between sciatic nerve injury and cognitive impairment among the sacral plexus nerves. A cross-sectional study from China found that as a branch of the sciatic nerve, the peroneal nerve exhibits decreased nerve conduction velocity in patients with MCI and AD, and the conduction velocity of the peroneal nerve is significantly correlated with the diagnosis of AD (Qian et al., 2022; Qian and Xiao, 2020). In addition, sciatic nerve injury can lead to neuropathic pain (Jones et al., 2018), and patients with neuropathic pain often report adverse emotional and cognitive consequences (Mendes et al., 2008; Begoña et al., 2018), such as anxiety, depression, sleep disorders (Radat et al., 2013; Attal et al., 2011), attention deficits, memory difficulties, and decreased decision-making ability (María and Rosa, 2005; Mccracken and Iverson, 2001).

Currently, cognitive function is clinically improved by stimulating the nerve branches of the lumbosacral plexus, which is still in the exploratory stage (Qu et al., 2016; Yamada et al., 2004). However, acupuncture at acupoints where the lumbosacral plexus nerves are distributed can improve cognitive function, usually through the synergistic action of multiple nerve branches. The sciatic nerve is the most commonly involved pathway. Acupoints such as SP6 (Sanyinjiao), KID3 (Taixi), KI1 (Yongquan), and GB30 (Huantiao) are mainly innervated by the branches of the sciatic nerve. Acupoints such as ST36 (Zusanli), LR3 (Taichong), and GB34 (Yanglingquan) are associated with the branches of the sciatic nerve. Evidence suggests that long-term acupoint massage (6 months) can improve cognitive function and activities of daily living in patients with mild to moderate dementia (Liang Ooi et al., 2023), with SP6 being a commonly used acupoint for intervention.

The cervical plexus comprises the anterior branches of the first to fourth cervical nerves, situated deep to the upper part of the sternocleidomastoid muscle, anterior to the origins of the scalene muscles and levator scapulae muscle (Usui et al., 2010). The primary cutaneous branches of the cervical plexus are (I) the lesser occipital nerve, originating from the C2 spinal nerve, distributed over the posterior aspect of the scalp and ear, providing skin sensation; (II) the transverse nerve of the neck. Composed of branches from C2 and C3, distributed over the anterior neck skin, providing skin sensation; (III) the supraclavicular nerve, formed by the combination of branches from C3 and C4, primarily distributed over the lower lateral neck, upper chest wall, and shoulder skin; (IV) the greater occipital nerve, composed of branches from C2 and C3, distributed over the scalp and adjacent skin. All of these nerves are cutaneous nerves. In addition, the cervical plexus gives off some muscular branches to supply the muscles of the neck region and diaphragm. The phrenic nerve, a motor branch of the cervical plexus, originates from the anterior branches of C3-C5 and provides motor innervation to the diaphragm (Waxenbaum et al., 2024; Kim et al., 2018; Paraskevas et al., 2014). Skin electrodes target the greater occipital nerve, establishing a communication gateway from the periphery to the brain via specific afferent fibers that project to the brainstem and synapse with the nucleus tractus solitarius (NTS) (Luckey et al., 2023; Adair et al., 2020), further activating the LC noradrenergic system of the brainstem, projecting to both the primary cortex and subcortical areas (Berridge and Waterhouse, 2003).

Branches of the cervical nerve, greater occipital nerve, and lesser occipital nerve are distributed under the acupoint of GB20 (Fengchi). Clinical studies have shown that massage of GB20 (Fengchi) and similar acupoints for 6 months can improve cognitive function and activities of daily living in patients with mild to moderate dementia (Liang Ooi et al., 2023). There is also a tiny occipital nerve distributed under the GB12 (Wangu) acupoint. Stimulation of this acupoint significantly improved learning ability in rats compared to the VD model group, with decreased mRNA expression levels of TNF-α, IL-6, and IL-1β (Fang and Sui, 2016). Animal experiments with laser acupuncture at DU20 (Baihui) have shown a significant increase in neuronal density in the CA1 and CA3 regions, increased activity of glutathione peroxidase (GSH-Px) and superoxide dismutase (SOD), decreased IL-6 and β-amyloid density ratio, exerting antioxidant and anti-inflammatory effects, thereby alleviating cognitive impairment in focal ischemic rats (Jittiwat, 2019). EA at DU20 (Baihui) and DU14 (Dazhui) enhances learning and memory abilities in rats, downregulating levels of inflammatory factors such as IL-1β, JAK2, and TNF-α (Han et al., 2017), alleviating hippocampal synaptic plasticity damage (Zheng et al., 2016), upregulating mRNA expression of BDNF and vascular endothelial growth factor (VEGF) (Kim et al., 2014), increasing regional cerebral blood flow (rCBF) (Han et al., 2017), promoting the regeneration of oligodendrocytes (Ahn et al., 2016), and increasing the number of proliferating cells and differentiated cells in the hippocampus and subventricular zone (Kim et al., 2014). Neuroimaging evidence indicates that EA at DU20 (Baihui) induces an increase in regional homogeneity (ReHo) in multiple areas, including the orbitofrontal cortex (OFC), midcingulate cortex (MCC), precentral cortex, and precuneus. It decreases ReHo values in the anterior cingulate cortex (ACC), supplementary motor area, thalamus, and cerebellum and decreases ReHo values in the precentral gyrus (preCUN) (Yu et al., 2019). Acupuncture at DU20 (Baihui), GB20 (Fengchi), and the Chorea-Tremor Controlled Zone reduced PD tremor by modulating the cerebello-thalamo-cortical circuit (decreased DC/ReHo and increased ALFF) and altering cognitive networks (DMN, visual areas, insula, PFC) (Li et al., 2018).

Thoracic nerves originate from the thoracic spinal cord. After exiting through the intervertebral foramen, they divide into anterior, posterior, meningeal, and communicating branches. Except for the first and parts of the twelfth pair, the anterior branches do not form plexuses and maintain segmental characteristics. The first to eleventh pairs are intercostal nerves, while the twelfth pair forms the subcostal nerves (Cramer, 2014).

Acupuncture or moxibustion at single or multiple points corresponding to the distribution of thoracic nerves can improve cognitive function. The location of the CV4 (Guanyuan) acupoint corresponds to the peripheral nerves of the T11-T12 segments. Studies have shown that moxibustion at CV4 (Guanyuan) and moxa smoke exposure both ameliorated cognitive deficits in APP/PS1 mice by normalizing tricarboxylic acid cycle flux and unsaturated fatty acid metabolism (Ha et al., 2019). Before the decline in cognitive abilities, there is a significant decrease in cerebral glucose metabolism (Shoshan-Barmatz et al., 2018). Neuronal functional activity requires a lot of energy, which is mainly provided by mitochondria. Mitochondrial defects can lead to synaptic dysfunction, neuroinflammation, and neuronal death (Arnold and Finley, 2023). Peripheral nerves from the T7-T8 segments innervate CV12 (Zhongwan). Using resting-state fMRI, researchers observed brain function changes during EA at CV4 (Guanyuan) and CV12 (Zhongwan) in 21 healthy volunteers. Stimulation significantly enhanced local connectivity in the ventromedial prefrontal cortex network, including the medial orbitofrontal and ventral anterior cingulate cortices, compared to the resting state (Fang et al., 2012).

The trigeminal nerve, the fifth cranial nerve, is located on the ventrolateral surface of the brainstem at the midpoint level. It comprises a thicker sensory root and a thinner motor root. Originating from the petrous part of the temporal bone, it terminates in the principal and spinal nuclei within the brainstem. The trigeminal nerve has three main branches—ophthalmic, maxillary, and mandibular—extensively distributed across the eye, forehead, nose, upper jaw, oral cavity, and cheek, and it innervates facial sensation and transmits sensory information from the dura mater (Kumada et al., 1977). It connects to the vasomotor center of the brainstem, including the lateral tegmental area, and is easily stimulated percutaneously (Goadsby et al., 1996; DeGiorgio et al., 2011).

Acupuncture at certain head and facial acupoints often coincide with the trigeminal nerve distribution. For example, ST7 (Xiaguan) is located at the branches of the facial and auriculotemporal nerves, with the mandibular nerve, including the inferior alveolar and lingual nerves. Other points like GB14 (Yangbai) and BL2 (Zanzhu) are distributed with the supraorbital and frontal nerves. According to meridian theory, the trigeminal nerve’s branches partial consistency with the routes of the three yang meridians of the hand and foot. The therapeutic effects of acupuncture at head acupoints seem to be associated with more inputs from the trigeminal nerve and the unique roles of the TG and trigeminal spinal nucleus (STN) in regulating meningeal function (Moskowitz et al., 1998). It is well-known that endogenous substances change after acupuncture. As a neuroregulatory therapy, trigeminal nerve stimulation is believed to treat VCI by increasing dopamine release in the hippocampus through the TG - corticotropin-releasing hormone (CRH) - dopamine transporter (DAT) - hippocampus (HPC) pathway, thereby enhancing hippocampal-dependent memory (Xu et al., 2023). Electrical acupuncture has been repeatedly shown to have effects on neuroprotection (Liu et al., 2017; Shi et al., 2018), synaptic plasticity (Gerrow and Triller, 2010), and regulation of neural cross-talk (Tu et al., 2019; Chang et al., 2018).

The vagus nerve (VN) is the tenth cranial nerve, with extensive distribution throughout the body, making it the longest and most widely distributed nerve among the cranial nerves. It originates from the medulla oblongata of the brainstem, passes through the jugular foramen into the carotid sheath, travels along the posterior groove between the carotid artery or internal carotid artery and internal jugular vein to reach the root of the neck, and then extends to the thorax and abdomen (Foley and DuBois, 1937). It comprises approximately 20% efferent fibers and 80% afferent fibers. Efferent or motor visceral pathways innervate organs below the neck, including the heart, lungs, and gastrointestinal tract, with the brain receiving information from the afferent projections of the vagus nerve. Afferent fibers project to the NTS and LC within the brainstem (Nomura and Mizuno, 1984), then directly or indirectly project to many brain areas, such as the midbrain, hypothalamus, amygdala, hippocampus, and frontal lobe (Lange et al., 2011; Carreno and Frazer, 2016). The auricular branch of the VN is the only afferent branch located on the body surface, innervating the skin around the external auditory meatus, inner ear pinna, and earlobe (Butt et al., 2020; Kiyokawa et al., 2014; Peuker and Filler, 2002). Moreover, the earlobe is entirely innervated by the auricular branch of the vagus nerve (Peuker and Filler, 2002). It contains approximately 80% of afferent fibers that transmit sensory information to the central nervous system, while the remaining fibers transmit motor information. A recent systematic review suggests (Robinson et al., 2012) that the auricular branch of the vagus nerve projects to the NTS, which further connects with other brain regions such as LC, lateral hypothalamus, amygdala, anterior cingulate cortex, insular cortex, and ventral tegmental area.

The EX-HN13 (YiMing) acupoint and auricular acupoint are commonly used acupoints in clinical practice for treating VCI. Stimulating these acupoints may activate pathways related to the vagus nerve. The EX-HN13 (YiMing) acupoint is located 1 cun posterior to TB-17 (YiFeng) at the inferior border of the mastoid process. The YiMing acupoint’s deep needling can stimulate the vagus nerve’s trunk. Some ear acupoints, such as heart (concha, CO15) and kidney (CO10), are widely distributed with the auricular vagus nerve. It is well-known that electrical stimulation of the vagus nerve and transcutaneous electrical stimulation of the auricular vagus nerve (taVNS) are two therapies that are quite similar in principle and mechanism of action. taVNS likely originated in China, with traditional Chinese medicine theory considering the relationship between “ear acupoints” and “disease” based on meridian theory and holistic concepts (Pj et al., 2015). In 1990, the World Health Organization recognized ear acupuncture as a microsystem of acupuncture that can positively affect systemic function regulation (World Health Organization, 1990).

Several potential acupuncture mechanisms involve other cranial nerves, such as the facial, optic, and olfactory nerves. Multiple facial acupoints involve the distribution of the facial nerve, such as ST7 (Xiaguan), which distributes the zygomaticotemporal branch of the facial nerve and fiber projections from the facial nerve nucleus. Facial nerve stimulation has been studied for increasing CBF in ischemic stroke (Borsody et al., 2013; Borsody et al., 2014). A study explored the minimum stimulation parameters for increasing CBF with facial nerve magnetic stimulation in Yorkshire pigs and tested safety, tolerability, and efficacy in healthy volunteers (Sanchez et al., 2018). The results indicated that continuous stimulation of the facial nerve at 1.6–1.8 Tesla power for approximately 10 min resulted in an average CBF of 32 ± 6%, with a ≥ 25% increase observed in 10 out of 31 volunteers. Some other cranial nerve pathways are gradually being discovered, such as the optic nerve and olfactory nerve. There is an association between the optic nerve and MCI, as retinal imaging through fundus photographs or optical coherence tomography (OCT) observes changes in the optic nerve head to identify early cognitive impairment (Gao et al., 2023). MCI patients exhibit significant progressive tau pathology and deposition of amyloid plaques in the olfactory area (CAm), with signal changes from the olfactory nerve possibly predicting the development of dementia (Bathini et al., 2019). Neural stimulation not only affects a series of changes in the peripheral and central systems but also heralds changes in the function of higher centers with changes in peripheral nerve signals. The nervous system operates like a precise, tightly linked network, where changes in any link may lead to effects on the periphery or even central nervous system changes. Many nerves, such as the optic nerve and olfactory nerve, cannot be stimulated through acupuncture, suggesting that improvements or new inventions are anticipated.